|Home | About | Journals | Submit | Contact Us | Français|
To identify the causative gene in an autosomal dominant limb-girdle muscular dystrophy (LGMD) with skeletal muscle vacuoles.
Exome sequencing was used to identify candidate mutations in the studied pedigree. Genome-wide linkage was then used to narrow the list of candidates to a single disease-associated mutation. Additional pedigrees with dominant or sporadic myopathy were screened for mutations in the same gene (DNAJB6) using exome sequencing. Skeletal muscle from affected patients was evaluated with histochemistry and immunohistochemical stains for dystrophy-related proteins, SMI-31, TDP43, and DNAJB6.
Exome analysis in three affected individuals from a family with dominant limb-girdle muscular dystrophy and vacuolar pathology identified novel candidate mutations in 22 genes. Linkage analysis excluded all variants except a Phe93Leu mutation in the G/F domain of the DNAJB6 gene, which resides within the LGMD 1E locus at 7q36. Analysis of exome sequencing data from other pedigrees with dominant myopathy identified a second G/F domain mutation (Pro96Arg) in DNAJB6. Affected muscle showed mild dystrophic changes, vacuoles, and abnormal aggregation of proteins, including TDP-43 and DNAJB6 itself.
Mutations within the G/F domain of DNAJB6 are a novel cause of dominantly-inherited myopathy. DNAJB6 is a member of the HSP40/DNAJ family of molecular co-chaperones tasked with protecting client proteins from irreversible aggregation during protein synthesis or during times of cellular stress. The abnormal accumulation of several proteins in patient muscle, including DNAJB6 itself, suggest that DNAJB6 function is compromised by the identified G/F domain mutations.
The autosomal dominant limb-girdle muscular dystrophies (LGMD 1) are characterized by post-natal onset of progressive weakness and muscle atrophy affecting proximal muscles of the upper and lower extremities1. To date, seven loci have been associated with LGMD 12, but additional genetic heterogeneity is likely. Causative genes have been identified for three of the best described syndromes: myotilin in LGMD 1A, lamin A/C in LGMD 1B, and caveolin-3 in LGMD 1C3–5. LGMD 1D (OMIM 602067) has been linked to 6q32 in a single family with significant cardiac involvement6. LGMD 1E (OMIM 603511) shows linkage to 7q36 and is characterized by the onset of proximal limb weakness in mid-adulthood and vacuoles on skeletal muscle pathology7, 8. The LGMD 1F locus at 7q32 is close to, but not overlapping, the LGMD 1E locus7, 9. With the exception of age of onset and possible anticipation, LGMD 1F is clinically and pathologically similar to LGMD 1E 10. The LGMD 1G locus resides on 4q21, and was described in a family with limb weakness accompanied by a striking limitation of finger and toe flexion11. Lastly, LGMD 1H is associated with 3p22 and characterized by mitochondrial abnormalities on biopsy, including ragged red fibers12.
Although clinical features or muscle biopsy findings can point to one LGMD 1 over another (e.g. dysarthria in LGMD 1A, cardiac conduction defects in LGMD 1B, muscle rippling in LGMD 1C), there is enough phenotypic overlap that achieving a specific genetic diagnosis can be difficult. If no mutation is found with candidate gene sequencing, linkage analysis can identify known or novel loci in larger families, and lead to candidate gene sequencing in the interval. The recent advance of whole-exome sequencing allows for an alternative approach, beginning with the simultaneous identification of mutations across the entire coding region of the human genome13. Bioinformatic filtering of the identified mutations, including the integration of linkage data, pathogenicity predictions, gene expression patterns, or protein function, can be helpful in distinguishing between likely disease-causing mutations and probable polymorphisms14.
We applied whole-exome sequencing to a small family resembling LGMD 1E in age of onset, pattern of weakness, and vacuolar pathology. Bioinformatic analysis of the variants integrated with linkage data identified a causative mutation in DNAJB6, a member of the HSP40 family of chaperone proteins. Exome sequencing in other pedigrees identified a second DNAJB6 mutation in an adjacent residue in the G/F domain of the protein. Finally, we observed that DNAJB6 forms large protein aggregates in muscle from patients with DNAJB6 mutations, analogous to those seen in other dominant myopathies with protein aggregates (such as myofibrillar myopathies). Together these findings provide strong evidence that DNAJB6 mutations are a novel cause of autosomal dominant myopathy.
