Mutations in the satellite cell gene MEGF10 cause a recessive congenital myopathy with minicores
- First Online:
- Cite this article as:
- Boyden, S.E., Mahoney, L.J., Kawahara, G. et al. Neurogenetics (2012) 13: 115. doi:10.1007/s10048-012-0315-z
- 905 Downloads
We ascertained a nuclear family in which three of four siblings were affected with an unclassified autosomal recessive myopathy characterized by severe weakness, respiratory impairment, scoliosis, joint contractures, and an unusual combination of dystrophic and myopathic features on muscle biopsy. Whole genome sequence from one affected subject was filtered using linkage data and variant databases. A single gene, MEGF10, contained nonsynonymous mutations that co-segregated with the phenotype. Affected subjects were compound heterozygous for missense mutations c.976T > C (p.C326R) and c.2320T > C (p.C774R). Screening the MEGF10 open reading frame in 190 patients with genetically unexplained myopathies revealed a heterozygous mutation, c.211C > T (p.R71W), in one additional subject with a similar clinical and histological presentation as the discovery family. All three mutations were absent from at least 645 genotyped unaffected control subjects. MEGF10 contains 17 atypical epidermal growth factor-like domains, each of which contains eight cysteine residues that likely form disulfide bonds. Both the p.C326R and p.C774R mutations alter one of these residues, which are completely conserved in vertebrates. Previous work showed that murine Megf10 is required for preserving the undifferentiated, proliferative potential of satellite cells, myogenic precursors that regenerate skeletal muscle in response to injury or disease. Here, knockdown of megf10 in zebrafish by four different morpholinos resulted in abnormal phenotypes including unhatched eggs, curved tails, impaired motility, and disorganized muscle tissue, corroborating the pathogenicity of the human mutations. Our data establish the importance of MEGF10 in human skeletal muscle and suggest satellite cell dysfunction as a novel myopathic mechanism.
KeywordsMEGF10Whole genome sequencingLinkage analysisCongenital myopathySatellite cellsCleft palate
Congenital myopathies result from a variety of genetic defects  that manifest histologically as a disturbance in the structure or organization of myofibers, evident on biopsy by abnormalities such as minicores, central cores, central nuclei, ragged red fibers, nemaline rods, or rimmed vacuoles . By contrast, muscular dystrophies are a class of myopathy in which normally developed muscle tissue undergoes destructive cycles of degeneration and regeneration, leading to distinct pathological hallmarks, including replacement of muscle tissue with fat and infiltration of endomysial connective tissue . These features are usually absent from congenital myopathies, and likewise, minicores and other myopathic disruptions are typically absent from dystrophic muscle. Multiminicore disease (MmD) is an autosomal recessive congenital myopathy characterized by predominantly axial and proximal muscle weakness, frequently in conjunction with progressive scoliosis and respiratory difficulties. The defining histological feature is the presence in a large proportion of muscle fibers of minicores, short regions of sarcomeric disorganization with few or no mitochondria . Mutations in the genes SEPN1 and RYR1 explain many but not all cases of MmD [5–8].
The discovery of the genetic basis for such Mendelian diseases has been facilitated by efficient genomewide sequencing methods, notably exome sequencing  and whole genome sequencing . Although selective sequencing of exons is currently cost-effective, sequencing of the entire genome offers the opportunity to identify non-exonic variants, extract higher resolution information on copy number and structural mutations, and potentially provide superior coverage of the exome than targeted capture methods achieve. Regardless of the choice of technology, either multiple probands with a genetically homogeneous disease  or linkage data  are generally required to filter variant lists sufficiently to implicate specific pathogenic mutations. Small families previously considered inaccessible to conventional genetic approaches can now be studied through a combination of linkage analysis and genomewide sequencing.
Materials and methods
Written informed consent was obtained for all subjects for participation in this study, which was approved by the Institutional Review Board of Children's Hospital Boston (CHB). Zebrafish experiments were conducted in compliance with the guidelines of the CHB Institutional Animal Care and Use Committee. Saliva or blood samples were collected and genomic DNA was isolated. Muscle biopsy slides were reviewed, and for subject 105-1, total RNA was extracted from muscle tissue. All six members of family 1030 were genotyped at 10,204 single nucleotide polymorphisms (SNPs) using the GeneChip Human Mapping 10K 2.0 Xba Array (Affymetrix). Genomewide multipoint parametric linkage scans were performed using MERLIN v1.1.2  with a recessive model, as described previously . Whole genome sequencing and preliminary analysis (initial mappings of reads, local de novo assembly, and variant calling) were performed by Complete Genomics, Inc. using a sequencing-by-ligation method with unchained reads . Polymerase chain reaction (PCR) and Sanger sequencing were performed according to standard protocols. Mutations were genotyped in control subjects using Custom TaqMan SNP Genotyping Assays (Applied Biosystems).
