Individual epigenetic status of the pathogenic D4Z4 macrosatellite correlates with disease in facioscapulohumeral muscular dystrophy
© Jones et al.; licensee BioMed Central. 2015
Received: 14 October 2014
Accepted: 11 March 2015
Published: 29 March 2015
Both forms of facioscapulohumeral muscular dystrophy (FSHD) are associated with aberrant epigenetic regulation of the chromosome 4q35 D4Z4 macrosatellite. Chromatin changes due to large deletions of heterochromatin (FSHD1) or mutations in chromatin regulatory proteins (FSHD2) lead to relaxation of epigenetic repression and increased expression of the deleterious double homeobox 4 (DUX4) gene encoded within the distal D4Z4 repeat. However, many individuals with the genetic requirements for FSHD remain asymptomatic throughout their lives. Here we investigated family cohorts of FSHD1 individuals who were either affected (manifesting) or without any discernible weakness (nonmanifesting/asymptomatic) and their unaffected family members to determine if individual epigenetic status and stability of repression at the contracted 4q35 D4Z4 array in myocytes correlates with FSHD disease.
Family cohorts were analyzed for DNA methylation on the distal pathogenic 4q35 D4Z4 repeat on permissive A-type subtelomeres. We found DNA hypomethylation in FSHD1-affected subjects, hypermethylation in healthy controls, and distinctly intermediate levels of methylation in nonmanifesting subjects. We next tested if these differences in DNA methylation had functional relevance by assaying DUX4-fl expression and the stability of epigenetic repression of DUX4-fl in myogenic cells. Treatment with drugs that alter epigenetic status revealed that healthy cells were refractory to treatment, maintaining stable repression of DUX4, while FSHD1-affected cells were highly responsive to treatment and thus epigenetically poised to express DUX4. Myocytes from nonmanifesting subjects had significantly higher levels of DNA methylation and were more resistant to DUX4 activation in response to epigenetic drug treatment than cells from FSHD1-affected first-degree relatives containing the same contraction, indicating that the epigenetic status of the contracted D4Z4 array is reflective of disease.
The epigenetic status of the distal 4qA D4Z4 repeat correlates with FSHD disease; FSHD-affected subjects have hypomethylation, healthy unaffected subjects have hypermethylation, and nonmanifesting subjects have characteristically intermediate methylation. Thus, analysis of DNA methylation at the distal D4Z4 repeat could be used as a diagnostic indicator of developing clinical FSHD. In addition, the stability of epigenetic repression upstream of DUX4 expression is a key regulator of disease and a viable therapeutic target.
KeywordsFSHD Muscular dystrophy DUX4 D4Z4 Epiallele Epigenetic modifier Disease modifier Decitabine DNA methylation
Each of the D4Z4 RUs within the 4q35 macrosatellite contains 3.3 kb of highly GC-rich (73%) DNA, encompassing >16 nucleosomes, with multiple repeat sequences normally associated with heterochromatin . Thus, FSHD1-sized deletions remove a substantial amount of regulatory heterochromatin from the 4q35 region, significantly altering the local epigenetic landscape of the contracted allele [20-22]. FSHD2 is also caused by the epigenetic disruption of the 4q35 D4Z4 array leading to aberrant gene expression; however, the dysregulation is not caused by the physical removal of regulatory heterochromatin as in FSHD1 but is due to mutations in gene(s) encoding the epigenetic machinery responsible for establishing and maintaining repression of the D4Z4 array [4,5]. More than 85% of FSHD2 cases analyzed to date are linked to mutations in the SMCHD1 gene [5,23-25], which encodes a chromatin remodeling protein required for normal DNA methylation levels and transcriptional repression at certain loci, including D4Z4 arrays [26-28]. In addition, mutations in the SMCHD1 gene increase the severity of FSHD1 [6,29], indicating that SMCHD1 is an epigenetic modifier of both forms of FSHD. Thus, epigenetic dysregulation of the 4q35 D4Z4 array, albeit through different mechanisms, links FSHD1 and FSHD2 [4,7,8].
A consequence of the epigenetic disruption at 4q35 in FSHD1 and FSHD2 is the increased expression and altered splicing of the double homeobox 4 (DUX4) gene to generate the DUX4-fl (DUX4-full length) mRNA in FSHD skeletal muscle, which results in aberrant expression of DUX4-FL and its downstream target genes with consequent pathology [17,30-36]. Although a copy of DUX4 resides within each RU of the D4Z4 array , only DUX4-fl transcribed from the distal-most 4q35 D4Z4 repeat is stably expressed in FSHD due to the presence of a polyadenylation signal (PAS) in a permissive 4A subtelomere-specific exon distal to the array, which is absent in 4B and other non-permissive subtelomeres . This distal third exon is spliced into the mRNA (thereby explaining the linkage of FSHD to the 4A-type subtelomeres) and translated to produce DUX4-FL protein [17,30]. However, DUX4-FL expression in FSHD is very low and shows cell-to-cell variability as <0.5% of the nuclei in FSHD1-derived myogenic cultures express DUX4-FL [30,33]. Although restricted to a small percentage of myonuclei at any one time, the aberrant expression of DUX4-FL is proposed to lead to progressive muscle atrophy and ultimately FSHD pathology [30-36,38-41]. Two studies have also reported expression of DUX4-fl mRNA and protein in some myogenic cells and muscle tissue from certain asymptomatic and healthy individuals [33,42], although at lower levels than in FSHD1 patients. Thus, DUX4-fl expression per se is not sufficient for developing clinical FSHD, suggesting the existence of disease modifiers both upstream and downstream of DUX4-FL.
As described above, one important class of disease modifier encompasses chromatin regulatory proteins, such as SMCHD1, that function to establish or maintain epigenetic repression of the D4Z4 array, thus affecting DUX4-fl expression. In addition, contracted D4Z4 arrays may be marked by different epigenetic states in different individuals due to shifts in the probabilistic establishment of these states during development, similar to the characteristics of metastable epialleles (reviewed in [43,44]). To investigate the role of epigenetic modifications in FSHD, we analyzed patterns of DNA methylation at the 4q35 D4Z4 array in family cohorts of myogenic cells from FSHD1-affected subjects, FSHD1-nonmanifesting carriers, and healthy controls. We determined that these cells have individual differences at the 4q35 D4Z4 array and these epigenetic differences affected the stability of DUX4 silencing. The patterns of DNA methylation at the distal, pathogenic D4Z4 repeat, as well as inducibility of DUX4-fl expression following epigenetic drug treatment, correlated with disease manifestation and offer an explanation for how individuals can be genetically FSHD1 yet clinically asymptomatic.
There are several key distinguishing aspects of our analysis. We studied our well-characterized FSHD1 family cohorts of myogenic cells derived from muscle biopsies [33,45,46], thus minimizing differences related to genetic background and also allowing the analysis of multiple cohorts of FSHD1-affected subjects and nonmanifesting carriers containing the same D4Z4 contraction. FSHD is a myopathy, and DUX4-fl expression is induced in differentiated myogenic cells ; thus, the use of these cells, as opposed to the lymphocytes used in most other studies, allowed analysis of epigenetic status and pathogenic gene expression in the most affected cell type. In contrast to earlier studies which analyzed very few CpGs, our study used bisulfite sequencing (BSS), enabling analysis of the methylation status for >50 CpGs each in both the gene body and 5′ promoter region of DUX4 . Importantly, our BSS amplifications were specific to the 4qA D4Z4 (4qA and 4qA-L BSS assays) or the 4q and 10q D4Z4 RUs (DUX4 5′ BSS assay). Our assays did not amplify and assess the numerous D4Z4 homologs from other regions of the genome that are not associated with or epigenetically dysregulated in FSHD [48,49]. Finally, we specifically analyzed the pathogenic distal-most D4Z4 repeat for both DNA methylation status and stability of epigenetic repression as indicated by DUX4-fl expression. This is in contrast to most other studies which have analyzed four centromere-proximal D4Z4 repeats (two from 10q, one from the contracted 4q, and one from the non-contracted 4q); these studies do not specifically assess the pathogenic chromosome and they focus on a region far from the site of stable DUX4-fl expression . Our unique approach provides the first epigenetic analysis of the distal DUX4 gene associated with FSHD and identifies distinct epigenetic characteristics of healthy, FSHD1-affected, and FSHD1-nonmanifesting states.
The frequency of DUX4-FL expression is stable in culture
Myogenic cells obtained from different individual donors have large differences in the frequency of DUX4-FL protein expression . Therefore, we first determined if DUX4-FL levels in myogenic cells were stable upon repeated culturing. Our earlier study  raised the possibility that DUX4-FL expression frequencies differed depending on the donor; however, that study examined DUX4-FL protein in only one culture for most donors and did not determine if the number of population doublings affected DUX4-FL expression. In addition, DUX4-FL expression in myogenic cells is almost exclusive to differentiated myocytes, as identified by expression of myosin heavy chain (MyHC) ; our previous study reported the number of DUX4-FL-positive nuclei per 1,000 total nuclei in the cultures and thus did not account for possibly differing extents of differentiation among different cultures. Thus, to extend our previous study, we examined DUX4-FL expression frequencies at different population doublings (PD) using a serial subculturing assay (see the ‘Methods’ section) with differentiated FSHD and unaffected cells derived from the biceps or deltoid muscles of multiple individual donors (Additional file 1: Table S1). Upon repeated subculturing, we found that the doubling times of these primary cultures in growth medium began to slow by PD approximately 55 to 60, therefore we limited DUX4-FL expression experiments to differentiated cultures derived from myogenic cells at PD ≤ ~47, which was prior to the replicative limit.
