Abstract
Objective: To investigate the molecular basis of autosomal dominant limb-girdle muscular dystrophy (AD-LGMD) in three large new families.
Methods andResults: Genome-wide linkage was performed to show that the causative gene in all three families localized to chromosome 21q22.3 (Zmax = 10.3; θ = 0). This region contained the collagen VI α1 and α2 genes, which have been previously shown to harbor mutations causing a relatively mild congenital myopathy with contractures (Bethlem myopathy). Screening of the collagen VI α1 and α2 genes revealed novel, causative mutations in each family (COL6A1—K121R, G341D; COL6A2—D620N); two of these mutations were in novel regions of the proteins not previously associated with disease. Collagen VI is a ubiquitously expressed component of connective tissue; however, both limb-girdle muscular dystrophy and Bethlem myopathy patients show symptoms restricted to skeletal muscle. To address the muscle-specific symptoms resulting from collagen VI mutations, the authors studied three patient muscle biopsies at the molecular level (protein expression). A marked reduction of laminin β1 protein in the myofiber basal lamina in all biopsies was found, although this protein was expressed normally in the neighboring capillary basal laminae.
Conclusions: The authors’ studies widen the clinical spectrum of Bethlem myopathy and suggest collagen VI etiology should be investigated in dominant limb-girdle muscular dystrophy. The authors hypothesize that collagen VI mutations lead to muscle-specific defects of the basal lamina, and may explain the muscle-specific symptoms of Bethlem and limb-girdle muscular dystrophy patients with collagen VI mutations.
The limb-girdle muscular dystrophies (LGMD) are a clinically and genetically heterogeneous group of disorders characterized by muscle weakness and wasting, and elevated creatine kinase levels. Over the last few years, great strides have been made in identifying causative genes involved in recessively inherited dystrophies. To date, more than ten different genes have been found.1-3⇓⇓ Many of these causative genes code for proteins that make up the large dystrophin-associated protein complex on the sarcolemmal membrane. Substantial progress also has been made in identifying genes for dominantly inherited dystrophies. Of the six dominant LGMD loci reported to date (LGMD1A, LGMD1B, LGMD1C, LGMD1D, LGMD1E, and EDMD2), four causative genes have been discovered. Mutations in myotilin, a sarcomeric protein that binds to α-actinin in the Z-disk, is mutated in LGMD1A.4 Mutations in caveolin-3, described in LGMD1C, are thought to disrupt oligocaveolae formation at the myofiber plasma membrane where it interacts with cell signaling molecules.5 Mutations in the gene encoding lamin A/C are thought to compromise the integrity of the nuclear lamina and account for two dominantly inherited dystrophies: LGMD1B and Emery–Dreifuss muscular dystrophy (EDMD2).6,7⇓ Despite these advances, most of the dominant LGMD genes appear to cause disease in just one or a few distinct families. We and others8,9⇓ have collected several families diagnosed with autosomal dominant limb-girdle muscular dystrophy (AD-LGMD) that do not map to pre-existing loci. Thus, there are still several families with AD-LGMD with unknown genetic etiologies.
Here we report novel mutations in two ubiquitously expressed collagen genes, collagen VI α1 (COL6A1) and VI α2 (COL6A2), in three unrelated families showing a clinically variable form of dominantly inherited LGMD. We show causality of the mutations by linkage studies, mutation data, and biochemical studies of the muscle. Despite ubiquitous expression of collagen VI genes, these missense mutations appear to specifically affect skeletal muscle, causing a progressive muscular dystrophy. Our results suggest that missense mutations in collagen VI genes do not overtly affect the secretion and deposition of collagen VI protein in the extracellular matrix (ECM) of myofibers, but rather cause disruptions in the myofiber basal lamina that lead to abnormalities of muscle development and failure of myofiber regeneration.
Patients and methods.
Family ascertainment and diagnosis.
Families 1 and 3.
Families 1 and 3 (figure 1) were evaluated by the same physicians. Family 1 was a white family from Mississippi; Family 3 was from both Mississippi and North Carolina. Twenty-two individuals (12 affected) from Family 1, and 20 individuals (9 affected) from Family 3 were examined in a standard format noting neonatal history, delay in milestones, motor difficulties in childhood, and age at onset of perceived motor problems. In addition, physical examination was performed stressing muscle strength using the MRC (Medical Research Counsel) scale in proximal and distal muscles as well as facial muscles and the presence or absence of contractures. We found that 12 of 22 individuals from Family 1 and 9 of 20 in Family 3 were affected. Age at examination among the 21 patients from both families ranged from 6 to 76 years. An infantile onset with delay in motor milestones was noted in 6 individuals in Family 1 and 1 individual in Family 3; motor difficulties began in childhood (<10 years old) in 5 patients in each family. Onset began in adult life in one patient in Family 1 and three patients in Family 3. Two patients (12 and 48 years old) from Family 1 and 1 patient from Family 3 (55 years old) were wheelchair-bound. In addition, 1 patient from Family 3 needed to walk with a cane at approximately 40 years of age.
