Specific Loss of Histone H3 Lysine 9 Trimethylation and
HP1c/Cohesin Binding at D4Z4 Repeats Is Associated
with Facioscapulohumeral Dystrophy (FSHD)
Weihua Zeng1, Jessica C. de Greef2, Yen-Yun Chen1, Richard Chien1, Xiangduo Kong1, Heather C.
Gregson1¤, Sara T. Winokur1, April Pyle3, Keith D. Robertson4, John A. Schmiesing1, Virginia E. Kimonis5,
Judit Balog2, Rune R. Frants2, Alexander R. Ball Jr.1, Leslie F. Lock1, Peter J. Donovan1, Silve `re M. van der
Maarel2, Kyoko Yokomori1*
1Department of Biological Chemistry, School of Medicine, University of California, Irvine, California, United States of America, 2Leiden University Medical Center, Center
for Human and Clinical Genetics, Leiden, The Netherlands, 3Institute for Stem Cell Biology and Medicine, Department of Microbiology, Immunology, and Molecular
Genetics, David Geffen School of Medicine, University of California, Los Angeles, California, United States of America, 4Department of Biochemistry and Molecular Biology,
University of Florida, Gainesville, Florida, United States of America, 5Division of Medical Genetics and Metabolism, Department of Pediatrics, University of California Irvine
Medical Center, Orange, California, United States of America
Facioscapulohumeral dystrophy (FSHD) is an autosomal dominant muscular dystrophy in which no mutation of pathogenic
gene(s) has been identified. Instead, the disease is, in most cases, genetically linked to a contraction in the number of 3.3 kb
D4Z4 repeats on chromosome 4q. How contraction of the 4qter D4Z4 repeats causes muscular dystrophy is not understood.
In addition, a smaller group of FSHD cases are not associated with D4Z4 repeat contraction (termed ‘‘phenotypic’’ FSHD),
and their etiology remains undefined. We carried out chromatin immunoprecipitation analysis using D4Z4–specific PCR
primers to examine the D4Z4 chromatin structure in normal and patient cells as well as in small interfering RNA (siRNA)–
treated cells. We found that SUV39H1–mediated H3K9 trimethylation at D4Z4 seen in normal cells is lost in FSHD.
Furthermore, the loss of this histone modification occurs not only at the contracted 4q D4Z4 allele, but also at the
genetically intact D4Z4 alleles on both chromosomes 4q and 10q, providing the first evidence that the genetic change
(contraction) of one 4qD4Z4 allele spreads its effect to other genomic regions. Importantly, this epigenetic change was also
observed in the phenotypic FSHD cases with no D4Z4 contraction, but not in other types of muscular dystrophies tested.
We found that HP1c and cohesin are co-recruited to D4Z4 in an H3K9me3–dependent and cell type–specific manner, which
is disrupted in FSHD. The results indicate that cohesin plays an active role in HP1 recruitment and is involved in cell type–
specific D4Z4 chromatin regulation. Taken together, we identified the loss of both histone H3K9 trimethylation and HP1c/
cohesin binding at D4Z4 to be a faithful marker for the FSHD phenotype. Based on these results, we propose a new model
in which the epigenetic change initiated at 4q D4Z4 spreads its effect to other genomic regions, which compromises
muscle-specific gene regulation leading to FSHD pathogenesis.
Citation: Zeng W, de Greef JC, Chen Y-Y, Chien R, Kong X, et al. (2009) Specific Loss of Histone H3 Lysine 9 Trimethylation and HP1c/Cohesin Binding at D4Z4
Repeats Is Associated with Facioscapulohumeral Dystrophy (FSHD). PLoS Genet 5(7): e1000559. doi:10.1371/journal.pgen.1000559
Editor: Anne C. Ferguson-Smith, University of Cambridge, United Kingdom
Received November 14, 2008; Accepted June 12, 2009; Published July 10, 2009
Copyright: ? 2009 Zeng et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported in part by NIH RO1 HD49488, NIH PO1 HD47675, and CIRM RC1-00110 to PD; the Netherlands Organization for Scientific
Research NWO (016.056.338) to SvdM; NIH GM59150, MDA4026, the David and Helen Younger Research Fellowship from the FSH Society (FSHS-DHY-001); and a
grant from the California Institute of Regenerative Medicine (RS1-00455-1) to KY. WZ is a recipient of the FSH Society David and Helen Younger Research
Fellowship (FSHS-DHY-002). RC is supported by NIH T32CA113265. None of the sponsors or funders listed above play any role in the design and conduct of the
study, in the collection, analysis, and interpretation of the data, and in the preparation, review, or approval of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
¤ Current address: Focus Diagnostics, Cypress, California, United States of America
FSHD is the third most common heritable muscular dystrophy
. It is characterized by progressive weakness and atrophy of
facial, shoulder, and upper arm musculature, which can spread to
the abdominal and foot-extensor muscles . It can be
accompanied by hearing loss and retinovasculopathy. The genetics
underlying FSHD are highly unusual, as no pathogenic muta-
tion(s) of a disease causing gene(s) has been identified. Instead, the
majority (.95%) of FSHD cases involve mono-allelic deletion of
D4Z4 repeat sequences at the subtelomeric region of chromosome
4q (termed ‘‘4q-linked’’ FSHD, FSHD1A (OMIM 158900);
designated as ‘‘4qF’’ in this study) . There are between one
and ten repeats in the contracted 4qter allele in FSHD patient
cells, in contrast to up to 11,150 copies in normal cells. In
addition, ,5% of FSHD cases are not associated with D4Z4
repeat contraction (termed ‘‘phenotypic’’ FSHD, FSHD2; referred
to as ‘‘PF’’ in this study), and their etiology remains undefined.
How contraction of the 4qter D4Z4 repeats causes muscular
dystrophy is not understood. A previous study reported the YY1-
nucleolin-HMGB2 repressor complex binding to D4Z4, and it was
postulated that reduction of the repeat number may result in
PLoS Genetics | www.plosgenetics.org1 July 2009 | Volume 5 | Issue 7 | e1000559
decreased repressor complex binding, leading to derepression of
neighboring genes . Consistent with this model, overexpression
of the neighboring 4q35 genes was demonstrated in the same
study, and the same group recently showed that muscle-specific
overexpression of the neighboring gene FRG1 indeed causes
muscular dystrophy in mice . Curiously, however, microarray
and quantitative expression studies by other laboratories revealed
that many genes located elsewhere in the genome important for
myoblast differentiation are dysregulated, but unanimously
provided no evidence for abnormal upregulation of FRG1 and
other 4q35 genes in FSHD [5–7]. Furthermore, the model cannot
explain the mechanism of phenotypic FSHD in which there is no
D4Z4 repeat contraction.
Cytological analyses revealed that the 4q telomeric region
uniquely associates with the nuclear periphery, consistent with the
hypothesis that this region is heterochromatic [8,9]. However,
since the D4Z4 repeat contraction in 4qF did not lead to any
significant localization changes, the functional relevance to FSHD
remains uncertain [8,9].
A recent study demonstrated that the 4qter D4Z4 region is
hypermethylated at the DNA level in normal cells, but is
hypomethylated in both 4q-linked and phenotypic FSHD .
This was the first evidence that 4qter D4Z4 is also involved in
phenotypic FSHD. DNA methylation is an important mechanism
for epigenetic regulation of gene transcription, and is generally
associated with transcriptional silencing . Thus, the results
suggested that the D4Z4 repeat array organizes a transcriptionally
suppressive heterochromatic environment, which is disrupted in
FSHD. However, DNA hypomethylation, more severe than that
seen in FSHD, at D4Z4 was also observed in another hereditary
disorder, the ‘‘immunodeficiency, centromere instability and facial
anomalies (ICF)’’ syndrome, due to a mutation in DNA
methyltransferase 3B (DNMT3B) [10,12]. Since the clinical
presentation of ICF syndrome shares no similarity with the FSHD
disease phenotype , the relevance of DNA methylation
changes in FSHD is unclear and the molecular events underlying
the D4Z4-linked disease process remain an open question.
Here we report the characterization of the chromatin of the 4q
and 10q D4Z4 repeats and a comparison between normal, FSHD
and other muscular dystrophy cells. Our results demonstrate that
there is a distinct change of histone modification and downstream
factor binding that is specifically associated with both 4q-linked
and phenotypic FSHD, suggesting that epigenetic alteration plays
a critical role in FSHD pathogenesis.
We examined the histone modification status of the D4Z4
region and whether this is altered in FSHD using chromatin
immunoprecipitation (ChIP). Analysis of 4q D4Z4 chromatin has
been difficult since D4Z4 repeat sequences are present in a
similarly large cluster on both chromosomes 4q and 10q (though
FSHD is associated only with D4Z4 contraction at 4q) . In
addition, D4Z4-like repeats are present on several other
chromosomes . We identified and used primer pairs that
amplified products exclusively from chromosomes 4 and 10, but
not from any other chromosome (Figure 1A). This was
confirmed using DNA from somatic cell hybrids carrying
individual human chromosomes as templates (Figure 1B and
1D). Furthermore, the regions amplified by the ‘‘Q-PCR’’
primer pairs contain specific nucleotide polymorphisms that
allow us to distinguish 4q- and 10q-derived D4Z4 sequences
(Figure 1C) . Thus, in this study, the ChIP DNA amplified
by Q-PCR primer pairs was cloned and sequenced to identify
the chromosome of origin (Table S1).