Families with dominantly-inherited or sporadic myopathy with unknown genetic cause were identified from the Washington University Neuromuscular Clinic and Genetics Project. Charts, clinical records, and pathology slides were reviewed and individuals were re-examined. All participants gave written informed consent, and all study procedures were approved by the Human Studies Committee at Washington University.
Indexed genomic DNA (gDNA) libraries were prepared from patient gDNA using the TruSeq DNA Preparation kit (Illumina) and exome-captured using the TruSeq Exome Enrichment kit (Illumina) according to manufacturer’s protocol. Sequencing was performed with 100bp paired-end reads on a HiSeq2000 (Illumina). Reads were aligned to the human reference genome with NovoAlign (Novocraft Technologies). Variants were called with SAMtools15 and annotated with SeattleSeq. Pathogenicity predictions for variants were obtained from SeattleSeq and SIFT16. Sequencing coverage depth was calculated using BEDTools17 and genomic coordinates provided by Illumina. Control exomes were from patients undergoing exome analysis for diseases other than myopathy. All reported genomic locations are from GRCh37/hg19. Segregation of mutations was assessed with standard PCR-based sequencing using Primer3Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) for primer design, an Applied Biosystems 3730 DNA Sequencer for sequencing, and LaserGene SeqMan Pro version 8.0.2 (DNAStar) for tracing analysis.
Genome-Wide Human SNP 6.0 arrays (Affymetrix, Santa Clara, CA) were run according to the manufacturer’s protocols. Quality control and genotyping used Genotyping Console version 18.104.22.1684. Linkage analyses used the easyLINKAGE Plus interface (v.5.08beta)18,19 and assumed autosomal dominant inheritance with complete penetrance, a disease allele frequency of 0.01%, no phenocopies, and Affymetrix “Caucasian” allele frequencies. 2-point logarithms of the odds (LOD) scores were calculated using FastLink v4.120 while GeneHunter v2.1r5 was used for multipoint analysis and haplotype construction21.
Muscle tissue was processed as previously described22. In brief, cryostat sections of rapidly frozen muscle were processed for sections and fixed in acetone. For TDP-43 immunostaining, sections were incubated in 4% paraformaldehyde prior to acetone. Antibodies used were mouse anti-DNAJB6 (Sigma Chemicals St. Louis, MO), SMI-31 (Covance, Berkley, CA) and TDP-43 (BC001487, Rabbit, Protein Tech Group, Chicago IL).
We identified a Caucasian family (Family 1, Fig 1A) with dominantly-inherited limb-girdle muscular dystrophy (i.e. LGMD 1) and prominent vacuoles on muscle pathology (see description below). DNA from three affected individuals underwent whole-exome capture followed by next-generation sequencing. An average of 10 Gb of sequence covering the target exome was generated per individual, resulting in an average coverage depth of 50× per targeted base (Supplementary Table 1). In each individual, ~14,000 single nucleotide polymorphisms (SNPs) and small insertions/deletions (indels) were identified in the targeted intervals. Based on the rarity of the disease and dominant inheritance pattern, we applied a filtering pathway (Table 1) to select heterozygous, autosomal, non-synonymous SNPs and coding indels that had not previously been reported to dbSNP131 or found in the 1000 Genomes Project (release August 2010).
Across the entire exome, a total of 22 variants fulfilling these criteria were shared by all three exomes (20 SNPs and 2 indels, Supplementary Table 2). The most intriguing candidate variation was a heterozygous T>C at chr7:157,160,108 (Supplementary Fig 1), producing a p.Phe93Leu substitution (c.277T>C in NM_058246) in a highly conserved residue within the G/F domain of DNAJB6 (Fig 1C). The mutation was confirmed by PCR-based sequencing (Fig 1B) and segregates with disease in the broader pedigree (Fig 1A). This mutation was considered the most likely causative variant for two reasons. First, DNAJB6 lies within the recently narrowed linkage interval for families with LGMD 1E7, whose clinical and pathological phenotype shows considerable overlap with Family 1 (see Clinical Phenotypes below). Second, DNAJB6 is a muscle-expressed member of the HSP40/DNAJ family of molecular chaperones23 and mutations in three other chaperone proteins (VCP, BAG3, and CRYAB) cause dominantly-inherited myopathies with vacuoles on pathology24–26.
