Oey et al. Epigenetics & Chromatin (2015) 8:54
Genetic andepigenetic variation
amonginbred mouse littermates: identication
ofinter-individual dierentially methylated
Harald Oey1,2†, Luke Isbel1†, Peter Hickey3, Basant Ebaid1 and Emma Whitelaw1*
Background: Phenotypic variability among inbred littermates reared in controlled environments remains poorly
understood. Metastable epialleles refer to loci that intrinsically behave in this way and a few examples have been
described. They display diﬀerential methylation in association with diﬀerential expression. For example, inbred mice
carrying the agouti viable yellow (Avy) allele show a range of coat colours associated with diﬀerent DNA methylation
states at the locus. The availability of next-generation sequencing, in particular whole genome sequencing of bisul-
phite converted DNA, allows us, for the ﬁrst time, to search for metastable epialleles at base pair resolution.
Results: Using whole genome bisulphite sequencing of DNA from the livers of ﬁve mice from the Avy colony, we
searched for sites at which DNA methylation diﬀered among the mice. A small number of loci, 356, were detected
and we call these inter-individual Diﬀerentially Methylated Regions, iiDMRs, 55 of which overlap with endogenous
retroviral elements (ERVs). Whole genome resequencing of two mice from the colony identiﬁed very few diﬀerences
and these did not occur at or near the iiDMRs. Further work suggested that the majority of ERV iiDMRs are metastable
epialleles; the level of methylation was maintained in tissue from other germ layers and the level of mRNA from the
neighbouring gene inversely correlated with methylation state. Most iiDMRs that were not overlapping ERV insertions
occurred at tissue-speciﬁc DMRs and it cannot be ruled out that these are driven by changes in the ratio of cell types
in the tissues analysed.
Conclusions: Using the most thorough genome-wide proﬁling technologies for diﬀerentially methylated regions, we
ﬁnd very few intrinsically epigenetically variable regions that we term iiDMRs. The most robust of these are at retroviral
elements and appear to be metastable epialleles. The non-ERV iiDMRs cannot be described as metastable epialleles at
this stage but provide a novel class of variably methylated elements for further study.
Keywords: Metastable epiallele, Genetic variation, Inbred mice, DNA methylation
© 2015 Oey et al. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://
creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided
you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate
if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/
zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Phenotypic variation in traits like weight and size within
inbred mouse colonies has intrigued geneticists for dec-
ades [1, 2]. Inbred mice are presumed to be virtually iso-
genic, and observed variation, therefore, attributable to
other factors such as stochastic or environmental events.
e precise mechanisms underlying such phenotypic
variation are still unclear but some of the variability is
likely to be reﬂected in, and possibly driven by, the epi-
genome and some is likely to be driven by genetic diﬀer-
ences. Human twin studies have shown that epigenomes
diﬀer slightly within monozygotic twin pairs [3, 4] but the
signiﬁcance of these diﬀerences remains unclear. While
monozygotic twins arise from the same zygote, litter-
mates in inbred mouse colonies arise from independent
Epigenetics & Chromatin
†Harald Oey and Luke Isbel equally contributed to this work
1 Department of Genetics, La Trobe Institute for Molecular Science, La
Trobe University, Bundoora, Melbourne, VIC 3086, Australia
Full list of author information is available at the end of the article
Page 2 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
gametes, providing opportunities for genetic diﬀerences
that result from germline mutations.
Some parts of the genome, such as the telomeres, are
known to be variable in length between inbred littermates
[5–7]. It has also been shown that some DNA copy num-
ber variants persist, despite careful inbreeding . Spon-
taneous germline mutations will also occur. In humans,
whole genome sequencing of trios has been used to esti-
mate that such mutations occur at a rate of 1.20 × 10−8
mutations per base per generation . While the corre-
sponding rate in mice was previously believed to be sig-
niﬁcantly higher , recent estimates suggest they are
similar [11, 12]. Now, whole genome sequencing can be
used to investigate such variation directly. is technol-
ogy has recently been used to characterize the genomes
of some common inbred mouse strains, revealing exten-
sive genetic variation between strains . However, the
extent of variation within an inbred strain has not previ-
ously been investigated using a whole genome approach.
e Avy mouse line has been used as a model of epige-
netic metastability for many years [14–19]. e founder
mouse was discovered 50 years ago in a litter from a
C3H/HeJ colony because of its unexpected yellow coat
. An intracisternal A particle (IAP) retrotranspo-
son was found to have integrated upstream of the agouti
gene. e original mouse was backcrossed for many gen-
erations to C57BL/6J, and has been maintained on that
background in the heterozygous state (Avy/a). Littermates
range in colour from yellow, through mottled (yellow and
brown patches) to pseudoagouti (brown). e coat colour
inversely correlates with the DNA methylation state of a
promoter within the IAP LTR (long terminal repeat) [18,
21]. e methylation state of the locus within an individ-
ual is conserved across tissue types suggesting establish-
ment very early in embryonic development . When
active, this promoter drives constitutive transcription of
agouti and results in a yellow coat. is locus is one of
only three or four classic murine metastable epialleles,
in which a variable phenotype correlates directly with
epigenetic state [22–24]. More recently, it has been pro-
posed that such loci are relatively frequent, in the thou-
sands [25, 26].
We have sequenced the genomes of two littermates
from the Avy mouse colony, one with a yellow coat and
one with a pseudoagouti coat, and searched for dif-
ferences between the two, both at the Avy locus, and
genome-wide, and conﬁrm that genetic diﬀerences are
unlikely to be involved in the variable coat colour. To
discover novel loci that display epigenetic metastability,
we used whole genome bisulphite sequencing (WGBS)
of the livers of ﬁve Avy/a mice and searched for regions
of signiﬁcant variability in DNA methylation. We found
a small number of loci that behave like metastable
epialleles, the most robust are associated with the ERV
family of retrotransposons. Most other variable loci are
associated with regions identiﬁed by others as tissue-spe-
ciﬁc DMRs, i.e. they display variable DNA methylation
across tissues .
Whole genome sequencing
Whole genome sequencing was carried out using the
Illumina sequencing by synthesis technology to 40-fold
coverage in two inbred males (one yellow and one pseu-
doagouti) and the genomes were searched for variants
against the C57BL/6J reference genome (mm9). Variants
that were identiﬁed included both those that diﬀered
between the two mice (e.g. heterozygous in one, wild
type in the other) (Table1) and those for which the mice
did not diﬀer but diﬀered against the reference genome
(i.e. heterozygous or homozygous in both mice) (Table2).
Variant calls at the C3H/HeJ region containing agouti
were excluded from these counts. No diﬀerences between
the two mice were seen in this region. Genome-wide, a
total of 985 single nucleotide variants (SNVs) were found
that diﬀered between the two mice (Table1; Additional
ﬁle1: Table S1) and as expected, the majority of these
were located in either intergenic or intronic regions (607
and 324, respectively) (Table1). Only 11 of the variants
were located inside exons, and of these, seven were pre-
dicted to result in amino acid changes and four were pre-
dicted to be silent (Table1).