D4Z4 contains both heterochromatic and euchromatic
We found trimethylation of H3K9 (H3K9me3) and H3K27
(H3K27me3) at D4Z4, both of which frequently represent
transcriptionally repressive heterochromatin [15,16], as well as
H3K4 dimethylation (H3K4me2) and H3 acetylation (H3Ac),
which mark transcriptionally permissive euchromatin 
(Figure 2A). H3K9me3 signals were confirmed by two different
antibodies specific for H3K9me3 which have slightly different
binding preferences  (Figure 2A, lanes 10–14). Recent
studies demonstrated that H3K9me3 can also be associated with
transcriptionally active gene regions [18,19]. However, no
significant H3K4me3, which is coupled to transcription-associ-
ated H3K9me3 , was detected using the same primer pairs
(Figure 2A, lane 4). Although it is possible that H3K4me3 may
be present elsewhere in the D4Z4 repeat, it is at least not present
within the promoter and 59 regions of the putative open reading
frame (ORF) for DUX4 where 4qHox and the Q-PCR primers
bind (Figure 1A). Furthermore, double-ChIP analysis revealed
that H3K9me3 coincides with H3K27me3, but not H3K4me2,
suggesting that the D4Z4 repeat cluster contains a distinct
heterochromatic domain marked by both H3K9me3 and
H3K27me3 as well as a euchromatic domain containing
H3K4me2 (Figure 2B). Notably, PCR amplification of the first
proximal D4Z4 repeat revealed that this end is euchromatic,
consistent with a previous report that the region proximal to the
D4Z4 repeat is euchromatic  (Figure 1A and Figure 2C). Both
H3K4me2 and H3K9me3 are present at D4Z4 in human
embryonic stem (hES) cells, suggesting that D4Z4 chromatin
domains are marked by these histone modifications early in
development andare maintained
(Figure 2D). This is in contrast to H3Ac, which is absent in
hES cells and appears to be added at later stages (compare
Figure 2A, lane 7 and Figure 2D, lane 5). Taken together, unlike
the previous model that implies that D4Z4 is a uniformly
Most cases of facioscapulohumeral muscular dystrophy
(FSHD) are associated with a decrease in the number of
D4Z4 repeat sequences on chromosome 4q. How this
leads to the disease remains unclear. Furthermore, D4Z4
shortening is not seen in a small number of FSHD cases,
and the etiology is unknown. In the cell, the DNA, which
encodes genetic information, is wrapped around abundant
nuclear proteins called histones to form a ‘‘beads on a
string’’–like structure termed chromatin. It became appar-
ent that these histones are modified to regulate both
maintenance and expression of genetic information. In the
current study, we characterized the chromatin structure of
the D4Z4 region in normal and FSHD patient cells. We
discovered that one particular histone modification
(trimethylation of histone H3 at lysine 9) in the D4Z4
repeat region is specifically lost in FSHD. We identified the
enzyme responsible for this modification and the specific
factors whose binding to D4Z4 is dependent on this
modification. Importantly, these chromatin changes were
observed in both types of FSHD, but not in other muscular
dystrophies. Thus, this chromatin abnormality at D4Z4
unifies the two types of FSHD, which not only serves as a
novel diagnostic marker, but also provides new insight
into the role of chromatin in FSHD pathogenesis.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org2 July 2009 | Volume 5 | Issue 7 | e1000559
transcriptionally repressive domain , we found that D4Z4
repeats are composed of both euchromatic and heterochromatic
domains with possibly the proximal repeats being euchromatic.
Importantly, the presence of both 4q- and 10q-specific
nucleotide polymorphisms (Figure 1C) was confirmed by
sequencing of the ChIP DNA, indicating that a similar spectrum
of histone modifications are present in the 4q and 10q D4Z4
regions (Table S1).
H3K9me3 is specifically lost in both 4q-linked and
We next examined the chromatin modifications in FSHD
patient-derived primary cells compared to normal cells from
healthy individuals. The H3K9me3 signal at D4Z4 was signifi-
cantly decreased in D4Z4-contracted FSHD myoblasts and
fibroblasts while H3K27me3 and H3K4me2 remained unaffect-
ed (Figure 3A and 3B, 4qF). Importantly, the loss of H3K9me3 is
site-specific because no significant change was observed at the
ribosomal DNA (rDNA) region (Figure 3A, lower panels, and
Figure 3B, lanes 7–11) or in the amount of total H3K9me3
detected by western blot (data not shown). Similarly, no loss of
H3K9me3 was observed at other repeat sequences, including
chromosome 1 a-satellite and satellite 2, chromosome 4 a-
satellite, NBL2, DXZ4, and RS447, in FSHD patient cells
compared to normal cells (Figure S1). The failure to detect
H3K9me3 at D4Z4 is not due to an insufficient number of D4Z4
copies since the ChIP signals were normalized to input DNA to
reflect D4Z4 repeat number changes, and the loss of H3K9me3
was also observed in phenotypic FSHD (PF) cells with no repeat
contraction. It is unlikely to be the result of a drastic change in
antibody accessibility since H3K27me3, which resides in the
same region according to the double-ChIP results (Figure 2B), is
unchanged (Figure 3A and 3B). The persistence of H3K27me3
at D4Z4 also eliminates the possibility that only one allele is
intrinsically organized as heterochromatin and deletion of this
particular allele leads to FSHD. This is in agreement with
previous observations that there is no clear paternal or maternal
bias of disease transmission suggestive of imprinting, which could
differentially organize the chromatin structure of the two alleles
[2,20]. Consistent with this, no significant difference in
subnuclear localization of the two 4qter regions was found by
the previous FISH analyses [8,9].
Interestingly, the total numbers of D4Z4 repeat copies (i.e. the
numbers of 4q and 10q repeats combined) are comparable
between normal and FSHD patient cells (Figure 3B, bottom
panel). Since the analysis in normal cells indicate that 10q D4Z4
also contains similar H3K9me3 modification (see above), the low
level of H3K9me3 ChIP signal in FSHD patient cells cannot
simply be attributed to the chromatin change at 4q D4Z4. This
suggests that the loss of H3K9me3 also occurs at 10q D4Z4. This
is further supported by the fact that both 4q and 10q
polymorphisms were found in the residual H3K9me3 ChIP Q-
PCR products of PF (KII-I) and 4qF (RD217) samples (Table S1).
The results provide the first evidence that 10q D4Z4 chromatin is
co-regulated with 4q D4Z4 chromatin and undergoes similar loss
of H3K9me3 in FSHD.
The loss of H3K9me3 at D4Z4 was observed not only in FSHD
patient myoblasts and fibroblasts, but also in lymphoblasts,
indicating that this chromatin change is not a mere non-specific
epiphenomenon associated with the dystrophic state of the muscle
cell (Figure 3A and 3B and Figure 4A). Presently, we have
examined 14 normal and 14 FSHD patient cell samples of
different origins and obtained consistent results. Importantly, no
significant loss of H3K9me3 at D4Z4 was observed in cells from
Duchenne muscular dystrophy (DMD), limb-girdle muscular
(OPMD), and inclusion body myopathy associated with Paget’s
diseaseof boneand frontotemporal
(Figure 3C). Therefore, the loss of H3K9me3 at 4q and 10q
D4Z4 appears to be a specific change uniquely associated with
both 4q-linked (4qF) and phenotypic (PF) FSHD.
Figure 1. Specific PCR amplification of D4Z4 repeat sequences. (A) A schematic diagram of the 4qter D4Z4 repeat region and a single D4Z4
repeat. PCR products for Q-PCR and 4qHox primer pairs are indicated by black bars. The DUX4 ORF and a GC-rich sequence homologous to the low-
copy repeat HHSPM3  are shown. (B) PCR analysis of a DNA mapping panel consisting of genomic DNA isolated from mouse and hamster somatic
cell hybrids containing individual human chromosomes using the 4qHox and Q–PCR primer pairs. The ‘‘B’’ PCR primer pair also binds to a region
within D4Z4. However, it amplified not only chromosomes 4 and 10, but also several other chromosomes presumably due to crossreactivity to other
D4Z4-like repeat sequences, and therefore, was not used for the experiments. For control PCR reactions, primers corresponding to the mouse b-
globin locus  were used for the chromosome 1, 16, 17, 20, and 21 hybrids, while primers for hamster rDNA regions were utilized for the other
hybrids. PCR analysis of additional mouse somatic cell hybrids for human chromosomes 4, 10, 13, 14, 15, and 21 also yielded similar results (data not
shown). (C) Sequence polymorphisms between 4q and 10q D4Z4 [14,70]. The nucleotide positions (nt) of the sequence polymorphisms are based on
AF117653 in the GenBank/EMBL Nucleotide Sequence Database. (D) PCR analysis using the 4qA161-1 primer pair against genomic DNA from mouse
somatic cell hybrids containing human chromosomes 4, 10, 13, 14, 15, and 21.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org3 July 2009 | Volume 5 | Issue 7 | e1000559
Loss of H3K9me3 is distinct from DNA hypomethylation
DNA and heterochromatic histone methylation are often co-
regulated . Although DNA methylation is more frequently a
downstream consequence of H3K9 methylation , DNA
methylation in some instances was shown to promote H3K9me3
. Thus, we next addressed whether the loss of H3K9me3 is
simply a downstream event of DNA hypomethylation previously
observed in FSHD and clinically unrelated ICF syndrome cells .
