Human Molecular Genetics Advance Access originally published online on November 24, 2004
Human Molecular Genetics 2005 14(2):279-293; doi:10.1093/hmg/ddi025
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Human Molecular Genetics, Vol. 14, No. 2 © Oxford University Press 2005; all rights reserved
Dominant collagen VI mutations are a common cause of Ullrich congenital muscular dystrophy
1Cell and Matrix Biology, Murdoch Childrens Research Institute and Department of Paediatrics, University of Melbourne, Royal Children's Hospital, Parkville, Victoria 3052, Australia, 2Department of Cell and Molecular Biology, Section of Molecular Pathogenesis, Lund University, 221 84 Lund, Sweden, 3Institute for Neuromuscular Research, The Children's Hospital at Westmead and Department of Paediatrics and Child Health, University of Sydney, Australia and 4Department of Paediatric Neurology, University Hospital, Leuven, Belgium
* To whom correspondence should be addressed. Tel: +61 393456650; Fax: +61 393457997; Email: shireen.lamande{at}mcri.edu.au
Received June 1, 2004; Accepted November 11, 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mutations in the three collagen VI genes COL6A1, COL6A2 and COL6A3 cause Bethlem myopathy and Ullrich congenital muscular dystrophy (UCMD). UCMD, a severe disorder characterized by congenital muscle weakness, proximal joint contractures and marked distal joint hyperextensibility, has been considered a recessive condition, and homozygous or compound heterozygous mutations have been defined in COL6A2 and COL6A3. In contrast, the milder disorder Bethlem myopathy shows clear dominant inheritance and is caused by heterozygous mutations in COL6A1, COL6A2 and COL6A3. This model, where dominant mutations cause mild Bethlem myopathy and recessive mutations cause severe UCMD was recently challenged when a patient with UCMD was shown to have a heterozygous in-frame deletion in COL6A1. We have studied five patients with a clinical diagnosis of UCMD. Three patients had heterozygous in-frame deletions in the N-terminal region of the triple helical domain, one in the
1(VI) chain, one in 2(VI) and one in 3(VI). Collagen VI protein biosynthesis and assembly studies showed that these mutations act in a dominant negative fashion and result in severe collagen VI matrix deficiencies. One patient had recessive amino acid changes in the C2 subdomain of 2(VI), which prevented collagen VI assembly. No collagen VI mutations were found in the fifth patient. These data demonstrate that rather than being a rare cause of UCMD, dominant mutations are common in UCMD, now accounting for four of the 14 published cases. Mutation detection in this disorder remains critical for accurate genetic counseling of patients and their families. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Collagen VI is a ubiquitous extracellular matrix protein that is made up of three distinct subunits,
1(VI), 2(VI) and 3(VI) (1
,2). Mutations in the three collagen VI genes COL6A1, COL6A2 and COL6A3 cause Bethlem myopathy (MIM 158810
[OMIM]
) and Ullrich congenital muscular dystrophy (UCMD, MIM 254090
[OMIM]
) and identify collagen VI as a critical component of skeletal muscle. The collagen VI chains all have a central triple helical domain of 335–336 amino acids with repeating Gly-Xaa-Yaa sequences, and C- and N-terminal globular domains that are predominantly composed of 
200 amino acid subdomains with homology to von Willebrand factor type A domains (3–6). Intracellular collagen VI assembly is complex, initially involving association of the three chains and folding into a triple helical heterotrimeric monomer. The monomers then align in an antiparallel staggered manner to form disulfide bonded dimers that then align laterally to form tetramers, also stabilized by disulfide bonds (1,2). Tetramers are secreted and associate end-to-end to form the characteristic beaded microfibrils.
Bethlem myopathy was the first disorder to be associated with collagen VI mutations (7). It is a relatively mild dominantly inherited disorder with slow progression, characterized by early childhood onset of proximal muscle weakness and commonly, flexion contractures that mainly affect the elbows, ankles and fingers (8). The 10 distinct mutations that have been characterized in the three collagen VI genes include single amino acid substitutions (7,9–11), splice site mutations resulting in in-frame deletions (12–14) and a COL6A1 splice site mutation that leads to a single base deletion from the mRNA, a codon reading frameshift and the introduction of a premature stop codon (15). The effect of some of the Bethlem myopathy mutations on collagen VI biosynthesis, assembly and structure has been studied. Both the 1(VI) premature stop codon mutation and the 1(VI) exon 14 deletion result in secretion of reduced amounts of structurally normal collagen VI, either because mutant chains are not produced (15) or because they are incorporated into monomers that fail to form dimers or tetramers and are not secreted (13). In contrast, chains containing glycine substitutions towards the N-terminal region of the triple helical domain are incorporated into tetramers and microfibrils, and in these cases Bethlem myopathy results from the presence of mutant collagen VI in the extracellular matrix (16).
UCMD shows a much more severe and progressive course than Bethlem myopathy. Its hallmarks are congenital muscle weakness, proximal joint contractures and marked distal joint hyperextensibility with normal intelligence (17–20). Many patients have congenital hip dislocations and torticollis, high-arched palate, prominent heels, scoliosis and a ‘sandpaper’ skin rash. Motor milestones are delayed, and some patients are never able to walk independently. The respiratory muscles are also affected and some patients require night-time ventilation support by their second decade. Recessive mutations, either homozygous or compound heterozygous, in COL6A2 and COL6A3 have now been described in nine UCMD patients (21–25). Eight of the 11 distinct mutations lead to the introduction of premature stop codons, and in three patients with recessive COL6A2 premature stop codon mutations, the 2(VI) mRNA undergoes nonsense-mediated mRNA decay leading to very low or no 2(VI) mRNA or protein expression (26). Nonsense-mediated mRNA decay presumably also occurs in the other UCMD patients with premature stop codon mutations although this has not yet been demonstrated experimentally. Since all three chains are required to form a collagen VI monomer, these patients are unable to assemble or secrete functional collagen VI protein and collagen VI is almost absent from muscle and fibroblasts (21). The other three recessive UCMD mutations are in-frame deletions (22,25). Their effect on collagen VI biosynthesis and assembly has not been studied in detail.