To obtain additional genetic evidence for, or against, the DNAJB6 variant, we performed genome-wide 2-point linkage analysis using 11 individuals (5 affected, 6 unaffected). The maximum LOD score (2.53 with theta=0.000) was found at SNP_A-2312119 on 7q36, and approached the maximum LOD score predicted for this pedigree by simulations (2.57 at theta=0.000, Supplementary Fig 2A). This marker is within the minimal LGMD 1E locus and located in the gene immediately telomeric to DNAJB6. Multi-point analysis showed a LOD score of 3.00 across this region and negative LOD scores across the LGMD 1F locus at 7q32 (Supplementary Fig 2B, C). A disease-associated haplotype from 120 cM to the telomere was identified and a single recombination event narrowed the minimal interval to 2.1 Mb (chr7:157,090,296 to the telomere, Supplementary Fig 3). Although two other SNPs showed 2-point LOD scores >2.0 (SNP_A-8427187 at 1p36, LOD 2.31; SNP_A-8425136 at 20q13, LOD 2.05), they did not overlap with any of the 22 candidate variants identified by exome sequencing. This linkage data strongly supports the DNAJB6 mutation as the disease-causing mutation in Family 1.
A strict criteria for whether a putative mutation causes a mendelian disorder is the identification of additional mutations in the same gene producing a similar phenotype. To obtain this level of certainty, we analyzed the exome sequencing data from an additional 13 individuals (8 with dominantly-inherited myopathy and 5 with sporadic vacuolar myopathy) for mutations in DNAJB6. The proband of an African American family (Family 2, Fig 1A) was found to have C>G at chr7:157,160,118 (Supplementary Fig 1), producing a p.Pro96Arg substitution (c.287C>G). Remarkably, this mutation is located only 2 amino acids away from the mutation found in Family 1 and is also in the G/F domain of the protein. This variant was confirmed with PCR-based sequencing and segregated with disease in participating individuals (Fig 1A and B).
Neither DNAJB6 mutation was found in 150 control exomes performed at our institution, in which the relevant exon was covered >100×. They were also absent from 1,383 European-American and 1,081 African-American exomes from the NHLBI Exome Sequencing Project in which coverage of the relevant exon was >100× per individual27. Furthermore, neither was identified in the 1094 individuals of various ancestries sequenced as part of the 1000 Genomes Project 28.
Five affected individuals from Family 1 were evaluated (Fig 1A and Table 2). Beginning in the 4th decade, all five individuals developed trouble climbing stairs or getting up from the floor. Examinations showed limb-girdle weakness proportional to disease duration, but there was no bulbar weakness or dysarthria. In two patients, the quadriceps group was relatively preserved compared to the hamstrings. Mild heel-cord contractures were seen only in II:3, but finger and toe flexion was normal. None had cardiac or pulmonary involvement, and none had muscle rippling or mounding. Creatine kinase (CK) levels were elevated in 4 of 5 subjects (2–5 times the upper limit of normal) and have remained elevated in those with serial testing. EMG/NCS had been performed in 2 patients, demonstrating clear myopathic changes with fibrillation potentials and positive sharp waves in one. Patients progressed at a similar rate, typically remaining capable of ambulation but requiring a wheelchair for community mobility within 20 years of symptom onset. Diagnostic muscle biopsies from three individuals (II:2, II:3, II:7) showed a chronic myopathy with vacuoles and are further described below.
Three affected individuals from Family 2 were evaluated and found to have a distal-predominant myopathy (Fig 1 and Table 2). Except for the variable ages of onset (18–35 years), the pattern of weakness and pace of progression have been similar across all three siblings. Weakness began in the distal lower extremities and presented with tripping. CK levels were elevated and EMG/NCS was consistent with myopathy. Weakness has gradually progressed to include the hands and proximal legs, with loss of ambulation after disease duration of 20–30 years. No cardiac or pulmonary involvement has been detected.
