Homozygosity for a splice site mutation of the COL1A2gene yields a non-functional proα2(I) chain and an EDS/OI clinical phenotype
- *MRC Connective Tissue Genetics Group, Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QP, UK
- †Dingley Child Development Centre, Battle Hospital, Oxford Road, Reading, Berks RG3 1AG, UK
- ‡Institute of Medical Genetics, University Hospital of Wales, Heath Park, Cardiff CF4 4XN, UK
- Dr Nicholls,
Editor—Type I collagen is the major structural protein of skin, bone, tendon, ligaments, and cornea. It is a heterotrimer of two α1(I) chains and one α2(I) chain. Mutations in one of its two structural genes (COL1A1,COL1A2) underlie the inherited disorders osteogenesis imperfecta (OI) and Ehlers-Danlos syndrome types VIIA and B (EDS VIIA, EDS VIIB).1 Mutations which inhibit the processing of the protein precursor, procollagen, cause the ligamentous laxity and extreme joint hypermobility of EDS VII. Structural abnormalities and/or reduced production of type I collagen lead to the increased bone fragility, thin skin, blue sclerae, dentinogenesis imperfecta, and presenile deafness of osteogenesis imperfecta. The OI phenotype can vary from perinatally lethal to mild disease depending upon the nature of the mutation. Among the mild group are patients who are heterozygous for a non-functional COL1A1allele.2-5 Homozygosity for a non-functionalCOL1A1 allele appears to be incompatible with life.6 Several years ago we described a male infant who was totally deficient in α2(I)chains owing to homozygosity for a non-functional COL1A2allele.7-9 He had the severe, progressively deforming type of osteogenesis imperfecta (OI type III). His heterozygous third cousin parents, however, showed no overt clinical symptoms but were considered to be prematurely osteoporotic followingx ray examination. Subsequently, two further patients lacking proα2(I) chains from type I collagen were reported.10 11 However, their clinical phenotypes lacked any of the skeletal abnormalities of osteogenesis imperfecta but included generalised joint hypermobility, skin hyperextensibility, and scarring typical of the Ehlers-Danlos syndrome. We have now identified another patient homozygous for a mutation yielding non-functionalCOL1A2 alleles whose clinical phenotype showed generalised joint hypermobility and foot deformities reminiscent of Ehlers-Danlos syndrome associated with pale blue sclerae and a mild increase in bone fragility characteristic of osteogenesis imperfecta.
The proband is the first child of first cousin parents (fig 1). Marked ligamentous laxity and muscle hypotonia had been first noted at her premature (28 weeks' gestation) birth. When ambulation was significantly delayed she was investigated for muscular dystrophy but, following a normal muscle biopsy, the motor delay was attributed to her ligamentous laxity. On clinical examination at 9 years of age she was of average height but with marked generalised joint laxity, pes planus, and valgus heels leading to a secondary shortening of the achilles tendon. Her skin was normal, her sclerae were pale blue, and there was dental overcrowding but no dentinogenesis imperfecta. The limbs were straight and normal both on examination and byx ray. There was a history of recurrent patellar dislocations and fractures of the skull, clavicle, fingers (three), and a toe following separate minimal traumas. An asymptomatic early systolic murmur present on initial presentation had spontaneously resolved. Her cardiovascular system, ECG, and echocardiography were normal. Her mother showed some joint laxity, but her father was not available for examination. She has two younger sisters who have both progressed normally. A paternal cousin was reported to have joint hypermobility and fractures following minimal injury.