This model, where dominant collagen VI mutations cause mild Bethlem myopathy and recessive mutations cause the more severe disorder UCMD, was recently challenged when a patient with severe UCMD was shown to have a heterozygous in-frame deletion in the COL6A1 gene (27). No other collagen VI mutations were found. The genomic deletion was not present in either parent indicating that it was a de novo mutation in the patient. Protein biosynthetic studies provided critical insights into the consequences of this mutation that explain how a severe UCMD phenotype can result from a heterozygous mutation. Mutant 1(VI) chains, with a 33 amino acid deletion towards the N-terminal region of the triple helical domain, were able to assemble with normal 2(VI) and 3(VI) to form monomers, then dimers and tetramers, and some mutant-containing tetramers were secreted. The stoichiometry of the multistep collagen VI intracellular assembly process means that only 1/16 of the tetramers produced by this patient could be composed entirely of normal chains. Collagen VI was almost completely absent from the fibroblast matrix, indicating that the effect of the mutation was further amplified in the microfibril assembly step where the tetramers join end-to-end via their N-terminal regions. Thus, in this case, UCMD resulted from a heterozygous mutation that exerted a dominant negative effect on collagen VI structure and assembly.
In this study, we have examined a group of five patients with a clinical diagnosis of UCMD. Surprisingly, three patients had dominant heterozygous in-frame deletions in the N-terminal region of the triple helical domain. Detailed collagen VI protein biosynthesis and assembly studies showed that these mutations act in a dominant negative fashion and result in severe collagen VI matrix deficiencies. One patient had recessive amino acid changes in the C-terminal C2 subdomain of the 2(VI) chain, which prevented assembly of collagen VI monomers. No collagen VI mutations were found in the fifth patient. These data indicate that dominant mutations are a common cause of UCMD and emphasize the importance of mutation detection for accurate genetic counseling for patients and their families.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Five patients with UCMD
The five patients included in this study, UCMD1, 2, 3, 4 and 5, all presented with the signature clinical features of UCMD, including congenital muscle weakness, proximal joint contractures and marked distal joint hyperextensibility. UCMD1, 2, 4 and 5 have a striking similarity in their facial appearance. They have a round face, small chin and lips, irregular crowded teeth, large round eyes and prominent ears. In contrast, UCMD3 has small eyes and obvious facial muscle weakness. All five patients show signs of skin abnormalities. UCMD1, 2, 3 and 4 have abnormal scarring, UCMD2 and 4 have mild hyperkeratosis of the upper arms and UCMD5 has a reddish papular rash over the face and arms.
In addition to these features, UCMD1, the first child of healthy unrelated parents, had congenital bilateral hip dislocation, scoliosis and torticollis. Clinical evaluation at 6 months of age showed severe hyperlaxity and muscle atrophy. Creatine kinase levels were normal and electromyography (EMG) showed a myogenic pattern. An MRI scan at 5 years showed no brain abnormalities and cardiac function was also normal. This patient was able to sit at 11 months, and walked at 21 months. She required night-time ventilation assistance from age 7. At age 8, she was extremely dystrophic, with growth delay despite gastrostomy feeding, was able to walk independently for short distances but unable to climb stairs. She had severe hyperlaxity of the fingers, contractures of the hips, knees, elbows and ankles and severe progressive scoliosis.
UCMD2 is now 11 years old and the only child of healthy non-consanguineous parents. She presented with congenital hypotonia, proximal joint contractures, distal joint hyperextensibility, scoliosis, torticollis and bilateral hip dislocations. She was able to sit at 13 months, crawl at 15 months and walk at 17 months, although she lost independent ambulation at 10 years. Creatine kinase levels were normal, and a muscle biopsy at age 8 showed a dystrophic pattern with an increase in endomysial fibrosis and an increased number of internal nuclei. She has hip, knee, elbow and ankle contractures, marked hyperlaxity of the fingers, a progressive scoliosis and a severe decline in respiratory function, with vital capacity <40%.
UCMD3, the second child of healthy unrelated parents, is currently 12 years old and presented at birth with the typical clinical features of UCMD including muscle weakness, hip contractures and distal hyperlaxity. She was able to sit at 7 months and walk at 18 months with a waddling, unsteady gait. A muscle biopsy at 3 years showed variation in fiber size and increased endomysial fibrosis. Creatine kinase was normal. She developed early and progressive retractions of the Achilles tendons and has twice had surgery to release the contractures. At 7 years she was severely dystrophic with striking muscle hypotonia, marked hyperlaxity of the fingers, scoliosis and contractures of the ankles and hips. She has difficulty closing her mouth and feeding and has restricted respiratory function. Her intellectual development is normal.
UCMD4, the first and only child of healthy unrelated parents, presented at birth with hypotonia, joint laxity and hip dislocation. He was able to sit at 11 months and walk independently at 3 years, but lost ambulation at 4 years of age. Progressive joint contractures were present from age 2. Currently 19 years old, he has severe joint contractures, severe muscle wasting, scoliosis, torticollis, striking hyperlaxity of the fingers and is severely dystrophic, despite gastrostomy feeding. He required nocturnal ventilation from 7 years of age and currently also receives ventilation during part of the day. An EMG at age 3 showed a myogenic pattern, and muscle biopsy demonstrated a severe increase of fibrosis, variation in fiber size and small clusters of atrophic fibers. Creatine kinase was normal, and there was no heart involvement. Brain MRI was normal.
