Aberrant neuromuscular junctions and delayed terminal muscle ﬁber maturation in a -dystroglycanopathies

Recent studies have revealed an association between post-translational modiﬁcation of a -dystroglycan ( a -DG) and certain congenital muscular dystrophies known as secondary a -dystroglycanopathies ( a -DGpathies). Fukuyama-type congenital muscular dystrophy (FCMD) is classiﬁed as a secondary a -DGpathy because the responsible gene, fukutin , is a putative glycosyltransferase for a -DG. To investigate the pathophysiology of secondary a -DGpathies, we proﬁled gene expression in skeletal muscle from FCMD patients. cDNA microarray analysis and quantitative real-time polymerase chain reaction showed that expression of developmentally regulated genes, including myosin heavy chain ( MYH ) and myogenic transcription factors ( MRF4 , myogenin and MyoD ), in FCMD muscle ﬁbers is inconsistent with dystrophy and active muscle regeneration, instead more of implicating maturational arrest. FCMD skeletal muscle contained mainly immature type 2C ﬁbers positive for immature-type MYH. These characteristics are distinct from Duchenne muscular dystrophy, suggesting that another mechanism in addition to dystrophy accounts for the FCMD skeletal muscle lesion. Immunohistochemical analysis revealed morphologically aberrant neuromuscular junctions (NMJs) lacking MRF4 co-localization. Hypoglycosylated a -DG indicated a lack of aggregation, and acetylcholine receptor (AChR) clustering was compromised in FCMD and the myodystrophy mouse, another model of secondary a -DGpathy. Electron microscopy showed aberrant NMJs and neural terminals, as well as myotubes with maturational defects. Functional analysis of NMJs of a -DGpathy showed decreased miniature endplate potential and higher sensitivities to d -Tubocurarine, suggesting aberrant or collapsed formation of NMJs. Because a -DG aggregation and subsequent clustering of AChR are crucial for NMJ formation, hypoglycosylation of a -DG results in aberrant NMJ formation and delayed muscle terminal maturation in secondary a -DGpathies. Although severe necrotic degeneration or wasting of skeletal muscle ﬁbers is the main cause of congenital muscular dystrophies, maturational delay of muscle ﬁbers also underlies the etiology of secondary a -DGpathies.


INTRODUCTION
Fukuyama-type congenital muscular dystrophy (FCMD; MIM 253800) is an autosomal recessive muscular dystrophy and the second most common childhood muscular dystrophy in Japan, following Duchenne muscular dystrophy (DMD) (1). Clinical manifestations of FCMD include severe congenital muscular dystrophy from early infancy, cobblestone lissencephaly and eye malformation. We previously isolated the responsible gene for FCMD, termed fukutin (2,3). Recently, it has been postulated that fukutin modulates the glycosylation of a-dystroglycan (a-DG), a major component of the dystrophinglycoprotein complex (4,5). FCMD is classified as one of the congenital muscular dystrophies, such as laminin-a2-deficient congenital muscular dystrophy (MDC1A) (6).
Recently, FCMD has also been classified as a secondary a-DGpathy, as mutations in genes encoding glycosyltransferases result in hypoglycosylated a-DG (7). a-Dystroglycan binds to extracellular matrix proteins such as laminin, agrin and perlecan, which are important in maintaining muscle cell integrity (8). Hypoglycosylated a-DG provokes the posttranslational disruption of dystroglycan -ligand interactions in the skeletal muscle of patients, leading to the severe phenotypes of congenital muscular dystrophies (7). Other glycosyltransferases include POMGnT1 (protein O-mannose b-1, 2-N-acetylglucosaminyltransferase 1), POMT1 and POMT2 (protein O-mannosyltransferases 1 and 2), fukutin-related protein (FKRP) and LARGE; mutations in these genes induce human muscle -eye -brain disease, Walker -Warburg syndrome and congenital muscular dystrophy type 1C/1D, and mouse myodystrophy, respectively (9 -14).
Primary characteristics of the so-called 'muscular dystrophy' such as DMD include necrotic change and active regeneration of muscle fibers. From infancy, DMD patients usually show dystrophic change in skeletal muscle, accompanied by elevation of serum creatine kinase (CK) levels. However, DMD patients usually maintain their gait until early adolescence. In contrast, FCMD patients show severe phenotypic characteristics from very early infancy, and few patients can acquire gait regardless of serum CK levels (1). Skeletal muscle fibers in FCMD are extremely small, irregular in cell size and architecturally disorganized, and extensive fibrosis prevails from the early infantile stage. However, only a small number of muscle fibers show severe necrotic change or active myofibril regeneration, and satellite cells are also fewer than those of DMD (1,15,16). These phenotypic differences promote the hypothesis that another mechanism may also account for the pathophysiology of secondary a-DGpathies.
Although expression profiling of skeletal muscle from patients with DMD, MDC1A and a-sarcoglycanopathy have been described (17 -19), no similar analysis has been reported for FCMD and other secondary a-DGpathies. To investigate the molecular mechanism of FCMD and other secondary a-DGpathies, we profiled gene expression in FCMD skeletal muscle using cDNA microarray and subsequent quantitative real-time polymerase chain reaction (PCR). Here we demonstrate that aberrant neuromuscular junctions (NMJs) and maturational delay of muscle fibers are significant to the mechanism underlying secondary a-DGpathies.