We found that H3K9me3 is largely intact at D4Z4 in ICF cells,
though there appears to be an increase in H3K4me2 and H3Ac, as
indicated by ChIP analysis (Figure 3C and Figure 4A). Similarly,
H3K9me3 is unaffected at another non-satellite repeat sequence
called NBL2 in ICF cells, which was also shown to be DNA-
hypomethylated in these cells (Figure 4B) . Furthermore, no
significant loss of H3K9me3 was observed in cells from a clinically
unaffected individual with significant DNA hypomethylation at
D4Z4 (Figure 3B, KI-II) . Finally, treatment of cells with 5-
Azacytidine (5-AzaC), which blocksDNA methylation, did not affect
H3K9me3 despite the significant reduction of DNA methylation at
D4Z4(Figure5E).Takentogether,DNA methylation isnotrequired
for H3K9me3 at D4Z4, and H3K9me3 loss clearly distinguishes
FSHD from ICF, implying that loss of H3K9me3 at D4Z4, rather
than DNA hypomethylation, is causally involved in FSHD.
HP1c and cohesin are specifically recruited to D4Z4,
which is lost in FSHD
What happens as a result of the loss of H3K9me3 at D4Z4? To
investigate the consequences of H3K9me3 loss in FSHD, we
Figure 2. D4Z4 chromatin contains both euchromatic and heterochromatic histone modifications. (A) Antibodies specific for H3K4me2,
H3K4me3, H3K9me3, H3K27me3, H3Ac, and acetylated H4 (H4Ac), as well as control preimmune IgG, were used for ChIP in HeLa cells. The ChIP DNA
was amplified using 4qHox primers and primers specific for regions on chromosomes 10 and 19 containing short Alu repeat sequences. The presence
of H3K9me3 was confirmed by two different antibodies (lanes 10–14) . (B) Double-ChIP analysis of D4Z4 histone modifications. H3K9me3 ChIP (1st
ChIP) was eluted and followed by the second (2nd) ChIP reactions using antibodies specific for H3K4me2, H3K9me3, H3K27me3, or preimmune IgG.
The ChIP DNA was amplified using 4qHox primers. (C) The proximal region of the D4Z4 cluster is euchromatic. ChIP analysis of the first proximal D4Z4
repeat using the 4qA161-1 primer pair (See Figure 1D for sequence amplification specificity) was performed in HeLa, normal human fibroblasts (FB),
myoblasts (MB), and lymphoblasts (LB). (D) Histone ChIP in human ES cells. ChIP DNA derived from the ES cell lines H1 and H9 was amplified by
4qHox primers. Antibodies used for ChIP are indicated at the top.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org4 July 2009 | Volume 5 | Issue 7 | e1000559
Figure 3. Histone H3 lysine 9 trimethylation is specifically lost at D4Z4 in both 4q-linked and phenotypic FSHD. (A) ChIP analysis of
H3K9me3 and H3K27me3 at D4Z4 in normal and FSHD (4qF) myoblasts. The rDNA region (445/446) serves as a positive control. (B) H3K9me3 is
specifically lost in FSHD fibroblasts. Endpoint PCR analysis with 4qHox primers by agarose gel electrophoresis and quantitation of real-time PCR with
Q-PCR primers are shown. The rDNA region, which was positive for HP1 and cohesin binding, was used for comparison (445/446) (See Figure 5). PCR
signals were normalized with preimmune, input, and no template PCR signals. Primary cells derived from healthy (normal) (H), phenotypic (PF), and
4q-linked (4qF) individuals were analyzed as indicated. D4Z4 repeat numbers for 4q and 10q alleles as well as the total D4Z4 repeat numbers are
shown in the table. The asterisk indicates a clinically unaffected individual with DNA hypomethylation at D4Z4, whose two offspring developed
phenotypic FSHD (KII-I and KII-II). (C) H3K9me3 ChIP analysis of different muscular dystrophy patient cells. The graph contains one 4qF (508) and two
PF (Rf394.2 and Rf394.3) patient fibroblast samples, five OPMD patient fibroblast samples carrying alanine repeat insertions in the PABPN1 gene (376,
395, 396, 54030922, and 203241), four DMD patient fibroblast samples with mutations in the dystrophin gene (d1137.5, 6103, 5639.1, and dl90.3),
three LGMD patient fibroblast samples with heterozygous mutations in the LMN gene (00–288, 01–196, 99–305) [55,56], two ICF patient fibroblast
samples , and four IBMPFD patient samples (two fibroblast and two lymphoblast) with mutations in the VCP gene (JH-FIB, MJ-FIB, 307/98, and RS-
LCL) . The KI-I (normal) fibroblast sample serves as a control. H3K9me3 was also retained in two additional control fibroblast samples (302, 557/96)
(data not shown).
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org5 July 2009 | Volume 5 | Issue 7 | e1000559
examined factors that bind to this region. Heterochromatin binding
protein HP1is recruited to heterochromaticregionsby direct binding
to the methylated H3K9 residue and plays an important role in
transcriptional silencing [24,25]. Swi6, an HP1 homolog in S. pombe,
was also shown to recruit the essential sister chromatid cohesion
complex ‘‘cohesin’’ to the pericentromeric heterochromatin where it
mediates centromeric sister chromatid cohesion critical for mitosis
[26,27]. Although the study in yeast indicated that cohesin does not
play any role in transcriptional repression at heterochromatic regions
, HP1 and cohesin are valid candidates for the downstream
effectors of H3K9me3 at D4Z4 in human cells. In mammals, there
are three HP1 variants: HP1a, HP1b and HP1c. We found that
HP1c specifically binds to D4Z4 (Figure 5A). Cohesin binding to
D4Z4 was also detected using antibodies against two of its subunits
(i.e., hSMC1 and hRad21), indicating the presence of the holo-
complex (Figure 5B). Cohesin binding was observed in both
undifferentiated myoblasts and differentiated (mitotically inactive)
myotubes, suggesting a role beyond mitosis at this site (Figure 5C).
Importantly, similar to H3K9me3, HP1c and cohesin binding was
also compromised at D4Z4, but not at the rDNA, DXZ4, and
chromosome 1 a-satellite and satellite 2 repeat regions where
H3K9me3 appears intact, in both 4qF and PF cells (Figure 5D and
Figure S2). The results indicate that H3K9me3, HP1c and cohesin
form heterochromatin at D4Z4, and suggest that the loss of HP1c
and cohesin binding to D4Z4 is a significant downstream
consequence of the loss of H3K9me3 in FSHD. Similar to
H3K9me3, treatment of cells with 5-AzaC did not affect cohesin
and HP1c binding to D4Z4, further separating H3K9me3 and
HP1c/cohesin binding from DNA methylation (Figure 5E).
SUV39H1 is responsible for H3K9me3, which is necessary
but not sufficient for HP1c/cohesin recruitment to D4Z4
The methyltransferase responsible for H3K9me3 and the
relationship between H3K9me3, HP1c and cohesin were
addressed using small interfering RNAs (siRNAs). SiRNA against
SUV39H1, which has no effect on SUV39H2, abolished
H3K9me3 at D4Z4 but not at rDNA, suggesting that SUV39H1
has a non-redundant function at D4Z4 (Figure 6A). Supporting
this notion, depletion of G9a, another H3K9 methyltransferase,
decreased H3K9me3 at the c-Myc region , but had no effect
at D4Z4 or rDNA (Figure 6A). Abolishment of H3K9me3 by
SUV39H1 depletion also impaired HP1c and cohesin binding at
D4Z4 but not at rDNA, confirming that SUV39H1-mediated
H3K9me3 is necessary for HP1c and cohesin binding specifically
at D4Z4. Neither HP1c nor cohesin depletion affected the level of
H3K9me3 at D4Z4, placing them downstream of H3K9me3 (see
Figure 7A, lane 5).
Interestingly, HP1c and cohesin binding to D4Z4 is significantly
low in normal lymphoblasts, even with intact H3K9me3 at D4Z4
(Figure 4A, lane 4), when compared to other cell types (compare
Figure 6B to Figure 5). This is not due to a general decrease of
HP1c and cohesin binding in lymphoblasts since HP1c and
cohesin binding was clearly observed at four other repeat
sequences tested (i.e., rDNA, a-satellite and satellite 2 on
chromosome 1, and DXZ4) in both normal and FSHD
lymphoblasts, similar to myoblasts and fibroblasts (Figure 6B
and Figure S2). Furthermore, the total level of H3K9me3 is
comparable between HeLa and both normal and FSHD
lymphoblasts (Figure 6B, lanes 11–13). The results indicate that
H3K9me3 is not sufficient and suggest that an additional factor(s),
which may be expressed in a cell type-specific manner, is required
for HP1c and cohesin binding to D4Z4. The requirement for an
additional factor(s) is also supported by the observation that not all
H3K9me3-positive repeat sequences are bound by HP1c and
cohesin, even in the same cell sample (Figure 6C).