Deltoid biopsies from Family 1 (II:2, II:3, II:7) were obtained and reanalyzed. All three showed moderate variation in fiber sizes, myofibers with internal nuclei, and variable involvement of different fascicles (Fig 2A). Many of the small fibers were angular and some contained characteristic rimmed vacuoles on gomori trichrome and congo red staining (Fig 2B). The vacuoles were negative for non-specific esterase or acid phosphatase. No necrotic fibers or inflammation was present. Two of the three biopsies had scattered fibers with absent cytochrome oxidase staining. Immunostaining for dystrophin, α-, β-, γ-, and δ-sarcoglycan, α2-laminin, caveolin-3, dysferlin, and desmin were normal in all biopsies. Aggregates of SMI-31 and TDP-43 were present in all three biopsies (Fig 2C and F).
Mutations in other proteins involved in protein folding or degradation (e.g. VCP, CRYAB, and BAG3)24,26,29 also cause dominant myopathies characterized by abnormal protein aggregates or vacuoles. Often, skeletal muscle shows accumulation of the mutated protein as inclusions24,25,30,31. We reasoned that DNAJB6 protein would similarly accumulate in patients harboring DNAJB6 mutations. Indeed, immunostaining for DNAJB6 highlighted multiple fibers with subsarcolemmal accumulation and sarcoplasmic inclusions in all three patients heterozygous for the Phe93Leu mutation (Fig 2D and F). Some DNAJB6 positive inclusions were located around and within vacuoles. Inclusions of DNAJB6 were not seen in muscle biopsies from 4 control patients (two without neurologic disease and two with amyotrophic lateral sclerosis) (Fig 2G). Increased DNAJB6 immunostaining was seen also in small fibers from two patients with sporadic inclusion body myositis (Fig 2H), suggesting that DNAJB6 accumulation may play a role in other myopathies characterized by abnormal protein aggregation and vacuoles. No member of Family 2 consented to biopsy.
Using exome sequencing, bioinformatic filtering, and linkage analysis, this study provides genetic and pathologic evidence that mutations in DNAJB6 are a novel cause of dominantly-inherited myopathy. DNAJB6 encodes a highly conserved member of the HSP40/DNAJ family of co-chaperones that function with HSP70 proteins to protect client proteins from irreversible aggregation during protein synthesis or during times of cellular stress32. DNAJ family members regulate the ATPase activity of HSP70 family members via their J-domain (Fig 1C) and are thought to help direct client protein specificity for HSP70 chaperone function33. DNAJB6 is a ubiquitously expressed HSP40 and is one of the more abundant DNAJB family members in skeletal muscle23.
The functional role of DNAJB6 and the identity of its client proteins are not fully understood in mammalian cells. Genetic knockout of DNAJB6 leads to the accumulation of toxic keratin-18 containing inclusion bodies in embryonically derived chorionic cells 34. Other studies have suggested that DNAJB6 facilitates ubiquitin-dependent degradation of select substrates such as β-catenin and even mitochondria via mitophagy 35, 36. DNAJB6 is also present in the core of Lewy bodies and is upregulated in astrocytes of patients with Parkinson’s disease 37.
Lastly, DNAJB6 plays an important role in modulating the aggregation and toxicity of polyglutamine (poly-Q) aggregates in both drosophila and mammalian cell lines 38–40. Recent work has further delineated this role by identifying DNAJB6 as one of the most potent suppressors of poly-Q aggregation amongst the DNAJ proteins, a function that requires the C-terminal domain (amino acids 152–232)40. A differential role for the two isoforms of DNAJB6 has also been suggested by experiments demonstrating that DNAJB6a (the longer, nuclear-localized isoform) effectively suppresses nuclear poly-Q protein aggregation but is ineffective with cytoplasmic aggregation, while DNAJB6b (the shorter, cytoplasmic isoform) is a potent suppressor of cytoplasmic poly-Q aggregation40.
