Autosomal-Dominant Distal Myopathy Associated with a Recurrent Missense Mutation in the Gene Encoding the Nuclear Matrix Protein, Matrin 3

Abstract

Distal myopathies represent a heterogeneous group of inherited skeletal muscle disorders. One type of adult-onset, progressive autosomal-dominant distal myopathy, frequently associated with dysphagia and dysphonia (vocal cord and pharyngeal weakness with distal myopathy [VCPDM]), has been mapped to chromosome 5q31 in a North American pedigree. Here, we report the identification of a second large VCPDM family of Bulgarian descent and fine mapping of the critical interval. Sequencing of positional candidate genes revealed precisely the same nonconservative S85C missense mutation affecting an interspecies conserved residue in the MATR3 gene in both families. MATR3 is expressed in skeletal muscle and encodes matrin 3, a component of the nuclear matrix, which is a proteinaceous network that extends throughout the nucleus. Different disease related haplotype signatures in the two families provided evidence that two independent mutational events at the same position in MATR3 cause VCPDM. Our data establish proof of principle that the nuclear matrix is crucial for normal skeletal muscle structure and function and put VCPDM on the growing list of monogenic disorders associated with the nuclear proteome.

Main Text

The distal myopathies are a clinically and pathologically heterogeneous group of genetic disorders in which the distal muscles of the upper and the lower limbs are selectively or disproportionately affected.^1,2 Comparison of the different forms of distal myopathy has shown considerable phenotypic variability, both in terms of the age at onset and the pattern of muscle involvement. Some subtypes of distal myopathy clinically or genetically overlap with the inclusion body myopathies (summarized in MIM 605637 for IBM3) and the limb-girdle muscular dystrophies (summarized in MIM 159000 for LGMD1A and MIM 253600 for LGMD2A). Distal myopathies are caused by mutations in genes encoding proteins associated with the plasma membrane (dysferlin [MIM 603009]^3,4 and caveolin-3 [MIM 601047]⁵), with the sarcomere (titin [MIM 188840]⁶ and myosin [MIM 160760]⁷), and with posttranslational protein modifications (UDP-N-acetylglucosamine-2-epimerase [MIM 603824]).⁸

A decade ago, we described in this journal an adult-onset autosomal-dominant distal myopathy often complicated by vocal cord paralysis (vocal cord and pharyngeal weakness with distal myopathy, VCPDM [MIM 606070]).⁹ Genome-wide linkage analysis in a single large North American pedigree revealed that the VCPDM locus mapped to a 12 cM (11.8 Mb) interval on chromosome 5q between microsatellite markers D5S1995 and D5S436. First symptoms developed at the age of 35–57 years and the most frequent initial manifestation was ankle dorsiflexion weakness and foot drop. The disease had a slowly progressive course with involvement of the feet and hands and eventually an effect on shoulder and pelvic muscles. Vocal cord or swallowing dysfunction occurred in most cases. Electromyogram (EMG) and nerve conduction studies indicated a myopathy, and the levels of serum creatine phosphokinase (CPK) were usually only mildly or moderately elevated (at maximum 8× the upper limits of normal). Analysis of muscle biopsies showed myopathic changes including variations in fiber size, fiber splitting, and subsarcolemmal rimmed vacuoles.

We were now able to expand the family because another two siblings were now definitely afflicted with the disease. These individuals belong to an additional branch of the family and had not been included in the 1998 paper (for a comparison of clinical data, see Table 1). Analysis of microsatellite markers and haplotype reconstruction disclosed a critical recombination event between D5S2009 and D5S21116 in both patients. On the basis of the available genomic sequence of chromosome 5q31, a set of additional densely spaced short tandem repeat (STR) polymorphisms was deduced. Using these markers, we mapped the recombination event between AC008667A and AC008667C, thereby reducing the candidate interval from 11.8 Mb to 5.9 Mb. By using newly established STRs, we could also narrow down the site of the recombination at the centromeric border of the VCPDM region. The refined critical region spanned 5.37 Mb between STR markers sara2AC and AC008667C (Figure 1, for details referring to newly established STR markers, see Table S1 available online). We then checked a series of ten families with autosomal-dominant distal myopathy with clinical features resembling VCPDM for compatibility with linkage to the VCPDM region on chromosome 5q31. We identified a multigenerational Bulgarian family (Figure 2) with a typical VCPDM phenotype (for a comparison of clinical data, see Table 1). With a set of 12 available DNA samples from this family, genotyping and haplotype reconstruction of STR markers D5S458, D5S500, D5S594, and D5S2116 established linkage to the VCPDM interval. A maximum two-point LOD score (Zmax) of 3.35 was obtained at recombination fraction (θ) 0 for marker D5S500, but no new recombination events, allowing a further restriction of the VCPDM interval, were identified. VCPDM-disease-associated haplotypes differed between the North American and Bulgarian samples. The review board for medical ethics at Aachen University of Technology gave approval for sample collection and molecular genetic studies in conjunction with local approval in the United States and Bulgaria.