Four custom morpholinos (Gene Tools, LLC) were designed against the intron 4–exon 5 (MO1), intron 5–exon 6 (MO2), exon 8–intron 8 (MO3), and exon 17–intron 17 (MO4) junctions. These morpholinos were intended to induce skipping of the targeted exons, homologous to human exons 7, 8, 10, and 19. MO1 and MO2 were designed against the 3′ splice acceptor side of the intron, and a full exonic skip would induce a frameshift and premature truncation of megf10, whereas MO3 and MO4 were designed against the 5′ splice donor side of the intron, and a full exonic skip would induce an in-frame deletion of the exons homologous to those mutated in family 1030. Zebrafish injections, birefringence assays, histochemistry, and immunohistochemistry were performed as described previously [16,17]. For electron microscopy (EM), ultrathin sections were stained with uranyl acetate and lead citrate and were observed with an FEI Tecnai Spirit at 80 kV.
Subject 105-1 was a female Caucasian with distant Portuguese ancestry, diagnosed with MmD. No one else in her family was affected. At several weeks of age, she presented with hypotonia, initially manifesting as a weak neck and floppy head. A markedly high arched palate and failure to thrive as a result of poor feeding were noted, as were blue-tinged sclera and hyperflexible joints. She had frequent acute episodes of pulmonary infection as a child but reported no chronic breathing problems and breathed unassisted at night. Serum CK level at age 10 years was mildly elevated at 302 U/L (upper limit of reference range 105 U/L), and mild scoliosis was reported at the same time, which progressed to 20° by age 12 years. Throughout childhood, she continued to exhibit a mild, non-progressive, primarily proximal weakness (including facial weakness), as well as deficits in motor performance. Muscle biopsy showed marked fiber size variation, internalized nuclei, and slightly increased endomysial connective tissue, as well as prominent patches of decreased NADH activity within myofibers, consistent with moth-eaten fibers or minicores (Online Resource 2). Electron microscopy showed focal areas of Z-band disarray devoid of mitochondria and extending across a few sarcomeres in the longitudinal axis, confirming the presence of minicores.
Family 1030 produced positive LOD scores at sixteen genomic loci, ten of which approached the theoretical maximum LOD score of 1.329, and which collectively contained over 1,000 candidate genes (Online Resource 3). Both SEPN1 and RYR1 were excluded, as were ACTA1 and TPM3, in which mutations cause congenital fiber type disproportion [18,19], a myopathy clinically similar to MmD. The linkage intervals included the myopathy genes MYOT and CRYAB , but neither of these contained coding or splice site mutations by Sanger sequencing.
Whole genome sequencing and mutation analysis
Whole genome sequencing of one affected subject from family 1030 yielded a mean coverage depth of 74-fold, with 95.9% of the genome fully called (Online Resource 4). We screened all 16 linked regions for genes containing either homozygous or compound heterozygous variants that altered the amino acid sequence or consensus splice site sequence of an encoded protein. Four linked genes contained putative compound heterozygous, nonsynonymous mutations absent from dbSNP v131. Sanger sequence analysis of three called variants in ZNF99 showed that all three were homozygous for the reference sequence in all family members; ZNF99 is a highly repetitive gene and the variant calls may have been due to misaligned reads. Mutations in LOC100128682 and LAMA5 were likewise excluded, as they failed to segregate with the phenotype in the entire family. By contrast, the heterozygous missense mutations c.976T > C (p.C326R) and c.2320T > C (p.C774R) in MEGF10, the sole remaining candidate, were confirmed by Sanger sequencing to be in trans in affected subjects and to segregate with the myopathy in all six family members.
