Transcriptional regulation differs in affected facioscapulohumeral muscular dystrophy patients compared to asymptomatic related carriers

Abstract

Facioscapulohumeral muscular dystrophy (FSHD) is a progressive muscle disorder that has been associated with a contraction of 3.3-kb repeats on chromosome 4q35. FSHD is characterized by a wide clinical inter- and intrafamilial variability, ranging from wheelchair-bound patients to asymptomatic carriers. Our study is unique in comparing the gene expression profiles from related affected, asymptomatic carrier, and control individuals. Our results suggest that the expression of genes on chromosome 4q is altered in affected and asymptomatic individuals. Remarkably, the changes seen in asymptomatic samples are largely in products of genes encoding several chemokines, whereas the changes seen in affected samples are largely in genes governing the synthesis of GPI-linked proteins and histone acetylation. Besides this, the affected patient and related asymptomatic carrier share the 4qA161 haplotype. Thus, these polymorphisms by themselves do not explain the pathogenicity of the contracted allele. Interestingly, our results also suggest that the miRNAs might mediate the regulatory network in FSHD. Together, our results support the previous evidence that FSHD may be caused by transcriptional dysregulation of multiple genes, in cis and in trans, and suggest some factors potentially important for FSHD pathogenesis. The study of the gene expression profiles from asymptomatic carriers and related affected patients is a unique approach to try to enhance our understanding of the missing link between the contraction in D4Z4 repeats and muscle disease, while minimizing the effects of differences resulting from genetic background.

Facioscapulohumeral muscular dystrophy (FSHD) is an autosomal dominant disorder, the locus of which has been mapped to the subtelomeric portion of chromosome 4, at 4q35 (1). This region is characterized by a series of 3.3-kb repeats termed D4Z4. The D4Z4 array may vary from 11 to more than 100 units in the general population, whereas most FSHD patients have a partial deletion of an integral number of these repeats, and exhibit 10 or fewer units (2). Similarly sized D4Z4 regions are observed in affected relative members within families.

FSHD is typically characterized initially by facial muscle weakness. During progression, weakness and atrophy of shoulder girdle muscles is observed in almost all cases. A gradual spread to abdominal and foot-extensor muscles, followed by clinical involvement of upper arm and pelvic girdle muscles, is seen in most patients. Asymmetry of muscle involvement is common. Among the extramuscular features that may be associated with FSHD are retinal abnormalities and high-tone hearing loss (3). Depression, muscle pain, and fatigue are also often observed among FSHD patients (4–6).

Although it is not possible to predict the course of the disease, there tends to be an inverse relationship between the residual repeat size, the age at onset, and the severity of the disease. Patients with 1 to 3 repeat units are usually very severely affected, whereas patients with 4 to 10 repeats tend to have a milder course (7).

FSHD is also characterized by interfamilial and intrafamilial variability, with severity ranging from asymptomatic carriers to loss of ambulation (8). Some families also show clinical anticipation, although all affected members carry the same deleted fragment (9, 10). Males are on average more often and more severely affected than females, with ≈20% of individuals related to FSHD patients remaining asymptomatic (11, 12). Furthermore, these nonpenetrant cases seem to be more common in particular families (13).

Several observations have suggested that FSHD is caused by a complex and uncommon mechanism. Despite the 98% of homology between the D4Z4 repeats at 4q35 and 10q26, and the equal frequency of translocations observed between these 2 regions, FSHD is uniquely associated with contractions on chromosome 4. Contraction of a translocated 4-type allele on chromosome 10 does not cause FSHD (14–16). As monosomy of 4qter is not associated with FSHD, it is believed that the presence of a small number of D4Z4 repeats is crucial to the process leading to the disease (17). In addition, studies of 4 different polymorphic markers showed that FSHD is restricted to one specific haplotype at 4q35ter, 4qA161 (18). In addition, the proximal unit of D4Z4 is significantly hypomethylated in affected and asymptomatic carriers, while in type-2 FSHD, a form of the disease that is not linked to contractions of D4Z4 repeats at 4q35, both alleles are significantly hypomethylated (19).