Table 1.

Comparison of Clinical Features

Family	Patient	Age at Onset (years)	First Symptom	Distal weakness		Shoulder Weakness	Swallowing/Vocal Cord Dysfunction	CPK
				Legs	Hands
NAa		Mean: 45 (Range: 35–57)	Legs: 7/11 Hands: 2/11 Pharyngeal: 2/11	12/12	10/12	4/12	9/12	Normal, 8×
NA	DM73.3	48	Legs	+	+	−	−	2×
NA	DM73.4	45	Legs	+	+	−	+	2×
BG	IV.9	43	Legs	+	+	−	−	ND
BG	V.6	47	Legs	+	+	+	+	ND
BG	V.9	40	Legs	+	+	+	+	2×
BG	V.11	54	Legs	+	+	−	ND	ND
BG	V.13	52	Legs	+	+	−	ND	ND
BG	V.15	42	Legs	+	−	−	ND	ND
BG	V.17	46	Hands	−	+	−	ND	ND
BG	V.23	49	Legs	+	+	+	−	4×
BG	V.27	40	Legs	+	+	+	+	3×
BG	V.30	36	ND	+	−	−	ND	2×

NA, North American family; BG, Bulgarian family; ND, not determined; CPK, creatine phosphokinase;

^aData summarized from Feit et al.⁹

Figure 1.

Refinement of the VCPDM Interval in the North American VCPDM Family

A subset of markers used in this study is given in centromeric to telomeric orientation. The physical map positions are according to the UCSC March 2006 freeze. Patient V.34 carries the full-blown disease haplotype. Recombination events in patient V.37 and patient DM73.3 (newly identified patient from the North American family) excluded markers sara2AC and AC008667C (printed in red), respectively, delimiting a 5.37 Mb region.

Figure 2.

Pedigree of the Bulgarian VCPDM Family

Individuals of whom DNA samples were available are marked by asterisks.

The May 2004 version of the human genome assembly displayed 56 genes represented in the NCBI Reference Sequence (RefSeq) collection between the newly established borders of the VCPDM critical region. The most apparent candidate gene within this interval was myotilin (MYOT [MIM 604103]). Myotilin contributes to sarcomere assembly and promotes actin crosslinking and sarcomeric integrity,¹⁰ and MYOT mutations have been shown to cause Limb-Girdle Muscular Dystrophy Type 1A (LGMD1A),¹¹ a small fraction of myofibrillar myopathy cases (MFM [MIM 609200]),¹² and spheroid body myopathy [MIM 182920].¹³ However, we found no MYOT abnormalities on the genomic, transcript, and protein level in the North American VCPDM family.¹⁴ Moreover, no MYOT mutation was detected in the Bulgarian pedigree by sequencing of all coding exons. These results provided strong evidence that myotilinopathies and VCPDM are not allelic disorders. Therefore, we prioritized the remaining annotated sequences with respect to skeletal muscle biology and disease. We used information from publicly available databases and characterized proteins in silico by BLAST alignments to protein database entries and by comparison of domain composition and organization with the SMART algorithm. We also investigated positional candidate genes for structural and functional homology with the known proteins involved in distal myopathies and related disorders. The entire coding and adjacent intronic regions of 20 candidate genes for a distal myopathy were analyzed by direct sequencing (Table 2).

Table 2.

Positional Candidate Genes Analyzed

Gene Name	Expression in Skeletal Muscle	Entrez Gene Identifier	OMIM Number	Strand	Genomic Position		Number of Coding Exons
					Start	End
SEC24A	+	10802	607183	+	134012378	134091500	23
C5orf14	+	79770	−	+	134237359	134265222	5
DCOHM	+	84105	609836	+	134268709	134326235	4
LOC153328	unknown	153328	−	+	135198264	135236335	6
FBXL21	+	26223	609087	+	135300162	135305266	4
SPOCK	+	6695	602264	−	136338886	136862917	10
KLHL3	+	26249	605775	−	136981088	137099678	15
C5orf5	+	51306	609371	−	137301541	137396701	21
BRD8	+	10902	602848	−	137520472	137542257	29
C5orf6	+	51307	609372	+	137701603	137713315	4
C5orf7	+	51380	609373	+	137716184	137800615	24
HSPA9B	+	3313	600548	−	137918923	137939014	17
SIL1	+	64374	608005	−	138310309	138561964	9
MATR3	+	9782	164015	+	138657254	138694029	14
PAIP2	+	51247	605604	+	138705418	138733308	3
PACAP	+	51237	609447	−	138751156	138753504	4
DNAJC18	+	202052	−	−	138775279	138803038	8
TMEM173	+	340061	612374	−	138835733	138842476	6
UBE2D2	+	7322	602962	+	138920935	138988202	7
PSD2	+	84249	−	+	139155590	139204232	14