Aberrant muscle regeneration is suggested by gene expression profiling of FCMD skeletal muscle
Gene expression profiling of FCMD skeletal muscle was performed using a custom cDNA microarray. Clustering analysis showed similar overall expression profiles of muscle from four FCMD patients, aged 20 days to 1 year, 6 months (Fig. 1A). This similarity is independent of age and histology of the muscle specimen in our samples.
We analyzed individual genes showing distinct expression patterns in FCMD skeletal muscle compared with normal children or DMD patients. Most genes encoding muscle components were down-regulated in FCMD. Among these, myosin light chain 1, 3 and 4 (myl1, 3 and 4) were up-regulated in DMD skeletal muscle, in contrast with FCMD (Fig. 1B). Expression of the developmentally regulated myosin heavy chains (MYHs), MYH1, MYH2 and MYH7 (slow, adult-type), was down-regulated in FCMD but not in DMD, whereas expression of MYH8 (fast-type) showed no significant change in FCMD compared with DMD or normal controls. Slow-type MYHs (MYH1, MYH2 and MYH7) are present in mature muscle fibers and crucial for sarcomere assembly to maintain muscle integrity, whereas fast-type or developmental MYHs (MYH3, MYH4 and MYH8) are seen in early immature myoblasts or in regenerating fibers. These observations suggest that expression of mature muscle components is suppressed in FCMD skeletal muscle at all ages examined.
With regard to muscle fiber differentiation, myogenic factors including MyoD, myf5 and myogenin (myf4) showed insufficient signal for the analysis. It is noteworthy, however, that MRF4 (myf6) was down-regulated in FCMD. Expression of the alpha-type cholinergic receptor (CHRNA), which is known to be regulated by MyoD and MRF4 (20,21), was much higher in FCMD patients than in normal controls.
We next performed real-time quantitative PCR to further investigate skeletal muscle differentiation. We compared mRNA expression in FCMD muscle with normal or DMD skeletal muscle, as DMD is a good example for active regeneration, in which expression of muscle component and myogenic factor mRNA expression is expected to be up-regulated. Although CHNRA was up-regulated in DMD, as predicted, its expression was even higher in FCMD ( Fig. 2A and B). Among these cholinergic receptor subtypes, gamma-type cholinergic receptor (CHNRG), which is a component of fetal isoforms, was up-regulated, whereas epsilontype cholinergic receptor (CHNRE), which only composes adult isoforms (22), was down-regulated in FCMD (Fig. 2B). MYH slow-type (MYH7) was down-regulated in FCMD, consistent with the microarray analysis, whereas expression of fast-type MYH (MYHIIb) was not altered in FCMD. Interestingly, although MyoD and myogenin were up-regulated in both DMD and FCMD, MRF4 was downregulated in FCMD muscle but up-regulated in DMD ( Fig. 2A and B). MRF4 expression is known to be up-regulated in the late phase of muscle regeneration or differentiation, followed by sequential expression of MyoD, myf5 and myogenin, indicating significant roles in terminal differentiation (20,21). These results suggest that FCMD skeletal muscle undergoes an unbalanced differentiation process.