Figure 1. Pedigrees of three families studied with autosomal dominant limb-girdle muscular dystrophy (Bethlem myopathy). □ = unaffected; ▪ = affected; = status unknown.
Many affected members from both families had muscle cramps, pain, and weakness. Clinical examination revealed diffuse muscle weakness in Family 1, both proximal and distal muscles being weak in all 12 affected individuals. Proximal accentuation of weakness was noted with four patients having severe leg weakness, and three with severe arm weakness, with MRC grade less than 3/5. Mild facial weakness was noted in four affected individuals. All nine affected individuals in Family 3 had proximal weakness; in four it was severe in the legs, and in two severe in the arms. Even though only six patients had distal weakness, this was severe in some. Facial weakness was not apparent in any affected members from Family 3.
At the outset, none had any contractures of shoulders, hips, spine, or neck. Also notable is that 5 patients in Family 1 and 4 patients in Family 3 did not have any contractures at ages ranging from 8 to 76 years. The interphalangeal joints were most frequently affected by contractures (4/12 in Family 1 and 5/9 in Family 3), followed by ankles (3/12 in Family 1 and 2/9 in Family 3) and elbows (2/12 in Family 1 and 3/9 in Family 3). Knee contractures occurred in three patients, all in Family 3. In many patients, contractures were limited to ankles (a frequent site of contractures in many neuromuscular diseases) or were subtle so they could be easily missed. In many interphalangeal contractures could only be discovered by the inability to extend fingers passively once the wrist was extended.
Electromyographs from 13 affected individuals typically showed myopathic motor unit potentials with sparse fibrillations. Three muscle biopsies (one patient from Family 1 and two from Family 3) showed necrosis, fiber size variation, fiber splitting, internal nuclei, increased endomysial connective tissue, and fat infiltration. Creatine kinase levels were elevated, ranging between 485 and 1750 IU (normal < 200 IU). Electrocardiograms performed on two affected individuals from each family were normal. No patients from either family had any clinical evidence of heart disease.
Family 2.
In this white family from Ohio, the clinical presentation was similar to that described above (see figure 1). Twenty-one family members were examined; 16 were affected. Detailed examination was performed on 3 patients: an 8-year-old girl, an 11-year-old girl, and a 56-year-old man. Both girls were reportedly floppy in infancy. Neither was noted to have torticollis nor contractures in early life. Age of walking was mildly delayed (16 months). Gait abnormalities and the use of the Gowers maneuver were noted at 2 to 3 years of age. There were difficulties with falling, in running, and with performing other physical activities. The girls showed mild proximal weakness, mild atrophy, and normal reflexes. With the exception of passive dorsiflexion of the ankles, no contractures were present. Progression of muscle weakness and atrophy was slow. The 56-year-old man uses a wheelchair for community ambulation and a walker for short distances. Weakness was moderate, atrophy striking, and contractures present in fingers, knees, and elbows. Pes cavus was present. He had frequent muscle cramps. Deep tendon reflexes were diminished to absent. He was unable to whistle and had weak sternomastoids and trapezi, but other cranial and sensory nerves were intact. Intellect was preserved and scoliosis was absent. His cardiac disease was in proportion to his age, lipids, sex, and lifestyle. He noted hearing loss in the past few years. There were no problems with wound healing or excessive hyperextensibility other than laxity of the metacarpophalangeal joint of one thumb.
Overall, symptoms typically included difficulty running and jumping, muscle pain, and cramps. Similar to Families 1 and 3, affected individuals in Family 2 displayed variable ages of onset and variable degrees of disease severity. Most patients showed symptoms in infancy or childhood. In addition to the proximal weakness, pain and weakness in the extremities (arms, hands, and feet) was found in 4 of 16 family members. Because of leg and hip-girdle weakness, 3 of 16 individuals required the use of a wheelchair. Contractures of the elbows, toes, and fingers were seen in some individuals, whereas several other individuals did not have contractures of any kind. Congenital neck torticollis was found in two individuals. Serum creatine kinase levels ranged from normal (95 IU) to mildly elevated (513 IU). Muscle biopsies from three patients (not available for histochemical analyses) revealed a marked loss of myofibers, variability of fiber size and shape, increased central nuclei, extensive infiltration of fat and connective tissue, as well as several degenerating fibers. All individuals complaining of weakness showed myopathic features on electromyogram. Cardiac problems were not associated with the disease.