With respect to the variants that did not diﬀer between
the two mice but diﬀered from the reference genome,
while such variants are not expected to account for phe-
notypic variation between the sequenced littermates, the
heterozygous variants are likely to be polymorphic within
the colony. Most of the homozygous variants are likely to
represent mutations that have arisen and spread within
the Avy strain.
To ascertain the false discovery rate of the variant calls,
a random set of 105 variants (taken from Additional ﬁle1:
Table S1) was picked for Sanger sequencing. 96 of these
could be PCR ampliﬁed and sequencing was carried out
in both littermates. Of these 96, 87 were validated using
Sanger sequencing (Additional ﬁle1: Table S2). All vari-
ants for which one of the two mice was homozygous were
found to represent true positives (out of 24 tested) and 34
variants that were heterozygous in one mouse and wild
type in the other (out of 36 tested) were conﬁrmed (Addi-
tional ﬁle1: Table S2).
e parents of the two sequenced mice were also tested
to determine the proportion that represents de novo
mutations. Data were obtained for 32 of the 34 variants
that were heterozygous in one mouse and wild type in the
other (data not shown). Four variants were unique to one
Page 3 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
of the oﬀsprings (and absent in the parents) and likely to
represent germline mutations. is number can be used
to obtain a crude estimate of mutation rate (see “Meth-
ods”). A mutation rate of 9.9×10−9 was obtained, which
is similar to that reported using whole genome sequenc-
ing for humans .
e two genomes were also searched for copy num-
ber variations (CNVs) and polymorphic retrotransposon
insertions. A single large CNV (Additional ﬁle2: Fig. S1)
and 10 retrotransposon insertions were identiﬁed that
diﬀered between the mice. e latter were either L1 or
MTA elements (Additional ﬁle3: Fig. S2). With respect
to the former, PCR ampliﬁcation across the breakpoint
showed that it was not linked to the Avy phenotype
(n=12, Additional ﬁle2: Fig. S1b). is CNV has previ-
ously been reported in the C57BL/6J strain .
Whole genome methylation
To identify regions that were diﬀerentially methyl-
ated among littermates from the Avy colony, we carried
out whole genome bisulphite sequencing on DNA from
the livers of ﬁve adult males. e Avy colony was main-
tained using Avy/a crossed to a/a mating pairs. ree of
the mice were a/a. e remaining two were both Avy/a,
one had a yellow coat (Y) and one a pseudoagouti coat
(Ψ). e bisulphite converted genomes were sequenced
and more than 70% of the CpGs were covered by at least
6 reads (Fig.1a). is is within the recommended range
for the identiﬁcation of diﬀerentially methylated regions
in WGBS data . e global CpG methylation levels
were calculated using 10kb windows and were found to
be similar across the mice, with a median methylation of
e Avy locus serves as a positive control for a meta-
stable epiallele with the LTR of the IAP expected to be
unmethylated in the Yellow mouse and methylated in
the Pseudoagouti mouse. Indeed, this was found to be
the case. Interestingly, the diﬀerence in methylation was
not limited to the IAP long terminal repeat (LTR), but
extended approximately 1kb outside the repeat (Fig.2),
consistent with .
Regions ofvariable methylation amongindividuals;
inter‑individual DMRs, iiDMRs
We then searched the genome for additional regions
where the individual mice diﬀered from one another
Table 1 Variants that are polymorphic between litter-
Distribution of the variant calls against C57BL/6J reference genome that diers
between the two Avy littermates. The genetic dierences between the Yellow
and the Pseudoagouti mouse are divided relative to their genic positions. Exonic
mutations have been subdivided into those that are synonymous and those that
Splice junction 1
Upstream (<2 kb) 18
Downstream (<2 kb) 16
Table 2 Polymorphic variants in the Avy colony that are
Distribution of variant calls against C57BL/6J for which the Avy littermates have
the same zygosity. The genetic dierences between C57BL/6J and the two
sequenced mice are divided relative to their genic positions. Exonic mutations
have been subdivided into those that are synonymous and those that are not
Both mice heterozy‑
gous Both mice homozy‑
Intergenic 734 2926
Intronic 345 1891
Exonic 21 49
Non-synonymous (12) (19)
Synonymous (7) (30)
Splice junction 0 9
Upstream (<2 kb) 7 79
Downstream (<2 kb) 11 65
UTR 12 36
Total 1130 5055
Fig. 1 WGBS methylation of the ﬁve individuals. a The coloured bars
indicate read depth and the Y-axis shows the percent of global CpGs
in each category. b Box and whisker plot showing the percentage of
all CpGs that are methylated for each of the ﬁve individuals, using
10 kb windows across the genome
Page 4 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
in their methylation. Yellow Avy mice become obese as
adults and the mice used in this study were 22weeks of
age. e Yellow mouse had a bodyweight 1.6 times that
of the average of the remaining four (50.7g versus the
average 31.5g±3.1s.d. for the remainder). To minimize
potential confounding eﬀects, the yellow mouse was
excluded from the diﬀerential methylation calling pre-
sented below. For interest, the methylation (and expres-
sion) values for the yellow mouse are shown in all ﬁgures.
Diﬀerentially methylated regions were located by ﬁrst
extracting the CpGs with Chi-squared P values that
support a diﬀerence in methylation (p<0.05). Diﬀeren-
tially methylated regions were deﬁned as those loci that
had (1) at least six adjacent CpGs (allowing for 10% of
CpGs being uninformative), (2) with a diﬀerence of at
least 20% between the weighted averages of the highest
and lowest methylated individual and (3), no more than
500bp between adjacent CpGs. Sites overlapping simple
repeats were excluded. A total of 356 regions were identi-
ﬁed and clustered by methylation levels (Additional ﬁles
1, 4: Table S3, Fig. S3). We call these loci iiDMRs ,
inter-individual DMRs. We noticed that mouse C57.1
was responsible for approximately half of all the identi-
ﬁed loci and had consistently higher methylation values
at these regions. In the absence of a clear understanding
of this, loci that were generated due to high methylation
in C57.1 are indicated (Additional ﬁle1: Table S3).