Final maturation step is retarded in FCMD skeletal muscle
To investigate how differentiation is impaired, we examined histological specimens of FCMD skeletal muscle. Marked interstitial tissues with numerous small, round-shaped immature fibers and some necrotic fibers increased with age were seen in FCMD skeletal muscle specimens. Interstitial tissue is prominent from early infancy and progresses with age ( Fig. 3A -C), and skeletal muscle from an FCMD fetus also shows rich interstitial tissues (Fig. 3E). Although necrotic change in muscle fibers is not so marked as in DMD fibers, DMD muscle shows less marked fibrosis and more mature fibers, despite more active necrotic and regenerating processes (Fig. 3D). Overall, FCMD muscle is reminiscent of fetal muscle; skeletal muscle from a normal fetus appears rich in fibrous tissues and small, round-shaped immature myotubes (Fig. 3F). Muscle fiber type is easily identified by ATPase staining. Normally, type 2C fibers are mainly seen in fetal muscle fibers or in regenerating fibers. However, in ATPase-stained cryospecimens, FCMD muscle showed a significantly higher percentage of undifferentiated type 2C muscle fiber contents relative to DMD or control samples (P , 0.005) ( Fig. 3G and H, Table 1).
Using immunohistochemical analysis, we examined MYH subtypes to confirm the differentiation impairment in FCMD and in myodystrophy mouse (myd ), which is another model of secondary a-DGpathies. In normal muscle from agematched controls, no staining of developmental or neonatal MYH ( Fig. 3I and J) was seen. In contrast, FCMD and myd muscle fibers stained positively for developmental and neonatal MYHs ( Fig. 3M and N). These positive fibers corresponded with those staining positive for fast-type MYHs in a serial section (Fig. 3M -O, arrows). Similar staining patterns were observed in skeletal muscle from an FCMD fetus. It is unlikely that all fibers showing developmental MYH expression are derived from regenerating fibers, as few active regenerating or necrotic fibers are seen in the hematoxylin and eosin (HE) specimen at any ages ( Fig. 3A -C). Similar staining patterns were observed in skeletal muscles from an FCMD fetus and adult myd (data not shown). It is unlikely that all fibers showing developmental MYH expression are derived from regenerating fibers, as few active regenerating or necrotic fibers are seen in the HE specimen ( Fig. 3A -C).
These results induce the possibility that maturation might be slowed or arrested in FCMD and myd skeletal muscles, and possibly this is common in secondary a-DGpathies. It also implies that secondary a-DGpathies have more complex etiology than the so-called 'muscular dystrophy', and that may be partly explained by a maturational defect.

NMJ abnormalities induce maturational delay in secondary a-DGpathies
Microarray analysis showed a reduction in MRF4 expression in FCMD. Using immunochemistry, we further investigated MRF4 expression in FCMD and in myd. Immunoreactivity against MRF4 was reduced dramatically in FCMD muscle fibers (Fig. 4A). In normal skeletal muscles, anti-MRF4 antibody yielded strong signals, which co-localized with the nucleus and NMJs (Fig. 4A, upper columns). MRF4 in FCMD muscles showed weak signals which were not merged with NMJ ( Fig. 4A, lower columns). Similar results were obtained in myd (data not shown). Regarding the fact that MRF4 is required at the time and place of NMJ development during skeletal muscle differentiation (23), these results prompt the hypothesis that the differentiation process of muscle fibers arrests at this point in secondary a-DGpathies.
We next examined the morphology of NMJs in both FCMD and myd by staining acetylcholine receptor (AchR) in NMJs with anti-a-bungarotoxin (Fig. 4B). Almost all the NMJs of FCMD and myd showed sparse, weak staining (Fig. 4B, lower columns), in contrast with the dense pattern in normal skeletal muscle (Fig. 4B, upper columns). In normal skeletal muscles, the borders of positive signals were characteristically flared because of multiple layers of synaptic folds, whereas borders in FCMD and myd appear smooth and simple, and synaptic folds-particularly secondary folds-were seldom observed. This signal pattern reflects deteriorated or nondeteriorated cluster of AChR on NMJs in secondary a-DGpathies.
Electron microscopic examination of these secondary a-DGpathies revealed aberrant NMJ lesions with abnormal neural endings. NMJs with fewer synaptic folds and secondary clefts were seen in all NMJs of FCMD and myd ( Fig. 5A -F). In addition, the muscle fibers showed characteristics of immaturity, consistent with our hypothesis that the myotubes are maturationally arrested ( Fig. 5G and H). These fibers are distinct from the active regenerating
We performed functional analysis of the morphologically aberrant NMJs in secondary a-DGpathy by measuring miniature endplate potential (MEPP) and endplate potential (EPP) of myd mice ( Table 2). The amplitudes of MEPP were markedly lower in myd mice than in normal littermates (P , 0.005). In contrast, quantal content of EPP was increased in myd (P , 0.005). The reduction of MEPP amplitude could be compensated by the increased quantal content, and the safety margin of neuromuscular transmission is considered to be maintained in myd mice. The number of endplates recorded in myd mice was much fewer than in normal littermates. However, the amount of d-Tubocurarine that can inhibit the muscle contraction induced by the EPP was distinctively low for myd muscle relative to that of normal littermate ( Table 2). These findings implicate, combined with the morphological observation, that most of the endplates in myd are not adequately innervated, but a small number of NMJs functionally compensate the low MEPP amplitude to maintain the safety margin of neuromuscular transmission.