Some studies have proposed that FSHD is caused by the transcription of a putative gene encoded by the D4Z4 repeats, termed double homeobox 4 (DUX4) (20, 21). The over-expression of DUX4 is generally toxic to cells, leading to apoptosis and activation of PITX1 (paired-like homeodomain transcription factor 1). Changes in DUX4 or PITX1, both homeodomain proteins, could explain several of the key features in FSHD, including the left–right asymmetry, atypical inflammatory responses, and defects in myoblasts reported in FSHD patients (20, 21). Another model suggests that over-expression of critical genes upstream of the D4Z4 repeats at 4q35 in FSHD patients causes a loss of position-effect variegation (22, 23). The findings that a repressor complex composed of YY1, HMGB2, and nucleolin bind to D4Z4 and that a nuclear matrix attachment site (S/MAR), located immediately upstream of D4Z4, dissociates from the nuclear matrix in FSHD patients, are consistent with a model for dysregulation of genes in cis as a primary event in FSHD (24, 25). Data consistent with such dysregulation have not been supported by several studies using gene array or quantitative RT-PCR, however.

Despite the substantial effort to elucidate the molecular mechanism underlying FSHD, the exact mechanism is still unclear. It is also significant that neither of the proposed mechanisms can explain the clinical variability characteristic of FSHD, clinical anticipation, or why some individuals with 10 or fewer D4Z4 repeats remain asymptomatic. We postulated that these asymptomatic individuals might provide important clues about the pathogenesis of FSHD and the molecular mechanisms to suppress it. To test this, we compared the gene-expression profiles from affected individuals, asymptomatic carriers, and normal controls through microarray analysis, looking for genes that might be implicated in suppression or enhancement of the disease expressivity. We focused our studies on samples from related individuals to minimize variations caused by genetic background. The contribution of microRNAs to FSHD was also evaluated.

Results

Expression Profiles in Affected Patients and Asymptomatic Carriers.

The gene-expression profiles from 3 related members (affected, asymptomatic carrier, and control) from 5 different families were compared through microarrays. Using the criteria described in Materials and Methods for microarray data analysis to identify the differentially expressed genes among affected individuals, asymptomatic carriers, and normal controls, 180 loci-annotated probes were found to be significantly dysregulated. When we compared the expression levels between the affected and asymptomatic, 147 probes were significantly dysregulated, of which 13 were up-regulated in affected patients compared to asymptomatic relatives [supporting information (SI) Tables S1 and S2]. In comparisons of affected individuals and healthy controls, only 56 probes were differentially expressed, of which 20 were up-regulated in affected relative to controls (Table S3 and Table S4). Comparisons of asymptomatic carriers and healthy controls identified 12 probes with a significant fold-change, all of which were up-regulated. Surprisingly, 5 of these probes represent genes from chromosome 4q (Table 1). The only gene similarly and significantly dysregulated in affected and asymptomatic carriers relative to control is IGHA1 (see Table 1), encoding an Ig heavy chain, suggesting that its up-regulation might be related to the presence of the FSHD repeat contraction in these individuals.

Table 1.

Expression fold-changes of significantly dysregulated genes in asymptomatic vs. control

Probe set	Gene	Loci	Fold-change (asy/cont)	Fold-change (asy/aff)	Fold-change (aff/cont)
217022_s_at	IGHA1	14q32.33	13.85	0.69	19.98
203915_at	CXCL9	4q21	11.44	1.85	6.19
211122_s_at	CXCL11	4q21.2	10.66	8.52	1.25
210163_at	CXCL11	4q21.2	8.91	12.13	0.73
205890_s_at	UBD	6p21.3	8.60	8.13	1.06
1557690_x_at	NPAS2	2q11.2	4.62	4.99	0.92
1558340_at	DIXDC1	11q23.1	4.00	5.16	0.78
1565935_at	LOC91431	4q25	3.98	4.65	0.86
216968_at	MASP2	1p36.3-p36.2	3.98	4.60	0.87
228230_at	PRIC285	20q13.33	3.95	4.32	0.91
206835_at	STATH	4q11-q13	3.92	3.11	1.26
243874_at	LPP	3q28	3.49	4.87	0.72

Underlined fold-changes represent those deemed significant.