DUX4-FL expression in differentiated myogenic cell cultures by individual donor and muscle of origin
Eco RI/ Bln I fragment sizes
#DUX4-FL + nuclei per 1,000 nuclei in MyHC + cells (average ± SE ( n ))
29 kb 4A161
0.095 ± 0.028 (14)**
0.17 ± 0.09 (4)
53 kb 4A161
34 kb 4B163
0.00 ± 0.00 (14)**
0.015 ± 0.015 (4)
53 kb 4A161
25 kb 4A161
0.79 ± 0.21 (14)**
2.14 ± 0.84 (4)*
>112 kb 4B168
>112 kb 4A161
0.12 ± 0.08 (14)**
0.00 ± 0.00 (4)*
>112 kb 4A166H
19 kb 4A161
3.71 ± 0.63 (14)**
4.76 ± 0.97 (4)**
87 kb 4A-L161
97 kb 4B163
0.021 ± 0.015 (12)**
0.012 ± 0.012 (4)**
>112 kb 4A161
90 kb 4A161
0.00 ± 0.00 (7)**
>112 kb 4B168
Consistent with our earlier work , we also detected a low frequency of DUX4-FL expression in nuclei within differentiated (MyHC-positive) cells from two of the four healthy (non-FSHD) donors (Table 1). Cells from these two unaffected donors showed a weak trend to higher DUX4-FL expression after repeated subculturing (R 2 = 0.31 for 09Ubic and 0.26 for 17Ubic). As with our previous study investigating DUX4-FL expression in large single cultures of myogenic cells from nine of the Wellstone Center cohorts (03, 07, 09, 12, 15, 16, 17, 18, 20) , for each of the three donor families (07, 09, 17), the average frequency of DUX4-FL-expressing nuclei was higher in differentiated cells from the FSHD donor than from the unaffected donor across multiple cultures (Table 1, n = 4 to 14); this difference reached significance (P < 0.05, t-test) in every case except 07Adel vs. 07Udel (P < 0.15) (Table 1). Thus, the percentage of myonuclei that expressed DUX4-FL varied among cell cultures isolated from different individuals but remained relatively stable among different cultures derived from the same donor biopsy. In cultures from all individuals tested, derived from 13 different biopsies, the number of DUX4-FL expressing nuclei remained stable upon repeated subculturing, indicating that the mechanisms regulating DUX4-FL expression are similarly stable in myocyte cell culture.
Myogenic cells derived from FSHD1-affected subjects are significantly hypomethylated at the distal D4Z4 unit of a contracted 4q array compared with the non-contracted allele and healthy controls
We used the BSS assays described above to compare DNA methylation profiles (Figure 3) in myogenic cells from eight familial cohorts (03, 07, 09, 12, 16, 17, 19, and 21) representing clinically affected (manifesting) FSHD1 subjects that showed low (03A, 07A), mid-level (09A), and high (17A) percentages of DUX4-FL expressing myonuclei, and healthy controls (U). In addition, we assayed peripheral blood mononuclear cells (PBMCs) from three familial cohorts (39, 41, and 51) (Figure 4). In subjects with only one 4qA allele (Additional file 1: Table S1), all of the 4qA BSS data was derived from a single allele. Similarly, in subjects with one 4qA-L allele (Additional file 1: Table S1), all of the 4qA-L BSS data was derived from a single allele. In subjects with two 4qA alleles, 50% of the BSS sequences are expected to originate from each of the two 4q alleles (although the precise percent may differ due to random sampling fluctuations). Thus, for FSHD1 subjects, 50% of the sequences are expected to originate from the pathogenic D4Z4 RU and 50% from the non-contracted distal D4Z4 RU. However, to prevent high and variable methylation at the non-contracted allele from masking or diluting the signal from the contracted allele, we used a statistical mixture-model to estimate the average percent methylation for just the least-methylated of the 4qA or 4qA-L alleles (see the ‘Methods’ section). As expected, the cells from unaffected subjects were hypermethylated (on average 71% methylation across the region for myocytes, 62% for PBMCs) and the cells from 11 FSHD1-affected subjects were hypomethylated (on average 7% for myocytes, 14% for PBMCs). However, despite a >50-fold range in DUX4-FL expressing myonuclei between the FSHD1 samples (Figure 1 and ), there were only small differences in average 4qA DNA methylation (03A = 5.8%, 07A = 17.8%, 09A = 6.7%, and 17A = 9.2%) for the contracted 4qA chromosomes analyzed for each subject. BSS analysis of the DUX4 5′ region supported these results (Additional file 1: Figure S1). Cells from FSHD1-affected subjects displayed higher overall average methylation at the DUX4 5′ region than at the 4qA region, but this is to be expected because the non-contracted 4q and both 10q chromosomes are included in analysis of the 5′ region; moreover, since any D4Z4 repeat (not just the distal-most) may be amplified in this PCR assay, the contracted 4qA allele makes a proportionately smaller contribution to the overall methylation.
Overall, in cells from FSHD1-affected subjects, the contracted 4qA allele is specifically hypomethylated and the non-contracted allele remains hypermethylated. DNA methylation levels at the distal D4Z4 unit are dramatically higher for healthy than for FSHD1-affected cells (P = 2 × 10−12, likelihood ratio test (LRT)), correlating with the correspondingly lower numbers of DUX4-FL expressing myonuclei in healthy cells. However, DNA methylation levels alone do not explain differences in the number of DUX4-FL expressing myonuclei among cells from different FSHD1-affected subjects or explain why so few FSHD1-affected myonuclei in a culture express DUX4-FL. Since DNA methylation is only one component of the epigenetic regulation, it is likely that there are additional differences in the overall chromatin state that can account for these changes in expression levels and frequency.
Myogenic cells from FSHD1-nonmanifesting subjects have intermediate DNA methylation levels at the distal DUX4 on the contracted 4q allele
Comparison of percent DNA methylation between cells derived from FSHD1-affected and nonmanifesting familial cohorts
Eco RI/ Bln I (kb)
14.9 and 16.9
7.3 and 4.9
In summary, higher DNA methylation levels at the distal 4q35 D4Z4 unit on the contracted 4qA allele were significantly correlated with decreased FSHD disease severity in individuals who shared the same FSHD1 deletion (P = 0.004 by a nonparametric sign test, for any choice of which subject to include for the two cases of two affected or two nonmanifesting subjects in a family). This increased level of DNA methylation in nonmanifesting vs. manifesting subjects was also significant in a parametric linear mixed-effects analysis (see the ‘Methods’ section), in which levels for nonmanifesting carriers of FSHD1 contractions are slightly but significantly higher than for manifesting subjects (P = 0.02, LRT), but significantly lower than for healthy controls (P = 1 × 10−7, LRT). Notably, there was no significant difference between myogenic cells and blood cells (P = 0.53, LRT), which makes blood samples appealing as a less-invasive alternative to muscle biopsies, at least for studies of DUX4 methylation.
We conclude that, with respect to the pathogenic distal D4Z4 repeat on the contracted 4qA allele (when appropriate), healthy subjects display DNA hypermethylation, FSHD1 subjects manifesting weakness display hypomethylation, and FSHD1-nonmanifesting subjects display intermediate levels of methylation, slightly but significantly higher than those of FSHD1-affected subjects.
Stability of epigenetic repression is variable between myogenic cells derived from FSHD1-affected and nonmanifesting subjects
As seen previously for DUX4-FL protein expression (Figure 1), initial DUX4-fl mRNA levels for the five cohorts analyzed were variable among the FSHD1 cells, while healthy control cells expressed DUX4-fl at much lower levels. FSHD1-affected and control cells were treated with Decitabine, TSA, or both, and DUX4-fl expression was assayed by qRT-PCR (Figure 8). DUX4-fl was detected in FSHD1-affected cells from both cohorts and, at much lower levels, in healthy controls, consistent with our previous study . Surprisingly, Decitabine treatment of healthy cells, which are hypermethylated at the 4q35 D4Z4 array, only mildly induced DUX4-fl levels and the absolute levels never approached those found in Decitabine-treated cells from FSHD1-affected subjects (Figure 8). Similarly, treatment with TSA had no effect on DUX4-fl levels in any of the healthy controls. Interestingly, the combination of Decitabine and TSA treatment had a small effect on induction in two of the five healthy lines (09U, 4.7-fold; 07U, 10.2-fold); however, again, the resulting DUX4-fl levels were well below those of the treated cells from all five FSHD-affected subjects (Figure 8). To control for efficacy of drug treatment, we assayed the expression of the ankyrin repeat domain 1 (ANKRD1) gene, which is known to be epigenetically regulated in myocytes , and found that Decitabine/TSA treatment significantly induced its expression equally in both unaffected and affected cells (Additional file 1: Figure S6). Thus, with respect to DUX4-fl expression, healthy control cells are refractory to these epigenetic drug treatments, suggesting that normal repression of the non-contracted D4Z4 array is very stable.
Conversely, Decitabine treatment of FSHD1-affected cells, which are already hypomethylated compared with controls at the distal D4Z4 RU (Figures 3 and 4), significantly induced DUX4-fl in four of the five subjects (03A, 50-fold; 07A, 120-fold; 17A, 3.2-fold; 19A, 122-fold) with three of the five showing >50-fold induction. The lone cell line (09A) that did not show induction by Decitabine had the highest levels of DUX4-fl mRNA in the untreated sample, and >40-fold more than its corresponding control cell line (09U), suggesting that these cells may have already reached the maximum level of epigenetic relaxation attainable. Of the five cohorts, only 03A, which expressed the lowest levels of DUX4-fl of all the untreated FSHD-affected cells, showed induction by TSA alone. We conclude that myogenic cells from FSHD1-affected subjects have less stable epigenetic repression of DUX4 than their healthy counterparts, and although the majority of cells do not express DUX4-fl, they are epigenetically poised for DUX4-fl expression.