Hypoglycosylation of a-DG as the etiology of non-clustering AChR in NMJs
We performed immunostaining to examine core a-DG in muscle fibers. In normal skeletal muscles, a-DG localized to the NMJ and sarcoplasmic membrane (Fig. 6A). In contrast, FCMD and myd showed substantial a-DG on the sarcoplasmic membrane, but only weak signals were observed in thin NMJs, indicating a failure of a-DG aggregation (Fig. 6A, normal NMJs, arrows; FCMD and myd, arrowheads). We also examined staining of glycosylated a-DG (IIH6). As expected, we saw no signal on NMJs or on the sarcoplasmic membrane in FCMD and myd (data not shown), implying that glycosylation Figure 3. HE and ATPase stains of biopsied FCMD skeletal muscle, used for microarray analysis. Each specimen shows marked fibrosis with numerous small immature muscle fibers, which is seen from early infancy (A, 20 days; B, 7 months; C, 1-year 6 months), and progresses with age. DMD muscle (D, 5 years) shows less marked fibrosis and less frequent immature fibers despite more active necrotic and regenerating processes. Note that the pathological findings of FCMD skeletal muscles are similar to those of fetal skeletal muscles (E, FCMD fetus, 19 weeks; F, normal fetus, 21 weeks). Also note many undifferentiated immature type 2C fibers stained darkly for ATPase under both alkaline (pH 10.4) (G) and acid (pH 4.6) (H) pre-incubations. Immunostaining for MYH subtypes shows positive staining of developmental and neonatal MYH and decreased staining of slow-type MYH, which are distinct from normal muscles (normal child muscles, 1 year, I-L; FCMD, 1 year, M-P in sequential cryosections). Scale bars ¼ 100 mm. is crucial for a-DG aggregation and also for the subsequent clustering of AChR in NMJs. a-DG is expressed on both the muscle peripheral membrane and the peripheral nerve terminal at NMJs (24). Thus, we examined whether a pre-synaptic or post-synaptic lesion contributes to aberrant NMJ formation.
Staining for synaptophysin at the pre-synaptic region or for fasciculin at the synaptic gap showed abnormal patterns similar to that of a-bungarotoxin (Fig. 6B). These observations indicate that NMJ abnormalities in secondary a-DGpathies may arise not only at the post-synaptic muscle peripheral membrane, but also by pre-synaptic hypoglycosylated a-DG. Utrophin and dystrophin are expressed abundantly in preand post-synaptic regions of mature NMJs and suggested to play an important role for synaptic maturation and the maintenance of NMJs (25). To analyze aberrations of the distribution of utrophin and dystrophin, we performed immunostaining for utrophin and dystrophin in NMJs. Examination under confocal microscopy allowed a precise view of both proteins on the sarcoplasmic membrane. In NMJs from a normal sample, utrophin strongly stains exclusively at fine primary and secondary synaptic folds, tangled with dystrophin staining just beneath the muscle peripheral membrane (Fig. 6C, left column; Fig. 6D, upper columns). In contrast, NMJs from secondary a-DGpathies show thinner, fold-less and weak signals for both utrophin and dystrophin (Fig. 6C, right column; Fig. 6D, middle and lower (green), a-bungarotoxin staining of AChR on NMJ (red) and DAPI-stained nuclei (blue) in normal and FCMD skeletal muscles. In normal muscle, NMJs stain strongly, merging with MRF4 staining and DAPI (arrows, upper columns). In FCMD, the staining pattern of MRF4 in the nucleus of muscle fibers is markedly decreased and no merging stain with NMJ is seen (arrowhead). (B) Compared with normal AChR on NMJs (red) stained by a-BTX (upper columns), scattered, fold-less staining pattern is present in both FCMD and myd (lower columns). Scale bars ¼ 5 mm.