We searched for possible genetic explanations for the
diﬀerential methylation using the list of 985 and 1130
polymorphisms that were found to be diﬀerent between
the two sequenced mice (Table1), or were heterozygous
in both mice (Table2), respectively. No iiDMRs directly
overlapped with a variant and only two iiDMRs were
within 1kb of a SNV. Similarly, none of the iiDMRs were
within 10kb of the 10 transposable elements that were
found to be polymorphic in the colony (Additional ﬁle3:
We initially focused on iiDMRs that overlapped ERVs
because the best characterized previously reported met-
astable epialleles, agouti viable yellow, axin fused and
Cdk5rap, are associated with IAPs. We found that 55 of
the 365 diﬀerentially methylated regions overlapped with
ERVs. We refer to these as ERV iiDMRs. e methylation
at each region behaved independently with respect to the
methylation at other ERV iiDMRs within the same mouse
and no single mouse (of all ﬁve mice) was consistently
more or less methylated at these elements than any other
mouse (Fig.3a; Additional ﬁle1: Table S4). IAPs make
up the majority of ERV iiDMRs and RLTR4s may also be
overrepresented in this list. RLTR4s are also referred to
in the literature as murine leukaemia virus (MLV) type
retrotransposons. In general, iiDMRs that overlap with
ERVs had a greater range of methylation levels across
individuals than the non-ERV iiDMRs (Fig. 3b). ose
IAP elements that had an internal sequence had a greater
range than lone IAP LTRs (Additional ﬁle5: Fig. S4). e
presence of an internal sequence would be expected in
recently integrated elements.
Clonal bisulphite sequencing was used to validate
methylation levels at one ERV iiDMR designated ERV
iiDMR 7, using the same DNA samples used for WGBS
and from the two mice with the most extreme methyla-
tion states (Fig.4). is ERV has been reported by others
to inﬂuence transcription of the Slc15a2 gene .
Fig. 2 DNA methylation at Avy. The weighted average DNA methyla-
tion levels of single CpG dinucleotides in the yellow mouse (blue)
and the pseudoagouti mouse (red) show changes extending out
from the IAP insertion, which is upstream of the agouti gene. Data are
shown only when more than ﬁve reads cover a CpG. Ectopic agouti
transcripts originate from the LTR element (green)
Fig. 3 Variable DNA methylation at ERVs. a Heatmap represent-
ing the 55 iiDMRs overlapping ERV elements. For these sites, the
weighted average CpG methylation for each mouse is shown.
Unsupervised clustering was performed. Data for the yellow mouse
are shown but were not used to identify the diﬀerentially methyl-
ated sites. b The range of methylation at each ERV iiDMR (n = 55)
for the ﬁve mice is shown and compared with that of all 301 iiDMRs
generated from Additional ﬁle 1: Table S3 after removal of the ERV-
associated loci. ERV iiDMRs have a signiﬁcantly greater range (T test,
p value <0.05)
Page 5 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
Evidence suggesting thatthese ERV iiDMRs are metastable
It has been shown that methylation levels at metastable
epialleles correlate across diﬀerent germ layers within an
individual and are, therefore, likely to be set prior to dif-
ferentiation of the three germ layers . We used clonal
bisulphite sequencing to examine the methylation levels
for three of the ERV iiDMRs; 24, 11 and 27, in spleen,
derived from mesoderm, from the same mice used to
generate the liver (endodermal) data. DNA methylation
levels across individuals correlated with that found in
Two of the classic IAP metastable epialleles, Avy and
AxinFu, were originally identiﬁed because of altered
expression patterns among inbred littermates. Using
reverse transcriptase quantitative PCR (RTqPCR), we
determined the expression of the genes adjacent to the
IAP-associated loci, ERV iiDMR 7 and ERV iiDMR 24.
e genes are slc15a2 and 2610035D17Rik, respectively.
As the Slc15a2 gene is not expressed in liver, we carried
out these experiments in spleen. An inverse correlation
was seen between the level of methylation and expression
of these genes (Fig.6a, b). is experiment was also car-
ried out on genes adjacent to two ERV iiDMRs associated
with RLTR4 elements, ERV iiDMR 11 and ERV iiDMR
27. e methylation state at ERV iiDMR 11 did not
inversely correlate with expression of the gene in which
it is located, Ccdc21 (Fig.6c). An inverse correlation was
observed between methylation at ERV iiDMR 27 and
expression of the adjacent Pik3c3 gene (Fig.6d). ese
results support the ability of the methodology to identify
metastable epialleles and show that for ERV iiDMRs the
DNA methylation level often inversely correlates with
transcription of an adjacent gene.
Dierentially methylated regions thatare not ERVs
Excluding the ERV iiDMRs, 301 other regions satisﬁed
the criteria for a locus that is variably methylated across
individuals (Additional ﬁle 1: Table S3). Interestingly,
many (156/301) of these non-ERV iiDMRs overlapped
with short regions that are diﬀerentially methylated
across tissues within an individual mouse, termed tissue-
speciﬁc diﬀerentially methylated regions, tsDMRs .
tsDMRs are conserved regions involved in transcrip-
tional regulation, mainly enhancers.
To validate methylation changes at these non-ERV
iiDMRs nine loci were randomly selected and pyrose-
quencing was used to asses DNA methylation in liver
(endodermal), cerebellum (ectodermal) and spleen (mes-
odermal), representing the three germ layers, from the
same ﬁve mice. e diﬀerential methylation validated in
liver at only ﬁve out of the nine loci (Additional ﬁle5: Fig.
S4a–e). No diﬀerences were seen at four of the nine loci in
any tissue across the ﬁve mice (Additional ﬁle5: Fig. S4f–
i). is relatively poor validation rate might be associated
with the lack of replicates, even though sequencing was
carried out at high coverage . In this study, the experi-
mental design necessitates no biological replicates. Alter-
natively, around half of these sites might be false positives.
At the ﬁve loci that did validate, the diﬀerential meth-
ylation was not seen in the cerebellum (ectoderm) or the
spleen (mesoderm) (Additional ﬁle 5: Fig. S4a–e), sug-
gesting that the establishment of these diﬀerent meth-
ylation states does not occur early in development and
raising the possibility that cell type ratio changes have
occurred in the liver. ese loci mostly show a modest
range in DNA methylation across individuals compared
to the ERV iiDMR group (Fig. 3b). Validation of this
group awaits the development of better technologies for
detecting small changes in DNA methylation.
isis the ﬁrst report of the use of whole genome rese-
quencing (approx 40× coverage) to investigate sequence
diﬀerences between two inbred mouse littermates. Four
of the randomly selected variants were absent in both par-
ents, representing likely germline mutations. is is con-
sistent with a mutation rate of 9.9 × 10−9, which agrees
with previous estimates. In addition, ~2000 SNPs were
identiﬁed that are either heterozygous in both mice or dif-
fer in zygosity between mice, a reminder that in inbred
colonies those mutations that have arisen in the recent
Fig. 4 Methylation variability at ERV iiDMR 7 validated using an
independent method. A screenshot of the WGBS methylation at ERV
iiDMR 7 is shown for the ﬁve mice (left). On the Y-axis 0 represents no
methylation, 1 represents 100 % methylation and the solid lines indi-
cate the 50 % methylation position. The coordinates of the ERV iiDMR
7 overlaps an IAP LTR, indicated in dark grey. Methylation levels from
clonal bisulphite sequencing (primer sequences are in Additional
ﬁle 1: Table S5) on the two extreme samples (yellow and C57.1 mouse
DNA) conﬁrmed the diﬀerential methylation (right). Each sample is
represented by at least 11 clones, ﬁlled in circles represent methylated
CpGs from each sequenced clone. Asterisk indicates T test, p value
Page 6 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
past are in many cases not ﬁxed in the population despite
inbreeding. e relatively small number of SNVs in cod-
ing sequences (n=32) and the failure to detect any muta-
tions close to the Avy allele reassure us that the variable
coat colour among Avy littermates is an epigenetic event.