Cohesin plays an active role in HP1c recruitment to D4Z4
Similar to the recruitment of cohesin to pericentromeric
heterochromatin in S. pombe [26,27], HP1 is required for cohesin
Figure 4. H3K9me3 at D4Z4 is maintained in ICF patient cells. (A) ChIP analysis by endpoint PCR using 4qHox primers was performed using
normal, ICF, and 4q-linked FSHD (4qF) lymphoblasts with antibodies specific for H3K4me2, H3K9me3, and H3Ac, and preimmune IgG as indicated at
the top. The rDNA region (445/446) serves as a positive control. ChIP analysis by real-time PCR using Q-PCR primers for H3K4me2, H3K9me3, and
H3Ac is also shown. Similar results were obtained with ICF fibroblasts (Q-PCR results are shown at the bottom right). (B) H3K9me3 is intact at the
NBL2 repeat region in ICF cells. Similar ChIP analysis was performed using PCR primers specific for the NBL2 repeat sequence.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org6 July 2009 | Volume 5 | Issue 7 | e1000559
binding at D4Z4 (Figure 7A). Interestingly, depletion of HP1c
alone abolished cohesin binding at D4Z4, indicating that HP1a
and HP1b cannot compensate for this function of HP1c at this
site. In contrast, depletion of HP1c had no effect on cohesin
binding to the rDNA region, a-satellite and satellite 2 repeats on
chromosome 1, and DXZ4, most likely due to functional
redundancy with other HP1 variants (Figure 7A (lanes 7–12)
and B). Consistent with this notion, HP1a binding was detected at
the a-satellite repeat, but not at D4Z4 (Figure 5A, lane 4; data not
shown). Thus, HP1c is uniquely involved in heterochromatin
formation at D4Z4.
We found that the cohesin loading factor Scc2  also binds to
D4Z4, which was significantly decreased by depletion of HP1c to
an extent similar to the decrease caused by depletion of Scc2 itself
(Figure 7C). Consistent with this, we found an interaction between
the endogenous HP1c and Scc2 by in vivo coimmunoprecipitation
(co-IP) (Figure 7D). Although weak, the interaction is specific and
partially resistant to a 1 M salt wash (Figure 7D, ‘‘eluate’’). We
found that HP1c mainly interacts with Scc2, rather than cohesin
(Figure 7D). Although it was originally shown that HP1 interacts
with cohesin in S. pombe , the interaction of Scc2 with HP1
variants was reported in human cells  and more recently in S.
pombe . Interestingly, CTCF, another factor recently shown to
recruit cohesin to its binding sites [32–34], interacts preferentially
with cohesin but not Scc2 (Figure 7D), suggesting distinct modes of
cohesin recruitment by these factors.
In S. pombe, cohesin is downstream of HP1, and does not play
any role in HP1 recruitment . Interestingly, we found that
depletion of hSMC1 impairs HP1c binding to D4Z4 (Figure 7A).
Similarly, depletion of Scc2 abolished D4Z4 binding of not only
cohesin but also HP1c. Thus, the results provide the first evidence
for an active role of cohesin in heterochromatin organization. This
appears to be context-dependent, since the rDNA region, a-
satellite and satellite 2 repeats on chromosome 1, and the DXZ4
region showed no effect on HP1c binding following depletion of
hSMC1 or Scc2 (Figure 7A and 7B).
Figure 5. The binding of HP1c and cohesin to D4Z4 is lost in FSHD patients. (A) HP1c, but not HP1a, binds to D4Z4 in HeLa cells. 4qHox
endpoint PCR of ChIP DNA using antibodies against HP1c and HP1a is shown. Immunoprecipitation with protein A beads alone serves as a negative
control. (B) Comparison of ChIP analyses using antibodies specific for two different subunits of cohesin (hSMC1 and hRad21). Preimmune IgG and
protein A beads alone were used as negative controls. Two different amounts of ChIP DNA were used for endpoint PCR with 4qHox primers as
indicated. The remainder of the cohesin ChIP experiments were carried out using anti-hRad21 antibody. (C) Cohesin binds to the 4qHox region in
undifferentiated and differentiated primary human myoblasts. Cohesin ChIP was compared to that of condensin, another major SMC-containing
complex, and protein A beads control. (D) ChIP PCR analyses using 4qHox primers of HP1c and cohesin binding in H, PF and 4qF fibroblasts as in
Figure 3B. Representative samples of the 4qHox PCR products on an agarose gel are shown. PCR primers corresponding to the rDNA locus serve as
positive (445/446) and negative (347/348) controls for HP1 and cohesin binding. Real-time PCR analysis using Q-PCR primers of HP1c and cohesin
ChIP is shown underneath. A similar loss of HP1c and cohesin was also observed in 4qF myoblasts (data not shown). (E) The effect of DNA
hypomethylation on cohesin and HP1c binding and H3K9me3. HeLa cells were treated with 5-AzaC and ChIP-PCR assays were performed using
antibodies specific for Rad21 (‘‘cohesin’’), HP1c and H3K9me3 and Q–PCR primers specific for D4Z4. Hypomethylation of DNA was confirmed by
MeCIP using antibody specific for 5-methylcytidine. The ChIP and MeCIP signal intensity was normalized by genomic DNA input control and pre-
immune control. No significant decrease of cohesin and HP1c binding and H3K9me3 was observed. Western analysis of cohesin, HP1c and H3K9me3
levels in untreated and 5-AzaC-treated cells is also shown.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org7 July 2009 | Volume 5 | Issue 7 | e1000559
In this study, we found that the loss of histone H3K9me3 and its
cell type-specific downstream effectors HP1c and cohesin from
D4Z4 repeats is the unifying molecular change in FSHD
(Figure 8A). Importantly, this change was observed in both 4qF
with D4Z4 contraction and PF without D4Z4 contraction. It was
not found in ICF syndrome, despite its apparent similarity to
FSHD with regard to D4Z4 DNA hypomethylation, or in other
types of muscular dystrophies tested. This tight phenotype-
epigenotype correlation strongly suggests that the loss of
H3K9me3 at D4Z4 is critically involved in FSHD pathogenesis.
Our results define a novel diagnostic marker for FSHD, and
provide the first direct evidence for the specific changes of D4Z4
chromatin that are linked to FSHD.
D4Z4 repeat clusters consist of euchromatic and
heterochromatic domains, and only H3K9me3, but not
H3K27me3, is lost from the heterochromatic domains in
Although D4Z4 was thought to be a uniformly transcriptionally
repressive domain [3,10], we found that D4Z4 regions contain a
mixture of euchromatic and heterochromatic histone modifica-
tions; specifically, H3K4me2 and H3Ac as well as H3K9me3 and
H3K27me3. These euchromatic and heterochromatic modifica-
tions are present in distinct domains within D4Z4 repeat clusters
with the first proximal repeat being euchromatic (Figure 8B).
Interestingly, only H3K9me3 is lost in FSHD, but not H3K27me3
from the heterochromatic region (Figure 8C). Thus, the chromatin
change in FSHD is not a total loss of transcriptionally repressive
heterochromatin. This is consistent with the fact that there
apparently is no significant compensatory increase of euchromatic
modifications, suggestive of expansion of euchromatic domains
within D4Z4, in FSHD.
Loss of H3K9me3 and D4Z4 contraction
PF and 4qF are genetically distinct. While the etiology of PF is
unknown, our results revealed a correlation between the repeat
contraction and the loss of H3K9me3 at D4Z4 in 4qF patient cells.
This raises the possibility that repeat contraction leads to the loss
of H3K9me3 at D4Z4 in 4qF. It is also formally possible that the
upstream event that initially caused the repeat contraction might
have also caused the loss of H3K9me3. It is less likely that the loss
of H3K9me3 is the cause of repeat contraction, since there is no
repeat number instability in phenotypic FSHD despite the similar
loss of H3K9me3. Detection of H3K9me3 at D4Z4 in hES cells
and multiple cell types indicates that H3K9me3 at this region is
normally established early during development at a pluripotent
stage, and is maintained throughout multi-lineage differentiation.
Figure 6. SUV39H1 HMTase is solely responsible for H3K9me3 at D4Z4, which is necessary, but not sufficient, for the recruitment of
HP1c and cohesin. (A) SUV39H1 is responsible for H3K9me3 and HP1c/cohesin association at D4Z4. HeLa cells were treated with siRNA specific for
SUV39H1, G9a, or control siRNA, and ChIP analysis using 4qHox primers was performed for the presence of cohesin, HP1c and H3K9me3 (lanes 1–16).
Preimmune IgG serves as a negative control. The rDNA (445/446) and c-Myc regions were used for comparison. Western-blot analysis of G9a and
SUV39H1 siRNA depletion is also shown (lanes 17–21). Depleted proteins are indicated at the top and proteins detected by western blot analysis are
indicated on the left. a-tubulin serves as a loading control. (B) HP1c and cohesin binding to D4Z4 is cell type-specific. ChIP analysis of D4Z4 and rDNA
regions was performed using normal and 4qF lymphoblasts (lanes 1–10). Western blot analysis comparing the level of H3K9me3 between HeLa and
lymphoblasts (256 (normal) and B8-1 (4qF)) is also shown (lanes 11–13). Coomassie staining of core histones is included as a loading control. (C) Not
all H3K9me3-positive repeats are bound by HP1c and cohesin. Six different repeat sequences (as in Figure S1) were tested for cohesin and HP1c
binding in HeLa cells. While H3K9me3 was detected at all six repeat sequences tested, cohesin and HP1c binding was found at only three repeats (a-
sat and sat2 on chromosome 1 and DXZ4).