1284
Human Molecular Genetics, 2006, Vol. 15, No. 8 columns). The utrophin signal was exclusively seen on NMJs, but weakly seen on muscle sarcoplasmic membrane in FCMD (Fig. 6C, right column), which is usually seen in regenerating muscle fibers (26). NMJs from fetal wild-type or myd showed similar staining patterns as expected (data not shown), suggesting that this staining pattern is indicative of immature muscle fibers. This observation is consistent with our hypothesis that NMJ formation in secondary a-DGpathies is developmentally arrested during myotube maturation in the fetal phase.

DISCUSSION
FCMD has long been classified as 'muscular dystrophy', although the clinical characteristics differ from those of DMD. Muscular dystrophy is defined generally by necrotic change and active regeneration of muscle fibers. However, FCMD muscle in infantile stage seems more likely to have additional features, implying a more complex pathogenesis for FCMD. Indeed, our microarray analysis showed that general expression profiling clusters FCMD and DMD distinctively. Expression of mature muscle components was surprisingly low in FCMD, indicating less active regeneration of muscle fibers. We also saw similar expression profiles among all FCMD patients in our samples, indicating that FCMD is a chronic rather than progressive disorder, at least at the infantile period. We confirmed maturational delay and aberrant NMJs in FCMD skeletal muscle fibers by expression profiling, morphological and histochemical analysis, and electrophysiological examination. These findings are common to secondary a-DGpathies but are not seen in DMD. In this study, we demonstrate that the etiology of secondary a-DGpathy skeletal muscle abnormalities may stem from maturational arrest caused by aberrant NMJs in addition to dystrophy. MDC1A, clinical characteristics of which are similar to FCMD, is described with an initial phase of necrosis and regeneration in the early steps of the disease (6). Although aberrant NMJ is also seen in MDC1A (27), the muscle fragility due to defect in the component of basement membrane mainly affects the phenotype and causes necrosis and regeneration.
A considerable body of evidence indicates that muscle differentiation ceases at NMJ formation in FCMD. First, immature type 2C fibers are predominant in FCMD (Table 1). At the initiation of muscle differentiation, satellite cells proliferate to become myoblasts, fuse to organize myotubes. These early myotubes contain type 2C fibers. Following NMJ formation, immature fibers are induced to differentiate further into mature muscle fibers such as type 1, 2A and 2B. Second, down-regulated expression of matured MYHs in FCMD also suggests arrest at this stage. Following completion of NMJ formation, embryonic, neonatal-type MYHs in immature myofibers are replaced by adult-type, slow MYHs, which are induced by extracellular matrix (ECM) components, growth factors or programmed cell differentiation (28).
Third, MRF4, which is postulated to be induced by AChR clustering in NMJ formation, is down-regulated in FCMD. Normally, during the early phase of muscle development, a series of myogenic regulatory genes such as myf5, MyoD and myogenin are sequentially expressed, followed by MRF4 up-regulation just after NMJ formation. AChRs bind to myotubes, associate with specific factors and trigger a signal to induce expression of myogenic factors including MRF4, which is suggested to play a crucial role in muscle terminal maturation and maintenance (20,21,29). Therefore, MRF4 is distinct from MyoD, myogenin and myf5 in that it is expressed mainly in matured myotubes and myofibers. MRF4 has been reported to co-localize with AChR in NMJs and to function in terminal muscle maturation and fiber maintenance (30). It is reasonable to assume that MRF4 function is required at the time and place of NMJ development during skeletal muscle differentiation (23). These results prompt the hypothesis that the differentiation process of muscle fibers arrests at this point in FCMD and myd.
Fourth, the fact that high CHRNG and low CHRNE expression in FCMD muscle relative to that in age-matched normal control or DMD muscle is striking, although the age of the DMD patient was slightly higher than that of the FCMD patients. It clearly indicates that most of the AChR in FCMD muscle is fetal type. It also supports our hypothesis that skeletal muscle in secondary a-DGpathy is immature and that delay of differentiation might be involved with maturation defect of NMJs.