Missense mutations in MEGF10 in congenital myopathy patients
Amino acid substitution
Total controls genotyped
Controls with European ancestry
c.211C > T
c.976T > C
c.2320T > C
By Sanger sequencing, we screened the entire open reading frame of MEGF10 in 190 additional patients with unexplained myopathies, as well as selected exons in another 82 patients. One subject was identified with only a single demonstrable heterozygous mutation; subject 105-1 carried a novel c.211C > T (p.R71W) mutation that was absent from dbSNP v135, 1,000 Genomes data, ESP data, 56 internal genomes, and 699 genotyped unaffected control subjects. The p.R71 residue is conserved in 39 of 41 species (two species of fish having different hydrophilic residues, in contrast to the hydrophobic mutant residue tryptophan). The c.211C > T mutation was heterozygous in an unaffected parent of subject 105-1, so it was not de novo dominant. Sanger sequencing further revealed that the subject had no coding or splice site mutations in SEPN1 or RYR1. To search for a second recessive mutation in MEGF10, we sequenced its entire open-reading frame in reverse-transcriptase PCR (RT-PCR) amplicons derived from skeletal muscle RNA. Expression of both alleles at the c.211C > T site in exon 4 was verified by amplification and sequencing of full-length cDNA. However, we did not identify a heterozygous exonic copy number change, inversion, or other structural mutation. Nevertheless, because the p.R71W mutation is a nonconservative substitution not found in control subjects, there were no nonsynonymous mutations in SEPN1 or RYR1 in this patient, and the clinical and histological characteristics were similar to those of family 1030 patients, subject 105-1 may carry a second mutation in MEGF10 that we were unable to detect.
Knockdown of megf10 in zebrafish
To provide additional evidence for the pathogenicity of MEGF10 mutations, we knocked down expression of full-length zebrafish megf10 using four antisense splice-blocking morpholino oligonucleotides (MOs), referred to as MOs 1, 2, 3, and 4 (Online Resource 5). We compared the morphant phenotypes induced by injection of these MOs to uninjected, wild-type fish and fish injected with a standard control morpholino (COMO). The ability of the four MOs to block splicing was verified by performing RT-PCR on total RNA harvested from treated zebrafish (Online Resource 6). All MOs induced the formation of mis-spliced products to some extent, but they generally produced multiple bands not necessarily corresponding in size to the amplicon predicted to result from full skipping of the targeted exon.
In morphant fish with severely curled tails, particularly those treated with MO2, the organization of muscle tissue was irregular, with the tail somites showing a less distinct V-shaped chevron pattern than control fish (Fig. 3b). This apparent muscle defect was further investigated using birefringence, the ability of highly organized sarcomeres to rotate polarized light, which can be used to assess muscle integrity . Fish treated with 6 ng of MO1 showed a reduction in birefringence, and 6 ng of MO2 induced a more dramatic decrease in birefringence that corresponded in its location to the disruption of somite structure evident by brightfield microscopy. The abnormal tail morphology and reduced birefringence in megf10 morphants closely resembled the phenotypes of morphant fish with deficits in known muscular dystrophy gene orthologs [16,23–25].
MEGF10 encodes multiple epidermal growth factor (EGF)-like domains 10 (MEGF10), which contains 17 atypical EGF-like domains, each with eight cysteine residues . The p.C326R and p.C774R mutations substitute the fourth and sixth cysteines of EGF-like domains six and sixteen, respectively, for arginines. All eight cysteines in other proteins containing eight-cysteine atypical EGF-like domains form disulfide bridges , indicating the mutations in family 1030 likely disrupt MEGF10 protein structure. Similarly, the p.R71W mutation in subject 105-1 is a nonconservative substitution within the emilin domain, which may function in protein–protein interactions . All three mutations were predicted by PolyPhen  to be “probably damaging”, and the two cysteine substitutions were predicted to disrupt annotated disulfide bond sites, as expected.
Nonsense and frameshift mutations in MEGF10 were recently determined to be the cause of a severe congenital myopathy with diaphragmatic weakness, areflexia, and dysphagia. One reported family was compound heterozygous for the c.2320T > C (p.C774R) missense mutation we identified in family 1030 and a frameshift mutation . The phenotypes described therein are generally similar to those of families 1030 and 105, but are more severe. This difference is consistent with the total loss of detectable MEGF10 expression observed in the reported families, compared to the expected misfolding of MEGF10 protein due to disruption of disulfide bonds in family 1030. Accordingly, affected members of family 1030 may retain a small amount of residual MEGF10 function, though genetic background or other factors could also account for clinical heterogeneity between families. Moreover, minicores were not noted in the biopsies from the reported families with null mutations [30,31] but were a prominent feature in our patients with missense mutations. Unstructured minicores in MmD may result from primary defects affecting intracellular redox state or regulation of excitation–contraction coupling at the triad . The molecular mechanisms by which SEPN1 and RYR1 mutations result in minicores in MmD have not been elucidated, and likewise, the relationship between MEGF10 mutations and minicore formation is unclear. Further characterization is required to determine the effect of missense mutations on MEGF10 protein function and sarcomeric structure and to investigate whether MEGF10 shares a functional pathway with SEPN1 or RYR1.