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org8 July 2009 | Volume 5 | Issue 7 | e1000559
The fact that H3K9me3 is lost even in lymphoblasts in FSHD
patients indicates that this establishment process during early
development may have gone awry.
Interestingly, our results indicate that contraction of one allele
not only triggers the histone modification change (loss of
H3K9me3) on the disease allele, but also affects H3K9me3 levels
on other non-contracted 4q and 10q D4Z4 alleles, suggesting a
functional communication between these homologous sequences
perhaps reminiscent of transvection in Drosophila  (Figure 8B).
This is in contrast to DNA hypomethylation, which appears to be
restricted to the disease chromosome in FSHD [10,13]. The
dominant effect of contraction of one 4q D4Z4 allele on
H3K9me3 at other D4Z4 alleles is consistent with the dominant
nature of the disease and is in agreement with our results
indicating that DNA hypomethylation is not required for the loss
of H3K9me3. This strongly argues against the theory that only the
contracted D4Z4 allele is involved in FSHD pathogenesis .
Rather, it is possible that both alleles of 4q D4Z4 as well as 10q
D4Z4 may be involved in the disease process. Consistent with the
coordinated chromatin changes observed, somatic pairing of 4q
and 10q D4Z4 has been reported . Although the mechanism is
currently unclear, the results provide the first evidence that the
initial genetic change (repeat contraction) spreads its effect to other
genomic regions in 4qF. A similar coordinated loss of H3K9me3
Figure 7. D4Z4-specific co-recruitment of HP1c, cohesin, and cohesin loading factor Scc2. (A) Binding of HP1c and cohesin to D4Z4 is
interdependent. ChIP analysis of HeLa cells after individual depletion of the cohesin subunit hSMC1, HP1c, or the cohesin loading factor Scc2 by
siRNA as indicated (lanes 1–12). Cohesin and HP1c binding was compared between D4Z4 and rDNA (445/446). Real-time PCR analysis using Q–PCR
primers is shown underneath. Western blot analysis of hSMC1, HP1c and Scc2 depletion is also shown (lanes 13–16). (B) HP1c and cohesin binding do
not affect each other at other repeat sequences. Realtime PCR analysis of Rad21 (‘‘cohesin’’), HP1c and H3K9me3 ChIP DNA from HeLa cells treated
with control, SMC1, HP1c, or Scc2 siRNA as indicated using Q-PCR primers specific for D4Z4, a-sat and sat2 repeat sequences on chromosome 1, and
DXZ4 (as in Figure S2). (C) Scc2 binding to D4Z4 is compromised by HP1c depletion. Realtime PCR analysis of Scc2 ChIP DNA from HeLa cells treated
with control, HP1c, or Scc2 siRNA as indicated using Q-PCR primers specific for D4Z4. (D) Coimmunoprecipitation (co-IP)–western blot analysis of
cohesin and Scc2 interaction with HP1c. HeLa nuclear extracts were used for co-IP using antibody specific for Scc2 or cohesin (Rad21) as previously
described [53,61]. After low-salt washes, precipitated materials were eluted with 1.0 M KCl (‘‘wash’’) and further eluted with 2.0 M guanidine-HCl
(‘‘eluate’’). Eluted proteins were analyzed by SDSPAGE and western blotting using antibody specific for HP1c. For comparison, a similar co-IP analysis
was performed and probed with antibody specific for CTCF.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org9 July 2009 | Volume 5 | Issue 7 | e1000559
Figure 8. Schematic models of chromatin changes and the possible consequences in FSHD. (A) Schematic summary of the cell type-
specific chromatin assembly at D4Z4 and its loss in FSHD. HP1c and cohesin are co-recruited to D4Z4 that harbors SUV39H1-dependent H3K9me3 in
certain cell types, including myoblasts and fibroblasts. In lymphoblasts, however, despite the presence of H3K9me3, HP1c and cohesin fail to
associate with D4Z4 raising the possibility that HP1c and cohesin are involved in cell type-specific chromatin organization and that a putative cell
type-specific factor(s) (or modification(s)) required for their recruitment may not be present in lymphoblasts. Thus, while loss of H3K9me3 at D4Z4 in
FSHD has no consequence at D4Z4 in lymphoblasts, it leads to abolishment of HP1c/cohesin binding in myoblasts, resulting in a detrimental effect
on chromatin organization leading to muscular dystrophy. (B) Coordinated loss of H3K9me3 on 4q and 10q D4Z4 in 4qF and PF. H3K9me3 (shown by
black triangles) clustered in the subdomains of D4Z4 repeat regions (distribution hypothetical) in normal cells is lost in both types of FSHD. (C) A
possible model for the spreading of the epigenetic change at D4Z4 to other genomic regions in FSHD. HP1c and cohesin may contribute to the
physical interactions of the heterochromatic D4Z4 region with other genomic regions leading to the spreading of the silencing effect to putative
target genes in normal cells. In FSHD, the loss of H3K9me3 (but not H3K27me3), HP1c, and cohesin from D4Z4 results in loss of chromatin interaction
and derepression of these genes leading to muscular dystrophy.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org10 July 2009 | Volume 5 | Issue 7 | e1000559
at 4q and 10q D4Z4 was observed in PF, further emphasizing the
significance of this phenomenon.
Regulation of the SUV39H1 activity at D4Z4
We identified the
SUV39H1, but not other HMTases, to be responsible for D4Z4
H3K9me3 (Figure 8A). This raises the possibility that misregula-
tion of this enzyme activity is linked to the etiology of FSHD.
However, no mutation in SUV39H1 itself (either at the promoter
or gene region) in FSHD patient cells was found . Consistent
with this, the total level of H3K9me3 in the nucleus is similar
between normal and FSHD cells. This suggests that a specific
cofactor of SUV39H1, possibly important for its recruitment,
and/or a specific histone demethylase acting at D4Z4, may be
compromised in FSHD. It is plausible that PF results from a
genetic mutation of such a factor. Further investigation of the site-
specific SUV39H1 (or antagonizing histone demethylase) regula-
tion will be important to understand FSHD’s etiology and
pathogenesis, and may shed new light onto the yet to be identified
cause of PF. It is also interesting to note that there is a slight but
consistent decrease in HP1c binding to other repeat sequences
tested in PF, but not 4qF, cells (Figure S2). Although the
significance of this small decrease is currently unclear, this may
reflect the distinct etiologies of PF and 4qF and may provide
another clue to identify the genetic defect in PF.
HP1c and cohesin as cell type–specific downstream
We established the loss of H3K9me3 at D4Z4 to be the
signature change in both types of FSHD, but how does this
epigenetic change lead to muscular dystrophy? We identified two
major downstream effectors of H3K9me3, the heterochromatin
binding protein HP1c and cohesin, whose binding to D4Z4 is
H3K9me3-dependent and, consequently, is severely compromised
in FSHD. The data presented here argue for both factors having a
role in FSHD pathogenesis. Importantly, while H3K9me3 at
D4Z4 is seen in all cell types tested, the binding of HP1c and
cohesin to D4Z4 is cell type-specific, suggesting that their binding
is involved in cell type-specific chromatin organization (Figure 8A).
This restricted HP1c/cohesin binding to D4Z4 may explain the
tissue-specific FSHD disease phenotype, as their loss may be
particularly deleterious to muscle function.
Interestingly, recent evidence suggests that cohesin is also
involved in gene regulation. Although initially identified as a factor
essential for mitosis, discoveries of mutations of cohesin compo-
nents and the essential cohesin chromatin loading factor NIPBL/
Scc2 in the developmental disorder Cornelia de Lange Syndrome
(CdLS) strongly suggested the involvement of cohesin in
developmental gene regulation [37–39]. The sequence-specific
DNA binding transcription factor CTCF was found to recruit
cohesin to many of its binding sites, where cohesin is involved in
CTCF-dependent transcriptional regulation [32–34]. Accumulat-
ing evidence indicates that gene regulation can be affected by
physical interaction between two distant chromosomal regions in
cis and in trans in mammalian cells [40–43]. CTCF is known to be
one such factor that exerts its transcriptional activity by directing
long-distance chromatin interactions and loop formation, for
example, in imprinting and X inactivation [44,45]. Thus, the
discovery that cohesin is an important mediator of CTCF
transcriptional function raised the intriguing possibility that
cohesin may dictate gene expression by facilitating such higher-
order chromatin organization. Recent reports support this notion
for cohesin at certain CTCF binding insulator sites [46,47].
Similar to what was proposed for sister chromatid cohesion ,
cohesin may trap two distant chromatin fibers inside of its ring.