We found 356 regions that vary in methylation across the
ﬁve inbred individuals and call these iiDMRs. 55 of these
356 loci overlapped with ERVs and these showed the largest
variability in DNA methylation across the mice. Given that
the few classic metastable epialleles identiﬁed in the mouse
prior to this report are linked to transcriptionally active ret-
rotransposons, this is not surprising. Despite the identiﬁca-
tion of 55 ERV iiDMRs, our statistical calling procedures
could not be implemented at all loci, e.g. approximately half
of the ~12,000 annotated IAP elements in the mm9 mouse
reference genome failed to meet coverage requirements. So
55 ERV iiDMRs are likely to be a twofold underestimate of
the ERV iiDMRs in the mouse genome.
We established a statistical method of calling iiDMRs that
required six CpGs less than 500bp apart in an attempt to
reduce false positives and this condition will bias our data-
set to regions with at least that density of CpGs. e use of
biological replicates is recommended for single CpGs reso-
lution . However, we could not use biological replicates
(every mouse will, by deﬁnition, be diﬀerent at these loci).
ree of the IAP elements found in this study to be iiD-
MRs have been reported previously to inﬂuence transcrip-
tion of adjacent gene expression, Slc15a2 and Polr1a ,
and Cdk5rap1 . Only the last of these has previously
been reported to be a metastable epiallele. Most of the ERV
Fig. 5 Methylation state at ERV iiDMRs in liver is conserved in spleen. Shown is a UCSC genome browser screen shot of three variably methyl-
ated loci, iiDMR 24 (a), iiDMR 11 (b) and iiDMR 27 (c), on the Y-axis 0 represents no methylation, 1 represents 100 % methylation and the solid lines
indicate the 50 % methylation position. Shown also are the underlying repeat elements. Clonal bisulphite sequencing (primer sequences are in
Additional ﬁle 1: Table S5) from spleen revealed the same pattern of diﬀerential methylation across the ﬁve mice for ERV iiDMR 24 and for the two
most extremes of methylation for ERV iiDMR 11 and ERV iiDMR 27. Each sample is represented by at least seven clones, ﬁlled in circles represent
methylated CpGs. Asterisk indicates T test, p value <0.05
Page 7 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
iiDMRs reported here are poorly annotated with respect to
their ability to inﬂuence transcription of adjacent genes but
are located well within an interval potentially able to drive
gene expression, as exempliﬁed by the IAP at Avy, which lies
approximately 100kb from the agouti coding sequence .
Clonal bisulﬁte sequencing using unique primers that
ﬂank the repeat element allowed us to reanalyse these
ERV iiDMRs in another tissue. is is diﬃcult to accom-
plish with other targeted approaches that are limited by
amplicon size, such as pyrosequencing or methylation-
sensitive high-resolution melt analysis, hence the use of
clonal bisulﬁte sequencing. e majority of ERV iiDMRs
that were tested here for metastability (i.e. showed the
same methylation state in a diﬀerent tissue and aﬀected
expression) turned out to be metastable epialleles.
Others have carried out a bioinformatic screen of IAPs
that possess active promoter histone marks, H3K4me3, and
identiﬁed 143 potential metastable epialleles in the mouse,
only three of which overlap with our ERV iiDMR dataset
. A follow-up study, by the same group, searched for
DNA methylation variability among inbred mice using
an IAP enrichment method and report thousands of
diﬀerentially methylated loci , none of which overlap
with those reported here. is lack of overlap is likely to be
the result of the diﬀerences between techniques.
It has previously been reported that the methylation
state of two metastable epialleles, Avy and AxinFu, are set
independently of each other, even when both alleles are
present in the same individual . Despite the under-
lying IAP LTR sequences at these two loci being identi-
cal, the programming of each appears to be independent.
is is consistent with the ERV iiDMRs identiﬁed in this
study for which no obvious bias towards hyper or hypo-
methylation in any one individual can be detected.
In humans, genetic diﬀerences confound the approach
used in this study. Despite this, a recent attempt has
been made to use genome-wide bisulphite sequencing
to identify metastable epialleles. ey identiﬁed 109 loci
that were candidate metastable epialleles with discordant
inter-individual DNA methylation. is group of regions
was enriched in retrotransposon-derived elements,
including LINEs and HERVs .
e fact that metastable epialleles produce oﬀspring
with a range of phenotypes despite “isogenicity”, might
Fig. 6 Expression of loci adjacent to ERV iiDMRs. The average expression from four technical replicates is shown for two genes, Slc15a2 (a) and
2610035D17Rik (b), in which ERV iiDMR 7 and ERV iiDMR 24, respectively, are located. The location of each IAP is indicated relative to the exonic and
intronic sequences of genes, indicated by bars connected by lines. Also shown embedded in each expression bar is the liver methylation level of the
iiDMR taken from Figs. 4 and 5. The average expression from two technical replicates is shown for two genes, Ccdc21 (c) and Pik3c3 (d), associated
with ERV iiDMR 11 and ERV iiDMR 27, respectively. The ERV iiDMR 11 RLTR4 is located in intron 3 of Ccdc21 while the ERV iiDMR 27 RLTR4 is located
approximately 5 kb upstream of the Pik3c3 transcription start site. Error bars indicate the SEM for each technical replicate
Page 8 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
enable genetically closed colonies, e.g. those geographi-
cally isolated, to cope with ﬂuctuating environmental
conditions. For example, one could envisage a situation in
which “yellower” mice are ﬁtter than pseudoagouti litter-
mates, such as a change in habitat from grasses to desert.
ese animals would maintain the genetic stock during
hot periods. On the other hand, such variability in a con-
stant environment is likely to be detrimental, by reducing
the number of successful oﬀspring in any litter.
Either way, the small number of such elements in the
genome (~50 in our strain) makes it unlikely that meta-
stable epialleles are major drivers of evolution.
Others have identiﬁed diﬀerences in patterns of ret-
rotransposons across the genomes of 17 inbred strains
and found at least 25,000 that are polymorphic [13, 35].
ese inbred strains have been maintained as independ-
ent colonies for around 100 years (equivalent to ~400
generations). However, without a better understanding
of selective pressures, diﬀerent rates in diﬀerent mouse
strains, diﬀerent rates for diﬀerent classes of retrotrans-
posons and the number of breeding pairs used over this
period, estimating insertion rates is not feasible.
e identiﬁcation of 10 polymorphic transposon inser-
tions between two individuals has helped us to rule out
such events as contributing to methylation changes but
for the reasons stated above, the data cannot be used to
accurately estimate insertion rates in this strain. e 10
repeats that were found to diﬀer between the individuals
are most likely polymorphic in the colony. Of the ten, ﬁve
were heterozygous in one mouse and homozygous in the
other and the insertions could, therefore, not have hap-
pened in the parents of the probands. e remaining ﬁve
insertions were heterozygous in one individual and absent
(i.e. wild type) in the other. As was shown for the analo-
gous germline SNVs (only 4 SNVs of 32 candidates were
found to represent germline mutations), most of these are
likely to be polymorphic in the colony and the combined
insertion rate (L1, MTA and MT2) is, therefore, consider-
ably lower than a maximum of ﬁve per generation.
