Ehlers-Danlos syndrome type VIIB. Deletion of 18 amino acids comprising the N-telopeptide region of a pro-alpha 2(I) chain

A patient with Ehlers-Danlos syndrome Type VIIB was found to have an interstitial deletion of 18 amino acids in approximately half of the pro-alpha 2(I) chains of Type I procollagen. Analysis of pepsin-solubilized tissue and fibroblast collagen revealed an abnormal additional chain, alpha 2(I)', which migrated in sodium dodecyl sulfate-5% polyacrylamide gel electrophoresis between the normal alpha 1(I) and alpha 2(I) chains. The apparent ratio of normal alpha 1(I):mutant alpha 2(I)':normal alpha 2(I) was 4:1:1. Procollagen studies and enzyme digestion studies of native mutant collagen suggested defective removal of the amino propeptide. Sieve chromatography of CNBr peptides from purified alpha 2(I)' chains revealed the absence of the normal amino telopeptide fragment CB 1 and the appearance of a larger new peptide of approximately 60 residues (CB X). Compositional and sequencing studies of this peptide identified normal amino propeptide sequences. However, the most carboxyl-terminal tryptic peptide of CB X differed substantially in composition and sequence from the expected and was found to have an interstitial deletion of 18 amino acids corresponding to the N-telopeptide of the pro-alpha 2(I) chain. This deletion removes the normal sites of cleavage of the N-proteinase and also removes a critical cross-linking lysine residue. The 18 amino acids deleted correspond exactly to the residues encoded by exon 6 of the pro-alpha 2(I) collagen gene (COL 1 A2), and, therefore, the protein defect may be due to a genomic deletion, or alternatively, an RNA splicing defect.     

Multiexon deletion in the procollagen III gene is associated with mild Ehlers-Danlos syndrome type IV

We have characterized a deletion of approximately 9 kilobases which spans from intron 33 to exon 48 of one pro-alpha 1 (III) collagen allele in a patient with Ehlers-Danlos syndrome type IV. The mutation results in the production of an in-frame species of mRNA which lacks the sequences corresponding to residues 595-1,008 of the triple-helical domain. Thus, half of the pro-alpha 1 (III) chains synthesized by the patient's fibroblasts are nearly 30% shorter than normal. The procollagen III molecules composed of either three normal length or three shortened chains are thermally stable and efficiently secreted. In contrast, the procollagen III molecules that contain one or two shortened chains are unstable and are not secreted. Failure to secrete unstable molecules and a residual functional role of the shortened but stable homotrimers may explain the somewhat milder phenotype of this individual compared with that of another Ehlers-Danlos type IV patient bearing a deletion of similar size in the amino-terminal portion of the alpha 1 (III) collagen chain.     

Isolation and characterization of a cDNA clone for the amino-terminal portion of the pro-alpha 1(II) chain of cartilage collagen

We have isolated a cDNA clone (pRcol 2) which is complementary to the 5'-terminal portion of the rat pro-alpha 1(II) chain mRNA. A synthetic oligonucleotide was used both as a primer for cDNA synthesis and as a probe for screening a cDNA library. The probe was a mixture of sixteen 14-mers deduced from an amino acid sequence present in the amino-terminal telopeptide of the rat cartilage alpha 1(II) chain. This primer was chosen so that the resulting cDNA would contain the sequence of the 5' end of the mRNA. The nucleotide sequences of the cDNA were determined and compared with that of three other interstitial procollagen chain mRNAs (pro-alpha 1(I), pro-alpha 2(I), and pro-alpha 1(III) chain mRNA). pRcol 2 contains a 521-base pair (bp) insert, including 153 bp of the 5' untranslated region plus 368 bp coding for the signal peptide, the amino-terminal propeptide, and a part of the telopeptide. The signal peptide of the type II collagen chain is composed of about 20 amino acids. There is little homology between the amino acid sequence of the signal peptide in the pro-alpha 1(II) chain and that of three other interstitial procollagen chains. The NH2-terminal propeptide is deduced to contain short nonhelical sequences at its amino and carboxyl ends and an internal helical collagenous domain comprising 25 repeats of Gly-X-Y with one interruption. There is a strong conservation of the amino acid sequence of the carboxyl-terminal part of the NH2-terminal propeptide in the pro-alpha 1(II), pro-alpha 1(I), and pro-alpha 2(I) chains. Type II collagen mRNA does not contain a sequence corresponding to a uniquely conserved nucleotide sequence around the translation initiation site which occurs in mRNA for other procollagen chains.     