UCMD5 is the third child born to unrelated parents, and there is no history of neuromuscular disease. He presented in the neonatal period with subluxated and dysplastic hips, contractures at the ankles, knees and elbows, hyperextensible fingers and a fracture of the distal femoral epiphysis, likely sustained during delivery. He crawled independently at 16 months of age and walked with assistance by 2 years of age. He has normal cognitive development. Creatine kinase, echocardiography, respiratory function and EMG have remained normal. Muscle biopsy performed at 2 years of age demonstrated variability in fiber size, increased interstitial connective tissue and scattered regenerating fibers consistent with a mild dystrophic pathology. Currently 12 years of age, he has generalized poor muscle bulk and generalized muscle weakness, more marked proximally. He has a high palate, cervical kyphosis, a mild thoracic scoliosis, hyperextensible wrists and long tapering fingers, and very prominent heels.
UCMD patient fibroblasts produce normal levels of collagen VI mRNAs
Eight of the 11 recessive mutations that have been characterized in UCMD patients lead to the introduction of premature stop codons in the 2(VI) and 3(VI) mRNAs (21–25). The most common consequence of premature termination mutations in many genes is rapid degradation of the mutant mRNA by a process known as nonsense-mediated mRNA decay (28,29), and this is also the case for collagen VI. The three 2(VI) UCMD premature stop codon mutations that have been examined result in either the complete absence or a severe reduction in the level of the affected mRNA seen on northern blots (26). Nonsense-mediated mRNA decay was also seen in a Bethlem myopathy patient with a heterozygous 1(VI) premature stop mutation (15). To determine whether our UCMD patients had mutations that resulted in mRNA degradation, total RNA from fibroblasts was separated on a denaturing agarose gel, transferred to nitrocellulose, and hybridized simultaneously to radiolabeled 1(VI), 2(VI) and 3(VI) cDNAs. All three collagen VI mRNAs were present in these patients, and their relative expression levels were similar to that of the control (Fig. 1), indicating that unlike the majority of UCMD cases reported to date, these patients do not have premature stop codon mutations leading to mRNA decay or mutations that prevent transcription of one of the genes.
|
UCMD1, 2, 4 and 5 have a collagen VI deficiency
To confirm that a collagen VI protein deficiency was the underlying cause of the UCMD phenotype in these five patients, we examined the accumulation of collagen VI microfibrils in the fibroblast extracellular matrix. Confluent cultures were incubated for 2 days in the presence of sodium ascorbate to promote deposition of a collagenous extracellular matrix, then stained prior to fixation with a specific collagen VI antibody. This ensured that only extracellular collagen VI was detected. Collagen VI was almost completely absent from the UCMD2 extracellular matrix, and when compared with controls, UCMD1, 4, and 5 matrices contained significantly less collagen VI (Fig. 2A). Collagen VI staining in the UCMD3 fibroblast matrix was not significantly different to the controls. To determine whether collagen VI synthesis and secretion were altered in these patients, the amount of collagen VI in the fibroblast culture medium was compared by immunoblotting. Collagen VI tetramers were not detected in UCMD2 culture medium, UCMD1, 4 and 5 secreted reduced amounts of collagen VI and UCMD3 produced similar amounts of collagen VI to controls (Fig. 2B). Together, these data confirmed that at least four of the five patients, UCMD1, 2, 4 and 5, had collagen VI abnormalities.
|
Immunostaining of a muscle biopsy from UCMD5 with perlecan and collagen VI antibodies demonstrated that the collagen VI was abnormally localized in this patient (Fig 2C). In normal muscle, the collagen VI co-localized with perlecan in the basement membranes surrounding the muscle fibers. Collagen VI was also present in the interstitial extracellular matrix between fibers in control muscle. However, in UCMD5 collagen VI was not concentrated in the basement membranes, but was present in the interstitial extracellular matrix between individual muscle fibers and surrounding fiber bundles (Fig. 2C). This is consistent with the observed reduction in collagen VI specifically at the basement membrane, which has been reported in other UCMD patients (22
,27).
Abnormal collagen VI intracellular assembly in UCMD1, 2, 4 and 5
The finding that UCMD1, 2, 4 and 5 fibroblasts had normal amounts of 1(VI), 2(VI) and 3(VI) mRNAs but reduced extracellular collagen VI, suggested that underlying structural mutations affect the assembly and secretion of collagen VI tetramers. To examine collagen VI biosynthesis and assembly, confluent fibroblast cultures were labeled overnight with 35S-methionine and the collagen VI in the cell and medium fractions was immunoprecipitated with an antibody made to the recombinant 3(VI) N1 subdomain. The individual collagen VI chains were resolved on 3–8% gradient polyacrylamide gels under reducing conditions (Fig. 3A). In UCMD2, 3(VI) chains were immunoprecipitated from the cell fraction; however, co-immunoprecipitated 1(VI) and 2(VI) chains were barely detectable (Fig. 3A, lane 4), suggesting that the mutations in this patient affected the ability of the mutant chains to assemble with normal chains to form a triple helical collagen VI monomer. Consistent with this, only minute amounts of collagen VI were immunoprecipitated from UCMD2 medium (Fig. 3A and B, lane 11). In addition to the normal 1(VI) and 2(VI) chains, the UCMD1 cell sample, contained a faster migrating band (Fig. 3A, lane 3), suggesting that either the 1(VI) or 2(VI) chain contained a deletion. In the UCMD4 cell sample, the relative intensities of the 1(VI) and 2(VI) bands were reversed (Fig. 3A, lane 6), suggesting an underlying deletion in the 2(VI) chain which normally migrates more slowly than the 1(VI) chain (26). In both patients, co-immunoprecipitation of the mutant chains with the 3(VI) antibody indicated that they were able to associate with normal chains. No migration differences of the individual chains were obvious in UCMD3 or 5 samples.
|
The ability of the UCMD collagen VI to assemble into dimers and tetramers was assessed by analysis of the immunoprecipitated collagen VI on non-reducing composite agarose/acrylamide gels. In both control and UCMD cultures, tetramers were the predominant secreted form of collagen VI (Fig. 3B). However, UCMD1, 4 and 5 cells showed intracellular collagen VI assemblies that migrated with a molecular mass consistent with a collagen VI dimer (
1000 kDa, Fig. 3B, lanes 3, 6 and 7). This indicated that the mutations in these three patients did not affect the assembly of collagen VI into monomers or dimers, but interfered with the association of dimers to form disulfide bonded tetramers. Dimers were also apparent in the medium of UCMD1, 4 and 5 (Fig. 3B, lanes 10, 13 and 14), indicating that some of the collagen VI secreted by these patients is structurally abnormal and probably contains mutant chains. No differences in collagen VI biosynthesis and assembly could be detected in UCMD3 (Fig. 3A and B).