Similarly, four family cohorts of myogenic cells from FSHD1-affected and nonmanifesting subjects were assayed for their response to Decitabine and/or TSA treatment (Figure 9). Again, Decitabine induced DUX4-fl expression in cells from FSHD1 individuals manifesting weakness in all four cohorts (15A, 28A, 29A, and 30A), while TSA alone had little to no effect. For 29A, the combination of Decitabine and TSA synergistically induced DUX4-fl expression. In parallel, cells from familial nonmanifesting subjects were subjected to the same set of drug treatments and assayed for DUX4-fl expression. For cells from nonmanifesting subject 29B, the pattern of induction was similar, although less pronounced, to that for cells from FSHD1 subject 29A. However, cells from nonmanifesting subjects 15B, 28B, and 30B showed little to no response to Decitabine or TSA, either alone or in combination.
In addition to FSHD-dependent changes in DNA methylation and histone acetylation states, changes in histone methylation at the FSHD locus have also been reported. These changes include reduced histone H3 lysine 9 tri-methylation (H3K9me3) and loss of its binding protein, heterochromatin protein 1 (HP1) [21,49]. Reducing the levels of H3K9me3 with chaetocin (CH), an inhibitor of the SUV39H1 methyltransferase responsible for establishing H3K9me3, induces DUX4-fl expression in immortalized human KD3 myoblasts [49,73,74]. Therefore, we assessed DUX4-fl induction by CH in these cohorts of FSHD-affected and nonmanifesting cells (Figure 9). Similar to treatment with Decitabine, treatment with CH alone induced DUX4-fl expression, and the combination of both increased expression even further. Again, for each treatment, cells from FSHD1 subjects manifesting weakness had higher DUX4-fl levels than cells from their nonmanifesting relatives with the identical 4qA allele. Thus, the repression of DUX4-fl in cells from nonmanifesting carriers is more refractory to induction by epigenetic drugs than in cells from their clinically affected relatives, despite sharing the same D4Z4 contraction.
Patterns of DNA methylation at the pathogenic D4Z4 correlate with disease outcome in FSHD and can distinguish between FSHD1-affected, FSHD1-nonmanifesting, and healthy controls
Studies investigating FSHD1 families have identified asymptomatic individuals that share the same FSHD1 genetic diagnosis as their affected relatives yet report no noticeable muscle weakness [25,33,54,56-58]. Similarly, larger studies of normal individuals with no known FSHD relatives revealed that there are many individuals - reportedly approximately 1% to 3% of certain populations - that fit the current FSHD1 genetic diagnostic criteria yet show no clinical manifestation of the disease [60,75]. It is established that the overall epigenetic status of the 4q35 D4Z4 macrosatellite is distinctly altered between FSHD-affected and healthy control subjects [4,20,21,49,50,76]. Therefore, we hypothesized that epigenetic changes, including DNA methylation at the 4q35 D4Z4 array and stability of epigenetic repression of the DUX4-fl mRNA, between individuals could account, at least in part, for the wide variability in clinical presentation of FSHD and similarly for the large number of asymptomatic individuals that fit the genetic criteria for FSHD1 [1,12,15,17,60,75,77]. Supporting this hypothesis, we found that myogenic cells from these FSHD1-nonmanifesting subjects have an intermediate epigenetic status at the pathogenic distal 4q35 D4Z4 repeat that is not as relaxed as that found in FSHD1 subjects manifesting weakness, but not as repressed as that found in healthy control subjects. In addition, DNA methylation levels at this region correlate with clinical disease, showing significant differences between the high methylation levels of healthy controls, the intermediate levels of FSHD1-nonmanifesting subjects, and the low levels of FSHD1-affected subjects. These differences in DNA methylation levels were significant in both a simple paired comparison between family members and also in a mixed-effect model including all samples (Figure 7).
This conclusion is in general agreement with a very recent publication that utilized the methyl-sensitive Southern blot method to investigate combined 4q and 10q D4Z4 DNA methylation levels at the proximal D4Z4 RU in FSHD1-affected and asymptomatic/nonpenetrant (comparable to our nonmanifesting) individuals . The authors found that for those genetically FSHD1 subjects carrying 7 to 10 RUs at their shortest FSHD-permissive allele, affected subjects have significantly less DNA methylation than predicted based on their 4q and 10q D4Z4 array sizes, while asymptomatic subjects do not. This was interpreted as suggesting that for 7 to 10 RUs, additional factors beyond array size are likely involved in determining methylation levels, and clinical severity, for those with borderline contracted alleles . This is in agreement with our finding that DNA methylation levels on the contracted allele for nonmanifesting subjects differ significantly from those for FSHD1-affected and healthy controls, representing an intermediate level of DNA methylation and epigenetic stability.
In light of this, there are several distinguishing features of our study. We show that in FSHD1 subjects, DNA methylation levels are altered specifically at the contracted distal 4qA D4Z4 RU, and these alterations correlate with disease severity. Importantly, our study goes beyond assaying CpG methylation levels in these subjects and shows that differential DNA methylation is functionally relevant, correlating with general epigenetic repression or relaxation of the contracted 4q35 D4Z4 array, as assayed by the expression of DUX4-fl. Myogenic cells from FSHD1-nonmanifesting subjects, which have intermediate DNA methylation at the distal 4q35 D4Z4 RU of the contracted allele, exhibit greater repression of DUX4-fl than cells from FSHD1-affected subjects, but less repression than healthy control cells. Interestingly, there is also variability in epigenetic repression among FSHD1-affected cells isolated from different subjects, suggesting that an individual’s epigenetic status may be an important aspect of clinical progression as well as disease presentation.
Considering that stable pathogenic DUX4-fl expression originates in the distal D4Z4 RU and extends to the permissive A-type subtelomere, it stands to reason that the distal unit on the contracted 4qA allele is the most critical region to analyze. However, due to technical limitations, all previous FSHD epigenetic studies had focused either on the proximal, non-pathogenic 4q/10q D4Z4 RU or on the random analysis of all 4q/10q D4Z4 RUs [4,20,25,50,51,76]. Our findings for this distal unit confirm earlier reports that hypomethylation in FSHD1 is restricted to the contracted 4q allele in subjects disomic for chromosome 4 type D4Z4 arrays  and moreover offer improved resolution of the allele-specific DNA methylation in two ways: first, in case of 4qA/4qA-L genotypes, the methylation of the two alleles is measured independently; second, for 4qA/4qA genotypes the measurement of methylation at multiple CpG sites per molecule allows us to estimate average methylation for each allele separately, rather than simply measuring the average methylation for both alleles combined.
The epigenetic status of the 4q35 distal D4Z4 region, as assayed by CpG methylation and DUX4-fl mRNA induction in response to epigenetic drugs, not only differs strongly between FSHD1-affected subjects and healthy controls, and between FSHD1-nonmanifesting subjects and healthy controls, but also differs between FSHD1-affected and FSHD1-nonmanifesting subjects within families (Figures 7 and 9). In fact, DNA methylation analysis of the distal 4qA D4Z4 RU could be used effectively as an FSHD biomarker that distinguishes healthy subjects from FSHD1-affected or FSHD1-nonmanifesting subjects. Within families, analysis of DNA methylation alone can generally distinguish between FSHD-affected and FSHD-nonmanifesting relatives (Table 2; cohorts 15, 28, 29, 30, 46, 47, 48, and 49); however, the differences in methylation levels between these genetically FSHD1 groups, while significant at the population level, are smaller than the differences found between either of the groups and healthy controls (Figure 7; Additional file 1: Table S2). Occasional families in which differences between affected and nonmanifesting subjects are minimal (for example, cohort 43), and variability in methylation levels between families, suggest that epigenetic factors in addition to DNA methylation are involved in determining if a subject will be clinically affected or disease nonmanifesting. Still, from a diagnostic standpoint, when combined with a clinical evaluation, this DNA methylation analysis will clearly identify both FSHD1-affected and FSHD1-nonmanifesting subjects from healthy (or non-FSHD) controls; the presence or absence of clinical symptoms consistent with FSHD will differentiate the two hypomethylated groups.
The current diagnostic techniques for FSHD1 include pulsed-field gel electrophoresis (PFGE) and molecular combing [78,79]. These tests can be diagnostic for FSHD1 in a patient with clinical symptoms if a contraction of the 4q35 D4Z4 array is identified ranging between 1 and 10 D4Z4 RUs in cis with an A-type subtelomere ; however, many people with RUs in the higher range (7 to 10 D4Z4 RUs) do not show any clinical manifestation of disease [20,33]. Therefore, PFGE and molecular combing have much less prognostic value for patients possessing D4Z4 contractions at the high end of the FSHD1 range. However, the epigenetic status of the distal D4Z4 RU does correlate with clinical manifestation and thus may be of more prognostic value.