We detected the same c.254C→G change in both families in exon 2 of the MATR3 gene (MIM 164015) (Figure 3A, for primer sets used to analyze the MATR3 gene, see Table S2), whereas no pathogenic mutations were identified in any of the other candidate genes. This mutation leads to an amino acid substitution from serine to cysteine at position 85 (S85C). We confirmed segregation of this change in both families. This mutation has not been deposited in the most recent release (build 129) of dbSNP, and 572 healthy controls (300 German, 164 white Northern American, and 108 Bulgarian) were negative for this nucleotide substitution. Analysis of haplotypes for closely flanking and intragenic SNP and STR markers showed no evidence of haplotype sharing between the North American and the Bulgarian VCPDM families (Figure 3B, for details referring to SNP and STR markers used, see Table S1). This indicates that these two kindreds with VCPDM are not closely related and suggests that the same MATR3 mutation arose independently in each of them.

Figure 3.

Identification of the VCPDM Mutation

(A) DNA sequence traces illustrating the c.254C→G, S85C mutation in the MATR3 gene (upper panel) and the corresponding wild-type sequence (lower panel). The position of the single nucleotide exchange is marked with an arrowhead.

(B) Comparison of disease-associated haplotypes. The disease-associated haplotype in the North American family is shown under the column heading “NA”; the disease-associated haplotype in the Bulgarian pedigree is shown under the heading “BG.” The phases of alleles have been determined for all markers. Intragenic markers and the mutation in exon 2 of the MATR3 gene are boxed. The closest flanking and intragenic markers that show different alleles for the North American and Bulgarian samples are printed in red.

(C) Multiple sequence alignment of human matrin 3 and related sequences. The mutant serine 85 residue is marked by an arrowhead.

Matrin 3 is a 130 kDA (847 amino acid) nuclear protein with nuclear import and export motifs and several binding sites for DNA and RNA^15,16 (Figure S1A). Matrin 3 is highly conserved through evolution (96% sequence homology between human and rat orthologs) and expression is observed in a variety of tissues including skeletal muscle and lymphoblasts (HUGE protein database and Figures S1B and S1C). Matrin 3 is a component of the nuclear matrix, an insoluble network that is dispersed throughout the nucleus and that is operationally defined as being resistant to high salt or detergents.¹⁷ Although the nature of the nuclear matrix is still under debate,¹⁸ it has achieved prominence given that its components are associated with the protein machinery for transcription, RNA splicing, and DNA replication.¹⁹ Moreover, accumulating experimental evidence suggests that the positioning of individual chromosomes in the interphase nucleus²⁰ is achieved by attachment of chromatin loops to the nuclear matrix at the so-called matrix attachment regions (MARs).^21,22 Although experimental evidence for a causal relationship is still lacking, the 3D organization of the genome in the interphase nucleus is believed to be an important factor for the orchestration of gene expression in the mammalian genome.²³

The nuclear matrix is structurally and functionally connected with the nuclear lamina, the nuclear proteome associated with the inner leaflet of the nuclear envelope.^24–26 The role of the nuclear lamina in human diseases has been established during the last 15 years,²⁷ and the growing list of diseases of the nuclear envelope also includes neuromuscular conditions such as Emery Dreyfuss muscular dystrophy (MIM 310300; MIM 181350),^28,29 limb-girdle muscular dystrophy type 1B (MIM 159001),³⁰ and Charcot-Marie-Tooth neuropathy type 2B1 (MIM 605588).³¹ Additional evidence for a role of the nuclear matrix in the pathophysiology of the skeletal muscle comes from very recent work on facio-scapulo-humeral dystrophy (FSHD [MIM 158900]) and X-linked myopathies. First, the partial deletion of the D4Z4 repeat array that causes FSHD³² results in delocation of an adjacent MAR from the nuclear matrix. This alteration is expected to change chromatin structure and to release transcriptional repression of neighboring genes implicated in the genesis of FSHD.³³ Second, mutations in the FHL1 gene (MIM 300163), encoding four-and-a-half LIM domain 1, cause clinically variable types of X-linked myopathies (MIM 300695).^34–36 FHL2, a homolog of FHL1, has been reported to bind to the nuclear matrix protein NP220 and the splicing factor PSF, a known matrin 3 interaction partner.^37,38