The categories of biological processes that our results identified as altered in FSHD were categorized in the DAVID program as described in Materials and Methods. The biological processes most affected in FSHD patients, compared to controls, are involved in histone acetyltransferase and synthesis of GPI-anchors (Table 2). In asymptomatic carriers, however, there is a clear prevalence of processes related to chemokines (see Table 2), and in affected versus asymptomatic samples, most of the categories are related to regulation of transcription (see Table 2).

Table 2.

Gene ontology (GO) categories enriched in the set of differentially expressed genes (P < 0.05) ranked by the fold-enrichment

GO category	Fold enrichment
Affected vs. Control
Histone acetyltransferase binding	80.6
GPI anchor metabolic process	32.9
GPI-anchor biosynthesis	32.3
Phosphoinositide metabolic process	16.7
Lipoprotein metabolic process	13.4
Glycerophospholipid metabolic process	12.2
Glycan structures– biosynthesis 2	11.1
Membrane lipid biosynthetic process	9.5
Nucleolus	7.0
Membrane lipid metabolic process	6.0
Cell adhesion	5.3
Intracellular non-membrane-bound organelle	2.1
Nonmembrane-bound organelle	2.1
Asymptomatic vs. Control
CXC chemokine	190.8
Small chemokine, C-X-C	153.4
Small chemokine, interleukin-8-like	65.2
Chemokine activity	58.8
Chemokine receptor binding	57.6
Inflammatory response	51.1
SCY	46.8
G-protein-coupled receptor binding	38.4
Locomotory behavior	20.2
Inflammatory response	12.8
Behavior	12.0
Response to wounding	9.0
Immune response	8.4
Immune system process	6.3
Secreted	4.9
Direct protein sequencing	3.7
Affected vs. Asymptomatic
Ligase activity, forming phosphoric ester bonds	48.1
Double-stranded RNA binding	18.2
Domain: helix-loop-helix motif	7.9
Basic helix-loop-helix dimerisation region (bHLH)	7.4
Helix-loop-helix DNA-binding	6.9
HLH	6.8
DNA-binding region: basic motif	6.3
Immune response	6.0
Membrane lipid biosynthetic process	5.2
Compositionally biased region: pro-rich	2.8
Immune response	2.8
RNA-binding	2.5
Immune system process	2.4
RNA binding	2.3
Defense response	2.2

Expression of 4q35 Genes.

To check if the genes previously reported to be dysregulated from the region in cis to 4q35 are specifically dysregulated in our affected-patient population, we compared their expression among our 3 groups of samples. There was no hybridization signal for 2 genes: DUX4 and FRG2 (Table S5). There were no significant alterations in transcriptional level among affected, asymptomatic, and controls for FRG1, PDLIM3, and ANT1 (see Table S5). However, as reported elsewhere (21), the probes on U133 Plus 2.0 chip for the DUX4 and FRG1 genes do not target the chromosome 4q35 genes specifically. The only significant differences in this set of genes were for ANKRD37 and F11 (see Table S5), which were significantly dysregulated only in affected patients, compared to asymptomatic carriers. Their products could be involved in the pathogenesis of FSHD.

Validation of Microarray Results.