Our results contrast with a recent study by Gaillard et al. , in which D4Z4 DNA methylation levels at the 3′ end of D4Z4s (near our 4qA BSS assay) were reported to be unchanged between FSHD1-affected, asymptomatic, and control cells while DNA methylation changes at the D4Z4 5′ region (similar to our DUX4 5′ BSS assay) could at best only distinguish some FSHD1-affected cells from some unaffected cells, grouping FSHD1 asymptomatic and healthy subjects together. Surprisingly, the authors report D4Z4 DNA methylation levels for FSHD1-asymptomatic cells that were equivalent across the repeat to those found in healthy control cells . This discrepancy between the two studies must be addressed, as it has significant implications for both the clinic, with respect to diagnostics and potentially genetic counseling, and the lab, with respect to understanding disease establishment and mechanism as well as the design of therapeutic approaches. We have identified several critical technical differences between these two studies that can reconcile the data. First, we utilized familial cohorts of FSHD1 subjects with or without disease manifestations who all have D4Z4 repeat arrays of 5 to 8.5 RU (Table 2); the asymptomatic subjects analyzed in the Gaillard et al. study had 7 to 10 RU, which is the typical described range for asymptomatic subjects [56,57,75,80]. In our analysis, these FSHD1-affected subjects were analyzed separately (Figure 7) from FSHD1-affected subjects without familial nonmanifesting subjects, which tend to have smaller contracted alleles with less DNA methylation that could skew the analysis . Additionally, our methylation analysis and interpretation of the DUX4 gene body is based on the distal 4qA D4Z4 RU; thus, either 100% (4qA/B) or approximately 50% (4qA/A) of the assayed chromosomes are from the contracted 4qA array. Therefore, we have specifically analyzed the methylation status of multiple independent sequences from the FSHD1-associated D4Z4, which is important because in FSHD1 only the contracted 4q D4Z4 array shows epigenetic changes . In contrast, the Gaillard et al. study combined all FSHD1-affected subjects, regardless of repeat size or familial relationship, which potentially skewed the average methylation for FSHD1-affected subjects to be lower than if only FSHD1-affected subjects with similar repeat sizes as their FSHD1-asymptomatic subjects were analyzed. In addition, the BSS assays utilized by Gaillard et al., similar to our DUX4 5′ assay, do not distinguish between 4q and 10q D4Z4s and are therefore dominated by D4Z4 sequences derived from the expanded 4q/10q D4Z4 arrays, with sizes averaging between 25 and 60 RUs and potentially reaching over 100 RUs each, leaving D4Z4s from the much smaller contracted FSHD1-associated 4q D4Z4 array (n ≤ 11) as a clear minority in, and potentially altogether absent from, the assayed population. Therefore, in the analysis of ten randomly amplified D4Z4s, the impact of sequences from contracted 4qA alleles on the overall average methylation is expected to be small, and likely within the range of normal variation for the other alleles; thus, their analysis has severely limited statistical power. A further complication involves the sequence variability of BSS amplicons. 4q and 10q D4Z4 repeats have very few sequence polymorphisms , data supported by both of our BSS assays, which both show >99.8% identity to the expected reference sequence (Figures 3, 4, 5, and 6; Additional file 1: Figures S1 and S2), and others . The presence of numerous sequence polymorphisms affecting expected CpG dinucleotides in the Gaillard et al. BSS analysis strongly suggests that D4Z4s were amplified from non-4q/10q D4Z4 homologs . Considering that these D4Z4 homologs are not associated with FSHD or epigenetically altered in the disease , any inclusion of these sequences further complicates the methylation analysis, as it further dilutes the signal from the contracted 4qA allele (important for FSHD1) and also dilutes the signal from combined 4q/10q alleles (important for FSHD2). Thus, the discrepancy between our study and the Gaillard et al. study is likely due to differences in 1) class of subjects analyzed, 2) specificity of the BSS assays, and 3) statistical power of the analysis. It could be suggested that differences might result from our analysis being performed on fewer subjects; however, the fact that the smaller number of samples in our study produced much clearer and more significant differences actually highlights the power of our technique.
Overall, the DNA methylation results produced by our analysis are consistent with the majority of published literature for FSHD1-affected subjects and healthy controls, and the sequences analyzed are clearly specific for the FSHD1-associated D4Z4 array. Therefore, we conclude that FSHD1-nonmanifesting subjects have an intermediate DNA methylation state at the distal D4Z4 on the contracted 4qA allele that distinguishes them from FSHD1 subjects with muscle weakness and from healthy control subjects. In addition, this intermediate state is functionally relevant in that myocytes from FSHD1-nonmanifesting subjects exhibit more stable epigenetic repression than their counterparts from FSHD1-affected first-degree relatives. These different epigenetic states of the distal 4qA D4Z4 repeat can be used effectively as disease biomarkers that clearly distinguish between FSHD1 subjects and healthy controls regardless of any familial relation , have clinical implications for FSHD diagnostics and therapy development, and provide a basis for understanding the mechanism of disease establishment. For example, our results suggest that restoring even an intermediate level of DNA methylation or small increases in heterochromatinization of the D4Z4 array might be sufficient to lower DUX4-fl expression to a non-pathogenic level. In addition, DNA methylation has been found to decrease with age, and these aging-related changes are not global within a cell; some genomic regions change while others do not, and the changes are tissue-specific [81-83]. It is not known if the 4q35 D4Z4 array is susceptible to age-related changes in DNA methylation, but it is possible that the initial epigenetic status of contracted D4Z4 arrays could affect age-related demethylation and thus age of onset or severity of disease.
FSHD1-sized D4Z4 arrays have characteristics of metastable epialleles
The epigenome consists of DNA methylation, histone post-translational modifications, and histone variants throughout the genome that together form an integral component of gene regulatory mechanisms [84-86]. Initially established during development, the epigenome organizes chromatin to restrict or facilitate the access of regulatory factors to DNA. Epigenetic marks provide a mechanism for regulatory memory that is passed on to subsequent cellular generations and is vital for maintaining cell-type specific patterns of expression and repression. The epigenetic modifications at the 4q35 D4Z4 array are established during early development  and differ among individuals. Potentially, variable aspects of the contracted D4Z4 array such as size or inherited DNA methylation patterns, when combined with an individual’s expression level or functional status of chromatin-modifying proteins such as SMCHD1, could shift the establishment of D4Z4 epigenetic repression in either direction. Similarly, stress, nutrition, and exposure to other environmental factors during critical points in development could influence the overall epigenetic state at the D4Z4 arrays. Once established, the epigenetic state would persist and provide protection from or susceptibility to aberrant DUX4-fl expression in muscle.
In addition to the strong influence of epigenetic regulation, another important aspect of FSHD1 contracted D4Z4 regions is the variegated gene expression of DUX4-fl mRNA, as both traits are characteristic of metastable epialleles. Metastable epialleles (reviewed in [43,44]), as opposed to traditional alleles, have variable expressivity leading to phenotypic mosaicism between individuals, as well as variegated cellular expression leading to phenotypic mosaicism between cells. This variable expression is not due to genetic heterogeneity, but rather is dependent on the epigenetic state, which is established in a probabilistic manner during development and then maintained in subsequent cellular generations. FSHD presents clinically with great variability in age of onset, affected muscles, rate of progression, and ultimate severity, even within families and among monozygotic twins [87-91]. The variegated DUX4-fl expression patterns in FSHD1 myogenic cells and the variable clinical manifestation in genetically FSHD1 individuals appear consistent with the FSHD1-associated DUX4 allele functioning as a metastable epiallele .
This study was approved by the Johns Hopkins School of Medicine Institutional Review Board. Families with a member diagnosed with FSHD1 were invited to participate. Individuals were genotyped and considered to be affected with FSHD1 if a 4qA EcoRI/BlnI fragment <35 kb was identified using genomic DNAs isolated from peripheral blood mononuclear cells (PBMC) or considered to be healthy controls if they lacked a contracted 4qA allele (Additional file 1: Table S1). Haplotypes for both 4q alleles were determined for all subjects, as described . All FSHD1 individuals were examined by an experienced neuromuscular physician (KRW). FSHD1 individuals were further characterized as ‘manifesting’ disease (affected) if they had weakness in the distribution classic for FSHD (for example, face, shoulder girdle, and biceps) or ‘nonmanifesting’ if they had full strength in this distribution.
Myogenic cells derived from the biceps muscles of genetically FSHD1 subjects (03Abic, 07Abic, 09Abic, 12Abic, 17Abic, 15Abic, 15Bbic, 16Abic, 19Abic, 21Abic, 28Abic, 28Bbic, 29Abic, 29Bbic, 30Abic, and 30Bbic) and their healthy unaffected first-degree relatives (03Ubic, 07Ubic, 09Ubic, 12Ubic, 16Ubic, 17Ubic, 17Vbic, 19Ubic, and 21Ubic) were used in this study (as previously described, Homma et al. ). All cells were obtained from the Paul. D. Wellstone Muscular Dystrophy CRC for FSHD at the University of Massachusetts Medical School, Worcester, MA (http://www.umassmed.edu/wellstone/). Myogenic cells were selected by FACS for CD56 expression such that all cultures were >90% desmin-positive [33,45]. Myogenic cells were grown on gelatin-coated dishes in high serum growth medium for proliferation then switched to low serum differentiation medium to induce myotube formation [33,45]. As described , proliferation of primary cultures of human myogenic cells began to slow at 55 to 60 population doublings as cells neared replicative limits. Therefore, all cells were used at <30 population doublings, except where indicated in subculturing experiments when cultures were examined at up to 47 population doublings. For all subjects in cohorts 39, 41, 43, 46, 47, 48, 49, and 51, DNA methylation analysis was performed on genomic DNAs isolated from PBMCs collected under IRB-approved protocols at the appropriate institution.
Myogenic cells were cultured in growth medium on gelatin-coated plates to approximately 80% confluence, at which time cells were counted to calculate population doublings and passaged at 1:10 dilution. At each passage, cells were cultured in parallel on one 100-mm plate and one gelatin-coated four-well chamber slide. The culture in each plate was used to maintain myoblasts in growth medium for additional passaging, whereas the culture in each chamber slide was used to generate differentiated myotubes, which were analyzed for DUX4-FL and MyHC expression after 4 days in differentiation medium.
Stock solutions of 100 mM 5-Aza-2′-deoxycytidine/Decitabine, (Sigma-Aldrich A3656, St. Louis, MO, USA) in DMSO, 5 mM Trichostatin A solution (TSA, Sigma-Aldrich T1952), and 10 mM chaetocin (Sigma-Aldrich C9492) in DMSO were stored at −20°C and diluted with PBS just before adding to the culture. To minimize culturing artifacts, low passage (<30 population doublings) myoblast cultures were used for all experiments and culture pairs for affected vs. healthy or affected vs. nonmanifesting were within 1 passage of each other in all instances. Myoblasts were seeded on collagen-coated plates at a cell density of 1.6 × 103/cm2. Starting the following day, Decitabine (5 μM final concentration) was added daily for a total of 3 days. When used, TSA (200 nM final concentration) or chaetocin (50 nM final concentration) was added to the cells for the last 24 h prior to sampling.