Genotyping and linkage analysis.
An automated genome-wide search was performed on Family 1 using fluorescent dye-labeled dinucleotide repeat markers from PE Applied Biosystems (Prism Linkage Mapping Set, Version 2; Foster City, CA). This set normally contains 396 fluorescent markers spaced at a 10-centimorgan density. The marker set was substantially trimmed to 227 markers spaced at an average density of 15 to 20 centimorgans on all 22 autosomes. Genomic DNA from 22 family members was PCR amplified using oligonucleotides from all 227 markers and was electrophoresed on automated sequencers (Applied Biosystems 377). The GENESCAN and GENOTYPER software packages (Applied Biosystems/Perkin-Elmer) were used to generate genotypes, and linkage analyses were performed using FASTLINK for two-point and sliding three-point logarithm of odds (lod) scores. When evidence for linkage was found on chromosome 21q22, additional markers were analyzed proximal and distal to the original markers.
Linkage analysis was investigated in Families 2 and 3 at the 22q22.3 locus using the following dinucleotide repeat markers: D21S212, D21S1903, COL6A110 (Research Genetics). PCR amplified alleles were hand-scored from autorads and genotypes were generated for available family members. Linkage analysis was performed using FASTLINK similar to the method described above.
Accession numbers for COL6A1 and COL6A2 genes.
COL6A1.
The complete coding sequence (cDNA) of the COL6A1 gene was obtained from GenBank (joint GenBank sequences X15879, M20776 J04211, and X15880; 4115 base pair (bp) with ATG at 49 and TAG at 3,135). The exon/intron structure of the 3′ end of the gene is available through Genbank, but no data existed for exon/intron structure of the 5′ exons. To obtain intron/exon structure of COL6A1 gene, the COL6A1 5′ coding sequence was BLAST searched against all available GenBank sequences.11 Search results yielded a perfect match to the Homo sapiens chromosome 21q22.3, cosmid clone Q4H9 (AJ011932). Comparing the coding sequence of COL6A1 to this cosmid sequence enabled us to identify intronic sequence surrounding exons 1 to 20. Intronic/exonic sequence was previously available for the remainder of the COL6A1 gene: S75385 (exons 10–27), X99135 (exons 28, 29), X99136 (exons 30–35).
Single-strand conformational polymorphism (SSCP) analysis and DNA sequencing.
Two affected individuals from each of the three families and two unrelated normal individuals were chosen for SSCP analysis. Primers flanking all exons of the COL6A1 and COL6A2 genes were designed to PCR amplify genomic DNA. Denatured PCR products were run on two different types of mutation-detection SSCP gels: 1) 0.5 × TBE/5% (49:1) acrylamide 4 °C, and 2) MDE gel (FMC). Aberrant conformers (bands showing an electrophoretic shift in mobility by SSCP) were excised from dried gels, re-amplified, and sequenced in both directions (Beckman CEQ Dye Terminator Cycle Sequencing Kits; Beckman, Fullerton, CA). Sequences then were aligned with those of the wild-type genes (Sequencher, GeneCodes, Ann Arbor, MI). Primers for PCR amplification and cycle sequencing are available upon request.
Restriction enzyme and amplification-refractory mutation system (ARMS) test analysis of missense mutations.
The G1070A missense mutation identified in Family 1 eliminates a Bgl I site in exon 14 of the COL6A1 gene. A 132-bp fragment from exon 14 (COL6A1) was PCR amplified from all available family members and digested with Bgl I. This was predicted to yield cut fragments of 72 and 60 by and an uncut 132-bp fragment and in all affected individuals. The A41OG missense mutation identified in the third exon of the COL6A1 gene in Family 2 was investigated using the ARMS test.12 The primers used in the ARMS test were: forward, CTCACGCCCGCCGTGCCTGTTCCTGGCAGG; normal-R, AGTCGGTGTAGGTGCCCTTCCCAAAGTAAT; and mutant-R, AGTCGGTGTAGGTGCCCTTCCCAAAGTAAC. The A-G missense mutation was predicted to be amplified by the “forward” and “mutant-R” ARMS primer pair, but not the “forward” and “normal-R” primer pair. In Family 3, the G-A (D620N) missense mutation eliminated a Taq I site in exon 25 of the COL6A2 gene. Exon 25 of the COL6A2 gene was amplified and digested with Taq I (New England Biolabs, Beverly, MA). The G-A mutation in exon 25 (COL6A2) was predicted to yield an uncut fragment of 221-bp and cut fragments of 95- and 126-bp in all affected family members. These techniques were used to study all available family members in addition to 95 unaffected control individuals (190 chromosomes).