The most interesting dysregulated genes were in asymptomatic muscle biopsies relative to control (see Table 1) and were from chromosome 4q. It was necessary to validate their expression using all possible samples to determine if these genes would be candidates to suppress the FSHD phenotype in asymptomatic carriers. (Please note that RNA samples were not available from all of the original samples used in generating the microarrays for this purpose). The 4 dysregulated genes from chromosome 4q (CXCL9, CXCL11, LOC91431, and STATH) were selected along with 2 other genes, the expression of which appeared more homogeneous among the samples (PRIC285 and LPP), and also CXCL10, which is clustered between CXCL9 and CXCL11 on 4q21, for validation by quantitative RT-PCR. Comparisons of the expression levels of these genes in paired samples from the same family that showed CXCL9, CXCL10, CXCL11, STATH, and LOC91431 are significantly up-regulated in asymptomatic carriers compared to affected individuals (Table 3). However, when other samples from unrelated affected and asymptomatic individuals with different numbers of residual D4Z4 are included, there is no difference in the relative expression of those genes by RT-PCR in affected and asymptomatic groups (Table S6). Interestingly, a graph of these results, considering the fold-changes for affected and asymptomatic as a function of the residual number of D4Z4 repeats (Fig. 1), revealed a correlation between the expression levels of the chemokines and the residual number of D4Z4 in affected patients, but not in asymptomatic carriers. In contrast, negative correlations were found for LOC91431 in asymptomatic carriers (Fig. 2). Given the small sample size available for the validation through RT-PCR, some of the calculated correlations are not significant (P > 0.05).

Table 3.

Expression fold-change in affected and asymptomatic carriers relative to healthy controls

Gene	Fold-change/affected	Fold-change/asymptomatic
CXCL9	14.1 (11.2–17.8)	137.8 (113.5–167.4)
CXCL10	6.4 (5.5–7.4)	186.9 (154.7–225.8)
CXCL11	2.1 (1.7–2.5)	113.8 (102.7–126.2)
STATH	2.4 (2.0–2.8)	225.4 (206.2–246.2)
LOC91431	0.7 (0.6–0.8)	3.0 (2.8–3.3)
PRIC285	0.5 (0.4–0.6)	0.8 (0.7–0.9)
LPP	0.8 (0.7–1.0)	1.0 (0.8–1.1)

Only the samples analyzed on microarrays were considered.

Fig. 1.

Validation of the expression of the chemokines through RT-PCR. The expression levels of CXCL9, CXCL10, and CXCL11 were calculated relative to the mean value from normal controls in the function of residual number of D4Z4. (A) In affected patients, the rank correlations are 1.000* (CXCL9), 1.000* (CXCL10), and 0.800*** (CXCL11); while the line correlations are 0.933** (CXCL9), 0.730*** (CXCL10), and 0.805*** (CXCL11). (B) In asymptomatic carriers, the rank correlations are –0,464*** (CXCL9), –0.564*** (CXCL10), and –0.527*** (CXCL11); while the line correlations are –0.595*** (CXCL9), –0,004*** (CXCL10), and –0.191*** (CXCL11). *, P < 0.0001; **, P = 0.07; ***, P > 0.1.

Fig. 2.

Relative expression of the other genes validated through RT-PCR in function of the number of D4Z4 repeats. The calculated correlation coefficients for the expression level and the number of repeats for STATH (A) in affected patients are 0.8000*** (rank) and 0.2645***(line) and in asymptomatic carriers are –0.7537** (rank) and –0.4417***(line); for LOC91431 (B) in affected patients are 0.2000*** (rank) and 0.6908*** (line) and in asymptomatic carriers are –0.8208** (rank) and –0.8280** (line); for LPP (C) in affected patients are 1.000* (rank) and 0.7503*** (line) and in asymptomatic carriers are –0.6669*** (rank) and –0.7650*** (line); for PRIC285 (D) in affected patients are 0.2000*** (rank) and –0.1361*** (line) and in asymptomatic carriers are 0.5643*** (rank) and 0.4008*** (line). *, P < 0.0001; **, P = 0.08; ***, P > 0.1.

Characterization of Polymorphisms from 4q35.