Currently, the effect of the identified S85C alteration on matrin 3 structure and function is not clear. Alterations of amino acid residues involved in nucleic acid binding may seriously impair the matrin 3 function. But the residues in contact with RNA and DNA are not close to the position shown here as changed (Figure S1A). However, assessment of the biochemical severity of the S85C change and the location and context of the altered amino acid in the protein sequence suggests that the mutation has functional consequences. With the Grantham scale,³⁹ which categorizes codon replacements into classes of chemical dissimilarity between the encoded amino acids, the c.254C→G missense mutation in matrin 3 was designated to be moderately radical according to the classification proposed by Li et al.⁴⁰ Given that disease-causing mutations are likely to occur at regions encoding conserved amino acid residues across species,⁴¹ we studied cross-species conservation of regions surrounding the S85C alteration by using the ClustalW algorithm. We found that the residue S85 is conserved across a wide range of species including mammals, chicken, frog, and fish (Figure 3C). Further, we applied three in silico algorithms—the SIFT algorithm,⁴² the PolyPhen algorithm,⁴³ and the subPSEC algorithm^44,45—to predict the putative effect of the mutation on the protein. With SIFT, the S85C substitution was classified as intolerant according to the classification by Ng and Henikoff.⁴² PolyPhen designated the S85C alteration as possibly damaging according to the classification by Xi et al.⁴⁶ Finally, the subPSEC program predicted the S85C change to cause a deleterious effect on protein function with the criteria proposed by Brunham et al.⁴⁷

We had the opportunity to study a muscle biopsy of a Bulgarian VCPDM case. Morphological analysis of the frozen muscle biopsy specimen revealed end-stage myopathy. The muscle was largely replaced by proliferated fat and connective tissue (Figure 4A). We observed strikingly variable degrees of immunoreactivity of myonuclei for matrin 3 (Figures 4B and 4C), whereas all nuclei of a normal control muscle were strongly labeled after incubation with the matrin 3 antibody (Figure 4D). Autophagic vacuoles were encountered in several muscle fibers and myonuclei displayed various stages of nuclear degeneration (Figures 4E–4G). On the other hand, nuclei of cultured lymphoblasts obtained from a VCPDM patient did not disclose overt pathology (Figure S2A). However, fractionation of nuclear proteins showed that matrin 3 is almost exclusively confined to the insoluble fraction in VCPDM lymphoblasts, whereas in control lymphoblasts, a considerable proportion of matrin 3 occurs in the nucleic-acid binding protein fraction (Figure S2B). A similar shift of the subnuclear distribution was seen for the Myc-tagged S85C matrin 3 mutant compared to Myc-tagged wild-type matrin 3 upon overexpression in Cos7 cells. Immunofluorescence analysis of transfected matrin 3 in Cos7 cells ascertained that this observation did not arise from the formation of insoluble nuclear aggregates (Figures S2C and S2D). Although the present data suggest that muscle pathology in VCPDM might be related to changes of the matrin 3 expression level and/or intranuclear mobility functional studies are obviously needed to clarify how mutant matrin 3 affects skeletal muscle physiology.

Figure 4.

Muscle Biopsy Findings in VCPDM

The muscle biopsy specimen was taken from the left M. gastrocnemius of a 54-year-old male patient of the Bulgarian family.

(A) The muscle parenchyma is largely replaced by fat and connective tissue. The remaining muscle fibers show a high degree of caliber variability. Hypertrophic muscle fibers containing nonsubsarcolemmal nuclei are marked by arrows. A hematoxylin stain was used on the cryostat section. The scale bar represents 250 μm.

(B and C) Matrin 3 immunostaining (rabbit anti-matrin 3 antibody BL2526; Bethyl Labs, Montgomery, TX, USA) of cryosectioned muscle fibers showing variable degrees of immunoreactivity of myonuclei (black arrows, immunoreactive nuclei labeled by brown reaction product; white arrows, faintly stained or almost unlabelled myonuclei). Because of the vast proliferation of fat tissue, the cryostat sections had to be cut much thicker than normal, 15 μm, resulting in a blurry appearance of the nuclei. A hematoxylin counterstain was used. The scale bar represents 40 μm in (B) and 25 μm in (C). The inset in (C) shows a muscle fiber with intact sarcolemmal dystrophin immunoreactivity (Dys 1 antibody, Novocastra, Newcastle, UK) indicating preserved overall immunoreactivity of the biopsied tissue. The scale bar represents 50 μm.