Microfibril assembly defects in UCMD1, 4 and 5
Following secretion, collagen VI tetramers associate end-to-end to form long thin microfibrils. To visualize the collagen VI tetramers that were secreted by the UCMD fibroblasts, medium from control and UCMD cultures was examined by negative staining electron microscopy. Microfibril formation was then quantified by analysis of a large number of micrograph fields (Fig. 4A). The occurrence of ‘microfibrils’ containing 1–10 tetramers is shown as a percentage of the total number of microfibrils. Though tetramers were seen in the medium from UCMD2 cultures, they were extremely rare and microfibril formation was not quantitated. In control fibroblasts, 20% of the microfibrils were single tetramers and microfibrils containing up to 10 tetramers were seen (Fig. 4A). In contrast, medium from UCMD1, 4 and 5 cells contained a larger proportion of single tetramers, accounting for 40–60% of the ‘microfibrils’, and microfibrils containing more than six tetramers were not seen (Fig. 4A). This indicated that the mutations in these three patients led to a reduction in the efficiency of end-to-end assembly of tetramers, and suggested that tetramers containing mutant chains were secreted.
|
A striking finding was that some of the UCMD3 collagen VI microfibrils had formed unusual large aggregates (Fig. 4B). These kinds of aggregates were not seen in control medium where only individual microfibrils were observed (Fig. 4B) or in the medium from the other UCMD cultures. This finding suggested that the collagen VI from UCMD3 was structurally abnormal even though no other biosynthetic or assembly abnormalities had been found. Not all of the UCMD3 collagen VI had formed aggregates, and when UCMD3 microfibril formation was quantified using fields with individual non-aggregated microfibrils, it was similar to the controls (Fig. 4A).
UCMD1, 4 and 5 have dominant heterozygous deletions in the collagen VI triple helical domain
The mutations in UCMD1, 4 and 5 were localized using RT–PCR to amplify 1(VI), 2(VI) and 3(VI) mRNAs. In UCMD1, an RT–PCR product corresponding to exons 7–15 of the COL6A1 gene contained a band of the expected size and a smaller mutant band (Fig. 5A). The normal and mutant products were of similar intensity confirming that the mutant mRNA was stable. Direct sequencing of the two RT–PCR products demonstrated that the smaller product lacked the 27 nucleotides encoded by COL6A1 exon 12 (data not shown). This nine amino acid deletion, residues 311–319 (amino acids 55–63 of the triple helical domain), maintained the normal reading frame and was consistent with the protein analysis that had detected a faster migrating 1(VI) or 2(VI) chain (Fig. 3A). Exon skipping is most often the result of a consensus splice site mutation, and so genomic DNA from the patient was PCR amplified and directly sequenced. A heterozygous IVS12+1 g>a mutation in the obligatory splice site was identified (Fig. 5B). All but one of the UCMD cases that have been characterized to date have a recessively inherited disorder and either homozygous or compound heterozygous mutations inherited from their unaffected heterozygous parents. However, in this case neither unaffected parent carried the mutation indicating that this was a de novo change in the child (Fig. 5B). We searched for a second heterozygous mutation by sequencing the entire coding regions of the 1(VI), 2(VI) and 3(VI) mRNAs but failed to find one. Four single nucleotide changes that resulted in amino acid substitutions were detected (Table 1) as well as eight silent nucleotide changes (data not shown). The silent changes were considered non-pathogenic polymorphisms for two reasons. Firstly, they were all found in controls. Secondly, the mutation analysis was based on mRNA amplification, and with the exception of the fragment containing COL6A1 exon 12, only a single RT–PCR product was obtained for each primer set. This means that we can be sure that the silent changes did not alter mRNA splicing as has been observed in many genes (30). The amino acid substitutions were all found in both the heterozgyous and homozygous state in control genomic DNA and could also be excluded as pathogenic mutations. Together, these data demonstrate that the UCMD phenotype in this patient results from a single dominant de novo COL6A1 mutation.
|
|
Similarly, in UCMD4 an
2(VI) RT–PCR produced a band of the expected size as well as a smaller band of equal intensity (Fig. 5C). Sequencing of the smaller band indicated that it lacked the 54 nucleotides encoded by COL6A2 exon 6 as well as the 27 most 5' nucleotides of exon 7 (data not shown). This mutation also maintained the normal reading frame and predicted the synthesis of a mutant 2(VI) chain with a 27 amino acid deletion: residues 268–294 (amino acids 13–39 of the triple helix). To characterize the underlying mutation in this patient, genomic DNA was amplified using a primer located within COL6A2 intron 4 and a primer spanning the exon 8/intron 8 boundary. Under optimal conditions this primer set should amplify a band of 2488 bp. This band was not amplified in either control or UCMD4 PCRs presumably because of its large size. However, in UCMD4 a 1156 bp product that was not seen in control PCRs was amplified. Direct sequencing revealed that this product contained a 1332 bp deletion including the most 3' 1169 bases of intron 5, all of exon 6 and intron 6, and the most 5' 27 bases of exon 7 (Fig. 5D). The deletion juxtaposed intron 5 and exon 7 sequences and created a 3' AG splice site motif at the junction (Fig. 5D). Comparison of the mRNA and genomic deletions demonstrated that the remaining 18 bp of exon 7 functioned as an exon and indicated that the novel splice site generated by the deletion was recognized by the pre-mRNA splicing machinery. The strength of this new 3' splice site sequence was estimated using maximum entropy scoring software that is based on large data sets of human splice sites (31,32). The ideal MaxENT score for a 3' splice site is 13.59 and theoretically, the larger the MaxENT value, the more efficient the splicing (31,32). The normal 3' splice site of COL6A2 intron 6 has a score of 10.68; however, the novel splice site generated by the deletion has a MaxENT score of only 0.97 and would not have been predicted to function as a splice site in the absence of the mRNA expression data.