What causes aberrant NMJs in secondary a-DGpathies? Normally, NMJs are built by a dense tangle of sarcoplasmic membrane and neural endings through a layer of basement membrane. It is possible that connections between the neural terminal and glycosylated a-DG, made through a layer of basement membrane and mediated by molecules such as laminin or agrin, are important to normal NMJ formation and subsequent muscle differentiation (31,32). MDC1A, clinical characteristics of which are similar to FCMD (6), also shows aberrant NMJs with fewer synaptic folds (27). In contrast, Musk or rapsyn deficiency does not resemble severe muscular dystrophy in spite of the abnormal NMJ formation. The interaction of laminin and a-DG is thought to play an essential role in the transition of AChR microaggregates into macroaggregates at the developing NMJ, followed by the concentration of a-DG on NMJs (32,33). These facts lend support to our hypothesis that the laminin/a-DG interaction fulfills a pivotal function in normal NMJ formation.
Alternatively, it is also possible that attachment of neural terminals to NMJs is affected pre-synaptically in secondary a-DGpathies. Abnormalities in the pre-synaptic peripheral nerve would affect the neural endings of NMJs, leading to The amount of d-Tubocurarine that can inhibit the muscle contraction induced by the EPP. deficient differentiation signal transduction to FCMD muscle fibers and arrested post-synaptic muscle differentiation. It has been suggested that a-DG in the central or peripheral nervous system is hypoglycosylated in secondary a-DGpathies (34). O-Mannose-type glycoprotein is suggested to contribute to the stability and maintenance of muscle cell membrane, synaptic formation and myelination of peripheral nerves, although the precise mechanism of a-DG activity in peripheral nerve tissue is unclear. The dystrophin-glycoprotein complex on Schwann cells is also thought to be important in peripheral myelinogenesis, regeneration, differentiation, apoptosis and polarity of skeletal muscle cells (35). Although Ishii et al. (36) reported that electron microscopy of Schwann cells on FCMD muscle revealed no pathologic findings, neural transmission may be developmentally impaired and collapsed as a result of hypoglycosylated a-DG. Naturally, our data do not rule out the other possibilities for immaturity of the skeletal muscles of a-DGpathies. Although the expression profile and histochemical data indicate that skeletal muscles are in a persistent undifferentiated state, it is possible that the regeneration process fulfills a crucial function in the phenomenon observed in skeletal muscles of a-DGpathies. It is also possible that most of the muscle fibers are under denervation status, because motor neurons are unable to maintain strong attachments to myofibers, leading to a constant stimulation of denervation signal pathways. Although we demonstrated substantial evidences for the aberrant NMJ, the pathogenesis of skeletal muscles in a-DGpathies is likely to be a combination of these problems and of multifactorial origin.
Taken together, these findings show that muscle fibers in secondary a-DGpathies are developmentally arrested more to 'dystrophic', perhaps because hypoglycosylated a-DG precludes proper aggregation on NMJs, preventing AChR clustering. These defects may disrupt terminal muscle maturation, which is induced after innervation of neurons on the muscle peripheral membrane via a basement membrane layer. Dystrophic changes traditionally thought to underlie 'muscular dystrophy' are caused by attenuated physical connections between a-DG and the muscle basement membrane. We propose that the muscle lesion in secondary a-DGpathies is caused by complex pathogenesis, not only by dystrophic change but more importantly, maturational arrest resulting from chronically delayed terminal muscle fiber maturation and NMJ deficiency. To date, no clinical approaches to secondary a-DGpathies exist. These findings open a possible

1286
Human Molecular Genetics, 2006, Vol. 15, No. 8 avenue for treating muscle tissues in these congenital muscular dystrophies, via targeted induction of NMJ maturation in muscle fibers.