A base substitution at a splice site in the COL3A1 gene causes exon skipping and generates abnormal type III procollagen in a patient with Ehlers-Danlos syndrome type IV

The dermis of a child with Ehlers-Danlos syndrome type IV (EDS-IV) contained about 11% of the normal amount of type III collagen and cultured dermal fibroblasts produced a reduced amount of type III procollagen which was secreted poorly. Type III collagen produced by these cells contained normal and abnormal alpha-chains and cyanogen bromide peptides. The site of the structural defect in the abnormal alpha 1 (III) chains was localized to the region of Met797, which is at the junction of the two carboxyl-terminal CB5 and CB9 cyanogen bromide peptides. Chemical cleavage of heteroduplexes formed between EDS-IV mRNA and a normal cDNA clone covering the CB5 and CB9 region showed that about 100 nucleotides were mismatched. Sequencing of amplified and cloned cDNA spanning the mutant region revealed a 108 nucleotide deletion corresponding to amino acid residues Gly775 to Lys810. The deleted nucleotide sequence corresponded to sequences that, by analogy to the organization of the type I collagen genes, should be precisely encoded by exon 41 of the COL3A1 gene. Sequencing of amplified genomic DNA, prepared using disimilar amounts of primers specific for exons 41 and 42, displayed a base substitution (G-to-A) in the highly conserved GT dinucleotide of the 5' splice site of intron 41. Normal sequences were also obtained from the normal allele. It is likely that the GT-to-AT transition at the splice donor site of intron 41 generated an abnormally spliced mRNA in which sequences of exon 40 and 42 were joined together with maintenance of the reading frame. The corresponding peptide deletion included the cyanogen bromide cleavage site Met797-Pro798 and the mammalian collagenase cleavage site at Gly781-Ile782. These losses account for the resistance of EDS-IV collagen to cyanogen bromide and mammalian collagenase digestion. Cultured fibroblasts produced normal homotrimer, mutant homotrimer, and mixed heterotrimer type III collagen molecules. The mutant homotrimer molecules were the major pepsin-resistant species and about 69% of the alpha 1(III) mRNA was in the mutant form.     

Use of R-loop mapping for the assessment of human collagen mutations

R-loop mapping of DNA:RNA hybrids formed between mutant pro-alpha 2(I) mRNAs and appropriate human pro-alpha 2(I) genomic clones was employed to define the location of mutations which result in the synthesis of shortened pro-alpha 2(I) chains in skin fibroblasts from two variants of osteogenesis imperfecta. Hybridization of the genomic clone NJ-9 with pro-alpha 2(I) mRNA from a patient with a mild atypical form of the disease resulted in the identification of mutant pro-alpha 2(I) mRNA lacking the sequences which correspond to exon 11 of the pro-alpha 2(I) collagen gene. Exon 11, a 54-base pair exon, encodes amino acids 73 to 90 of the alpha 2(I) chain. Also, electron microscopy of R-loop structures formed between the genomic clone NJ-1 and mRNA from a variant with a perinatal lethal form of osteogenesis imperfecta visualized pro-alpha 2(I) mRNAs which did not hybridize to the sequences of exon 28, a 54-base pair exon coding for amino acids 448 to 465 of the alpha 2(I) chain. Moreover, nuclease S1 mapping of the variant's mutant pro-alpha 2(I) mRNA, employing the human pro-alpha 2(I) cDNA clone Hf-15, confirmed the location of the mismatch to the sequences corresponding to exon 28. Although the data do not determine the exact nature of the mutations, they illustrate the use of R-loop mapping as an alternative approach to S1 mapping analysis for the detection and localization of collagen mRNA deletions.     