To determine whether the unaffected parents of UCMD4 carried the deletion, genomic DNA was PCR amplified using a primer within COL6A2 intron 5 and a primer that spanned the exon 7/intron 7 boundary. This primer set was predicted to amplify a 1396 bp band from the normal allele and a 64 bp band from the mutant allele. As expected, both bands were amplified from UCMD4 genomic DNA although the larger band was relatively faint due to differences in the amplification efficiencies of the mutant and normal alleles (Fig. 5D). Only the 1396 bp normal band was amplified from the father's DNA; however, in addition to the normal product, a small amount of the 64 bp mutant band was amplified from the mothers DNA (Fig. 5D). The relative intensities of the two bands from UCMD4 and the mother were very different. In contrast to UCMD4, the major amplified product from the mother's DNA was the 1396 bp band from the normal allele, indicating that the normal allele was much more abundant than the mutant and that the mother was not heterozygous for the mutation. This finding was reproduced in multiple PCRs and suggested that despite being phenotypically normal, the mother of UCMD4 was a mosaic carrier of the deletion. To confirm that this was the only mutation in UCMD4, the entire coding regions of the three collagen VI mRNAs were sequenced. Nine silent nucleotide changes (data not shown) and nine changes that resulted in amino acid substitutions were detected (Table 1). Eight of the amino acid substitutions were found in controls and could be considered non-pathogenic; however, we did not find the 3(VI) R5Q substitution in 108 control chromosomes. This substitution is within the signal peptide and importantly, both sequences are predicted to function as signal peptides with a probability of 1.0 (SignalP 3.0). The ability of the alternative sequences to function as signal peptides was tested experimentally in in vitro coupled transcription and translation reactions performed in the presence of canine microsomal membranes. Both the normal and alternative signal peptides were able to direct translocation into the microsomal membranes and both signal peptides were cleaved (data not shown), providing strong evidence that that the R5Q substitution is a rare non-pathogenic polymorphism and does not contribute to the disease phenotype in this patient.
In UCMD5, an 3(VI) RT–PCR produced a band of the expected size as well as a smaller band (Fig. 5E) that lacked the 54 bases encoded by COL6A3 exon 16 (data not shown). This exon codes for residues 2058–2070 of the 3(VI) chain (amino acids 16–33 of the triple helix). When genomic DNA from the patient was PCR amplified and directly sequenced, a heterozygous IVS16+1 g>a mutation was identified in the obligatory 5' consensus splice site of intron 16 (Fig. 5F) that accounts for the observed exon skipping. Importantly, the mutation was not present in either of the unaffected parents (Fig. 5F), indicating that it was a de novo mutation in the patient. Sequencing of the coding regions of the three collagen VI mRNAs identified 17 silent nucleotide changes (data not shown) and eight changes that resulted in amino acid substitutions (Table 1). All of these changes were found in controls.
UCMD1, 4 and 5 thus have dominant heterozygous in-frame deletions in the triple helical domain of collagen VI. Although the mutations are in different chains, 1(VI) in UCMD1, 2(VI) in UCMD4 and 3(VI) in UCMD5, they are all located towards the N-terminal end of the triple helix, between amino acids 13 and 63 of the 335–336 residue triple helical domain.
Recessive COL6A2 mutations in UCMD2
The finding that only a minute amount of collagen VI was assembled and secreted by UCMD2 fibroblasts suggested that this patient had homozygous or compound heterozygous structural mutations in one of the collagen chains that prevented chain association. To characterize the mutations, the entire coding regions of the 1(VI), 2(VI) and 3(VI) mRNAs were amplified by RT–PCR and directly sequenced. Sixteen silent nucleotide changes were detected (data not shown) and excluded as the pathogenic mutations as they were all found in controls. In addition, there were 13 changes that resulted in amino acid substitutions (Table 1). Five of these, 1(VI) S890L, 3(VI) V2431D, T2927M, 2957insA and V2987M, have been previously reported as polymorphisms (10,27), and three, 1(VI) R850H, 2(VI) N227D, and 3(VI) P3011A, were commonly seen in controls. We did not find the remaining five changes in 108 control chromosomes. UCMD2 inherited 3(VI) A807T, A830S and K1088Q from her unaffected father who is also heterozygous for these changes (data not shown). They have also been found in a patient with a dominantly inherited muscular dystrophy where linkage to COL6A3 has been excluded (A. Lampe and K. Bushby, personal communication). These three changes are thus likely to be rare non-pathogenic polymorphisms. The two remaining changes, c.2510T>C (L837P) and c.2699delAAC (delN897), are both within the C2 subdomain of the 2(VI) chain and are likely to be the disease causing mutations in this patient. Consistent with this proposal, UCMD2 inherited the L837P mutation from her unaffected mother (Fig. 6A) and the N897 deletion from her unaffected father (Fig. 6B).