Human MEGF10 or mouse Megf10 protein has been described as an engulfment receptor expressed in the brain [33–35] and involved in phagocytosis of apoptotic neuronal corpses . One of the patients in family 1030 had overt cleft palate at birth, while a second patient had a midline ridge on her palate and a tendency to reflux liquids into her nose. All three affected subjects had hypernasal speech, which can be a sign of submucous cleft palate . Subject 105-1 had a high-arched palate and poor feeding as an infant. Of seven patients in four reported families with congenital myopathy and mutations in MEGF10, one patient had cleft palate and an additional three patients had high-arched palate . Although these clinical features are relatively nonspecific, occurring commonly in patients with myopathic facies, the expression of mouse Megf10 in brain [35,38] is consistent with a role in craniofacial development. In addition, Megf10 interacts with Ap2m1, one of four subunits of the assembly protein complex 2 (AP2) . Ablation of Ap2b1, which encodes another of the AP2 subunits, causes cleft palate in mouse . Furthermore, Megf10 in muscle is only expressed in cells that also express Pax7 , and Pax7−/− null mice exhibit facial malformations of the maxilla and nose due to a defect in cephalic neural crest development . Lastly, some morphant zebrafish treated with MO1 or MO2 showed small heads relative to control fish, further suggesting a possible role for megf10 in head development. Collectively, these data indicate that MEGF10 may influence palate formation or facial midline determination in humans, but more research is required to test this hypothesis.
In addition to brain, mouse Megf10 is expressed in muscle satellite cells [38,41], a population of committed, myogenic precursors that specifically express the transcription factor Pax7 and contribute to the regeneration of adult muscle in response to exercise, disease, or injury [42,43]. The importance of satellite cell function in mouse muscle maintenance has been demonstrated; Pax7−/− mice have a reduced number of satellite cells as embryos and very few or none as adults, and muscle regeneration is impaired in these mice . Recent studies further confirmed that Pax7+ satellite cells are completely required for muscle regeneration [45–47], suggesting that in humans, compromised satellite cell function could result in a myopathic phenotype. Expression of Megf10 in skeletal muscle is restricted to satellite cells and is low in resting muscle but increases dramatically in regenerating muscle upon activation of satellite cells by treatment with cardiotoxin . Megf10 regulates the differentiation program in skeletal myoblasts, as overexpression of Megf10 in C2C12 myoblasts increases proliferation and decreases differentiation, whereas knockdown of Megf10 in primary myoblasts and individual myofibers causes precocious differentiation and a reduction of activated satellite cells .
Consistent with the role of Megf10 in mouse myoblasts, our data establish that human MEGF10 is required for normal muscle function. Due to presumed disruption of disulfide bonds, the p.C326R and p.C774R mutations in family 1030 likely result in a loss of MEGF10 function, in which case we hypothesize that excessive differentiation of satellite cells could deplete the population of self-renewing cells available to regenerate damaged muscle fibers. Such impairment could potentially explain the dystrophic features observed in patient biopsy tissue at ages 8 and 9. Notably, a recent study found that Sepn1−/− null mice showed a decrease in the baseline number of satellite cells in adult skeletal muscle, as well as imperfect muscle regeneration in response to a first injury and an inability to regenerate muscle following a second injury. MmD patients with SEPN1 deficiency also had a reduction in their satellite cell population . These data suggest that both MmD and the myopathy described here, which share some clinical and histological characteristics, may arise through a common pathogenic mechanism of satellite cell dysfunction.
The authors thank the patients and their families for their participation in this study, as well as Hal Schneider, Laura Moody, Susan Kim, Sachiko Kajino, Kanako Goto, and Yukiko Hayashi for technical assistance, Michael Lawlor and Pankaj Agrawal for helpful discussions, Fedik Rahimov for critical reading of the manuscript, and Timothy Yu and Christopher Walsh for contribution of control genome sequence data. This work was supported by NIH K08 NS048180 (PBK), the Genise Goldenson Fund (PBK), a Children's Hospital Boston Pilot Grant (PBK), Muscular Dystrophy Association Research Grants 186796 (PBK) and 201302 (AHB), the Bernard F. and Alva B. Gimbel Foundation (LMK), NIH R01 AR044345 (AHB), the Lee and Penny Anderson Family Foundation (AHB), and the Professor-Dr.-Adolf-Schmidtmann-Stiftung (FD). Microarray genotyping and Sanger DNA sequencing experiments were performed in the Molecular Genetics Core Facility at Children's Hospital Boston, supported by NIH P30 HD18655 through the Intellectual and Developmental Disabilities Research Center and NIH P50 NS40828 through the Neuromuscular Disease Project. Electron microscopy was performed at the Harvard Medical School EM Core Facility with the assistance of Maria Ericsson, Louise Trakimas, and Elizabeth Benecchi.
Conflict of interest
The authors declare that they have no conflict of interest.
This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.