As it was previously reported that only contraction of D4Z4 repeats in 4qA161 haplotype was found to cause FSHD, while contractions in other common 4q haplotypes are nonpathogenic (18), we tested if the affected and related asymptomatic individuals, sharing the same FSHD contraction, have different haplotypes that could be associated with the expression of the disease. We examined several polymorphisms on 4qter: the subtelomeric variations distal to D4Z4 (A and B), which can be distinguished by the presence of a beta satellite DNA on A-type alleles, the G/C SNP within the most proximal D4Z4 unit, and a simple sequence-length polymorphism (SSLP) located 3.5-kb proximal to D4Z4. The polymorphisms, when compared in affected and related asymptomatic carriers, can determine their correlation with the pathogenicity of the contracted allele. All tested individuals (from families 1 to 5, Table S7) shared the 4qA allele and the same SNP in D4Z4 (G). In addition, the 4qA161 haplotype was carried in all families tested, with the exception of family 4, in which the affected individual has the 4qA161/4qA166 haplotypes, while the asymptomatic carrier has the 4qA159/4qA166 haplotypes (data not shown).

microRNAs in FSHD.

MicroRNAs (miRNAs) are a class of small noncoding RNAs that are increasingly recognized for their ability to regulate gene expression posttranscriptionally. Among other targets, miRNAs in skeletal muscle regulate the expression of transcription factors and signaling mediators important for muscle biology, which in turn can influence proliferation and differentiation during myogenesis (26, 27).

The expression profiles of miRNAs in 10 different groups of muscle disorders, including FSHD, were recently compared (28) and 62 miRNAs that were differentially expressed in FSHD muscle—all up-regulated—were identified. The predicted targets for these differentially expressed miRNAs were compared to the gene products identified here as down-regulated in FSHD patients, compared to healthy controls. MAMI (meta mir:target inference) was used to predict the targets for the dysregulated miRNAs. Several different miRNAs target the same gene (Table S8), such as ATG16L1, EPAS1, and PCDH9, which are down-regulated in biopsy samples from affected patients. These same genes, ATG16L1, EPAS1, and PCDH9, are also down-regulated in affected individuals relative to asymptomatic carriers (see Table S2), which suggests that their down-regulation by miRNAs may be linked to the pathophysiology of FSHD. Perhaps significantly, most of these predicted targets take part in histone acetylation and the synthesis of GPI anchors, both of which are dysregulated in affected patients vs. healthy controls (see Table 2).

Discussion

Although FSHD was among the first of the muscular dystrophies to have its locus mapped, the molecular mechanism leading to the disease is still unclear. The unspecific histological alterations in muscle and the high clinical variability observed among patients do not give any clues about a possible mechanism underlying FSHD. One possible way to understand the pathology would be to examine the differences in the expression of genes in patients with FSHD and in their close relations, who are asymptomatic carriers. The present study is unique in making such a comparison, which minimizes the effects deriving from differences in genetic background, and likely gives more consistent and reliable results with microarray analysis (29, 30). In addition, all muscle biopsies were collected by the same physician and were processed in the same way, keeping variations from technical sources to a minimum. The results support the evidence that the expression of some genes in FSHD, and particular those at chromosome 4q, is abnormal in FSHD patients.