A 19-base pair deletion in the pro-alpha 2(I) gene of type I procollagen that causes in-frame RNA splicing from exon 10 to exon 12 in a proband with atypical osteogenesis imperfecta and in his asymptomatic mother

Previous observations established that fibroblasts from a proband with atypical osteogenesis imperfecta synthesized about equal amounts of normal pro-alpha 2(I) chains and shortened pro-alpha 2(I) chains of type I procollagen. The pro-alpha 2(I) chains were shortened because of an in-frame deletion of most or all of the 18 amino acids encoded by exon 11 of the pro-alpha 2(I) gene. Here it was demonstrated that one of the proband's alleles for the pro-alpha 2(I) gene contained a 19-base pair deletion at the junction of intervening sequence 10 and exon 11 that produced an RNA splicing defect. Probe protection experiments did not reveal any evidence for use of cryptic splice sites, and they suggested that the major species of abnormally spliced pro-alpha 2(I) mRNA in the proband's fibroblasts was completely spliced from exon 10 to 12. The defect in RNA splicing is unusual among RNA-splicing mutations in producing an abnormal polypeptide chain that is used for protomer assembly. Since the probe protection experiments showed the same defect in the mRNA from the fibroblasts of the asymptomatic mother, the mutation was inherited in an autosomal dominant manner but showed variable phenotypic expression in the proband's family.     

Heterozygosity for a large deletion in the alpha 2(I) collagen gene has a dramatic effect on type I collagen secretion and produces perinatal lethal osteogenesis imperfecta

We characterized a de novo 4.5 kilobase pair deletion in the paternally derived alpha 2(I) collagen allele (COL1A2) from a patient with perinatal lethal osteogenesis imperfecta. The intron-to-intron deletion removed the seven exons which encode residues 586-765 of the triple helical domain of the chain. Type I procollagen molecules that contain the mutant pro-alpha 2(I) chain have a lower than normal thermal stability, undergo increased post-translational modification amino-terminal to the deletion junction, and are retained within the rough endoplasmic reticulum. The block to secretion appears to result from improper assembly of the triple helix, apparently a consequence of a disruption of charge-charge interactions between the shortened pro-alpha 2(I) chain and normal pro-alpha 1(I) chains. The lethal effect may be due to decreased secretion of normal collagen and secretion of a small amount of abnormal collagen that disrupts matrix formation.     

A base substitution at the splice acceptor site of intron 5 of the COL1A2 gene activates a cryptic splice site within exon 6 and generates abnormal type I procollagen in a patient with Ehlers-Danlos syndrome type VII.

The dermal type I collagen of a patient with Ehlers-Danlos type VIIB (EDS-VIIB) contained normal alpha 2(I) chains and mutant pN-alpha 2(I)' chains in which the amino-terminal propeptide (N-propeptide) remained attached to the alpha 2(I) chain. Similar alpha 2(I) chains were produced by cultured dermal fibroblasts. Amino acid sequencing of tryptic peptides, prepared from the mutant amino-terminal pN-alpha 2(I) CB1' peptide, indicated that five amino acids, including the N-proteinase (the specific proteinase that cleaves the procollagen N-propeptide) cleavage site, had been deleted from the junction of the N-propeptide and the N-telopeptide (the nonhelical domain at the amino-terminus of the alpha chains of fully processed type I polypeptide chains) of the mutant pro-alpha 2(I)' chain. The corresponding 15 nucleotides, which were deleted from approximately half of the alpha 2(I) cDNA polymerase chain reaction products, of the alpha 2(I) cDNA polymerase chain reaction products, were encoded by the +1 to +15 nucleotides of exon 6 of the normal alpha 2(I) gene (COL1A2). These 15 nucleotides were deleted in the splicing of alpha 2(I) pre-mRNA to mRNA as a result of inactivation of the 3' splice site of intron 5 by an AG to AC mutation and the activation of a cryptic AG splice acceptor site corresponding to positions +14 and +15 of exon 6. Loss of the N-proteinase cleavage site explained the persistence of the pN-alpha 2(I)' chains in the dermis and in fibroblast cultures. Collagen production by cultured dermal fibroblasts was doubled, possibly due to reduced feedback inhibition by the N-propeptides. In contrast to previously reported cases of EDS-VIIB, Lys5 of the N-telopeptide was not deleted and appeared to take part in the formation of intramolecular cross-linkages. However, increased collagen solubility and abnormal extraction profiles of the mutant type I collagen molecules indicated that collagen cross-linking was abnormal in the dermis. The proband and her son were heterozygous for the mutation. It is likely that the heterozygous loss of the N-proteinase cleavage site, with persistence of a shortened N-propeptide, was the major factor responsible for the EDS-VIIB phenotype.     