In addition to the retrotransposon-associated iiDMRs,
we ﬁnd a new class of variably methylated loci linked to
transcriptional regulatory elements. In general, these
loci had a smaller range in methylation across individu-
als than ERV iiDMRs and the low validation rate at these
loci likely reﬂect the limitations of identifying diﬀeren-
tially methylated regions using a single biological repli-
cate. It is possible that some loci in this group are driven
by individual diﬀerences in cell composition within each
mouse’s liver. is limitation extends to all studies using
complex tissue and even cells puriﬁed using antibodies to
surface marker proteins . Even in puriﬁed cells it is
diﬃcult rule out DNA methylation variation as a reﬂec-
tion of uncharacterised subpopulations .
However, it is unlikely that cell type ratio changes
underlie large changes in DNA methylation, as each PCR
clone and each deep sequencing read represent a single
cell and, therefore, changes in DNA methylation would
require an equally large change in cell type ratio. Either
way, non-ERV iiDMRs represent a novel class of inter-
individual DMRs and it will be of interest to study these
Using the most thorough genome-wide proﬁling tech-
niques for short regions that show diﬀerential epigenetic
state, we identify approximately three hundred intrinsi-
cally epigenetically variable loci and the most robust of
these are likely to be associated with recently integrated
Whole genome resequencing ofAvy littermates
Animal work was conducted in accordance with the Aus-
tralian code for the care and use of animals for scientiﬁc
purposes, and was approved by the Animal Ethics Com-
mittee of LaTrobe University (project reference number:
AEC 12-74). Two male mice heterozygous for the Avy
allele, a yellow and a pseudoagouti, were selected from
the Avy colony and DNA was extracted from tails for
whole genome sequencing. Tail DNA was also extracted
from the parents and used for downstream validations.
Whole genome libraries were prepared using a DNA
insert size of 480 bp and sequenced using 2 × 100 bp
paired reads on an Illumina HiSeq 2000 by the BGI
(Shenzhen, People’s Republic of China). A total of
~7×108 paired reads were sequenced for each genome.
e sequenced reads were aligned to the mouse genome
(NCBI37/mm9 assembly) using the program BWA, ver-
sion 0.6.2 , and the commands “bwa aln -I -R 500”
and “bwa sampe -a 510 -o 1000000”. e mapped reads
were coordinate-sorted and PCR duplicates removed
using the utility MarkDuplicates from the Picard pack-
age (http://picard.sourceforge.net). e reads were then
recalibrated by the GATK version 1.6-13  using the
tools RealignerTargetCreator (setting −rbs=10,000,000),
IndelRealigner (using indels from the ﬁle 20110602-calla-
ble-dinox-indels.vcf by Keane etal.  to deﬁne known
alleles), CountCovariates and ﬁnally TableRecalibration.
Single nucleotide variants and short indels were extracted
by passing the resulting bam-ﬁles through the following
pipeline utilizing Samtools, BCFtools and VCFtools [40,
41]: samtools mpileup –EDS –g | bcftools view -p 0.99 –
vcgN - | vcf-annotate –ﬁll-type -f StrandBias = 0.0001/
EndDistBias = 0.0001/MinDP = 14/MaxDP = 100/
MinMQ = 25/Qual = 10/MinAB = 6/VDB = 0/Gap-
Page 9 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
SnpGap=5/HWE=0.0001. Variants in which more than
90% of reads at the locus supported the variant genotype
were classiﬁed as homozygous while the remaining with at
least a frequency of 30% were classiﬁed as heterozygous.
e resulting variants were ﬁltered for overlap with ele-
ments annotated as simple repeats with a periodicity <9
in the UCSC Genome Browser [42, 43], homopolymer
runs >8 bp (plus 1 bp either end), dinucleotide repeat
runs >14bp (plus 1bp either end) and regions with an
average mapping quality score <40. Additionally, regions
where three or more heterozygous variant calls were
made within 10kb of each other, and where the variants
also overlapped elements annotated as segmental dupli-
cations or annotated repeats, were excluded.
To calculate the proportion of false-positive variant
calls, a random set of 102 variants were selected for valida-
tion by PCR ampliﬁcation followed by Sanger sequencing.
e distribution of the variant types selected for valida-
tion is listed in Additional ﬁle1: Table S2. For six targets,
PCR primers could not be designed, or PCR ampliﬁcation
failed to produce amplicons. ese were excluded.
e germline mutation rate was extrapolated from the
frequency of experimentally validated germline mutations
relative to the total number of potential germline mutations
from the genome-wide variant calls. Variants on chromo-
somes X and Y and variants at unplaced contigs located
on chromosomes annotated as “Random” (227 Mbp in
total) were excluded. Additionally, a total of 258 Mbp was
excluded due to repetitiveness, sequence composition or
insuﬃcient read coverage, leaving 2136 Mbp in which het-
erozygous variants could be called for this purpose. e
frequency was adjusted for the experimentally derived false-
positive discovery rate of 20%, and a false-negative rate of
23% (calculated based on variant calls overlapping known
variants at the genomic region around Agouti, which is het-
erozygous for C3H/HeJ) and adjusted for the false-positive
and negative rates reported for these variants .
CNVs were called by the program Control-FREEC
 using the settings coeﬃcientOfVariation = 0.05,
forceGCcontentNormalization=1, sex=XY. e result-
ing calls intersected with genes annotated in the UCSC
Genome Browser’s “UCSC Genes” database , and
calls that overlapped genes were scrutinized for pres-
ence of reads and read pairs supporting breakpoints at
the termini of each CNV. To visualize such breakpoints,
a dataset was created of discordantly mapped read pairs
combined with a dataset of soft-clipped reads (identiﬁed
from the SAM-ﬁle CIGAR string), as such reads and read
pairs are typically found adjacent to breakpoints.
A CNV that was found using this method was validated
by PCR using primers speciﬁc to the junction between the
tandemly repeated copies using the primers CNV1_F and
CNV_R (Additional ﬁle1: Table S5). e distribution of
this CNV within the colony was investigated by targeting
the junction by PCR in DNA extracted from 12 mice (6
yellow and 6 pseudoagouti). For each template, a control
primer (CNV_C), which together with CNV1_F ampliﬁes
the wild-type sequence at the 3′ end of the CNV, was used
in parallel to verify presence of the template.