|
No detectable collagen VI or biglycan mutations in UCMD3
UCMD3 had normal levels of the three collagen VI mRNAs and no intracellular assembly or secretion defects were detected. However, the large collagen VI aggregates that were seen in the culture medium by electron microscopy suggested that there was an underlying collagen VI abnormality and so the entire coding regions of the
1(VI), 2(VI) and 3(VI) mRNAs were sequenced. No mutations were found in this patient. Thirteen non-pathogenic silent polymorphisms were present (data not shown), as well as 10 changes that resulted in amino acid substitutions (Table 1). All of these changes were commonly found in controls and, in addition, six have previously been reported as polymorphisms (10,27). Together, these data suggested that the UCMD phenotype in this patient was caused by a mutation in the gene for a protein that interacts with collagen VI and is important in microfibril structure. One such candidate is the small leucine rich repeat proteoglycan biglycan. To determine whether a biglycan mutation was present in UCMD3 the seven coding exons of the BGN gene including the consensus splice sites were PCR amplified and directly sequenced. No mutations were found. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
UCMD has long been considered a recessively inherited condition (18
), and until recently all of the collagen VI mutations that had been characterized in this disorder were either homozygous or compound heterozygous recessive mutations in the COL6A2 or COL6A3 genes (21–25). In contrast, the milder disorder Bethlem myopathy shows clear dominant inheritance within families and is caused by heterozygous mutations in COL6A1, COL6A2 or COL6A3 (7,9,11–15). This collagen VI disease paradigm was challenged by the identification of a de novo heterozygous COL6A1 mutation in a patient with classical severe UCMD (27). Our work now demonstrates that three out of a group of five UCMD patients have heterozygous dominant mutations and indicates that rather than being a rare cause of UCMD (one out of 10 patients), dominant mutations are common in UCMD, now accounting for four of the 14 published cases. The incidence of recessive UCMD may have been overestimated because the parents of four of the 10 previously reported cases were consanguinous (21,25). This finding will have a major impact on the genetic counseling advice that parents of UCMD patients receive, and mutation detection is clearly critical for accurate advice to be provided.
The dominant UCMD mutations were all heterozygous in-frame deletions in the triple helical domain of collagen VI. While the mutations were in different chains, they were all located close to the N-terminal region of the triple helix, between amino acids 13 and 63 of the 335–336 residue helical domain (Fig. 7A), and this explains their similar effect on collagen VI intracellular and extracellular assembly. These dominant UCMD mutations identify, for the first time, a protein domain important for collagen VI tetramer formation, and provide new insights into the interacting protein domains that are involved in intracellular and extracellular collagen VI assembly. Interactions between the C-terminal globular domains of 1(VI), 2(VI) and 3(VI) lead to the formation of heterotrimeric monomers (16,33). Monomers then align in a staggered antiparallel manner with a 75 nm overlap in the triple helix to form dimers that are stabilized by disulfide bonds involving cysteine residues within the triple helix of one monomer and the C-terminal globular domain of the other (1) (Fig. 7B). Both 1(VI) and 2(VI) have a cysteine at amino acid 89 of the triple helix. Although it is not yet known which one of these cysteines forms the disulfide bridge that stabilizes the dimer, deletion of 1(VI) exon 14, which encodes amino acids 79–96 of the helix and includes cysteine 89, completely prevents dimer formation, indicating that the structure of the triple helix in this region is critical for assembly of dimers (13,27). Recent in vitro assembly experiments suggest that amino acids 106–113 of the 2(VI) helix may play a role in dimer formation by interacting with the 2(VI) C2 subdomain; however, this has yet to be confirmed in cells expressing all three collagen VI subunits (34). Our finding that dimers are able to form from monomers lacking amino acids 55–63 of the 1(VI) helical domain, 13–39 of the 2(VI) helical region or 16–33 of the 3(VI) helix, further delineates the region involved in dimer formation; triple helical amino acids 13–63 are not required, 79–96 are critical and 106–113 may also be important. This proposal is supported by the finding that dimers can form from monomers containing 1(VI) chains in which amino acids 13–45 of the triple helical region are deleted (27).
|
Collagen VI tetramers are generated by the parallel alignment of two dimers in register. Electron microscopy shows that the major, if not exclusive, contact regions are located in the outer 30 nm segments of the triple helix (35
). In the tetramer, these regions cross over in a scissor-like fashion, and are linked by a disulfide bridge between the cystseines at amino acid 50 of the 3(VI) helical domain. Surprisingly, glycine substitutions at amino acid 49 of the 1(VI) and 3(VI) helix, which interrupt the repeating Gly-X-Y sequence and are likely to perturb the structure of the helix in this region, do not affect assembly of disulfide bonded tetramers (16). Disturbing the structure of the helix just C- or N-terminal to 3(VI) Cys50 by the incorporation of chains with in-frame deletions does, however, impair tetramer formation. This suggests that a large portion of the N-terminal region of the triple helix that does not overlap in the dimer is important for efficient tetramer assembly.
These biosynthetic and assembly studies also allow us to understand why these dominant mutations result in the severe UCMD phenotype rather than the milder disorder Bethlem myopathy. Initial association of the chains to form 1(VI)/2(VI)/3(VI) heterotrimeric monomers results in two populations of monomers, one containing the normal chain and one containing the mutant (Fig. 7B). Mutant monomers are able to form dimers and this would be expected to produce three populations of dimers. One population, representing one-fourth of the dimers, is composed of two normal monomers that are able to form tetramers and are secreted. The remaining three-quacters of the dimers contain either one or two mutant monomers and their assembly into tetramers is impaired (Fig. 7B). The UCMD cells secrete reduced amounts of collagen VI and this indicates that a substantial portion of the mutant-containing molecules are degraded inside the cell. Some mutant-containing tetramers are secreted, however, and they have a reduced ability to associate end-to-end into microfibrils and so the effects of the mutation are amplified even further. In the microfibrils, the N-terminal domains of adjacent tetramers are predicted to overlap in a junctional complex, allowing multiple potential interactions with each other, with the C-terminal domains and with the N-terminal region of the triple helix of the adjacent tetramers (1,35). One region that is necessary for microfibril formation is the N-terminal N5 subdomain of the 3(VI) chain (36), although its interacting partner has not been identified. Reduced end-to-end association of tetramers in the UCMD patients with deletions in the N-terminal region of the triple helix demonstrates that this region of the helix is also important in the interactions leading to microfibril formation. These dominant UCMD mutations thus act at the level of tetramer and microfibril formation and because of this they exert a severe dominant negative effect on collagen VI structure and assembly and result in a dramatic reduction in the amount of extracellular collagen VI that accounts for the severe phenotype. In contrast, the Bethlem myopathy deletion of amino acids 79–96 of the 1(VI) triple helical region acts at the level of dimer formation (13,27). Both mutant and normal monomers are assembled, but the mutant monomers are unable to form dimers and are degraded within the cell. This results in a 50% reduction in the amount of collagen VI, but the protein that is produced is structurally normal and the phenotype relatively mild.