We failed to detect any significant binding of CTCF
concomitant with cohesin at D4Z4 (data not shown), which is
consistent with the fact that CTCF and heterochromatin are
mutually exclusive . However, cohesin may still function in a
similar manner mediating long-distance chromatin interactions,
together with HP1c in the case of D4Z4 heterochromatin. In
Drosophila, it was suggested that HP1 promotes interchromosomal
association of heterochromatin, which may be important for
coordinated gene silencing . Evidence for gene silencing by
association with distant heterochromatin was also found in
mammalian cells, in which the temporal association of the
terminal transferase (Dntt) gene with pericentromeric heterochro-
matin correlates with its silencing during thymocyte maturation in
mice . Thus, one possibility for the involvement of D4Z4
heterochromatin in gene regulation is that it makes contact with,
and represses, distant target genes via long-distance chromatin:
chromatin interactions by spreading a silencing effect in normal
cells (Figure 8C). H3K27me3 found in the same region may also
contribute to this by possibly recruiting the polycomb silencing
complex. We hypothesize that in FSHD the loss of H3K9me3, and
therefore of HP1c and cohesin, results in the loss of this chromatin
interaction, thereby causing abnormal derepression of these
distant target genes that leads to the dystrophic phenotype
(Figure 8C). There may be different sets of target genes for 4q
and 10q D4Z4, both of which would be affected in FSHD due to
the concomitant loss of H3K9me3. Interestingly, some evidence
for change in local higher-order chromatin organization and
nuclear matrix association in 4q-linked FSHD was recently
reported . However, this change appears to occur in the
nearby regions outside of the D4Z4 cluster, and how D4Z4
contraction affects this is unclear. The same phenomenon has not
been confirmed in phenotypic FSHD. In addition, since this
change was shown to be restricted to the contracted allele and not
other D4Z4 alleles, the relationship to the spreading of D4Z4
chromatin changes observed in the current study remains to be
investigated. Further studies to examine the possible chromatin
interactions and organization involving D4Z4 and their changes in
FSHD may provide critical insight into the mechanism of FSHD
Cells and DNA mapping panel
HeLa cells were grown as described previously . The
undifferentiated and differentiated normal myoblasts and the
FSHD patient myoblasts were grown in SkBM-2 (Skeletal Muscle
Cell Basal Medium, Cambrex Bio Science, NJ). Myoblast
differentiation was induced by 2% horse serum as previously
described . Five normal and five 4q-linked FSHD myoblast
lines were used. Control (KI-I, KI-II, NFGr), ICF (ICF1 and
ICF2), 4q-linked FSHD (91RD217, 423/16, F2625, 508) and
phenotypic FSHD (KII-I, KII-II, Rf394.2, RF394.3) fibroblasts
were grown in DMEM/F-12 (1:1) supplemented with 10% FBS,
penicillin/streptomycin, 2 mM GlutaMAX-I (Invitrogen-Gibco,
CA), 10 mM HEPES buffer and 1 mM sodium pyruvate [10,54].
For comparison among different muscular dystrophies, one 4q-
linked FSHD (508) and two phenotypic FSHD (Rf394.2 and
Rf394.3) patient fibroblast samples, five OPMD patient fibroblast
samples (376, 395, 396, 54030922, and 203241), four DMD
patient fibroblast samples (d1137.5, 6103, 5639.1, and dl90.3),
three LGMD patient fibroblast samples (00–288, 01–196, 99–305)
[55,56], two ICF patient fibroblast samples , and four
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org11 July 2009 | Volume 5 | Issue 7 | e1000559
IBMPFD patient samples (two fibroblast and two lymphoblast)
(JH-FIB, MJ-FIB, 307/98, and RS-LCL)  were used. Control
(256.1 LCL), ICF (10759 ICF LCL), and FSHD (B8-1)
lymphoblast cells were grown in RPMI-1640 supplemented with
10% FBS, penicillin/streptomycin, and 2 mM L-Glutamine
(Invitrogen-Gibco, CA). Human ES cells H1 and H9 were grown
as described . Mouse somatic cell hybrids containing
chromosome 4, 10, 13, 14, 15 or 21 (GM11687, 11688, 11689,
10479, 11715, 08854, respectively, from Coriell Cell Repositories,
Camden, NJ) were grown in DMEM/F-12 (1:1) medium with the
same supplements as the fibroblasts. Chromosomes 13, 14, 15, and
21 are known to contain D4Z4-like repeat sequences . The
NIGMS Human/Rodent Somatic Cell Hybrid Mapping Panel
#2, version 3 was from Coriell Cell Repositories, in which
chromosome 1, 16, 17, 20, and 21 hybrids are from mice while the
others are from Chinese hamsters.
Antigen affinity-purified rabbit polyclonal antibodies specific for
Rad21, hSMC1, hCAP-G, and the pre-immune IgG control were
published previously [53,61]. Antibodies against H3K4me2,
H3K4me3, H3K9me3, H3K27me3, H3 Ac, H4 Ac, HP1c,
SUV39H1, and G9a (Upstate Biotech, MA), against H3K9me3
(Abcam, Cambridge, MA) and against HP1a (Novus Biologicals,
CO) were used. Antibody against 5-methylcytidine was from
Eurogentec North America (San Diego, CA).
The ChIP analysis was performed as recommended by the
Upstate ChIP assay kit. Briefly, we crosslinked the cells with 1%
formaldehyde and used 16106cells for one histone ChIP and 36106
cells for the other ChIP assays. Protein A beads were preincubated
with 1 mg/ml BSA and 0.2 mg/ml ssDNA for 20 min at 4uC.
Typically, 4–8 mg of affinity-purified IgG was used per assay. The
mixtures of antibody and nuclear extracts pre-cleared with protein A
beads were incubated at 4uC overnight followed by precipitation
with protein A beads. After washing, immunoprecipitated materials
were eluted with 0.1 M NaHCO3and 1% SDS, and crosslinks were
reversed at 65uC for 4–6 hrs. Primer sequences are listed in Table
S2. PCR primers specific for chromosome 1 a-satellite (a-sat) and
satellite 2 (sat2), chromosome 4 a-satellite (a-sat), DXZ4, RS447,
and NBL2 sequences were used [6,12,62]. In addition, a PCR
primer pair specific for the c-Myc region was used as a control for
G9a depletion as previously described . The primers for rDNA
are located in the intergenic region. All of the end-point PCR
experiments were repeated at least three times. The endpoint gel
quantitation of the ChIP-PCR products was carried out using the
Gel-Doc Imager and Quantity One software (Bio-Rad). Real-time
Q-PCR primers were designed using Lasergene software. Q-PCR
wasperformed using theiCycleriQReal-timePCRdetection system
(Bio-Rad) with iQ SYBR Green Supermix (Bio-Rad). The ChIP
PCR signal was normalized by the subtraction of the preimmune
IgG ChIP PCR signal, which was further divided by input genomic
PCR (for normalization of different D4Z4 repeat numbers in
different cells)minusPCRwith notemplate.Resultswereanaverage
the normal control sample. Double-ChIP analysis was performed
according to the published protocol .
5-Azacytidine (5-AzaC) treatment and methylcytidine
ChIP (MeCIP) assay
The 5-AzaC treatment was performed as previously described
. Briefly, 50 mM of 5-AzaC was added to HeLa cells at 80%
confluency and after 24 hr incubation, the cells were harvested for
ChIP experiments. The MeCIP assay was performed according to
the published protocol . After the cell samples were harvested
and sonicated, they were treated with proteinase K overnight and
the DNA from these samples was purified by the QIAquick gel
purification kit (QIAGEN). Four mg of the purified DNA was used
per MeCIP assay. The DNA was denatured at 95uC for 10 min
and incubated with 4 ml antibody against 5-methylcytidine in
500 ml IP buffer (10 mM sodium phosphate, pH 7.0, 140 mM
NaCl, 0.05% Triton X-100) at 4uC for 2 hrs. The DNA: antibody
mixtures were further incubated with protein A beads at 4uC for
an additional 2 hrs. The beads were washed with 700 ml IP buffer
three times and treated with proteinase K at 50uC for 3 hrs.
Finally, the DNA was recovered using the gel purification kit and
analyzed by PCR.
HeLa cells were transfected three times 24 hours apart with
siRNAs at a final concentration of 10 nM using HiPerFect
Transfection Reagent per manufacturer’s instructions (Qiagen).
The target sequences for SUV39H1 and G9a were previously
described [66,67]. Other siRNA target sequences include hSMC1
(59-CACCATCACACTTTAATTCCA-39), HP1c (59-CTAAGT-
TAAATGAACATTTAA-39), Scc2 (59-CTAGCTGACTCTGA-
CAATAAA-39), and negative
GAACGTGTCACGT-39). Cells were used for ChIP and
western blot analyses at 48 hours after the third transfection.
Coimmunoprecipitation (co-IP)–western analysis
HeLa nuclear extracts were used for co-IP using antibody
specific for Scc2 or cohesin (Rad21) as previously described
[53,61]. Briefly, precipitated materials were washed four times
with a buffer containing 0.1 M KCl, then eluted with 1.0 M KCl
(‘‘wash’’) and finally eluted with 2.0 M guanidine-HCl (‘‘eluate’’).
Proteins in the wash and eluate fractions were precipitated by
trichloroacetic acid (TCA) and analyzed by SDSPAGE and
western blotting using antibody specific for HP1c.
es in normal and FSHD patient cells. PCR primers specific for
chromosome 1 a-satellite (a-sat) and satellite 2 (sat2) and
chromosome 4 a-satellite (a-sat) sequences were used. In addition,
DXZ4, RS447, and NBL2-specific primers were used. Although
sequences are unrelated, DXZ4 (on Xq23) and RS447 (primarily
on 4q16.1) are members of the macrosatellite repeat family similar
to D4Z4. NBL2 is in the acrocentric chromosomes and is known
to be DNA-hypomethylated in ICF syndrome patient cells (see
Figure 4B). The PCR primer sequences are listed in the
Supporting Table S2. Results of the endpoint PCR using 4qHox
primers and realtime PCR using Q-PCR primers for (A) myoblasts
(normal (N27) and 4qF (GM17940)), (B) fibroblasts (normal (KI-I),
PF (KII-I), and 4qF (RD217)), and (C) lymphoblasts (normal (256)
and 4qF (B8-1)) are shown.