To locate transposon insertions that were polymorphic
between the two genomes we used the tool RetroSeq 
with the options –discover -align -len 50 -q 28 –unmapped
and –call -reads 10 -depth 100 -hets -q 28. ree separate
instances of the program were run with diﬀerent transpo-
son consensus sequences obtained from RepBase  used
as input. For endogenous retroviruses (ERV) we used those
sequences annotated as Endogenous Retrovirus belonging
to the taxon Mus musculus. For long interspersed repeat 1
elements (L1), we used sequences annotated as L1 belong-
ing to the taxon Mus musculus, plus the following acces-
sions: L1Md_F_5end, L1Md_Gf_5end, L1_Mus1_3end,
L1_Mus2_3end. Finally, insertions were also called against
the Mammalian apparent LTR retrotransposon (MaLR)-
related MTa repeats using the accessions MT2A, MT2B,
MT2B1, MT2B2_LTR, MTA, MTAI, MTA_Mm_LTR,
MTB, MTB_Mm_LTR, MTC and MTC_I. During the ﬁnal
calling step, putative insertions that overlapped repeats
annotated in the UCSC RepeatMasker track as ERVK, L1
and ERVL or MaLR, respectively, were ﬁltered out.
e zygosity of each predicted insertion was then
determined by carefully scrutinizing the reads mapped
to each locus by visualizing the whole genome datasets
in the UCSC Genome Browser . Insertions that were
found to have diﬀering zygosity were validated by carry-
ing out local assembly of the discordant and soft-clipped
reads that had been mapped to that locus. Assembly
was performed by the program Velvet  using a hash_
length of 50 and –ins_length of 480, and the identity of
the inserted element was determined from the resulting
contigs by RepeatMasker (http://www.repeatmasker.org).
Whole genome bisulphite sequencing ofAvy littermates
Whole genome bisulphite sequencing was carried out by
Centro Nacional de Análisis Genomico (CNAG, Barce-
lona, Spain) and the data were processed and mapped,
as described previously . An iiDMR was deﬁned as a
region with at least six adjacent CpGs with a Chi-squared
p value <0.05 (allowing for a single CpG without signiﬁ-
cant p value), at most 500bp spacing each CpG and with
a diﬀerence of at least 20% between the weighted averages
of the individuals with highest and lowest DNA methyla-
tion. Methylation values for each CpG dinucleotide were
merged to obtain a single methylation value for each CpG.
e weighted average for a region was obtained by dividing
each CpG methylation score by the sum of the read cover-
ages across all CpGs in the region followed by multiplying
Page 10 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
each CpG by the read coverage at that individual CpG and
ﬁnally adding together each of the adjusted CpG scores to
obtain a ﬁnal score. ose CpGs with less than a 6 read cov-
erage were discarded and values for the remaining CpGs
used to identify iiDMRs using custom R scripts (available
Clonal bisulphite sequencing
Bisulphite treatment was performed on DNA samples
puriﬁed using phenol–chloroform-extracted DNA.
500ng of DNA was bisulphite converted using the Qia-
gen EpiTect Bisulphite Kit (Qiagen, Doncaster, VIC,
Australia) and single loci were ampliﬁed using prim-
ers designed to only amplify bisulﬁte-converted DNA
strands without CpGs in primer sequence. PCR product
was puriﬁed using the QIAquick PCR puriﬁcation kit
(Qiagen, Doncaster, VIC, Australia), then cloned using
the pGEM-T Easy Vector (Promega, Alexandria, NSW,
Australia). Clones were sequenced using e BigDye
Terminator v3.1 Cycle Sequencing Kit (Life technolo-
gies, Mulgrave, VIC, Australia) as per kit instructions.
Primers used for bisulphite sequencing are given in Addi-
tional ﬁle1: Table S5. To calculate statistical signiﬁcance,
a Student’s T test was used to compare the fractions of
methylated CpGs for an individual’s bisulﬁte PCR clones
(i.e. the per-clone methylation values) to those of another
Pyrosequencing ofbisulphite‑treated DNA
DNA was extracted by phenol–chloroform followed by
ethanol precipitation. Primer design, bisulphite conver-
sion and pyrosequencing were carried out by the Aus-
tralian Genome Research Facility. Average methylation
scores were collected from at least 4 CpGs per locus.
Reverse transcriptase quantitative polymerase chain
Total RNA was extracted from snap frozen tissues either
using TRIzol reagent (Life technologies, Mulgrave, VIC,
Australia) or the AllPrep DNA/RNA/Protein kit (Qiagen,
Doncaster, VIC, Australia) according to manufacturer
instructions. cDNA synthesis was carried out from total
RNA using the QuantiTect Reverse Transcription Kit
(Qiagen, Doncaster, VIC, Australia) and RTqPCR was
performed with the QuantiTect SYBR Green reagent
(Qiagen, Doncaster, VIC, Australia). Samples were run
on the CFX384 Touch Real-Time PCR Detection System
(Biorad, Gladesville, NSW, Australia), with the following
conditions: 95°C 10min, 39× cycles with 95°C 15s then
60°C 1min, with a ﬁnal step of 95°C 15s. Relative cDNA
abundance was calculated using the ∆∆CT method nor-
malizing to housekeeper gene expression indicated in the
ﬁgures. Primers are in Additional ﬁle1: Table S5.
Availability ofsupporting data
e data sets supporting the results of this article are
available in the Gene Expression Omnibus repository,
under the accession number GSE72177 (http://www.
Additional le 1: Table S1. Variant calls from whole genome sequenc-
ing of two Avy mice, one yellow and one pseudoagouti. The variant calls
and associated genomic coordinates are relative to the NCBI37/mm9
genome assembly. Table S2. Validation of variant calls. A set of 105 SNP
variants were selected at random from those that diﬀered between the
two mice or were heterozygous in both mice. Of the 96 that could be PCR
ampliﬁed and sequenced, those that did or did not validate are shown.
Table S3. Variably methylated regions designated iiDMRs. A list of 356
regions that display diﬀerential methylation between inbred individuals,
generated from 6 adjacent CpGs that support diﬀerential methylation
and a diﬀerence of at least 20 % between the highest and lowest value.
Values from the Yellow excluded from calling regions. Table S4. Variably
methylated regions that overlap ERV elements, designated ERV iiDMRs.
A list of 55 regions that display diﬀerential methylation between inbred
individuals and overlap with ERV elements from the UCSC mm9 repeat-
masker database. Diﬀerentially methylated regions were generated from 6
adjacent CpGs that support diﬀerential methylation and a diﬀerence of at
least 20 % between the highest and lowest value. Values from the Yellow
mouse were excluded from calling regions. Table S5. Primers used in the
Additional le 2: Fig. S1. (a) Read-coverage at a locus where a large
gain of ~ 200 Kb of DNA is polymorphic in the Avy colony. Protein-coding
genes are illustrated below the graph. (b) A total of six yellow (Y) and six
pseudoagouti (P) mice, as well as the two individuals whose genomes
were sequenced, were investigated for presence of the CNV using prim-
ers amplifying the junction between the copies. For each, a region not
aﬀected by the CNV was ampliﬁed to conﬁrm presence of the genomic
DNA template. A non-template control (N) was also included.