These studies also highlight the importance of screening cDNA for understanding and predicting the consequences of changes in genomic DNA. The most common consequence of a mutation in the highly conserved 5' GT and 3' AG splice site motifs is exon skipping (37), and this was the case for the 5' GT to AT splice site mutations that were identified in UCMD1 and 5. However, splice site mutations do not always result in single exon skipping and other outcomes include activation of cryptic splice sites, creation of a pseudo-exon within an intron, intron retention and skipping of multiple exons (37,38). Predicting the strength of splice sites and the outcome of splice site mutations is not always straightforward and the COL6A2 genomic deletion in UCMD4 emphasizes this. The deletion brings together sequences from intron 5 and exon 7 and provides a potential 3' AG consensus splice site (Fig. 5). A polypyrimidine tract is usually present immediately upstream of 3' splice sites (32). However, the new intron 5 3' splice site does not have a pyrimidine-rich sequence as only eight out of the 20 most 3' bases are pyrimidines, and as a result it has a low MaxENT splice site strength score that predicts it would not function as a splice site. In the absence of biological data, this genomic deletion would have been expected to result in the deletion of exon 6 and exon 7 sequences from the mRNA. The RT–PCR data clearly demonstrates that this is not the case and that the new low-scoring 3' splice site is used, leading to inclusion of the remaining 27 bases of exon 7 in the mRNA.
Only one of our five patients, UCMD2, had recessive collagen VI mutations. Both mutations, L837P and delN897, were in the C-terminal C2 subdomain of the 2(VI) chain (Fig. 7C) and are the first two C2 subdomain mutations to be characterized. Almost no collagen VI was assembled and secreted by UCMD2 fibroblasts suggesting that the mutations prevented the initial chain association event. There is mounting evidence that the collagen VI C-terminal globular domains are critical for chain association. The C1 subdomain of each chain contains a putative -helical coiled-coil oligomerization domain (39). These heptad repeat motifs have been shown to drive trimerization in transmembrane collagens (40,41), and in a reporter molecule containing a collagen-like sequence that does not spontaneously trimerize (39). The best experimental evidence for C-terminal chain association in collagen VI comes from analysis of naturally occurring and introduced triple helical glycine mutations. A glycine substitution towards the C-terminal region of the 3(VI) triple helix severely impairs association of the three chains and folding of the triple helix while similar mutations towards the N-terminal region of the triple helix do not affect assembly (16). One possibility is that in addition to sequences in the helix and the C1 subdomain, the C2 subdomain also contains sequence motifs that are important for chain association and the UCMD2 mutations disrupt these sequences and prevent the interactions that are necessary for assembly. This seems unlikely since constructs of the three collagen VI chains that contained just the C1 subdomain at the C-terminal region formed triple-helical monomers in in vitro transcription/translation experiments (33). A more plausible explanation is that the mutations cause the C2 subdomain to misfold, these misfolded chains are recognized by endoplasmic reticulum quality control mechanisms (42,43), are rapidly degraded and are therefore not available for assembly. Alignment of the protein sequence of the 2(VI) C2 subdomain with the von Willebrand factor A3-domain whose crystal structure has been determined (44) indicates that 2(VI) L837 is in the ßA strand, one of the six ß-strands that form the central ß-sheet of the A domain. Proline residues are disfavored in ß-sheets as they introduce a turn into the protein backbone (45,46), and this suggests that the L837P mutation is likely to destabilize the ßA-strand and result in misfolding of the C2 subdomain. The deleted residue, 2(VI) N897, is within the consensus sequence for N-linked oligosaccharide addition, NLT. While it is not known if this glycosylation site is normally utilized, N-linked oligosaccharides are commonly important for correct protein structure and their removal can result in misfolding and intracellular degradation (47). The absence of collagen VI triple helical monomers indicates that these mutant chains do not take part in protein assembly and so they do not have a dominant negative effect on collagen VI assembly or structure. This accounts for the lack of a disease phenotype in the parents of UCMD2 who each carry one of the mutations.