Myogenic cell cultures were fixed and co-immunostained for DUX4-FL and myosin heavy chain (MyHC). DUX4-FL was detected with either P4H2 mouse mAb as described  or rabbit mAb E5-5 (Epitomics, Burlingame, CA, USA) as described . MyHC isoforms were detected with either mouse mAb MF20 or mouse mAb F59 , which were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biology, Iowa City, IA, USA. Nuclei were stained with bisbenzimide. The number of DUX4-FL-positive nuclei was determined from manually scanning the entire culture area. The number of nuclei in MyHC-positive cells and the total number of nuclei was approximated for each cell strain by counting 10 random fields of known area at 10X magnification and extrapolating to the total area of the well. Nuclei of 60,000 to 150,000 were screened for each cell culture. Cultures were imaged with a Nikon E800 fluorescence microscope with Spot camera and software, version 4.6 (Diagnostic Instruments, Inc., Sterling Heights, MI, USA).
BSS DNA methylation analysis
For all subjects in cohorts 03, 07, 09, 12, 15, 16, 17, 19, 21, 28, 29, and 30, DNA methylation analysis was performed on genomic DNAs isolated from myocytes. For all subjects in cohorts 39, 41, 43, 46, 47, 48, 49, and 51, DNA methylation analysis was performed on genomic DNAs isolated from PBMCs. DNA methylation of the 4qA and 4qA-L distal regions was analyzed using the 4qA BSS and 4qA-L BSS assays, as described [23,48]. BSS analysis of 59 CpGs in the DUX4 promoter region (DUX4 5′ BSS assay) of 4q and 10q D4Z4 repeats was performed using primers BSS167F: TTTTGGGTTGGGTGGAGATTTT and BSS1036R: AACACCRTACCRAACTTACACCCTT, then followed by nested PCR with BSS475F: TTAGGAGGGAGGGAGGGAGGTAG and BSS1036R using 10% of the first PCR product. PCR products were cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA), sequenced, and analyzed using web-based analysis software BISMA (http://biochem.jacobs-university.de/BDPC/BISMA/)  with the default parameters.
Allele-specific DNA methylation estimation
The percentage of methylated CpG sites in a region can vary between alleles and can also vary between cells for the same allele. To prevent high methylation on the non-contracted 4qA allele from masking or diluting the signal for reduced methylation on the contracted 4qA allele (a weakness with methods that only measure overall average methylation ), we wish to estimate methylation for just the allele with lower methylation. For the purpose of distinguishing FSHD1-affected subjects from healthy controls, we proposed a simple score, the lower quartile (Q1) of percent methylation of all sequenced clones . If two alleles have non-overlapping ranges of methylation and are represented in roughly equal proportions, this will approximate the median for just the allele with lower methylation. But if two alleles have overlapping ranges of methylation, Q1 is biased toward underestimating the median for the allele with lower methylation. Likewise, akin to the extreme cases in which two alleles have identical distributions, Q1 will underestimate the median methylation in cases where only one allele is amplified by the PCR assay, for example, if the other allele is a 4B, 4A-L, or 4A166 haplotype, which may not be known in advance. To reduce this bias, here we use a parametric model-based method for estimating allele-specific methylation.
The distribution of counts of methylated CpG sites across clones is not satisfactorily modeled by a binomial distribution, as the observed variance is typically approximately four times greater than that of a binomial distribution with the same mean and N (where N is the number of CpG sites; N = 56 for the 4qA assay, and N = 30 for the 4qA-L assay) (Additional file 1: Figure S3). This overdispersion is not simply due to the presence of two alleles with different methylation probabilities, as it is also seen when restricting the analysis to samples for which all clones arise from a single 4qA allele (for example, 4qA/4qB genotypes). This overdispersion can also not be addressed by allowing site-specific methylation probabilities for each CpG site (as in ) since by a convexity argument the resulting Poisson binomial distribution has variance at most as large as a standard binomial distribution with the same mean and same N.
To account for the overdispersion, the number of methylated CpGs for each allele (i = 1, 2) was modeled as a beta binomial distribution, where each clone (indexed by j) from the allele has an average methylation probability p ij drawn independently from a beta distribution with parameters a i and b i , and the observed number of methylated CpGs follows a binomial distribution with probability p ij and sample size N. This distribution has the expected average CpG methylation fraction a i /(a i + b i ), with variance decreasing as a i + b i increases, approaching a binomial distribution in the limit of large a i + b i . A Bayesian two-component mixture model was used to infer the parameters of the beta binomial distributions for each of the alleles and to compute the posterior probability of each sequence having originated from each allele, based on the observed methylation data. Note that unlike refs [95,96] we model just the total count of methylated CpGs, and not site-specific methylation probabilities; we also differ in using full Bayesian inference rather than maximum likelihood estimation.
The beta binomials were re-parameterized by r i = log(a i /b i ) and s i = a i + b i for i =1, 2. To break the symmetry between the two alleles and impose a labeling of alleles so that r 1 ≤ r 2 we use a N(μ = 0, σ = 2) prior for the average of r 1 and r 2, and a zero-inflated gamma(k = 1, β = 0.5) distribution as a prior for the difference d = r 1 − r 2 ≥ 0. The zero-inflation puts a 0.5 prior probability mass on the difference being exactly zero, so the model can be used for 4qA/4qA, 4qA/4qB, or unknown genotypes. One could also adjust the prior based on known genotype data, or use the posterior probability that d > 0 as a measure of evidence for allele-specific methylation. We use a gamma(k = 1, β = 0.025) prior for s 1 and s 2. A small fraction of sequences are missing methylation data at a small number (1 to 3) of sites; N was decreased accordingly for these sequences. Posterior means for the parameters of interest were computed using Markov Chain Monte Carlo (MCMC), with the Rjags (v3-14) interface to the JAGS (v3.3.0) sampler. We used 1,000 MCMC steps for burn-in, followed by 30,000 MCMC steps for inference; convergence was monitored with the Gelman-Rubin diagnostic (PFSR < 1.01)  based on three chains run in parallel.
Additional file 1: Figure S4 (top) shows an example (16Abic) in which clones clearly separate into two clusters with distinct methylation percentages, and the two components of the mixture correspond to these two clusters, while allowing for slight deviations from 50% of clones in each cluster; Additional file 1: Figure S4 (bottom) shows an example (17Ubic) in which the clones do not clearly separate into two clusters, and the two estimated mixture components are nearly the same, with the allele of origin ambiguous for all clones; as the genotype of this sample is 4qA/4qB, we do not expect to see evidence of allele-specific methylation here. Bayesian allele-specific estimates depend on the prior probability distributions specified, but we confirmed that the reported differences between groups remained significant for other choices of parameters for the priors (twofold increase or decrease for standard deviation σ of normal prior and rate parameters β for gamma priors).
Comparisons of DNA methylation between disease classes
For comparisons of DUX4 gene body methylation between FSHD-affected, nonmanifesting, and control samples, we first used the procedure described above to estimate the average methylation percentage for the 4A allele with lowest average methylation. For FSHD1 samples, this is expected to be the contracted D4Z4 4A allele. We use the same procedure for control samples with no contracted alleles for uniformity. We likewise use this procedure for samples believed to have only one amplified 4A allele; in such cases, the two allele-specific methylation estimates are typically quite close (within a percent or two, although larger deviations did sometimes occur, particularly in blood, perhaps representing increased mixing of multiple cell lineages).
We used a linear mixed effect (LME) model to fit the values y = log(a/b) for each sample, with fixed effects for cell type (myocyte or blood) and disease class (FSHD-affected, nonmanifesting, or control), including interactions between them, and a random effect for family. We also included an additive fixed effect for assay type (4qA or 4qA-L), as these assess different CpG sites that may have different baseline methylation percentages; indeed, for the 4qA assay, there are variations in CpG methylation probabilities across the length of the sequence, with the central third of the CpG sites typically showing less methylation than the first third (Additional file 1: Figure S5). Because we had limited 4qA-L data, we did not attempt to model interactions between assay type and cell type or disease class here. For sample 17A, which had both 4qA and 4qA-L alleles, we used the 4qA assay as it gave a smaller value of y. This corresponded to the shorter allele (19 kb vs. 87 kb) as desired; however, in the absence of genotyping data, a known baseline difference in methylation between 4qA and 4qA-L alleles could be adjusted for in deciding which should be regarded as the less methylated allele.
Note that y is equal to the log odds ratio log(p/(1 − p)), where p is the average fraction of CpG sites methylated. This logit transformation avoids the compression of values near p = 0 and p = 1. Estimated means and confidence intervals were transformed back to percentages in figures and tables. Models were fit using the R package lme4 (v1.1-7), and likelihood ratio tests were used for assessing significance. Because FSHD-affected subjects with nonmanifesting first-degree relatives may as a group differ from other FSHD subjects (due, for example, to nonmanifesting individuals tending to have borderline D4Z4 repeat lengths), we performed these tests with FSHD subjects divided into two subgroups, allowing nonmanifesting subjects to be compared with just their affected relatives (subgroup FSHD(b)) in a joint model that also includes the other FSHD cases (subgroup FSHD(a)) (for these particular FSHD samples, the two subgroups did not differ significantly; P = 0.29 by LRT). Likelihood ratios were computed between the full model and models with two of the four disease-call subgroups collapsed, or with the two cell types collapsed, with the lme4 function ‘anova’.
Total RNAs were extracted using Trizol (Invitrogen, Carlsbad, CA, USA) and purified using the RNeasy Mini kit (Qiagen, Limburg, Netherlands) after on-column DNase I digestion. Total RNA (2 μg) was used for cDNA synthesis using Superscript III Reverse Transcriptase (Invitrogen), and 200 ng of cDNA were used for DUX4-fl qPCR analysis as described . All data were normalized to levels of 18S rRNA . Oligonucleotide primer sequences are provided in . For the analysis of ANKRD1 mRNA expression, 40 ng of cDNA were used with primers hANKRD1 For: GCCTACGTTTCTGAAGGCTG and Rev: GTGGATTCAAGCATATCACGGAA.