Additional le 3: Fig. S2. Transposon insertions that diﬀer between lit-
termates. The genomes of two agouti viable yellow mice, one with yellow
and one with pseudoagouti coat colour, were sequenced and searched
for retrotransposon insertions, relative to the C57/BL6 reference genome
(NCBI37/mm9 assembly). Insertions that diﬀered between the two mice
are presented below. In the ﬁgures, paired deep sequencing reads are
presented in the form of red bars (forward reads) and blue bars (reverse
reads) connected by black lines (the un-sequenced part of the insert).
Each ﬁgure is centred on the transposon insertion site, which is usually
deﬁned by truncated (soft clipped) reads and ﬂanked by un-paired or
discordantly mapped deep sequencing reads.
Additional le 4: Fig. S3. Candidate diﬀerentially methylated regions
between littermates. The mm9 genome were searched for sites that had
a methylation values signiﬁcantly diﬀerent between the Pseudoagouti,
C57.1, C57.2 and C57.3 mice with at least 6 adjacent CpGs and a range of
at least 20 %. For each site, the weighted average CpG methylation was
calculated and used for clustering (unsupervised).
Additional le 5: Fig. S4. The range of methylation at ERV iiDMRs that
overlap with IAP elements that have an internal sequence (n = 17) or are
lone IAP LTR elements (n = 10). IAPs with internal sequence elements
have a signiﬁcantly greater range (T-test, p-value > 0.05).
Additional le 6: Fig. S5. Validation of DNA methylation at random
non-ERV iiDMRs. WGBS weighted averages for DNA methylation values
are shown for the nine loci chosen for pyrosequencing (a-i). The average
pyrosequencing methylation level, from at least 4 individual CpGs from
each iiDMR, is shown for liver (L), cerebellum (C) and spleen (S). DNA from
these tissues was made using the ﬁve mice originally used for WGBS.
Methylation levels validated in liver DNA for ﬁve of nine loci (a-e).
Page 11 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
Avy: agouti viable yellow; CNV: copy number variations; CpG: cytosine guanine
dinucleotide sequence; ERVs: endogenous retroviral elements; iiDMRs: inter-
individual diﬀerentially methylated regions; IAP: intracisternal A particle; LTR:
long terminal repeat; RTqPCR: reverse transcriptase quantitative polymerase
chain reaction; SNV: single nucleotide variants; tsDMRs: tissue-speciﬁc diﬀeren-
tially methylated region; WGBS: whole genome bisulphite sequencing.
HO participated in the design of the study, helped draft the manuscript and
performed the statistical analysis. LI participated in the molecular assays and
drafted the manuscript. PH wrote custom R scripts for statistical analysis. BE
participated in molecular assays. EW conceived the study, and participated in
its design and coordination and helped to draft the manuscript. All authors
read and approved the ﬁnal manuscript.
1 Department of Genetics, La Trobe Institute for Molecular Science, La Trobe
University, Bundoora, Melbourne, VIC 3086, Australia. 2 Present Address:
University of Queensland Diamantina Institute, Translational Research Institute,
Princess Alexandra Hospital, Brisbane, QLD 4102, Australia. 3 Bioinformatics
Division, The Walter and Eliza Hall Institute of Medical Research, 1G Royal
Parade, Parkville, VIC 3052, Australia.
This study was supported by National Health and Medical Research Council of
Australia grant to EW (1058345) and the Victorian Life Sciences Computation
The authors declare that they have no competing interests.
Received: 8 October 2015 Accepted: 23 November 2015
1. Gärtner K. A third component causing random variability beside environ-
ment and genotype. A reason for the limited success of a 30 year long
eﬀort to standardize laboratory animals? Lab Anim. 1990;24(1):71–7.
2. Wong AHC, Gottesman II, Petronis A. Phenotypic diﬀerences in geneti-
cally identical organisms: the epigenetic perspective. Hum Mol Genet.
3. Kaminsky ZA, Tang T, Wang S-C, Ptak C, Oh GHT, Wong AHC, et al. DNA
methylation proﬁles in monozygotic and dizygotic twins. Nat Genet.
4. Fraga MF, Ballestar E, Paz MF, Ropero S, Setien F, Ballestar ML, et al. Epige-
netic diﬀerences arise during the lifetime of monozygotic twins. Proc Natl
Acad Sci USA. 2005;102(30):10604–9.
5. Starling JA, Maule J, Hastie ND, Allshire RC. Extensive telomere repeat
arrays in mouse are hypervariable. Nucleic Acids Res. 1990;18(23):6881–8.
6. Kipling D, Cooke HJ. Hypervariable ultra-long telomeres in mice. Nature.
7. Roberts A, Blewitt M, Youngson N, Whitelaw E, Chong S. Reduced dosage
of the modiﬁers of epigenetic reprogramming Dnmt1, Dnmt3L, SmcHD1
and Foxo3a has no detectable eﬀect on mouse telomere length in vivo.
8. Watkins-Chow DE, Pavan WJ. Genomic copy number and expres-
sion variation within the C57BL/6J inbred mouse strain. Genome Res.
9. Kong A, Frigge ML, Masson G, Besenbacher S, Sulem P, Magnusson G,
et al. Rate of de novo mutations and the importance of father/’s age to
disease risk. Nature. 2012;488(7412):471–5.
10. Li W-H, Ellsworth DL, Krushkal J, Chang BHJ, Hewett-Emmett D. Rates of
nucleotide substitution in primates and rodents and the generation-time
eﬀect hypothesis. Mol Phylogenet Evol. 1996;5(1):182–7.
11. Huttley GA, Wakeﬁeld MJ, Easteal S. Rates of genome evolution
and branching order from whole genome analysis. Mol Biol Evol.
12. Kumar S, Subramanian S. Mutation rates in mammalian genomes. Proc
Natl Acad Sci. 2002;99(2):803–8.
13. Keane TM, Goodstadt L, Danecek P, White MA, Wong K, Yalcin B, et al.
Mouse genomic variation and its eﬀect on phenotypes and gene regula-
tion. Nature. 2011;477(7364):289–94.
14. Chong S, Vickaryous N, Ashe A, Zamudio N, Youngson N, Hemley S, et al.
Modiﬁers of epigenetic reprogramming show paternal eﬀects in the
mouse. Nat Genet. 2007;39(5):614–22.
15. Cropley JE, Suter CM, Beckman KB, Martin DIK. Germ-line epigenetic
modiﬁcation of the murine Avy allele by nutritional supplementation.
Proc Natl Acad Sci. 2006;103(46):17308–12.
16. Kaminen-Ahola N, Ahola A, Maga M, Mallitt K-A, Fahey P, Cox TC, et al.
Maternal ethanol consumption alters the epigenotype and the pheno-
type of oﬀspring in a mouse model. PLoS Genet. 2010;6(1):e1000811.