Skin fibroblast cultures were established from both the proband and her mother. The cells were labelled with 14C-proline (1 μCi/ml).12 Proteins were harvested from the medium by ethanol precipitation and from the cells by lysis with 0.5 mol/l acetic acid/0.5% triton X-100 at 4°C. Medium and cell layer procollagens were converted to collagens by limited pepsin digestion of the native protein.12 Procollagen and collagen chains were analysed on 5% polyacrylamide gels containing 2 mol/l urea using the tris-glycine-SDS buffer system.13 Procollagen samples were reduced with 1% mercaptoethanol before electrophoresis. The SDS-polyacrylamide gels of the procollagens secreted by the proband's cells showed proα1(I) and α1(I) bands but no evidence of proα2(I), pNα2(I), or α2(I) bands (fig 2). After limited pepsin digestion the collagens from both medium and cell layer of the proband's cells showed only α1(I) chains. These migrated more slowly on SDS-PAGE than those from a control cell line (fig 2) indicative of increased post-translational modification. The chains of other collagens, for example, type III and type V, from the patient's cells were not affected.
Cytoplasmic RNA, isolated from fibroblasts by the NP 40 lysis technique,14 was reverse transcribed and PCR amplified to yield several overlapping fragments encoding the whole α2(I) chain. Products were analysed on agarose or polyacrylamide gels. The proband's RNA gave a product for each primer pair; however one product was slightly smaller than the corresponding fragment from the control RNA (fig 3A). Although this smaller fragment was the predominant product from the proband's RNA, two minor higher molecular weight bands visible in polyacrylamide gels suggested heteroduplex formation and implied the existence of a second molecular species in the RT-PCR reaction. This fragment pattern was consistent in replicate reverse transcriptions of duplicate RNA preparations from the proband (fig 3A). The proband's abnormal product was inserted into M13 vectors and the sequences of multiple clones compared to that of the wild type fragment from control RNA. All of the proband's clones examined had an identical mutant sequence which, compared to the normal control, was missing the last 17 bp of COL1A2 exon 46 (fig 3B). The resulting frameshift introduces a premature termination codon after three residues. To characterise the minor components in the proband's RT-PCR reaction, both higher molecular weight bands were excised from a polyacrylamide gel, eluted, and reamplified. Each gave two bands of approximately equal intensity on ethidium bromide stained polyacrylamide gels; one band corresponded to the deleted product, the other was the size of the normal fragment (fig 3C). Several M13 clones of each product were sequenced and yielded two distinct sequences, one showing the 17 bp deletion and the other the normal α2(I) mRNA sequence.
To assess the relative abundance of the different α2(I) mRNAs, equal amounts of the mother's, proband's, and control RNAs were RT-PCR amplified for only a limited number of cycles with one of the primers 33P end labelled. The products were analysed by polyacrylamide gel and autoradiography. In the mother, the normal sized product was predominant with the deleted product representing only a very minor component. The patient gave a similarly weak signal for the deleted product (fig 3D).
Genomic DNA from the proband, her mother, and a control were PCR amplified using primers located upstream of the deletion in exon 46 and downstream in exon 47 of COL1A2. Each DNA yielded single products of equal size. These were ligated into the M13 derivatives BM20, BM21 (Boehringer-Mannheim) for sequencing. Every clone from the proband had a C instead of T as the second base of IVS 46 (fig 4A). Clones of the mother's DNA showed two changes from the control. Half her clones showed the C for T+2 substitution seen in the proband, the others had the normal IVS46 T+2sequence but had an A instead of a G at position +42 of IVS 46. The T+2→C+2 mutation creates a new restriction site for the enzymes BbvI [GCAGC(N)8] and TseI [GC(A/T)GC]. These enzymes were used to digest amplified genomic DNA from the proband's blood and fibroblasts and blood from her mother, her two sisters, and four unrelated controls. The digests were electrophoresed on polyacrylamide gels (fig 4B). Both enzymes confirmed that the proband was homozygous for the mutation and the mother was a heterozygous carrier. The older of the proband's two sisters was also shown to be a carrier of the mutant allele but the second sister had two normal alleles.