Immunohistochemical studies.
Antibodies.
The following primary antibodies were used: laminin β1 (GibcoBRL; 3E5/H2/E7, gift from E. Engvall), laminin β2 (C4, gift from J. Sanes), laminin α2 (mAb1922, Chemicon, Temecula, CA; merosin, Novocastra, Newcastle Upon Tyne, UK), laminin α4 (gift from J. Miner), laminin α5 (4C7, GibcoBRL; HC7/B4, E. Engvall), laminin γ1 (2E8/F11, E. Engvall), collagen VI (5C6, Developmental Studies Hybridoma Bank [DSHB], University of Iowa; polyclonal, E. Engvall; 70XR95, Fitzgerald Industries), collagen III (mAb1343, Chemicon), collagen IV, dystrophin (d10), α-, β-, γ-, δ-sarcoglycan (Novocastra), β-dystroglycan (Novocastra), integrin β1 (AIIB2, DSHB, University of Iowa), integrin α7B (polyclonal, E. Engvall), perlecan (mAb 1948, Chemicon), fibronectin (HFN 7.1, DSHB, University of Iowa), nidogen/entactin, and embryonic myosin heavy chain.
Unfixed skeletal muscle from one affected individual from Family 1, two affected individuals from Family 3, an age-matched control, and a Duchenne muscular dystrophy (DMD) control were flash-frozen in liquid nitrogen-cooled isopentane sectioned on a cryostat (4 μm). Cryosections were incubated with diluted primary and Cy-3-conjugated secondary antibodies (Jackson Immunochemicals) as previously described. All slides were analyzed using a Nikon (FXA) microscope.
Immunoblots.
Frozen muscle biopsies from two individuals from Family 3, three patients with DMD, one patient with BMD, and four patients (two age-matched) (variable ranges of histopathology, no detectable biochemical abnormalities) were selected from our muscle bank. Muscle cryosections (<0.1 g) were boiled in protein loading buffer loaded on 3.5% to 12.5% sodium dodecyl sulfate-polyacrylamide gels. Fractionated proteins were transferred to nitrocellulose and incubated with diluted primary antibodies to collagen VI (Fitzgerald Industries) and collagen III (mAb1343, Chemicon International). Blots were then washed and incubated with affinity-purified HRP-conjugated anti-rabbit or anti-mouse secondary antibodies (Santa Cruz Biotechnology), followed by detection with ECL western blotting detection reagents (Amersham, Buckinghamshire, UK).
Results.
Three families diagnosed with AD-LGMD show genetic linkage to chromosome 21q22.3.
Clinical descriptions of each family are provided in Patients and Methods. A 22-member Mississippi pedigree (Family 1) carrying the diagnosis of AD-LGMD was initially studied by genetic linkage analyses. Selected for an automated genome-wide scan were 227 dye-labeled dinucleotide markers spaced at an average density of 15 to 20 centimorgans throughout all 22 autosomes in the genome. Markers were genotyped and lod scores were calculated for each locus using all 22 available family members. The maximum two-point lod scores for the genome-wide scan are plotted in figure 2. A significant lod score (≥3.0) was not identified through this first pass, but markers D1S450, D10S249, D12S86, and D21S1252 yielded lod scores greater than 1.0. We subsequently analyzed additional short tandem repeat (STR) markers that map proximal and distal to these loci. Multiple recombination events were identified at the chromosome 1, 10, and 12 loci yielding significant negative lod scores (<−2.0). However, no recombinantion events were observed at markers D21S212, D21S1903, and COL6A1 with the disorder in the family. A maximum lod score of 3.281 at θ = 0.0 was obtained at marker COL6A1 (see figure 2).
Figure 2. Maximum two-point logarithm of odds (Lod) score graph for the genome-wide linkage scan on Family 1. All 22 individuals from Family 1 were genotyped at 227 loci throughout the genome. Maximum two-point scores were calculated and the results were plotted as a function of marker location in centimorgans (cM). Chromosome numbers are indicated at the top of the plot. ⧫ = lod scores from first pass; ▪ = second pass.