(D) Normal control muscle immunostained with the matrin 3 antibody: Strong immunoreactivity of all myonuclei. A hematoxylin counterstain was used on the cryostat section. The scale bar represents 50 μm.

(E) Semithin section of biopsy tissue that was cut from the block of frozen tissue, thawed, fixed in 6% buffered glutaraldehyde, and embedded in epoxy resin. Two muscle fibers are surrounded by fibrotic connective tissue and fat cells. Dark, osmiophilic material has accumulated in both fibers. The arrow indicates a strongly osmiophilic deposit, most likely autophagic material. A paraphenylene diamine stain was used. The scale bar represents 20 μm.

(F and G) Electron microscopy of ultrathin sections of the resin-embedded tissue. In (F), this muscle fiber shows an irregularly shaped nucleus with loose chromatin and numerous small perinuclear autophagic vacuoles containing osmiophilic, often myelin-like material. The scale bar represents 3 μm. In (G), two irregular, preapoptotic myonuclei are shown. One of the nuclei embraces an autophagic vacuole (indicated by an arrow), probably as a result of sarcoplasmic invagination. The scale bar represents 2 μm.

A variety of mechanisms leading to myofiber degeneration could be involved in VCPDM pathogenesis: (1) modification of gene expression relative to abnormal chromatin organization: matrin 3 has been shown to regulate transcription rates through association with matrix attachment sites in vitro.⁴⁸ (2) Deregulation of nuclear mRNA export: together with p54nrb and PSF, matrin 3 controls the retention of hyperedited mRNAs in the nucleus to prevent their translation at the ribosome.⁴⁹ (3) Abnormal pre-mRNA splicing: matrin 3 was found to be coimmunoprecipitated with hLodestar/HuF2, together with CDC5L, hnRNP A1, hnRNP A2/B1, and chromokinesin, suggesting it is also involved in pre-mRNA splicing.^50,51 (4) Alterations of the nuclear proteome: matrin 3 also plays a role in the nuclear import of proteins,⁵² thereby potentially altering nuclear functions. As a matter of course, these putative disease mechanisms are not necessarily mutually exclusive.

To our knowledge, this study is the first report of the association of a mutation in a gene encoding a nuclear matrix protein with a human hereditary disease. Our observations of the same MATR3 gene mutation in two unrelated kindreds with VCPDM that results in a nonconservative substitution of an interspecies conserved amino acid and that was absent in control subjects strongly support the pathogenic significance of this mutation. Our discovery of MATR3 mutations as the cause of VCPDM will aid genetic testing and counselling of patients with distal myopathies. Moreover, the knowledge about the involvement of the nuclear matrix protein matrin 3 in an inherited myopathy provides a path to a better understanding of the development, degeneration, and regeneration of skeletal muscle. Identification of the underlying mechanisms of muscle fiber damage in inherited myopathies may hold promise for the development of therapeutic interventions in these (usually rare) entities and, perhaps, in the (usually more frequent) sporadic forms of skeletal muscle disorders. In this way, our data are relevant for patient and family care, for basic science, and—hopefully—future options to treat neuromuscular disorders.

Web Resources

The URLs for data presented herein are as follows:

• European Bioinformatics Institute Web site (for ClustalW), http://www.ebi.ac.uk/index.html

• GeneCards Human Gene Database (for data on positional candidate genes), http://www.genecards.org

• HUGE protein database (for matrin 3 [KIAA0723]), http://www.kazusa.or.jp/huge/index.html

• Human genome assembly (for chromosome 5q31), http://genome.ucsc.edu

• National Center for Biotechnology Information (NCBI; for data on positional candidate genes and matrin 3 orthologs (RefSeq, Pubmed, OMIM, Gene, Protein), http://www.ncbi.nlm.nih.gov

• NCBI blast server (for protein sequence alignments), http://blast.ncbi.nlm.nih.gov/Blast.cgi

• NCBI dbSNP repository (for single-nucleotide polymorphisms), http://www.ncbi.nlm.nih.gov/SNP

• PolyPhen algorithm (for prediction of the S85C mutation effect), http://genetics.bwh.harvard.edu/pph/

• SIFT algorithm (for prediction of the S85C mutation effect), http://blocks.fhcrc.org/sift/SIFT.html

• SMART program (for the analysis of protein domain structure and composition), http://smart.embl-heidelberg.de

• subPSEC algorithm (for prediction of the S85C mutation effect), http://www.pantherdb.org/tools/csnpScoreForm.jsp