The mutation that we have identified in this proband substitutes the obligate T+2 of the donor splice site -ctGTAAGT- ofCOL1A2 IVS46 with a C. When such mutations arise in collagen genes, the usual outcome is skipping of the preceding exon from the mRNA to yield a shortened protein which retains its reading frame and the collagen triplet (Gly-X-Y) amino acid sequence. However, in this case there was no evidence for an exon skipped product; instead the predominant product arose from use of a cryptic donor splice site using the second and third bases of a glycine codon 17 bp upstream of the normal splice junction as the obligate -GT- dinucleotide. Calculations according to the Shapiro and Senapathy rules15 yield a score of 79 for the normal and 60.8 for the mutant splice site sequences. The cryptic site generates a score of 72.8. There is a potential -GT- sequence closer to the original splice site which yields a score of 62 but this is not used. Thus, for most of the mature mRNA the last 17 bp of exon 46 are deleted and the resultant frameshift introduces a new termination codon just three codons further downstream. The weak signal from the proband and the lower yield of the deleted product from the heterozygous mother obtained by semi-quantitative RT-PCR of equal amounts of RNA indicate that mutant messenger RNA is less abundant than normal. Messenger RNAs containing premature termination codons are frequently destroyed by a process known as nonsense mediated decay.16 If any of the mutant message was translated, the proα2(I) chains would lack a C propeptide domain and thus would not be incorporated into triple helical molecules. As a consequence, the proband produces a type I collagen homotrimer molecule containing three α1(I) chains instead of the normal heterotrimer with two α1(I) and one α2(I) chains. The α1(I) homotrimer triple helices form less readily than heterotrimer molecules and are thus subjected to increased post-translational modification leading to slow migration of the chains on SDS-PAGE gels. Mutation of the obligate -GT- of a donor splice site should totally abolish splicing at that junction; however, in this case the faint heteroduplexes observed in the proband's RT-PCR contained some normal product, suggesting that a limited amount of normal splicing had occurred at the mutated site despite it having a lower score than alternative cryptic sites in the vicinity. Any protein produced from these normal transcripts was undetectable in the cell culture analyses.
Patients with a total α2(I) deficiency are extremely rare, this latest patient is only the fourth to be described, and they show markedly different clinical phenotypes. In the original case, homozygosity for a 4 bp deletion within the α2(I) C propeptide coding region meant that the proα2(I) chains synthesised could not be incorporated into type I collagen molecules and α1(I) trimers were formed.8 The patient had an extremely severe, progressively deforming osteogenesis imperfecta with multiple fractures, severe skeletal deformities, and popcorning of the epiphyses.7 His heterozygous third cousin parents who made half normal amounts of proα2(I) chains suffered from mild joint laxity but appeared clinically normal untilx ray examination indicated premature osteoporosis. Two later accounts of Japanese patients10 11 reported an absence of α2(I) chains in skin fibroblast cultures from patients who presented with joint hypermobility, hyperextensible skin, and scarring characteristic of the Ehlers-Danlos syndrome. Aortic regurgitation was also observed but no skeletal abnormalities were recorded. In the first of these cases the parents were second cousins suggesting possible homozygosity10; in the second the parents were unrelated and compound heterozygosity was postulated.11 Although a reduction in α2(I) mRNA was reported in one of these patients,11 no gene defects were defined. Why apparently similar protein abnormalities should produce such diverse clinical phenotypes is unclear. One could speculate that the unknown mutations in the Japanese patients affected a tissue specific gene promoter or enhancer limiting the expression of the defect to soft tissues and sparing the skeleton. However, in the other two cases the α2(I) defects result from structural abnormalities within theCOL1A2 gene and should be expressed wherever the gene is active. As such they might be expected to generate similar clinical phenotypes but our latest proband is much more mildly affected than the original patient and shows none of his skeletal deformity or severe osseous fragility. This could be the result of epigenetic factors or perhaps the minute amount of normal α2(I) detected as heteroduplexes in the RT-PCR reactions, although not as protein in cell cultures, is sufficient to ameliorate the clinical phenotype from a severe OI to a much milder EDS/OI.
The authors gratefully acknowledge the assistance of Mrs M Laidlaw in maintaining the fibroblast cultures.
↵§ Present address: Clinical Molecular Genetics Laboratory, Addenbrooke's Hospital, Hills Road, Cambridge, UK