Families 2 and 3 were diagnosed with AD-LGMD and exhibited a phenotype similar to that of Family 1. To determine if Families 2 and 3 were linked to chromosome 21q22.3, all available individuals from both families were genotyped with markers D21S212, D21S1903, and COL6A1. Maximum two-point lod scores calculated using FASTLINK were 5.035, θ = 0.0 (COL6A1) for Family 2, and 4.265, θ = 0.0 (D21S1903) for Family 3. These results, combined with those from Family 1 showed strong statistical support for all 3 families to be localized to chromosome 21q22.3 (D21S1903 − Zmax = 10.248 @ θ = 0.0).
Collagen VI missense mutations were identified in all three families with AD-LGMD.
SSCP analysis was used to screen all exons of the COL6A1 and COL6A2 genes of two affected individuals from each of three families. We identified an aberrant conformer in exon 14 of the COL6A1 gene in the two individuals from Family 1. Sequencing of aberrant conformers revealed a heterozygous G-A point mutation, nucleotide 1070, resulting in a glycine to aspartic acid change at amino acid position 341 (G341D). Heterozygosity for the G1070A mutation was confirmed by re-amplification of exon 14 and subsequent restriction enzyme digestion with Bgl I (figure 3A). In Family 2, an aberrant conformer in exon 3 of the COL6A1 gene was investigated by direct sequencing. We identified a heterozygous A410G nucleotide change, resulting in a lysine to arginine amino acid substitution (K121R). The A410G change was confirmed using the ARMS test (see figure 3B). In Family 3, we identified a heterozygous G1858A missense mutation in the COL6A2 gene, resulting in an aspartic acid to asparagine amino acid change (D620N). This G1858A mutation was confirmed through digestion of exon 3 with Taq I restriction enzyme (see figure 3C).
Figure 3. Confirmation and cosegregation of base changes in affected members of Bethlem myopathy families. Numbers listed above each lane correspond to the numbers assigned to individuals in figure 1. (A) The COL6A1, G1070A base change identified in Family 1 eliminates a Bgl I restriction enzyme site. The persistence of the 132-bp fragment after Bgl I digestion denotes the presence of the G1070A (G341D) base change. (B) The COL6A1, A410G base change in Family 2 was confirmed using the amplification-refractory mutation test (ARMS). The higher molecular weight band denotes the presence of A410G (K121R) mutation. (C) The COL6A2, G1858A base change identified in Family 3 eliminates a Taq I restriction enzyme site. The persistence of the 221-bp fragment after digestion denotes the presence of the G1858A (D620N) change in affected individuals.
All base changes co-segregated with the disease in all affected family members (see figure 3) and were not found in 95 unrelated control individuals (190 control chromosomes) (data not shown). Furthermore, all three base substitutions changed amino acid residues within the COL6A1–2 proteins. Alignment of the human COL6A1–2 protein sequences to those in mice showed evolutionary conservation of all three amino acid residues (data not shown). For these reasons, we feel that the G1070A and A410G changes found in the COL6A1 gene as well as the G1858A substitution found in the COL6A2 gene are the bona fide mutations causing AD-LGMD in each of the three families. The relative positions of these mutations are diagramed in the schematic of collagen VI shown in figure 4.
Figure 4. COL6A1, COL6A2, and COL6A3 missense and splicing mutations known to cause Bethlem myopathy. Open arrows indicate previously identified mutations; filled arrows indicate mutations identified in this study; the triple helical regions are represented by wavy lines; rectangular hatched regions denote globular domains.
The COL6A2-D620N mutation does not overtly affect the expression of ECM proteins, but does alter the expression of laminin β1 in the basement membrane of muscle fibers.
To investigate the consequence of collagen VI mutations on muscle, we processed a frozen muscle biopsy from an individual from Family 3 (COL6A2; D620N) for immunofluorescence studies. We investigated components of the muscle cytoskeleton using antibodies directed against components of the basal lamina (laminin α2, α4, α5, β1, β2, γ1, collagen IV, integrin α7B and β1, entactin/nidogen, perlecan), the extracellular matrix (collagen III and VI, fibronectin), the dystrophin associated complex (dystrophin, β-dystroglycan), and the sarcoglycan complex (α-, β-, γ-, and δ-, sarcoglycans). Control subjects included biopsies from an age-matched individual showing a mild myopathy but with normal findings for all muscular dystrophy proteins, and a patient with DMD. The specific antibodies used and their corresponding staining patterns are summarized in the table⇓.