The finding that the genes from the chemokine cluster on 4q21, that are located more than 100 Mbp from D4Z4 repeats, have a similar expression pattern (CXCL9 > CXCL10 > CXCL11) in all tested affected patients through RT-PCR suggests that these 3 chemokines may be under control of the same regulators in FSHD patients, in contrast to asymptomatic carriers. The IFN-γ-induced chemokines CXCL9, CXCL10, and CXCL11 are ligands for CXCR3 receptor, and are thought to play a key role in directing activated T cells and other cell types (such as natural killer cells and macrophages) to sites of inflammation. The up-regulation of these chemokines have not been described in either FSHD or other forms of muscular dystrophy (31–35), suggesting that their expression is unlikely to be linked to inflammatory cell infiltrates. Moreover, the muscle histology from asymptomatic carriers looks similar to the normal controls, not showing a major inflammatory component, while in affected patient muscle it is possible to observe general dystrophic features (Fig. S1). The fact that there was no difference in the expression levels of these chemokines as well as the other genes validated—STATH, LOC91431, LPP, and PRIC285—between affected patients and asymptomatic carriers when additional samples were included, lessens the likelihood that these genes could be suppressing the FSHD phenotype in asymptomatic carriers. On the other hand, our microarray and RT-PCR data suggest that the contraction of repeats in affected patients and asymptomatic carriers might disturb the gene expression in cis. This observation is supported by the finding that the contracted allele is significantly hypomethylated in both affected individuals and asymptomatic carriers (19). These studies did not look at the differential expression of FRG1 or DUX4 in FSHD patient biopsies. There was evidence from the microarray data that indicated that from chromosome 4q35, ANKRD37 and F11 genes are differentially expressed when compared with affected patients and asymptomatic individuals, suggesting that these 2 groups may have some specific differences on 4q35. In the 4 families tested, it appeared that the affected patients and related asymptomatic carriers share the same 4qA161 haplotype. Thus, these polymorphisms by themselves cannot explain the pathogenicity linked to the contraction of D4Z4 repeats, and other factors, perhaps acting in cis, are likely to be important in the disease mechanism. Further studies concerning the differences between affected patients and asymptomatic carriers are of utmost importance and will certainly help in our knowledge about factors necessary to trigger FSHD.

The data presented here indicated that the biological process that may be specifically impaired in FSHD patients involves the synthesis of GPI anchors. The structure of the GPI anchor is composed of short chains of sugars, specifically mannose and glucosamine, which are assembled in the endoplasmic reticulum and linked to the inositide residues of phosphatidylinositol. After synthesis, the entire glycolipid is transferred to C-terminal regions of proteins posttranslationally, thereby anchoring these proteins to the outer leaflet of the cell membrane, where they tend to associate with lipid rafts (36). GPI anchors also have roles in membrane diffusion, intracellular protein sorting, and signaling (37). A major subclass of rafts is caveolae, which are invaginations of the cell membrane characterized by the abundance of caveolin. Caveolae contain clusters of GPI-anchored proteins, the most well-studied of which is the folate receptor. However, evidence also exists for the presence of other GPI-anchored proteins, such as alkaline phosphatase, Thy-1, and prion PrP(C) in caveolae (38). No caveolae abnormalities in muscle plasma membranes from FSHD patients were found in one study (39). However, alterations in the sarcolemmal reorganization in muscle tissue from FSHD patients have been reported (40). The steps in GPI assembly, and the enzymes that carry them out, are highly conserved. The genes found to be down-regulated in biopsies from FSHD patients presented here, and which take part in this biological process, are involved in the catalytic steps of the GPI biosynthetic pathway. The importance of GPI anchoring in mammals is underscored by the fact that abrogation of GPI biosynthesis results in embryonic lethality (41). Defects in O-glycan biosynthesis are associated with congenital muscular dystrophies; further studies concerning alterations in the GPI anchor and other glycan structures from caveolae and other proteins associated with the sarcolemma in FSHD should be pursued.

The changes in gene expression observed here may also be linked to changes in histone acetylation, which occur only in affected patients and not in asymptomatic carriers. Histone modifications are likely to control the structure and/or function of the chromatin fiber. The pattern of histone modification may correlate with transcriptional repression/activation and also global chromatin dynamics (42). Although previous studies (43, 44) on histone H4 acetylation levels in the proximal region of D4Z4 did not find any difference between affected individuals and healthy controls, the specific state of the chromatin at the D4Z4 repeats of chromosome 4q35 has not yet been studied in affected patients because of the lack of specific primers, but such studies could yield interesting findings.

Our study suggests that changes in the expression of miRNAs may also be pathogenic in FSHD. Our results are consistent with another study that profiled the expression of miRNAs in FSHD, and suggest that the down-regulation of mRNA levels encoding ATG16L1, EPAS1, and PCDH9 in affected individuals, but not in asymptomatic carriers or healthy controls, may be mediated at least in part by increases in the levels of miRNA that target these gene products.