A mutation in the pro alpha 2(I) gene (COL1A2) for type I procollagen in Ehlers-Danlos syndrome type VII: evidence suggesting that skipping of exon 6 in RNA splicing may be a common cause of the phenotype.

Fibroblasts from a proband with Ehlers-Danlos syndrome type VII synthesized approximately equal amounts of normal and shortened pro alpha 2(I) chains of type I procollagen. Nuclease S1 probe protection experiments with mRNA demonstrated that the pro alpha 2(I) chains were shortened because of a deletion of most or all of the 54 nucleotides in exon 6, the exon that contains codons for the cleavage site for procollagen N-proteinase. Sequencing of genomic clones revealed a single-base mutation that converted the first nucleotide of intron 6 from G to A. Therefore, the mutation was a change, in the -GT-consensus splice site, that produced efficient exon skipping. Allele-specific oligonucleotide hybridizations demonstrated that the proband's mother, father, and brother did not have the mutation. Therefore, the mutation was a sporadic one. Analysis of potential 5' splice sites in the 5' end of intron 6 indicated that none had favorable values by the two commonly employed techniques for evaluating such sites. The proband is the fourth reported proband with Ehlers-Danlos syndrome VII with a single-base mutation that causes skipping of exon 6 in the splicing of RNA from either the COL1A1 gene or COL1A2 gene. No other mutations in the two type I procollagen genes have been found in the syndrome. Therefore, such mutations may be a common cause of the phenotype. The primers developed should be useful in screening for the same or similar mutations causing the disease.     

Multiexon deletion in an osteogenesis imperfecta variant with increased type III collagen mRNA

Recently, the dermal fibroblasts (ATCC CRL 1262) of a lethal perinatal variant of osteogenesis imperfecta have been used for the first molecular characterization of a collagen gene defect (Chu, M. L., Williams, C. J., Pepe, G., Hirsch, J. L., Prockop, D. J., and Ramirez, F. (1983) Nature (Lond.) 304, 78-80). These studies revealed that the patient was heterozygous for an internal deletion of approximately 500 base pairs in the pro-alpha 1(I) collagen gene, consistent with previous investigations indicating that CRL 1262 fibroblasts equally synthesized a normal and a shortened pro-alpha 1(I) chain (Barsh, G. S., and Byers, P. H. (1981) Proc. Natl. Acad. Sci. U.S.A. 78, 5142-5146). Cloning and analysis of the affected allele of CRL 1262 has now indicated that the deletion is contained between two introns of the pro-alpha 1(I) gene and results in the elimination of three exons of the triple helical domain. Furthermore, the termini of the rearrangement are located within two short inverted repeats suggesting that the self-complementary nature of these DNA elements may have favored the formation of a DNA secondary structure intermediate which, in turn, served as substrate for the deletion. Evidence are also presented for an elevated Type III collagen mRNA content in the patient fibroblasts.     