The mutations we have identified involve adjacent, highly conserved residues within the G/F domain of the DNAJB6 protein (Fig 1C), a domain shared by both isoforms and other DNAJB family members (Supplementary Fig 4). While the J-domain is essential for stimulating HSP70 ATPase activity and the C-terminal amino acids play a key anti-aggregation role, the precise role of the G/F domain is less understood. The mutations we have identified are likely to help clarify mammalian G/F domain function. Studies of this domain in lower organisms provide some initial information. In bacteria, the G/F domain is required for the recognition of partially unfolded client proteins and is important for the formation of stable DNAJB6/client protein complexes41. In S. cerevisiae, the G/F domain of Sis1 (a DNAJB6 ortholog) is required for the propagation of specific yeast prions 42, a process that depends on the proper interaction of Sis1 with soluble and aggregated forms of Sup35 and Rnq1 proteins. Furthermore, deletion or mutation of a 13 amino-acid motif within the Sis1 G/F domain blocks propagation of the Sup35 and Rnq1 prions43. This 13 amino acid motif is not present in DNAJA family members or other yeast HSP40 co-chaperones such as Ydj1 (Fig 1D, Supplementary Fig 4A). Yeast prions contain Q/N rich prion-like domains, a feature they share with TDP-4344–46. It is therefore notable that the human DNAJB6 mutations we have identified fall within this same 13 amino-acid motif (Fig 1D). This parallel is intriguing in light of the observation that TDP-43 aggregates were found in skeletal muscle from patients with DNAJB6 mutations.
Because DNAJB6 is ubiquitously expressed and implicated in other neurodegenerative diseases, it is interesting that patients with mutations show a selective disease of muscle. In vitro data suggests that the balance and availability of DNAJ family co-chaperones serve as the limiting factor on HSP70 chaperone activity47. Presumably therefore, small changes in the function of one chaperone (e.g. DNAJB6) could have global consequences on protein folding/handling within the cell. Post-mitotic differentiated skeletal muscle may be uniquely prone to such perturbations, as has been previously suggested for mutations in VCP which, like DNAJB6, is a ubiquitously expressed co-chaperone that can present with a phenotype limited to muscle48. The fact that DNAJB6 is a highly expressed DNAJB family member in skeletal muscle23 could also contribute to the vulnerability of this tissue. Alternatively, client proteins requiring DNAJB6 for proper homeostasis might differ based on tissue or be more toxic to skeletal muscle in an aggregated state (i.e. TDP-43). It is even possible that mutant DNJAB6 aggregation itself is pathogenic.
Future studies of DNAJB6 and its mutations will be directed at a deeper understanding of the molecular pathogenesis of these myopathies. We hope to clarify how G/F domain mutations affect DNAJB6 function, examine the possibility of distinct roles for DNAJB6 isoforms, and investigate the mechanisms underlying the muscle specificity of the disease.
The clinical phenotypes of our two DNAJB6 families share similarities and distinctions. Both families have onset of weakness in early adult-hood and gradual progression to severely impaired or lost ambulation in 20 or 30 years. In contrast to Family 1, where weakness occurs in a limb-girdle pattern, Family 2 has distal-predominant weakness. There is considerable precedent for this type of wide phenotypic spectrum among dominant myopathy disease genes. For example, mutations in myotilin, even identical ones, are associated with proximal-predominant (LGMD-1A), distal myopathy, and myofibrillar myopathy patterns49, 50. Similarly, mutations in VCP are known primarily for causing inclusion body myopathy with Paget’s disease of the bone and fronto-temporal dementia24–26. Recently however, mutations have been described in patients with distal myopathy51. Thus, it is not surprising to find DNAJB6 mutations producing either a limb-girdle or distal myopathy phenotypes.
As noted above, DNAJB6 resides within the 7q36 linkage interval previously reported for Finnish families with LGMD 1E7,8. These patients had proximal lower extremity weakness, a mean age of onset of 40 (range 28–60), mildly elevated CK and vacuolar pathology on muscle biopsy similar to the phenotype of Family 1. It will be interesting to see whether DNAJB6 mutations explain these pedigrees as well.
This work was supported by the BJC Institute for Clinical and Translational Sciences, the Children’s Discovery Institute, National Institutes of Health (NIH) Grants NS055980 and NS069669 (R.H.B), NS075094 (M.B.H.), AG031867 (C.C.W), the Neuroscience Blueprint Core Grant NS057105 to Washington University, the Hope Center for Neurological Disorders, the Muscular Dystrophy Association (R.H.B.) (C.C.W). R.H.B. holds a Career Award for Medical Scientists from the Burroughs Wellcome Fund. A.P. acknowledges the Washington University Neuromuscular Research Fund. We thank the Genome Technology Access Center (GTAC) in the Department of Genetics for help with genomic analysis, which is partially supported by NCI Cancer Center Support Grant #P30 CA91842 to the Siteman Cancer Center and by ICTS/CTSA Grant# UL1RR024992 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research.