17. Morgan HD, Sutherland HGE, Martin DIK, Whitelaw E. Epigenetic inherit-
ance at the agouti locus in the mouse. Nat Genet. 1999;23(3):314–8.
18. Waterland RA, Jirtle RL. Transposable elements: targets for early
nutritional eﬀects on epigenetic gene regulation. Mol Cell Biol.
19. Wolﬀ GL, Kodell RL, Moore SR, Cooney CA. Maternal epigenetics and
methyl supplements aﬀect agouti gene expression in Avy/a mice. FASEB
20. Dickies MM. A new viable yellow mutation in the house mouse. J Hered.
21. Blewitt ME, Vickaryous NK, Paldi A, Koseki H, Whitelaw E. Dynamic repro-
gramming of DNA methylation at an epigenetically sensitive allele in
mice. PLoS Genet. 2006;2(4):e49.
22. Rakyan VK, Blewitt ME, Druker R, Preis JI, Whitelaw E. Metastable epialleles
in mammals. Trends Genet. 2002;18(7):348–51.
23. Druker R, Bruxner TJ, Lehrbach NJ, Whitelaw E. Complex patterns of tran-
scription at the insertion site of a retrotransposon in the mouse. Nucleic
Acids Res. 2004;32(19):5800–8.
24. Weinhouse C, Anderson OS, Jones TR, Kim J, Liberman SA, Nahar MS, et al.
An expression microarray approach for the identiﬁcation of metastable
epialleles in the mouse genome. Epigenetics. 2011;6(9):1105–13.
25. Ekram MB, Kang K, Kim H, Kim J. Retrotransposons as a major source
of epigenetic variations in the mammalian genome. Epigenetics.
26. Ekram MB, Kim J. High-throughput targeted repeat element bisulﬁte
sequencing (HT-TREBS): genome-wide DNA methylation analysis of IAP
LTR retrotransposon. PLoS One. 2014;9(7):e101683.
27. Hon GC, Rajagopal N, Shen Y, McCleary DF, Yue F, Dang MD, et al. Epige-
netic memory at embryonic enhancers identiﬁed in DNA methylation
maps from adult mouse tissues. Nat Genet. 2013;45(10):1198–206.
28. Graubert TA, Cahan P, Edwin D, Selzer RR, Richmond TA, Eis PS, et al. A
high-resolution map of segmental DNA copy number variation in the
mouse genome. PLoS Genet. 2007;3(1):e3.
29. Ziller MJ, Hansen KD, Meissner A, Aryee MJ. Coverage recommendations
for methylation analysis by whole-genome bisulﬁte sequencing. Nat
30. Rebollo R, Karimi MM, Bilenky M, Gagnier L, Miceli-Royer K, Zhang Y, et al.
Retrotransposon-induced heterochromatin spreading in the mouse
revealed by insertional polymorphisms. PLoS Genet. 2011;7(9):e1002301.
31. Gemma C, Ramagopalan S, Down T, Beyan H, Hawa M, Holland M, et al.
Inactive or moderately active human promoters are enriched for inter-
individual epialleles. Genome Biol. 2013;14(5):R43.
32. Li J, Akagi K, Hu Y, Trivett AL, Hlynialuk CJW, Swing DA, et al. Mouse
endogenous retroviruses can trigger premature transcriptional termina-
tion at a distance. Genome Res. 2012;22(5):870–84.
33. Rakyan VK, Chong S, Champ ME, Cuthbert PC, Morgan HD, Luu KVK, et al.
Transgenerational inheritance of epigenetic states at the murine AxinFu
allele occurs after maternal and paternal transmission. Proc Natl Acad Sci.
34. Silver M, Kessler N, Hennig B, Dominguez-Salas P, Laritsky E, Baker M,
et al. Independent genomewide screens identify the tumor suppressor
VTRNA2-1 as a human epiallele responsive to periconceptional environ-
ment. Genome Biol. 2015;16(1):118.
35. Nellaker C, Keane T, Yalcin B, Wong K, Agam A, Belgard TG, et al. The
genomic landscape shaped by selection on transposable elements
across 18 mouse strains. Genome Biol. 2012;13(6):R45.
Page 12 of 12
Oey et al. Epigenetics & Chromatin (2015) 8:54
36. Stella C, Cazzola M, De Fabritiis P, De Vincentiis A, Gianni A, Lanza F, et al.
CD34-positive cells: biology and clinical relevance. Haematologica.
37. Wijetunga NA, Delahaye F, Zhao YM, Golden A, Mar JC, Einstein FH, et al.
The meta-epigenomic structure of puriﬁed human stem cell populations
is deﬁned at cis-regulatory sequences. Nat Commun. 2014;5:5195.
38. Li H, Durbin R. Fast and accurate short read alignment with Burrows-
Wheeler transform. Bioinformatics. 2009;25(14):1754–60.
39. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernyt-
sky A, et al. The genome analysis toolkit: a MapReduce framework
for analyzing next-generation DNA sequencing data. Genome Res.
40. Danecek P, Auton A, Abecasis G, Albers CA, Banks E, DePristo MA, et al.
The variant call format and VCFtools. Bioinformatics. 2011;27(15):2156–8.
41. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The
sequence alignment/map format and SAMtools. Bioinformatics.
42. Benson G. Tandem repeats ﬁnder: a program to analyze DNA sequences.
Nucleic Acids Res. 1999;27(2):573–80.
43. Meyer LR, Zweig AS, Hinrichs AS, Karolchik D, Kuhn RM, Wong M, et al.
The UCSC genome browser database: extensions and updates 2013.
Nucleic Acids Res. 2013;41(D1):D64–9.
44. Boeva V, Popova T, Bleak ley K, Chiche P, Cappo J, Schleiermacher G, et al.
Control-FREEC: a tool for assessing copy number and allelic content
using next generation sequencing data. Bioinformatics. 2011;28(3):423–5.
45. Keane TM, Wong K, Adams DJ. RetroSeq: transposable element
discovery from next-generation sequencing data. Bioinformatics.
46. Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, Walichiewicz J.
Repbase update, a database of eukaryotic repetitive elements. Cytogenet
Genome Res. 2005;110(1–4):462–7.
47. Zerbino DR, Birney E. Velvet: algorithms for de novo short read assembly
using de Bruijn graphs. Genome Res. 2008;18(5):821–9.
48. Harten SK, Oey H, Bourke LM, Bharti V, Isbel L, Daxinger L, et al. The
recently identiﬁed modiﬁer of murine metastable epialleles, Rearranged
L-Myc Fusion, is involved in maintaining epigenetic marks at CpG island
shores and enhancers. BMC Biol. 2015;13:21.
• We accept pre-submission inquiries
• Our selector tool helps you to ﬁnd the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research
Submit your manuscript at
Submit your next manuscript to BioMed Central
and we will help you at every step:
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at