Summary of immunofluorescence results using aCOL6A2(D620N) patient, a normal age-matched control subject, and a patient with DMD
We did not detect immunostaining differences in collagen VI (figure 5A), collagen III, or collagen IV (not shown). Normal immunostaining of collagen VI also was observed in another COL6A2-D620N family member and one COL6A1-G341D patient from Family 1. The expression of dystrophin and the dystrophin associated proteins were similar between the COL6A2 patient and the normal control, but were markedly reduced in the patient with DMD, as expected. However, dramatic differences were detected with laminin β1, a component of the basal lamina. The laminin β1 antibody showed strong immunostaining of the capillaries and the basal lamina of myofibers in the two control subjects. We found a significant reduction of laminin β1 in the basal lamina of our COL6A2 patient compared with DMD and normal control subjects, whereas capillary staining appeared normal (see the table and figure 5A). This myofiber-specific basal lamina deficiency was confirmed in another affected member of Family 3 and one affected member of Family 1 (3 of 3 biopsies tested). Muscle biopsies from Family 2 were not available for histochemical analyses. The myofiber-specific decrease in laminin β1 was seen with two different β1 antibodies. No other protein abnormalities were detected for any of the other proteins tested.
Figure 5. Protein studies of muscle biopsies from patients with collagen VI mutations. (A) Immunofluorescence analysis of muscle biopsies from a COL6A2 patient (D620N) (Family 3, III-1) and control biopsies (normal and dystrophin-deficient [DMD]) using antibodies to collagen VI and laminin β1. Despite the mutation in COL6A2, immunostaining of this protein in patient muscle appears normal. In DMD and normal muscle, myofiber basal lamina and capillaries show strong immunostaining for laminin β1. The COL6A2 (D620N) patient shows a marked reduction of laminin β1 staining in the myofiber basal lamina, but basal lamina of capillaries show normal immunostaining. Similar immunofluorescence patterns were observed using three different antibodies to collagen VI and two different antibodies to laminin β1. This suggests that the collagen VI mutations specifically disrupt the myofiber basal lamina, and not the capillary basal lamina. (B) Immunoblot analyses using antibodies to collagen VI and collagen III. Coomassie blue staining of myosin heavy chain shows equal loading of the gel. The amount of collagen VI protein parallels that of collagen III in all biopsies, suggesting that these proteins reflect the amount of fibrosis in the biopsy. The collagen VI mutation-positive patients show a similar amount of collagen VI and collagen III. Also note the relatively similar levels of COL6A1 and COL6A2 as compared with COL6A3 in each biopsy. These data argue against a major dominant-negative mode of biochemical pathology.
Western blotting confirms normal levels of collagen VI protein in patients with COL6A2–D620N mutations.
The results of the immunostaining suggested that collagen VI mutations did not affect the assembly and deposition of collagen VI at the ECM. To more quantitatively measure the expression of collagen VI, we processed frozen muscle biopsies for immunoblot analyses using antibodies to collagen VI. Two COL6A2 biopsies were compared with dystrophin-deficient (DMD) biopsies (four patients) and several control biopsies (five patients) that showed a wide range of histopathology, but with no detectable biochemical abnormalities. Polyclonal antibodies to collagen VI reacted equally to a polypeptide of 200 kd (α3) and two polypeptides of 140 kd (α1, α2). The level of collagen VI protein was variable between the DMD/BMD and normal control subjects, and the COL6A2 patients (see figure 5B). We felt that the variable amount could reflect the extent of connective tissue proliferation (fibrosis) in each biopsy. To test this we normalized the collagen VI signal to that of collagen III: collagen III is co-expressed in the ECM with collagen VI. Blots were stripped and reprobed using collagen III antibodies. In all biopsies, the amount of collagen VI was similar to that of collagen III, including the two COL6A2 mutation-positive patients (see figure 5B). These results are consistent with our immunofluorescence results and suggest that the overall expression of collagen VI protein is not overtly disrupted by mutations in collagen VI.
Regenerating myofibers are not observed in COL6A2 patient biopsies.
We hypothesized that an abnormal basal lamina caused by secondary reduction of laminin β1 could lead to an impaired regenerative capacity in patient muscle. To test this we used an antibody directed against embryonic myosin13,14⇓ to detect regenerating fibers in muscle biopsies from two COL6A2 patients (D620N), one COL6A1 patient (G341D) from Family 1, and control subjects. No embryonic myosin positive fibers were seen in the muscle biopsies from the COL6A1, COL6A2, and normal control patients, suggesting that myofiber regeneration was not a prominent process in this muscle (figure 6). DMD muscles showed numerous positive fibers.