Collagen VI microfibrils in the medium of UCMD3 cultures formed unusual large aggregates that were distinctly different from the individual microfibrils seen in controls and the other UCMD patients. This suggested that the collagen VI was structurally abnormal; however, we were unable to find a collagen VI mutation in UCMD3, even though the entire coding regions of the three collagen VI mRNAs, were sequenced. UCMD3 also produced normal levels of 1(VI), 2(VI) and 3(VI) mRNAs, indicating that there were no underlying premature stop codon mutations or mutations that prevented transcription of one of the collagen VI genes. The collagen VI genes have previously been excluded as the cause of UCMD in three families (19). A muscle biopsy was available from one of these patients and showed normal collagen VI staining (19), reminiscent of the normal levels of collagen VI staining in the UCMD3 fibroblast extracellular matrix. Aggregation of the collagen VI microfibrils in UCMD3 points to proteins that interact with collagen VI and are important in microfibril structure as the most obvious candidate disease genes in this patient. One such candidate was the small leucine rich repeat proteoglycan, biglycan. Biglycan binds close to the N-terminal region of the collagen VI triple helix (48) and also interacts with the critical skeletal muscle protein -dystroglycan (49), thus potentially forming a bridge between -dystroglycan and collagen VI. In addition, the biglycan null mouse has a muscular dystrophy phenotype (50), further implicating this molecule as a candidate disease gene in human muscular dystrophies. However, we did not find a biglycan mutation in UCMD3. It is thus likely that the UCMD phenotype in this patient is caused by a mutation in another, as yet unidentified gene.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
The patients
Five patients with a diagnosis of UCMD (UCMD1, 2, 3, 4 and 5) and their unaffected parents were studied. UCMD1, 2, 3 and 4 were from Belgium and UCMD5 from Australia. Dermal fibroblast cultures were established from skin biopsies of the patients and were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (Life Technologies) in 5% CO2 at 37°C as previously described (51
). Genomic DNA was extracted from peripheral blood samples obtained from the parents.
RNA extraction and northern blot
Total RNA was isolated from confluent fibroblasts using RNeasy® (Qiagen). Total RNA (5 µg) was separated under denaturing conditions on a 1% agarose gel and transferred to nitrocellulose (52). The filter was hybridized simultaneously to [-32P]dCTP-labeled 1(VI), 2(VI) and 3(VI) cDNA probes [P18, P1 and P24 (3)], washed, and specific hybridization visualized by autoradiography.
Immunostaining of fibroblasts
Human fibroblasts were grown to confluence in four-well chamber glass slides (Nunc) and then supplemented daily for 2 days with 0.25 mM sodium ascorbate. To visualize collagen VI that had been deposited into the extracellular matrix, cell layers were washed with cold PBS then incubated with the collagen VI antibody 3C4 (53) for 1 h at 4°C. Bound antibody was detected using AlexaFluor® 488 goat anti-mouse IgG (Molecular Probes), cell nuclei were stained with 0.2 µg/ml 4,6-diamidino-2-phenylindole (DAPI) and the cell layers were then fixed with 4% paraformaldehyde for 15 min. Slides were mounted in FluorSaveTMreagent (Calbiochem) and then viewed and photographed using a Zeiss fluorescence microscope.
Immunostaining of muscle biopsy
Frozen sections (8 µm) were cut from patient muscle biopsies using a CM1900 Cryostat (Leica), and were mounted onto Superfrost® Plus slides (Menzel-Glaser). The sections were blocked with 2% (w/v) bovine serum albumin (BSA) in PBS for 15 min at room temperature, and then incubated at 4°C overnight with a primary antibody solution containing rabbit anti-collagen VI antibody (70-XR95, Fitzgerald Industries International, 1 : 20 000 dilution) and rat anti-perlecan antibody (clone A7L6, Chemicon International, 1 : 80 000 dilution) in 2% BSA. Slides were washed with PBS (3x5 min) then incubated for 1 h at room temperature with a secondary antibody solution containing a 1:250 dilution of CyTM-3 conjugated AffiniPure goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) and a 1 : 200 dilution of AlexaFluor® 488 conjugated goat anti-mouse IgG (Molecular Probes) in 2% BSA. The sections were then washed with PBS (3x5 min) and coverslipped with Immumount mounting media (Thermo-Shandon). Images were obtained using an Olympus BX50F4 microscope at 40x magnification.
Collagen VI immunoblotting
Human fibroblasts were seeded at the same density in 10 cm2 dishes then grown until 2 days post-confluency. Cultures were incubated for 18 h in 0.5 ml of serum-free medium containing 0.25 mM sodium ascorbate, then 20 µl aliquots of the medium were separated under non-reducing conditions on composite gels. Proteins were electrophoretically transferred to ImmobilonTM-P Transfer Membrane (Millipore) and collagen VI detected with a polyclonal antibody (70-XR95, Fitzgerald Industries International) that recognizes all three collagen VI chains (9), peroxidase-conjugated swine anti-rabbit immunoglobulins (DAKO) and an enhanced chemiluminescence kit (ECL, Amersham Pharmacia Biotech).
Collagen VI biosynthetic labeling and analysis
Primary skin fibroblasts were grown to confluence in 10 cm2 dishes, incubated overnight in the presence of 0.25 mM sodium ascorbate, and then biosynthetically labeled for 18 h with 100 µCi/ml [35S]methionine (Tran35S-label 1032 Ci/mmol, ICN Pharmaceuticals, Inc.) in 750 µl of methionine-free and serum-free Dulbecco's modified Eagle's medium containing 0.25 mM sodium ascorbate. The medium was removed to a sterile tube, and protease inhibitors were added to the following final concentrations: 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride, 20 mM N-ethylmaleimide, and 5 mM EDTA. The cell layer was solubilized in 50 mM Tris/HCl, pH 7.5, containing 5 mM EDTA, 150 mM NaCl, 1% Nonidet P-40, 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride, and 20 mM N-ethylmaleimide. Cell lysates and medium samples were clarified by centrifugation and then collagen VI was immunoprecipitated overnight at 4°C using an antibody raised in rabbits to the recombinant 3(VI) N1 subdomain, and 100 µl of 20% protein A-Separose (Amersham Biosciences). The protein A-sepharose beads were washed twice with NET buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.1% Nonidet P-40) and then once with 10 mM Tris–HCl, pH 7.5, 0.1% Nonidet P-40 for 30 min each. Immunoprecipitated collagen VI was eluted into gel loading buffer at 65°C for 15 min and analyzed following reduction with 25 mM dithiothreitol by SDS–PAGE on 3–8% (w/v) gradient polyacrylamide gels. Collagen VI triple helical monomers, dimers and tetramers were analyzed on 2.4% (w/v) acrylamide, 0.5% (w/v) agarose composite gels