- BS PCR:
facioscapulohumeral muscular dystrophy
likelihood ratio test
myosin heavy chain
polymerase chain reaction
quantitative reverse transcriptase PCR
We thank members of the Sen. Paul D. Wellstone Muscular Dystrophy Cooperative Research Center for FSHD Research for deriving the original cultures of myogenic cells [33,45,46]: Dr. Genila Bibat (Kennedy-Krieger Institute and Johns Hopkins School of Medicine) for coordinating the clinical aspects of the study and Ms. Kendal Hanger (University of Massachusetts Medical School) for the preparation of myogenic cells from biopsies. The authors thank the participating subjects and their families, and Mr. Daniel P. Perez and the FSH Society for subject recruitment and funding of travel costs for subject participation, and are grateful to Dr. Stephen Tapscott and Dr. Linda N. Geng (Fred Hutchinson Cancer Research Center) for generously providing initial samples of DUX4-FL mAbs. Genotyping was performed by Steve Moore at the University of Iowa. This work was funded by grant R01AR062587 from the National Institute of Arthritis, Musculoskeletal and Skin Diseases (NIAMS) NIH awarded to PLJ, NIAMS grant R01AR060328 awarded to JBM, grant AFM15700 from the Association Française contra les myopathies awarded to PLJ and JBM, grant MDA216422 from the Muscular Dystrophy Association (MDA) awarded to JBM, grant #MDA216652 awarded to CPE, and the Senator Paul D. Wellstone Muscular Dystrophy Cooperative Research Center for FSHD Research grant U54HD060848 from the Eunice Kennedy Shriver National Institute for Child Health and Human Development. The authors thank the Chris Carrino Foundation for FSHD and the Thoracic Foundation (Boston, MA, USA) for their support. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
- Padberg GW. Facioscapulohumeral Disease [thesis]. Leiden, the Netherlands: Leiden University; 1982.Google Scholar
- Tawil R. Facioscapulohumeral muscular dystrophy. Neurotherapeutics. 2008;5:601–6.View ArticlePubMed CentralPubMedGoogle Scholar
- Padberg GW, van Engelen BG. Facioscapulohumeral muscular dystrophy. Curr Opin Neurol. 2009;22:539–42.View ArticlePubMedGoogle Scholar
- de Greef JC, Lemmers RJ, van Engelen BG, Sacconi S, Venance SL, Frants RR, et al. Common epigenetic changes of D4Z4 in contraction-dependent and contraction-independent FSHD. Hum Mutat. 2009;30:1449–59.View ArticlePubMedGoogle Scholar
- Lemmers RJ, Tawil R, Petek LM, Balog J, Block GJ, Santen GW, et al. Digenic inheritance of an SMCHD1 mutation and an FSHD-permissive D4Z4 allele causes facioscapulohumeral muscular dystrophy type 2. Nat Genet. 2012;44:1370–4.View ArticlePubMed CentralPubMedGoogle Scholar
- Sacconi S, Lemmers RJ, Balog J, van der Vliet PJ, Lahaut P, van Nieuwenhuizen MP, et al. The FSHD2 gene SMCHD1 is a modifier of disease severity in families affected by FSHD1. Am J Hum Genet. 2013;93:744–51.View ArticlePubMed CentralPubMedGoogle Scholar
- van der Maarel SM, Miller DG, Tawil R, Filippova GN, Tapscott SJ. Facioscapulohumeral muscular dystrophy: consequences of chromatin relaxation. Curr Opin Neurol. 2012;25:614–20.View ArticlePubMed CentralPubMedGoogle Scholar
- Himeda CL, Jones TI, Jones PL. Facioscapulohumeral muscular dystrophy as a model for epigenetic regulation and disease. Antioxid Redox Signal. 2014, In press.
- Prevalence of rare diseases: bibliographic data in Orphanet Report Series: rare diseases collection. [http://www.orpha.net/orphacom/cahiers/docs/GB/Prevalence_of_rare_diseases_by_alphabetical_list.pdf]
- Deenen JC, Arnts H, van der Maarel SM, Padberg GW, Verschuuren JJ, Bakker E, et al. Population-based incidence and prevalence of facioscapulohumeral dystrophy. Neurology. 2014;83:1056–9.View ArticlePubMedGoogle Scholar
- Wijmenga C, Hewitt JE, Sandkuijl LA, Clark LN, Wright TJ, Dauwerse HG, et al. Chromosome 4q DNA rearrangements associated with facioscapulohumeral muscular dystrophy. Nat Genet. 1992;2:26–30.View ArticlePubMedGoogle Scholar
- van Deutekom JC, Wijmenga C, van Tienhoven EA, Gruter AM, Hewitt JE, Padberg GW, et al. FSHD associated DNA rearrangements are due to deletions of integral copies of a 3.2 kb tandemly repeated unit. Hum Mol Genet. 1993;2:2037–42.View ArticlePubMedGoogle Scholar
- Schaap M, Lemmers RJ, Maassen R, van der Vliet PJ, Hoogerheide LF, van Dijk HK, et al. Genome-wide analysis of macrosatellite repeat copy number variation in worldwide populations: evidence for differences and commonalities in size distributions and size restrictions. BMC Genomics. 2013;14:143.View ArticlePubMed CentralPubMedGoogle Scholar
- Rossi M, Ricci E, Colantoni L, Galluzzi G, Frusciante R, Tonali PA, et al. The Facioscapulohumeral muscular dystrophy region on 4qter and the homologous locus on 10qter evolved independently under different evolutionary pressure. BMC Med Genet. 2007;8:8.View ArticlePubMed CentralPubMedGoogle Scholar
- Lemmers RJ, de Kievit P, Sandkuijl L, Padberg GW, van Ommen GJ, Frants RR, et al. Facioscapulohumeral muscular dystrophy is uniquely associated with one of the two variants of the 4q subtelomere. Nat Genet. 2002;32:235–6.View ArticlePubMedGoogle Scholar
- Lemmers RJ, Wohlgemuth M, van der Gaag KJ, van der Vliet PJ, van Teijlingen CM, de Knijff P, et al. Specific sequence variations within the 4q35 region are associated with facioscapulohumeral muscular dystrophy. Am J Hum Genet. 2007;81:884–94.View ArticlePubMed CentralPubMedGoogle Scholar
- Lemmers RJ, van der Vliet PJ, Klooster R, Sacconi S, Camano P, Dauwerse JG, et al. A unifying genetic model for facioscapulohumeral muscular dystrophy. Science. 2010;329:1650–3.View ArticlePubMedGoogle Scholar
- de Greef JC, Lemmers RJ, Camano P, Day JW, Sacconi S, Dunand M, et al. Clinical features of facioscapulohumeral muscular dystrophy 2. Neurology. 2010;75:1548–54.View ArticlePubMed CentralPubMedGoogle Scholar
- Hewitt JE, Lyle R, Clark LN, Valleley EM, Wright TJ, Wijmenga C, et al. Analysis of the tandem repeat locus D4Z4 associated with facioscapulohumeral muscular dystrophy. Hum Mol Genet. 1994;3:1287–95.View ArticlePubMedGoogle Scholar
- van Overveld PG, Enthoven L, Ricci E, Rossi M, Felicetti L, Jeanpierre M, et al. Variable hypomethylation of D4Z4 in facioscapulohumeral muscular dystrophy. Ann Neurol. 2005;58:569–76.View ArticlePubMedGoogle Scholar
- Zeng W, de Greef JC, Chen YY, Chien R, Kong X, Gregson HC, et al. Specific loss of histone H3 lysine 9 trimethylation and HP1gamma/cohesin binding at D4Z4 repeats is associated with facioscapulohumeral dystrophy (FSHD). PLoS Genet. 2009;5:e1000559.View ArticlePubMed CentralPubMedGoogle Scholar
- Balog J, Thijssen PE, de Greef JC, Shah B, van Engelen BG, Yokomori K, et al. Correlation analysis of clinical parameters with epigenetic modifications in the DUX4 promoter in FSHD. Epigenetics. 2012;7:1–6.View ArticleGoogle Scholar
- Mitsuhashi S, Boyden SE, Estrella EA, Jones TI, Rahimov F, Yu TW, et al. Exome sequencing identifies a novel SMCHD1 mutation in facioscapulohumeral muscular dystrophy 2. Neuromuscul Disord. 2013;23:975–80.View ArticlePubMedGoogle Scholar
- Winston J, Duerden L, Mort M, Frayling IM, Rogers MT, Upadhyaya M. Identification of two novel SMCHD1 sequence variants in families with FSHD-like muscular dystrophy. Eur J Hum Genet. 2015;23:67–71.View ArticlePubMedGoogle Scholar
- Lemmers RJ, Goeman JJ, Van Der Vliet PJ, Van Nieuwenhuizen MP, Balog J, Vos-Versteeg M, et al. Inter-individual differences in CpG methylation at D4Z4 correlate with clinical variability in FSHD1 and FSHD2. Hum Mol Genet. 2015;24:659–69.View ArticlePubMedGoogle Scholar
- Blewitt ME, Gendrel AV, Pang Z, Sparrow DB, Whitelaw N, Craig JM, et al. SmcHD1, containing a structural-maintenance-of-chromosomes hinge domain, has a critical role in X inactivation. Nat Genet. 2008;40:663–9.View ArticlePubMedGoogle Scholar
- Gendrel AV, Apedaile A, Coker H, Termanis A, Zvetkova I, Godwin J, et al. Smchd1-dependent and -independent pathways determine developmental dynamics of CpG island methylation on the inactive X chromosome. Dev Cell. 2012;23:265–79.View ArticlePubMed CentralPubMedGoogle Scholar
- Mould AW, Pang Z, Pakusch M, Tonks ID, Stark M, Carrie D, et al. Smchd1 regulates a subset of autosomal genes subject to monoallelic expression in addition to being critical for X inactivation. Epigenetics Chromatin. 2013;6:19.View ArticlePubMed CentralPubMedGoogle Scholar
- Larsen M, Rost S, El Hajj N, Ferbert A, Deschauer M, Walter MC, et al. Diagnostic approach for FSHD revisited: SMCHD1 mutations cause FSHD2 and act as modifiers of disease severity in FSHD1. Eur J Hum Genet. 2014.