In summary, there appears to be a profound difference in the transcriptional regulation between affected patients and asymptomatic carriers. The contraction of D4Z4 repeats seems to affect the transcription of genes at 4q, in both groups, but they suggest that factors in other regions of the genome are also associated with the pathogenesis of FSHD. Nevertheless, perturbations in histone acetylation may affect the chromatin structure specifically in the affected patients, leading to the pathogenic changes associated with the disease. Similarly, differences in the synthesis of GPI-anchored proteins may also be linked to muscle pathology in affected individuals. In addition, our results indicate that the miRNAs play a role in the regulatory network in FSHD, increasing the complexity of the mechanism underlying the disease. The study of expression profiles from asymptomatic carriers is a unique approach that has revealed these new and interesting findings, and in the future should contribute to a better understanding of FSHD.

Materials and Methods

Muscle Biopsies.

Muscle biopsies were taken from related members (affected, asymptomatic carrier, and normal control) belonging to 5 unrelated families (see families 1 to 5 in Table S7). Asymptomatic carriers were considered those older than 30 years old, without any FSHD clinical signal, but who share the deleted fragment in common with their clinically affected relative. Additional biopsies were also collected from single members of additional families (families 6 to 10 in Table S7). The asymptomatic carriers corresponding to number 16, 17, 19, and 20 samples also have at least one related affected who shares the same FSHD allele. The presence of mosaicism had been previously excluded through pulsed field-gel electrophoresis. For the microarray studies, muscle biopsies were taken from the biceps in 3 families and from the deltoid in the remaining 2 families, because the clinically affected patients had a severe atrophy. The same muscle was taken from all individuals within each family to minimize variability. One portion of each biopsy was frozen for molecular studies and another for histological analysis (see Fig. S1). All of the biopsies were taken at the Human Genome Research Center (University of São Paulo, Brazil) after informed consent approved by the institutional review board.

RNA Isolation.

The frozen muscle tissue was ground in dry ice in a mortar, and the total RNA was isolated from muscle tissue using the TRIReagent method (Molecular Research Center, Inc.) following the manufacturer's protocol. For details, see the SI Text.

Microarray Assay.

mRNA microarray was performed with the Human Genome U133 Plus 2.0 chip (Affymetrix), following the procedures for target labeling according to “One-Cycle Eukaryotic Target Labeling” protocol as described in the SI Text.

Microarray Data Analysis.

A probe-by-probe differential fold analysis was performed between any 2 sample populations—control vs. asymptomatic, control vs. affected, asymptomatic vs. affected—using a geometric logarithmic fold method. For details see the SI Text.

Gene Ontology Enrichment Analysis.

We performed a gene ontology (GO) enrichment analysis of the list of probes of interest using DAVID 2008 (http://david.abcc.ncifcrf.gov) data freeze June 2008. These lists were restricted to probes that had an Entrez gene assignment and duplicated/redundant Entrez genes were removed. The background gene set was 19,501 unique Entrez genes assayed on the Affymetrix Human Genome GeneChip Human Genome U133 Plus 2.0 microarray. We identified ontologic categories that had a Fisher exact P-value (EASE score) <0.05 (45). This analysis was performed on the significantly dysregulated genes.

RT-PCR Validation of Microarray Data.

The validation of microarray results was done through quantitative RT-PCR, using TaqMan Gene Expression Assays (Applied Biosystems) as described in the SI Text.

Characterization of 4q35 Alleles.

After isolation of DNA from peripheral blood, the DNA was digested with HindIII and separated by electrophoresis on agarose gels, followed by transfer to Hybond membrane (GE Healthcare) by Southern blotting. The membrane was hybridized with probes that specifically distinguish these alleles, as described in ref. 10. The D4Z4 SNP and the SSLP of all individuals studied in the present work were defined as in ref. 18.