Found at: doi:10.1371/journal.pgen.1000559.s001 (1.13 MB TIF)
H3K9me3 ChIP analysis of different repeat sequenc-
sequences. Rad21 and HP1c ChIP analysis of three repeat
sequences (a-sat and sat2 on chromosome 1 and DXZ4) in normal
and FSHD myoblasts, fibroblasts, and lymphoblasts as indicated.
Endpoint PCR using 4qHox primers and realtime PCR analysis
using Q-PCR primers are shown.
Found at: doi:10.1371/journal.pgen.1000559.s002 (0.70 MB TIF)
Cohesin and HP1c binding to different repeat
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org 12July 2009 | Volume 5 | Issue 7 | e1000559
with 4q- or 10q-specific nucleotide polymorphisms. Input and
ChIP DNA amplified by Q-PCR primer pairs was cloned and
sequenced to identify the chromosome of origin based on SNPs
that allow us to distinguish 4q- and 10q-derived D4Z4 sequences.
Found at: doi:10.1371/journal.pgen.1000559.s003 (0.05 MB
The number of input and ChIP DNA PCR clones
Found at: doi:10.1371/journal.pgen.1000559.s004 (0.06 MB
List of PCR primers used.
We would like to thank Drs. Paolo Sassone-Corsi, Peter Verrijzer, Tim
Osborne, and Melanie Ehrlich for critical reading of the manuscript. The
FSHD lymphoblast cell line (B8-1) was kindly provided by Drs. Ichizo
Nishino, Yukiko Hayashi, and Fumiko Saito-Ohara.
Conceived and designed the experiments: WZ HCG KY. Performed the
experiments: WZ JCdG YYC RC XK HCG ARB. Analyzed the data: WZ
JCdG YYC RC XK HCG STW ARB SMvdM KY. Contributed
reagents/materials/analysis tools: XK STW AP KR JAS VEK JB RRF
ARB LFL PJD SMvdM. Wrote the paper: WZ STW RRF ARB SMvdM
KY. Provided technical support for Alex Ball: LFL.
1. Padberg GW (1982) Facioscapulohumeral disease. [MD Thesis]. Leiden, The
Netherlands: Leiden University.
2. van der Maarel SM, Frants RR (2005) The D4Z4 repeat-mediated pathogenesis
of facioscapulohumeral muscular dystrophy. Am J Hum Genet 76: 375–
3. Gabellini D, Green MR, Tupler R (2002) Inappropriate gene activation in
FSHD: a repressor complex binds a chromosomal repeat deleted in dystrophic
muscle. Cell 110: 339–348.
4. Gabellini D, D’Antona G, Moggio M, Prelle A, Zecca C, et al. (2006)
Facioscapulohumeral muscular dystrophy in mice overexpressing FRG1. Nature
5. Winokur ST, Chen YW, Masny PS, Martin JH, Ehmsen JT, et al. (2003)
Expression profiling of FSHD muscle supports a defect in specific stages of
myogenic differentiation. Hum Mol Genet 12: 2895–2907.
6. Jiang G, Yang F, van Overveld PGM, Vedanarayanan V, van der Maarel SM,
et al. (2003) Testing the position-effect variegation hypothesis for facioscapulo-
humeral muscular dystrophy by analysis of histone modification and gene
expression in subtelomeric 4q. Hum Mol Genet 12: 2909–2921.
7. Osborne RJ, Welle S, Venance SL, Thornton CA, Tawil R (2007) Expression
profile of FSHD supports a link between retinal vasculopathy and muscular
dystrophy. Neurol 68: 569–577.
8. Masny PS, Bengtsson U, Chung SA, Martin JH, van Engelen B, et al. (2004)
Localization of 4q35.2 to the nuclear periphery: is FSHD a nuclear envelope
disease? Hum Mol Genet 13: 1857–1871.
9. Tam R, Smith KP, Lawrence JB (2004) The 4q subtelomere harboring the
FSHD locus is specifically anchored with peripheral heterochromatin unlike
most human telomeres. J Cell Biol 167: 269–279.
10. van Overveld PG, Lemmers RJ, Sandkuijl LA, Enthoven L, Winokur ST, et al.
(2003) Hypomethylation of D4Z4 in 4q-linked and non-4q-linked facioscapu-
lohumeral muscular dystrophy. Nat Genet 35: 315–317.
11. Li E (2002) Chromatin modification and epigenetic reprogramming in
mammalian development. Nat Rev Genet 3: 662–673.
12. Kondo T, Bobek MP, Kuick R, Lamb B, Zhu X, et al. (2000) Whole-genome
methylation scan in ICF syndrome: hypomethylation of non-satellite DNA
repeats D4Z4 and NBL2. Hum Mol Genet 9: 597–604.
13. de Greef JC, Wohlgemuth M, Chan OA, Hansson KB, Smeets D, et al. (2007)
Hypomethylation is restricted to the D4Z4 repeat array in phenotypic FSHD.
Neurol 69: 1018–1026.
14. Lemmers RJL, de Kievit P, van Geel M, van der Wielen MJ, Bakker E, et al.
(2001) Complete allele information in the diagnosis of facioscapulohumeral
muscular dystrophy by triple DNA analysis. Ann Neurol 50: 816–819.
15. Rice JC, Briggs SD, Ueberheide B, Barber CM, Shabanowitz J, et al. (2003)
Histone methyltransferases direct different degrees of methylation to define
distinct chromatin domains. Mol Cell 12: 1591–1598.
16. Peters AH, Kubicek S, Mechtler K, O’Sullivan RJ, Derijck AAHA, et al. (2003)
Partitioning and plasticity of repressive histone methylation states in mammalian
chromatin. Mol Cell 12: 1577–1589.
17. Bernstein BE, Kamal M, Lindblad-Toh K, Bekiranov S, Bailey DK, et al. (2005)
Genomic maps and comparative analysis of histone modifications in human and
mouse. Cell 120: 169–181.
18. Vakoc CR, Mandat SA, Olenchock BA, Blobel GA (2005) Histone H3 lysine 9
methylation and HP1gamma are associated with transcription elongation
through mammalian chromatin. Mol Cell 19: 381–391.
19. Lomberk G, Bensi D, Fernandez-Zapico ME, Urrutia R (2006) Evidence for the
existence of an HP1-mediated subcode within the histone code. Nat Cell Biol 8:
20. Tonini MM, Passos-Bueno MR, Cerqueira A, Matioli SR, Pavanello R, et al.
(2004) Asymptomatic carriers and gender differences in facioscapulohumeral
muscular dystrophy (FSHD). Neuromuscular Disorders 14: 33–38.
21. Jones PA, Baylin SB (2007) The epigenomics of cancer. Cell 128: 683–692.
22. Lachner M, Jenuwein T (2002) The many faces of histone lysine methylation.
Curr Opin Cell Biol 14: 286–298.
23. Sarraf SA, Stancheva I (2004) Methyl-CpG binding protein MBD1 couples
histone H3 methylation at lysine 9 by SETDB1 to DNA replication and
chromatin assembly. Mol Cell 15: 595–605.
24. Bannister AJ, Zegerman P, Partridge JF, Miska EA, Thomas JO, et al. (2001)
Selective recognition of methylated lysine 9 on histone H3 by the HP1 chromo
domain. Nature 410: 120–124.
25. Lachner M, O’Carroll D, Rea S, Mechtler K, Jenuwein T (2001) Methylation of
histone H3 lysine 9 creates a binding site for HP1 proteins. Nature 410:
26. Bernard P, Maure J-F, Partridge JF, Genier S, Javerzat J-P, et al. (2001)
Requirement of heterochromatin for cohesion at centromeres. Science 294:
27. Nonaka N, Kitajima T, Yokobayashi S, Xiao G, Yamamoto M, et al. (2002)
Recruitment of cohesin to heterochromatic regions by Swi6/HP1 in fission
yeast. Nat Cell Biol 4: 89–93.
28. Duan Z, Zarebski A, Montoya-Durango D, Grimes HL, Horwitz M (2005) Gfi1
coordinates epigenetic repression of p21Cip/WAF1 by recruitment of histone
lysine methyltransferase G9a and histone deacetylase 1. Mol Cell Biol 25:
29. Ciosk R, Shirayama M, Shevchenko A, Tanaka T, Toth A, et al. (2000)
Cohesin’s binding to chromosomes depends on a separate complex consisting of
Scc2 and Scc4 proteins. Mol Cell 5: 243–254.
30. Lechner MS, Schultz DC, Negorev D, Maul GG, Rauscher F Jr (2005) The
mammalian heterochromatin protein 1 binds diverse nuclear proteins through a
common motif that targets the chromoshadow domain. Biochem Biophys Res
Commun 331: 929–937.