- Snider L, Geng LN, Lemmers RJ, Kyba M, Ware CB, Nelson AM, et al. Facioscapulohumeral dystrophy: incomplete suppression of a retrotransposed gene. PLoS Genet. 2010;6:e1001181.View ArticlePubMed CentralPubMedGoogle Scholar
- Wuebbles RD, Long SW, Hanel ML, Jones PL. Testing the effects of FSHD candidate gene expression in vertebrate muscle development. Int J Clin Exp Pathol. 2010;3:386–400.PubMed CentralPubMedGoogle Scholar
- Wallace LM, Garwick SE, Mei W, Belayew A, Coppee F, Ladner KJ, et al. DUX4, a candidate gene for facioscapulohumeral muscular dystrophy, causes p53-dependent myopathy in vivo. Ann Neurol. 2011;69:540–52.View ArticlePubMed CentralPubMedGoogle Scholar
- Jones TI, Chen JC, Rahimov F, Homma S, Arashiro P, Beermann ML, et al. Facioscapulohumeral muscular dystrophy family studies of DUX4 expression: evidence for disease modifiers and a quantitative model of pathogenesis. Hum Mol Genet. 2012;21:4419–30.View ArticlePubMed CentralPubMedGoogle Scholar
- Geng LN, Yao Z, Snider L, Fong AP, Cech JN, Young JM, et al. DUX4 activates germline genes, retroelements, and immune mediators: implications for facioscapulohumeral dystrophy. Dev Cell. 2012;22:38–51.View ArticlePubMed CentralPubMedGoogle Scholar
- Krom YD, Thijssen PE, Young JM, den Hamer B, Balog J, Yao Z, et al. Intrinsic epigenetic regulation of the D4Z4 macrosatellite repeat in a transgenic mouse model for FSHD. PLoS Genet. 2013;9:e1003415.View ArticlePubMed CentralPubMedGoogle Scholar
- Caruso N, Herberth B, Bartoli M, Puppo F, Dumonceaux J, Zimmermann A, et al. Deregulation of the protocadherin gene FAT1 alters muscle shapes: implications for the pathogenesis of facioscapulohumeral dystrophy. PLoS Genet. 2013;9:e1003550.View ArticlePubMed CentralPubMedGoogle Scholar
- Gabriels J, Beckers MC, Ding H, De Vriese A, Plaisance S, van der Maarel SM, et al. Nucleotide sequence of the partially deleted D4Z4 locus in a patient with FSHD identifies a putative gene within each 3.3 kb element. Gene. 1999;236:25–32.View ArticlePubMedGoogle Scholar
- Kowaljow V, Marcowycz A, Ansseau E, Conde CB, Sauvage S, Matteotti C, et al. The DUX4 gene at the FSHD1A locus encodes a pro-apoptotic protein. Neuromuscul Disord. 2007;17:611–23.View ArticlePubMedGoogle Scholar
- Bosnakovski D, Xu Z, Gang EJ, Galindo CL, Liu M, Simsek T, et al. An isogenetic myoblast expression screen identifies DUX4-mediated FSHD-associated molecular pathologies. EMBO J. 2008;27:2766–79.View ArticlePubMed CentralPubMedGoogle Scholar
- Young JM, Whiddon JL, Yao Z, Kasinathan B, Snider L, Geng LN, et al. DUX4 binding to retroelements creates promoters that are active in FSHD muscle and testis. PLoS Genet. 2013;9:e1003947.View ArticlePubMed CentralPubMedGoogle Scholar
- Tassin A, Laoudj-Chenivesse D, Vanderplanck C, Barro M, Charron S, Ansseau E, et al. DUX4 expression in FSHD muscle cells: how could such a rare protein cause a myopathy? J Cell Mol Med. 2013;17:76–89.View ArticlePubMed CentralPubMedGoogle Scholar
- Broucqsault N, Morere J, Gaillard MC, Dumonceaux J, Torrents J, Salort-Campana E, et al. Dysregulation of 4q35- and muscle-specific genes in fetuses with a short D4Z4 array linked to facio-scapulo-humeral dystrophy. Hum Mol Genet. 2013;22:4206–14.View ArticlePubMedGoogle Scholar
- Rakyan VK, Blewitt ME, Druker R, Preis JI, Whitelaw E. Metastable epialleles in mammals. Trends Genet. 2002;18:348–51.View ArticlePubMedGoogle Scholar
- Dolinoy DC, Das R, Weidman JR, Jirtle RL. Metastable epialleles, imprinting, and the fetal origins of adult diseases. Pediatr Res. 2007;61:30R–7.View ArticlePubMedGoogle Scholar
- Homma S, Chen JC, Rahimov F, Beermann ML, Hanger K, Bibat GM, et al. A unique library of myogenic cells from facioscapulohumeral muscular dystrophy subjects and unaffected relatives: family, disease and cell function. Eur J Hum Genet. 2012;20:404–10.View ArticlePubMed CentralPubMedGoogle Scholar
- Rahimov F, King OD, Leung DG, Bibat GM, Emerson Jr CP, Kunkel LM, et al. Transcriptional profiling in facioscapulohumeral muscular dystrophy to identify candidate biomarkers. Proc Natl Acad Sci U S A. 2012;109:16234–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Himeda CL, Debarnot C, Homma S, Beermann ML, Miller JB, Jones PL, et al. Myogenic enhancers regulate expression of the facioscapulohumeral muscular dystrophy associated DUX4 gene. Mol Cell Biol. 2014;34:1942–55.View ArticlePubMed CentralPubMedGoogle Scholar
- Jones TI, Yan C, Sapp PC, McKenna-Yasek D, Kang PB, Quinn C, et al. Identifying diagnostic DNA methylation profiles for facioscapulohumeral muscular dystrophy in blood and saliva using bisulfite sequencing. Clin Epigenetics. 2014;6:23.View ArticlePubMed CentralPubMedGoogle Scholar
- Zeng W, Chen YY, Newkirk DA, Wu B, Balog J, Kong X, et al. Genetic and epigenetic characteristics of FSHD-associated 4q and 10q D4Z4 that are distinct from non-4q/10q D4Z4 homologs. Hum Mutat. 2014;35:998–1010.View ArticlePubMedGoogle Scholar
- van Overveld PG, Lemmers RJ, Sandkuijl LA, Enthoven L, Winokur ST, Bakels F, et al. Hypomethylation of D4Z4 in 4q-linked and non-4q-linked facioscapulohumeral muscular dystrophy. Nat Genet. 2003;35:315–7.View ArticlePubMedGoogle Scholar
- Gaillard MC, Roche S, Dion C, Tasmadjian A, Bouget G, Salort-Campana E, et al. Differential DNA methylation of the D4Z4 repeat in patients with FSHD and asymptomatic carriers. Neurology. 2014;83:733–42.View ArticlePubMedGoogle Scholar
- Hartweck LM, Anderson LJ, Lemmers RJ, Dandapat A, Toso EA, Dalton JC, et al. A focal domain of extreme demethylation within D4Z4 in FSHD2. Neurology. 2013;80:392–9.View ArticlePubMed CentralPubMedGoogle Scholar
- Ottaviani A, Schluth-Bolard C, Gilson E, Magdinier F. D4Z4 as a prototype of CTCF and lamins-dependent insulator in human cells. Nucleus. 2010;1:30–6.View ArticlePubMed CentralPubMedGoogle Scholar
- Ricci E, Galluzzi G, Deidda G, Cacurri S, Colantoni L, Merico B, et al. Progress in the molecular diagnosis of facioscapulohumeral muscular dystrophy and correlation between the number of KpnI repeats at the 4q35 locus and clinical phenotype. Ann Neurol. 1999;45:751–7.View ArticlePubMedGoogle Scholar
- Wohlgemuth M, Lemmers RJ, van der Kooi EL, van der Wielen MJ, van Overveld PG, Dauwerse H, et al. Possible phenotypic dosage effect in patients compound heterozygous for FSHD-sized 4q35 alleles. Neurology. 2003;61:909–13.View ArticlePubMedGoogle Scholar
- Tonini MM, Passos-Bueno MR, Cerqueira A, Matioli SR, Pavanello R, Zatz M. Asymptomatic carriers and gender differences in facioscapulohumeral muscular dystrophy (FSHD). Neuromuscul Disord. 2004;14:33–8.View ArticlePubMedGoogle Scholar
- Goto K, Nishino I, Hayashi YK. Very low penetrance in 85 Japanese families with facioscapulohumeral muscular dystrophy 1A. J Med Genet. 2004;41:e12.View ArticlePubMed CentralPubMedGoogle Scholar
- Sakellariou P, Kekou K, Fryssira H, Sofocleous C, Manta P, Panousopoulou A, et al. Mutation spectrum and phenotypic manifestation in FSHD Greek patients. Neuromuscul Disord. 2012;22:339–49.View ArticlePubMedGoogle Scholar
- Scionti I, Fabbri G, Fiorillo C, Ricci G, Greco F, D’Amico R, et al. Facioscapulohumeral muscular dystrophy: new insights from compound heterozygotes and implication for prenatal genetic counselling. J Med Genet. 2012;49:171–8.View ArticlePubMedGoogle Scholar
- Scionti I, Greco F, Ricci G, Govi M, Arashiro P, Vercelli L, et al. Large-scale population analysis challenges the current criteria for the molecular diagnosis of fascioscapulohumeral muscular dystrophy. Am J Hum Genet. 2012;90:628–35.View ArticlePubMed CentralPubMedGoogle Scholar
- Jones PA, Taylor SM. Cellular differentiation, cytidine analogs and DNA methylation. Cell. 1980;20:85–93.View ArticlePubMedGoogle Scholar