Figure 6. Muscle biopsies from patients with COL6A2 mutations lack of evidence of muscle regeneration. (top panel) Hematoxylin and eosin–stained muscle biopsy from a COL6A2 (D620N) patient shows increased endomysial connective tissue, fatty infiltration, and variability in fiber size. Despite these dystrophic features, embryonic myosin, a marker for regenerating fibers, is not detected in patient myofibers (middle panel). Compared with the COL6A2 patient (middle panel), large numbers of myofibers in the patient with DMD (bottom panel) show intense staining for embryonic myosin. Scale bar = 320 μm.
Discussion.
We conducted a genome-wide search using a 22-member family carrying the diagnosis of AD-LGMD. The results of our scan showed linkage to chromosome 21q22.3 (Zmax = 3.28, θ = 0). Two additional AD-LGMD families with a similar phenotype showed localization of the disease gene to the same region (Σmax = 10.28, θ = 0). Two of the three known collagen VI genes have been mapped to chromosome 21q22.3: COL6A1 and COL6A2.15,16⇓ These genes were considered excellent candidates for causing the disease observed in our three families, as mutations in these collagen VI genes have been previously reported to cause Bethlem myopathy, a relatively benign muscle disease with contractures.17-19⇓⇓ Using SSCP analysis and sequencing, we then identified novel mutations in collagen VI genes in each of the three families. A glycine to aspartic acid amino acid substitution in the triple helical (TH) domain of collagen α1 was identified in Family 1 (G341D). In Family 2, a lysine to arginine change (K121R) was identified in the amino-terminal globular domain of collagen VI α1. Finally, in Family 3 we identified an aspartic acid to asparagine change in the carboxy-terminal globular domain of COL6A2 (D620N). To our knowledge, this is the first report describing mutations in the globular domains of COL6A1 and COL6A2. The locations of these novel mutations relative to those previously described are shown in figure 4.
Bethlem myopathy is considered a relatively benign, nonprogressive congenital myopathy.20,21⇓ Nearly all patients reported to date have shown early onset flexion contractures of the fingers, wrists, elbows, and ankles. Serum creatine kinase levels are typically normal to slightly elevated, and muscle biopsy studies have shown nonspecific myopathic changes with some evidence of myofiber degeneration. The families reported here showed weakness that was variable and, in some cases, considerably more severe than typically described for Bethlem myopathy, with many features of a progressive dystrophy. Joint contractures were either absent or much milder in the three families presented here compared with typical Bethlem myopathy patients. Furthermore, whereas Bethlem myopathy patients typically present with symptoms in infancy, the age at onset in these three families varied from infancy, to early childhood, to adulthood. Although some patients presented with mild weakness giving rise to only limited functional impairment, others presented with a more severe, dystrophic-like weakness with symptoms including Gower’s maneuver, toe walking, and loss of ambulation. Muscle biopsy findings showed a loss of myofibers, evidence of fiber degeneration, and extensive fat and connective tissue infiltration. This unusual range of severity and subtle nature of contractures explains why a diagnosis of Bethlem myopathy was not originally considered for these families. Our studies lend support to emerging data showing that Bethlem myopathy patients may show a wide range of clinical variability, and may be less benign than previously thought in some patients or families.22,23⇓
Homozygous or compound heterozygous mutations in the COL6A2 gene recently were reported to cause Ullrich syndrome, a recessive congenital muscular dystrophy affecting connective tissue and muscle.24,25⇓ Four mutations have been described, all of which severely reduce the production of collagen VI in skeletal muscle (loss-of-function mechanism). A total of seven missense mutations have been described previously in dominant Bethlem myopathy patients—four in COL6A1, one in COL6A2, and two in COL6A3. Three of these mutations are glycine substitutions in the TH domains of the α1, α2, and α3 chains.17,19⇓ These missense mutations disrupt the Gly-X-Y motif that defines the TH domain, and have been proposed to disturb the formation of trimeric collagen VI and its downstream assembly into larger multimeric aggregates (dominant-negative biochemical defect). Three mutations are splice site mutations in the TH domain of COL6A1, which have been shown to decrease secretion of collagen VI in an in vitro transfection system (haploinsufficiency mechanism).23,26,27⇓⇓ The last mutation is a glycine to glutamic acid (G1679E) substitution in the amino globular domain of the α3 chain and has also been proposed to interfere with protein folding and secretion.18,28⇓ No immunohistochemical studies of mutation-positive patient muscle have been previously reported; thus, there has been no in vivo evidence for the dominant-negative or haploinsufficiency models previously proposed for Bethlem myopathy.