Heterozygous mutation in the G+5 position of intron 33 of the pro-alpha 2(I) gene (COL1A2) that causes aberrant RNA splicing and lethal osteogenesis imperfecta. Use of carbodiimide methods that decrease the extent of DNA sequencing necessary to define an unusual mutation.

Cultured skin fibroblasts from a proband with osteogenesis imperfecta were found to synthesize normal and shortened alpha 2(I) chains of type I procollagen. A cDNA library was prepared using mRNA isolated from the proband's fibroblasts. Partial nucleotide sequencing of five clones demonstrated that two clones lacked the 54 base pairs (bp) of coding sequences found in exon 33 of the pro-alpha 2(I) gene (COL1A2). To reduce the amount of nucleotide sequencing required, heteroduplexes were prepared from two of the clones, one normal and the other lacking exon 33, and reacted with a water-soluble carbodiimide under conditions in which nonbase-paired G and T nucleotides are specifically modified by the reagent. Analysis of the heteroduplexes by immunoelectron microscopy suggested that the sequence variation near the codons of exon 33 was the only sequence difference in the cDNA clones. Amplification of cDNA from the proband by polymerase chain reaction gave products of two sizes, one of the expected size for the normal sequence and the other of the expected size for a product lacking the 54 bp in exon 33. To define the mutation in genomic DNA, a 1.6-kilobase region spanning exons 32 and 34 was amplified by the polymerase chain reaction and DNA heteroduplexes were prepared from the products. The heteroduplexes were treated with a water-soluble carbodiimide and then used as templates for primer extension under conditions in which extension terminates at the site of a carbodiimide-modified base. The results suggested a mismatch near the exon-intron boundary of exon 33 and a second mismatch near the 3' end of intron 33. Nucleotide sequencing of the polymerase chain reaction products revealed a single-base substitution in one allele that changed the moderately conserved G at position +5 of the 5' splice site of intron 33 to an A. In addition, there was an apparently neutral single-base substitution that placed both a G and T at position +661 of intron 33. The results provide only the third example of a mutation in the G at the +5 position of an intron that causes aberrant RNA splicing. Also, the results demonstrate that use of techniques involving carbodiimide modification of DNA heteroduplexes can reduce the amount of nucleotide sequencing necessary to define mutations in large and complex genes.     

Guinea pigs possess a highly mutated gene for L-gulono-gamma-lactone oxidase, the key enzyme for L-ascorbic acid biosynthesis missing in this species.

Guinea pigs cannot synthesize L-ascorbic acid because of their deficiency in L-gulono-gamma-lactone oxidase, a key enzyme for the biosynthesis of this vitamin in higher animals. In this study we isolated the L-gulono-gamma-lactone oxidase gene of the rat and the homologue of this gene of the guinea pig by screening rat and guinea pig genomic DNA libraries in lambda phage vectors, respectively, using a rat L-gulono-gamma-lactone oxidase cDNA as a probe. Sequencing analysis showed that the amino acid sequence of the rat enzyme is encoded by 12 exons and that all the intron/exon boundaries follow the GT/AG rule. On the other hand, regions corresponding to exons I and V were not identified in the guinea pig L-gulono-gamma-lactone oxidase gene homologue. Other defects found in this gene homologue are a deletion of the nucleotide sequence corresponding to a 3' 84-base pair part of rat exon VI, a 2-base pair deletion in the remaining exon VI-related region, and nonconformance to the GT/AG rule at one of the putative intron/exon boundaries. Furthermore, a large number of mutations were found in the amino acid-coding regions of the guinea pig sequence; more than half of them lead to nonconservative amino acid changes, and there are three stop codons as well. Thus it is clear that the guinea pig homologue of the L-gulono-gamma-lactone oxidase gene exists as a pseudogene that randomly accumulated a large number of mutations without functional constraint since the gene ceased to be active during evolution. On the basis of the neutral theory of evolution, the date of the loss of L-gulono-gamma-lactone oxidase in the ancestors of the guinea pig was roughly calculated to be less than 20 million years ago.