- Yoshida M, Kijima M, Akita M, Beppu T. Potent and specific inhibition of mammalian histone deacetylase both in vivo and in vitro by trichostatin A. J Biol Chem. 1990;265:17174–9.PubMedGoogle Scholar
- Stresemann C, Lyko F. Modes of action of the DNA methyltransferase inhibitors azacytidine and decitabine. Int J Cancer. 2008;123:8–13.View ArticlePubMedGoogle Scholar
- Komashko VM, Farnham PJ. 5-azacytidine treatment reorganizes genomic histone modification patterns. Epigenetics. 2010;5:229–40.View ArticlePubMedGoogle Scholar
- Lin JC, Jeong S, Liang G, Takai D, Fatemi M, Tsai YC, et al. Role of nucleosomal occupancy in the epigenetic silencing of the MLH1 CpG island. Cancer Cell. 2007;12:432–44.View ArticlePubMedGoogle Scholar
- Si J, Boumber YA, Shu J, Qin T, Ahmed S, He R, et al. Chromatin remodeling is required for gene reactivation after decitabine-mediated DNA hypomethylation. Cancer Res. 2010;70:6968–77.View ArticlePubMed CentralPubMedGoogle Scholar
- Taddei A, Maison C, Roche D, Almouzni G. Reversible disruption of pericentric heterochromatin and centromere function by inhibiting deacetylases. Nat Cell Biol. 2001;3:114–20.View ArticlePubMedGoogle Scholar
- Yang XJ, Seto E. HATs and HDACs: from structure, function and regulation to novel strategies for therapy and prevention. Oncogene. 2007;26:5310–8.View ArticlePubMedGoogle Scholar
- Chambers AE, Banerjee S, Chaplin T, Dunne J, Debernardi S, Joel SP, et al. Histone acetylation-mediated regulation of genes in leukaemic cells. Eur J Cancer. 2003;39:1165–75.View ArticlePubMedGoogle Scholar
- Glaser KB, Staver MJ, Waring JF, Stender J, Ulrich RG, Davidsen SK. Gene expression profiling of multiple histone deacetylase (HDAC) inhibitors: defining a common gene set produced by HDAC inhibition in T24 and MDA carcinoma cell lines. Mol Cancer Ther. 2003;2:151–63.PubMedGoogle Scholar
- Cameron EE, Bachman KE, Myohanen S, Herman JG, Baylin SB. Synergy of demethylation and histone deacetylase inhibition in the re-expression of genes silenced in cancer. Nat Genet. 1999;21:103–7.View ArticlePubMedGoogle Scholar
- Juan AH, Derfoul A, Feng X, Ryall JG, Dell’Orso S, Pasut A, et al. Polycomb EZH2 controls self-renewal and safeguards the transcriptional identity of skeletal muscle stem cells. Genes Dev. 2011;25:789–94.View ArticlePubMed CentralPubMedGoogle Scholar
- Greiner D, Bonaldi T, Eskeland R, Roemer E, Imhof A. Identification of a specific inhibitor of the histone methyltransferase SU(VAR)3-9. Nat Chem Biol. 2005;1:143–5.View ArticlePubMedGoogle Scholar
- Rea S, Eisenhaber F, O’Carroll D, Strahl BD, Sun ZW, Schmid M, et al. Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature. 2000;406:593–9.View ArticlePubMedGoogle Scholar
- Ricci G, Scionti I, Sera F, Govi M, D’Amico R, Frambolli I, et al. Large scale genotype-phenotype analyses indicate that novel prognostic tools are required for families with facioscapulohumeral muscular dystrophy. Brain. 2013;136:3408–17.View ArticlePubMed CentralPubMedGoogle Scholar
- de Greef JC, Wohlgemuth M, Chan OA, Hansson KB, Smeets D, Frants RR, et al. Hypomethylation is restricted to the D4Z4 repeat array in phenotypic FSHD. Neurology. 2007;69:1018–26.View ArticlePubMedGoogle Scholar
- Wijmenga C, Frants RR, Hewitt JE, van Deutekom JC, van Geel M, Wright TJ, et al. Molecular genetics of facioscapulohumeral muscular dystrophy. Neuromuscul Disord. 1993;3:487–91.View ArticlePubMedGoogle Scholar
- Ehrlich M, Jackson K, Tsumagari K, Camano P, Lemmers RJ. Hybridization analysis of D4Z4 repeat arrays linked to FSHD. Chromosoma. 2007;116:107–16.View ArticlePubMed CentralPubMedGoogle Scholar
- Nguyen K, Walrafen P, Bernard R, Attarian S, Chaix C, Vovan C, et al. Molecular combing reveals allelic combinations in facioscapulohumeral dystrophy. Ann Neurol. 2011;70:627–33.View ArticlePubMedGoogle Scholar
- Tawil R, Forrester J, Griggs RC, Mendell J, Kissel J, McDermott M, et al. Evidence for anticipation and association of deletion size with severity in facioscapulohumeral muscular dystrophy. The FSH-DY Group. Ann Neurol. 1996;39:744–8.View ArticlePubMedGoogle Scholar
- Jung M, Pfeifer GP. Aging and DNA methylation. BMC Biol. 2015;13:7.View ArticlePubMed CentralPubMedGoogle Scholar
- Day K, Waite LL, Thalacker-Mercer A, West A, Bamman MM, Brooks JD, et al. Differential DNA methylation with age displays both common and dynamic features across human tissues that are influenced by CpG landscape. Genome Biol. 2013;14:R102.View ArticlePubMed CentralPubMedGoogle Scholar
- Horvath S, Zhang Y, Langfelder P, Kahn RS, Boks MP, van Eijk K, et al. Aging effects on DNA methylation modules in human brain and blood tissue. Genome Biol. 2012;13:R97.View ArticlePubMed CentralPubMedGoogle Scholar
- Bernstein BE, Meissner A, Lander ES. The mammalian epigenome. Cell. 2007;128:669–81.View ArticlePubMedGoogle Scholar
- Bonasio R, Tu S, Reinberg D. Molecular signals of epigenetic states. Science. 2010;330:612–6.View ArticlePubMed CentralPubMedGoogle Scholar
- Rivera CM, Ren B. Mapping human epigenomes. Cell. 2013;155:39–55.View ArticlePubMedGoogle Scholar
- Tawil R, Van Der Maarel SM. Facioscapulohumeral muscular dystrophy. Muscle Nerve. 2006;34:1–15.View ArticlePubMedGoogle Scholar
- Pandya S, King WM, Tawil R. Facioscapulohumeral dystrophy. Phys Ther. 2008;88:105–13.View ArticlePubMedGoogle Scholar
- Tawil R, Storvick D, Feasby TE, Weiffenbach B, Griggs RC. Extreme variability of expression in monozygotic twins with FSH muscular dystrophy. Neurology. 1993;43:345–8.View ArticlePubMedGoogle Scholar
- Griggs RC, Tawil R, McDermott M, Forrester J, Figlewicz D, Weiffenbach B. Monozygotic twins with facioscapulohumeral dystrophy (FSHD): implications for genotype/phenotype correlation. FSH-DY Group. Muscle Nerve. 1995;2:S50–5.View ArticleGoogle Scholar
- Tupler R, Barbierato L, Memmi M, Sewry CA, De Grandis D, Maraschio P, et al. Identical de novo mutation at the D4F104S1 locus in monozygotic male twins affected by facioscapulohumeral muscular dystrophy (FSHD) with different clinical expression. J Med Genet. 1998;35:778–83.View ArticlePubMed CentralPubMedGoogle Scholar
- Yoon S, Stadler G, Beermann ML, Schmidt EV, Windelborn JA, Schneiderat P, et al. Immortalized myogenic cells from congenital muscular dystrophy type1A patients recapitulate aberrant caspase activation in pathogenesis: a new tool for MDC1A research. Skelet Muscle. 2013;3:28.View ArticlePubMed CentralPubMedGoogle Scholar
- Miller JB, Crow MT, Stockdale FE. Slow and fast myosin heavy chain content defines three types of myotubes in early muscle cell cultures. J Cell Biol. 1985;101:1643–50.View ArticlePubMedGoogle Scholar
- Rohde C, Zhang Y, Reinhardt R, Jeltsch A. BISMA–fast and accurate bisulfite sequencing data analysis of individual clones from unique and repetitive sequences. BMC Bioinformatics. 2010;11:230.View ArticlePubMed CentralPubMedGoogle Scholar
- Fang F, Hodges E, Molaro A, Dean M, Hannon GJ, Smith AD. Genomic landscape of human allele-specific DNA methylation. Proc Natl Acad Sci U S A. 2012;109:7332–7.View ArticlePubMed CentralPubMedGoogle Scholar
- Dolzhenko E, Smith AD. Using beta-binomial regression for high-precision differential methylation analysis in multifactor whole-genome bisulfite sequencing experiments. BMC Bioinformatics. 2014;15:215.View ArticlePubMed CentralPubMedGoogle Scholar
- Gelman A, Rubin DB. Inference from iterative simulation using multiple sequences. Stat Sci. 1992:457–72.
- Bodega B, Ramirez GD, Grasser F, Cheli S, Brunelli S, Mora M, et al. Remodeling of the chromatin structure of the facioscapulohumeral muscular dystrophy (FSHD) locus and upregulation of FSHD-related gene 1 (FRG1) expression during human myogenic differentiation. BMC Biol. 2009;7:41.View ArticlePubMed CentralPubMedGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. 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.