Samples
All clinical materials were collected for diagnostic purposes. Four muscle specimens (biceps branchii) from FCMD patients (ages: 20 days, 7 months, 1-year 4 months, and 1-year 6 months) were used in the analysis. Genetic screening identified a homozygous retrotransposal insertion into the 3 0 untranslated region of fukutin in all FCMD patients (3). For the non-dystrophic muscle controls, muscle RNAs from two children (ages: 1 year) was used. These patients were selected based on normal laboratory findings, normal plasma CK levels and no histopathological evidence for muscular dystrophy. We also obtained myodystrophy (Large myd ) mice and control littermates (Large myd/þ or Large þ/þ ), aged 3 and 6 months, by mating heterozygous pairs provided by Jackson Laboratories.

RNA isolation and expression profiling
Generation of cDNA microarrays containing skeletal muscle transcripts has been reported previously (19). Using similar methods, we constructed a new cDNA chip containing 5600 genes expressed in skeletal muscle. RNA isolation, hybridization and detection methods also have been reported previously. Microarray experiments were carried out using a competitive hybridization method with two labeled targets: one for muscle RNAs from FCMD patients or normal children, and another for pooled muscle RNAs (Origene), which served as a template control for per-chip normalization. Each analysis was conducted at least twice. The hybridization intensities of each spot and the background intensities were calculated using a ScanArray 5000 microarray scanner with Quant Array software (Perkin-Elmer Life Science).

Microarray data analysis
Analysis of microarray data was performed using Genespring version 6.1 (Silicon Genetics) software. Data used for further analysis were calculated using a previously reported method (19). To avoid 'false-positive' signals, we excluded genes from the analysis for which average normal expression level constraints are under 500. We sorted 1790 genes from a total of 5600 for further analysis.

Quantitative real-time PCR
Two patients with FCMD (ages: 10 months and 1 year), two patients with DMD (ages: 1 and 7 years) and two normal children (ages: 1 and 2 years) were used for quantitative RT -PCR using skeletal muscle RNAs. Single-strand cDNA was produced with random primers, and quantitative real-time RT -PCR using SYBR-green was performed using the ABI Prism 7900 sequence detection system (Applied Biosynthesis). Data analysis was performed in duplicate experiments.
Statistical significance was evaluated using Student's t-test, and P , 0.05 was considered significant. The primers used for the experiments are shown in Supplementary Material, Table S1. Gapdh was used as an internal control.

Imaging analysis
HE staining and ATPase staining were performed on cryosections. ATPase staining was performed at pH 9.4-10.6 and 4.2 -4.6. For ATPase staining, average data from 10 DMD patients (ages: 3 -9) and 10 normal control cases (ages: around 1 year) were selected. Scion Image Beta 3b (Scion Corporation) was used for estimating the content of type 2C fibers, muscle fibers, adipose tissues and interstitial tissues ( Table 1). Statistical analysis was performed using Student's t-test.

Electron microscopy
Muscle specimens were obtained from four patients diagnosed as FCMD (ages: 6 months to 4 years old) and from two normal children (ages: 1 year). Five NMJs were found in three FCMD patients' skeletal muscle and four normal NMJs were examined in two normal children. Intercostal muscles and soleus muscles were dissected from myodystrophy and control mice, and cut into 1-mm thick cubes. Four NMJs were examined in myd and normal controls, respectively. Samples were fixed in 2% glutaraldehyde and embedded in epoxy resin as described previously (36). Ultrathin (50 -90-nm thick) sections were cut on an Ultracut S ultramicrotome (Reichert).

Electrophysiologic examination
Diaphragms with its motor nerve were dissected and used for the conventional intracellular microelectrode study (37). MEPPs, EPPs and resting membrane potentials (RMPs) were recorded. For EPP recording, the phrenic nerve was stimulated using a suction electrode at 0.5 Hz. d-Tubocurarine chloride (Curaren, Sigma) was used at a concentration sufficient to inhibit muscle contraction. The potentials were corrected for non-linear summation and the last 64 responses in a train of 114 were saved for later analysis. The quantal content m was calculated by the variance method. MEPP and EPP amplitudes were corrected to a standard RMP of 280 mV. To correct MEPP amplitude by the fiber diameter, the geometric mean of the shortest and longest diameters of muscle fibers was determined in 30 randomly selected muscle fibers in cryostat sections.