31. Fischer T, Cui B, Dhakshnamoorthy J, Zhou M, Rubin C, et al. (2009) Diverse
roles of HP1 proteins in heterochromatin assembly and functions in fission yeast.
Proc Natl Acad Sci 106: 8998–9003.
32. Wendt KS, Yoshida K, Itoh T, Bando M, Koch B, et al. (2008) Cohesin
mediates transcriptional insulation by CCCTC-binding factor. Nature 451:
33. Parelho V, Hadjur S, Spivakov M, Leleu M, Sauer S, et al. (2008) Cohesins
functionally associate with CTCF on mammalian chromosome arms. Cell 132:
34. Stedman W, Kang H, Lin S, Kissil JL, Bartolomei MS, et al. (2008) Cohesins
localize with CTCF at the KSHV latency control region and at cellular c-myc
and H19/Igf2 insulators. EMBO J 27: 654–666.
35. Duncan IW (2002) Transvection effects in Drosophila. Annu Rev Genet 36:
36. Stout K, van der Maarel S, Frants RR, Padberg GW, Ropers HH, et al. (1999)
Somatic pairing between subtelomeric chromosome regions: implications for
human genetic disease? Chrom Res 7: 323–329.
37. Krantz ID, McCallum J, DeScipio C, Kaur M, Gillis LA, et al. (2004) Cornelia
de Lange syndrome is caused by mutations in NIPBL, the human homolog of
Drosophila melanogaster Nipped-B. Nat genet 36: 631–635.
38. Tonkin ET, Wang TJ, Lisgo S, Bamshad MJ, Strachan T (2004) NIPBL,
encoding a homolog of fungal Scc2-type sister chromatid cohesion proteins and
fly Nipped-B, is mutated in Cornelia de Lange syndrome. Nat Genet 36:
39. Musio A, Selicorni A, Focarelli ML, Gervasini C, Milani D, et al. (2006) X-
linked Cornelia de Lange syndrome owing to SMC1L1 mutations. Nat Genet
40. Tolhuis B, Palstra RJ, Splinter E, Grosveld F, de Laat W (2002) Looping and
interaction between hypersensitive sites in the active beta-globin locus. Mol Cell
41. Horike SI, Cai S, Miyano M, Cheng JF, Kohwi-Shigematsu T (2005) Loss of
silent-chromatin looping and impaired imprinting of DLX5 in Rett syndrome.
Nat Genet 37: 31–40.
42. Spilianakis CG, Lalioti MD, Town T, Lee GR, Flavell RA (2005)
Interchromosomal associations between alternatively expressed loci. Nature
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org13 July 2009 | Volume 5 | Issue 7 | e1000559
43. Lomvardas S, Barnea G, Pisapia DJ, Mendelsohn M, Kirkland J, et al. (2006) Download full-text
Interchromosomal Interactions and Olfactory Receptor Choice. Cell 126:
44. Kurukuti S, Tiwari VK, Tavoosidana G, Pugacheva E, Murrell A, et al. (2006)
CTCF binding at the H19 imprinting control region mediates maternally
inherited higher-order chromatin conformation to restrict enhancer access to
Igf2. Proc Natl Acad Sci 103: 10684–10689.
45. Xu N, Donohoe ME, Silva SS, Lee JT (2007) Evidence that homologous X-
chromosome pairing requires transcription and Ctcf protein. Nat Genet 39:
46. Hadjur S, Williams LM, Ryan NK, Cobb BS, Sexton T, et al. (2009) Cohesins
form chromosomal cis-interactions at the developmentally regulated IFNG
locus. Nature Epub.
47. Mishiro T, Ishihara K, Hino S, Tsutsumi S, Aburatani H, et al. (2009)
Architectural roles of multiple chromatin insulators at the human apolipoprotein
gene cluster. EMBO J 28: 1234–1245.
48. Gruber S, Haering CH, Nasmyth K (2003) Chromosomal cohesin forms a ring.
Cell 112: 765–777.
49. Cho DH, Thienes CP, Mahoney SE, Analau E, Filippova GN, et al. (2005)
Antisense transcription and heterochromatin at the DM1 CTG repeats are
constrained by CTCF. Mol Cell 20: 483–489.
50. Li Y, Danzer JR, Alvarez P, Belmont AS, Wallrath LL (2003) Effects of tethering
HP1 to euchromatic regions of the Drosophila genome. Development 130:
51. Su RC, Brown KE, Saaber S, Fisher AG, Merkenschlager M, et al. (2004)
Dynamic assembly of silent chromatin during thymocyte maturation. Nat Genet
52. Petrov A, Pirozhkova I, Carnac G, Laoudj D, Lipinski M, et al. (2006)
Chromatin loop domain organization within the 4q35 locus in facioscapulohu-
meral dystrophy patients versus normal human myoblasts. Proc Natl Acad Sci
53. Gregson HC, Schmiesing JA, Kim J-S, Kobayashi T, Zhou S, et al. (2001) A
potential role for human cohesin in mitotic spindle aster assembly. J Biol Chem
54. Muchir A, van Engelen BG, Lammens M, Mislow JM, McNally E, et al. (2003)
Nuclear envelope alterations in fibroblasts from LGMD1B patients carrying
nonsense Y259X heterozygous or homozygous mutation in lamin A/C gene.
Exp Cell Res 291: 352–362.
55. Muchir A, Bonne G, van der Kooi AJ, van Meegen M, Baas F, et al. (2000)
Identification of mutations in the gene encoding lamins A/C in autosomal
dominant limb girdle muscular dystrophy with atrioventricular conduction
disturbances (LGMD1B). Hum Mol Genet 9: 1453–1459.
56. van der Kooi AJ, Bonne G, Eymard B, Duboc D, Talim B, et al. (2002) Lamin
A/C mutations with lipodystrophy, cardiac abnormalities, and muscular
dystrophy. Neurol 59: 620–623.
57. Hagleitner M, Lankester A, Maraschio P, Hulte ´n M, Fryns J, et al. (2008)
Clinical spectrum of immunodeficiency, centromeric instability and facial
dysmorphism (ICF-syndrome). J Med Genet 45: 93–99.
58. Watts GD, Wymer J, Kovach MJ, Mehta SG, Mumm S, et al. (2004) Inclusion
body myopathy associated with Paget disease of bone and frontotemporal
dementia is caused by mutant valosin-containing protein. Nat Genet 36:
59. Pyle AD, Lock LF, Donovan PJ (2006) Neurotrophins mediate human
embryonic stem cell survival. Nat Biotechnol 24: 344–350.
60. Lyle R, Wright TJ, Clark LN, Hewitt JE (1995) The FSHD-associated repeat,
D4Z4, is a member of a dispersed family of homeobox-containing repeats,
subsets of which are clustered on the short arms of the acrocentric chromosomes.
Genomics 28: 389–397.
61. Heale JT, Ball JAR, Schmiesing JA, Kim JS, Kong X, et al. (2006) Condensin I
interacts with the PARP-1-XRCC1 complex and functions in DNA single-
stranded break repair. Mol Cell 21: 837–848.
62. Chadwick BP (2008) DXZ4 chromatin adopts an opposing conformation to that
of the surrounding chromosome and acquires a novel inactive X-specific role
involving CTCF and antisense transcripts. Genome Res 18: 1259–1269.
63. Ju B-G, Solum D, Song EJ, Lee K-J, Rose DW, et al. (2004) Activating the
PARP-1 sensor component of the groucho/TLE1 corepressor complex mediates
a CaMKinase IId-dependent neurogenic gene activation pathway. Cell 119:
64. Hakimi MA, Bochar DA, Schmiesing JA, Dong Y, Barak OG, et al. (2002) A
chromatin remodeling complex that loads cohesin onto human chromosomes.
Nature 418: 994–998.
65. Weber M, Davies JJ, Wittig D, Oakeley EJ, Haase M, et al. (2005)
Chromosome-wide and promoter-specific analyses identify sites of differential
DNA methylation in normal and transformed human cells. Nat Genet 37:
66. Ait-Si-Ali S, Guasconi V, Fritsch L, Yahi H, Sekhri R, et al. (2004) A Suv39h-
dependent mechanism for silencing S-phase genes in differentiating but not in
cycling cells. EMBO J 23: 605–615.
67. Lee DY, Northrop JP, Kuo MH, Stallcup MR (2006) Histone H3 lysine 9
methyltransferase G9a is a transcriptional coactivator for nuclear receptors. J Biol
Chem 281: 8476–8485.
68. Hewitt JE, Lyle R, Clark LN, Valleley EM, Wright TJ, et al. (1994) Analysis of
the tandem repeat locus D4Z4 associated with facioscapulohumeral muscular
dystrophy. Hum Mol Genet 3: 1287–1295.
69. Sawado T, Igarashi K, Groudine M (2001) Activation of b-major globin gene
transcription is associated with recruitment of NF-E2 to the beta-globin LCR
and gene promoter. Proc Natl Acad Sci 98: 10226–10231.
70. Deidda G, Cacurri S, Piazzo N, Felicetti L (1996) Direct detection of 4q35
rearrangements implicated in facioscapulohumeral muscular dystrophy (FSHD).
J Med Genet 33: 361–365.
Epigenetic Change in FSHD
PLoS Genetics | www.plosgenetics.org14 July 2009 | Volume 5 | Issue 7 | e1000559