To determine if the COL6A2-D620N mutation affected the production, or accumulation, or both, of wild type collagen VI oligomeric protein, we performed immunologic studies of patient muscle biopsies using antibodies directed against collagen VI. Immunostaining and immunoblot analyses of collagen VI was similar in COL6A2 patients compared with control subjects. These results argue against the dominant-negative and haploinsufficiency mechanisms previously proposed by others. We then investigated if the COL6A2 mutation produced deleterious effects on proteins associated with collagen VI in the extracellular matrix (collagen III and fibronectin), basal lamina (laminins α2, α4, α5, β1, β2, γ1; integrins β1 and α7B; nidogen) and underlying membrane cytoskeleton (dystrophin; β-dystroglycan; α, β, γ, and δ-sarcoglycans).
We detected a specific and dramatic reduction of laminin β1 in the myofiber basal lamina of two COL6A2 and one COL6A1 patients from two families compared with dystrophin-deficient and normal control subjects. Laminin β1 staining of the capillary basal lamina was normal, suggesting a myofiber-specific effect of COL6A2 mutations on laminin β1 expression. All other components of the basal lamina and the membrane cytoskeleton appeared indistinguishable from control subjects. Our data are consistent with a case report that identified an age-related progressive deficiency of laminin β1 in muscle biopsies of four patients from a Bethlem myopathy family diagnosed through genetic linkage analyses but without identified mutations.29 These data suggest that collagen VI mutations cause abnormalities of the myofiber basal lamina by disrupting direct or indirect protein interactions between collagen VI and laminin β1. Consistent with this model, immunoelectron microscopy studies have shown localization of collagen VI to the ECM, immediately adjacent to the basal lamina.30 Recent yeast two-hybrid studies have suggested that laminin β1 and collagen VI may not directly interact30; however, yeast two-hybrid experiments are known to be problematic when applied to extracellular or membrane proteins.
Type VI collagen is thought to help anchor the basal lamina to the extracellular matrix by interacting with collagen IV, which in turn is thought to bind laminin β1.30 We did not detect immunostaining differences in the collagen IV protein. This implies that there may be protein(s) mediating the binding of laminin β1 and collagen VI in the myofiber, which are different than those in the capillaries, or that the binding of laminin β1 to collagen IV requires the participation of specific domains on collagen VI.
Two additional publications have reported reduced laminin β1 expression in patient muscle biopsies. One report described an age-dependent laminin β1 deficiency in families presenting with a slowly progressive, early onset, autosomal dominant myopathy with proximal muscle weakness, calf hypertrophy, contractures, spinal rigidity, and cardiac conduction defects.31 Another group reported laminin β1 deficiency in 3 of 18 patients whose clinical features were consistent with a slowly progressive, adult-onset LGMD.32 Notably, no mutation studies were conducted on any of these previously identified cases. Our studies suggest that the clinical spectrum of Bethlem myopathy is wider than originally described; therefore, it may be helpful to investigate collagen VI genes in patients showing laminin β1 protein deficiency and other AD-LGMD families with unknown genetic etiologies.
Our results provide a model for the pathogenesis for the muscle-specific phenotype caused by missense mutations of ubiquitously expressed collagen VI. These missense mutations of collagen VI, an extracellular matrix protein, may mediate a reduction in laminin β1 protein, specifically in the myofiber basal lamina. This basal lamina defect is likely particularly deleterious to myofibers, possibly leading to abnormal development and failed regeneration of muscle. Consistent with a functional defect of the basal lamina, we found no regenerating myofibers in patient biopsies, despite the presence of degenerating myofibers, and the abundant presence of these in dystrophic control subjects. As other studies in Bethlem myopathy have reported regenerating fibers and given the occurrence of central nuclei in a small minority of the biopsies studied here, we hypothesize that regeneration may occur, but only at a reduced rate relative to normal muscle. Further, more conclusive evidence for a causal relationship between the collagen VI abnormality and apparent absence of regeneration must await more directed experiments in animal models for this disease.
Acknowledgments
Supported by a graduate student fellowship from the Western PA American Heart Association (P.C.S.) and the NIH (National Institute of Neurological Disorders and Stroke 2 R01 NS29525–08) (E.P.H.).
Acknowledgment
The authors thank the families for their participation, Eva Engvall for her constructive criticism and willingness to supply antibodies, Jeffrey Trent and Dietrich Stephan for supplying expertise and equipment for automated linkage analyses, Molly Fuller for help with immunoblots, and Jeffrey H. Miner for supplying antibodies to laminin α4. The collagen VI, fibronectin, integrin β1 antibodies developed by E. Engvall, R.J. Klebe, and C.H. Damsky were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biologic Sciences, Iowa City, IA.
- Received July 31, 2001.
- Accepted November 6, 2001.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan