Hum Genet (1992) 89:480-484

9 Springer-Verlag1992

Original investigations Mild dominant osteogenesis imperfecta with intrafamilial variability: the cause is a serine for glycine al(I) 901 substitution in a type-I collagen gene Monica Mottes 1, Antonella Sangalli 1, Maurizia Valli 2, Macarena Gomez Lira t, Ruggero Tenni 3, Piera Buttitta 4, Pier Franco Pignatti 1, and Giuseppe Cetta 2 lIstituto di Scienze Biologiche, University of Verona, Strada Le Grazie, 1-37134 Verona, Italy 2Dipartimento di Biochimica, Centro per lo Studio delle Malattie del Tessuto Connettivo, University of Pavia, Pavia, Italy 3Istituto Pluridisciplinare di Fisiologia Umana, University of Messina, Messina, Italy 4Ospedale dei Bambini "G. Di Cristina', 1-90134 Palermo, Italy Received November 25, 1991 / Revised February 2, 1992

Summary. The molecular defect responsible for a case of mild osteogenesis imperfecta (OI) with repeated femural fractures was investigated. The proband and his mother, who presented minor OI signs but no bone fractures, were shown to produce normal and abnormal type-I procollagen molecules in their dermal fibroblasts. The molecular defect was localized in about half of the proband's pro a l ( I ) m R N A molecules by chemical cleavage with piperidine of hydroxylamine-reacted m R N A : cDNA heteroduplexes. The corresponding region was reversetranscribed and amplified by polymerase chain reaction (PCR). Cloning and sequencing of the amplified products revealed in both subjects a G-to-A transition in the first base of codon 901 of the ~l(I) triple helical domain, which led to a serine for glycine substitution. Allele-specific oligonucleotide hybridization to amplified genomic D N A from fibroblasts and leukocytes confirmed the heterozygous nature of both patients and proved the absence of mosaicism. The presence of the mutation was excluded in other healthy family members, who were reported to have bluish sclerae. The mild phenotypic outcome of this newly characterized mutation contradicts previous findings on glycine substitutions in the C-terminal region of collagen triple helix, most of which caused lethal OI.

Introduction The molecular biology of osteogenesis imperfecta (OI) has been elucidated through the characterization of many (more than 80 to date) different mutations in the type-I procollagen structural genes ( C O L I A 1 and COL1A2; for a review see Kuivaniemi et al. 1991). Mutations that Correspondence to: M. Mottes

affect quantitatively the production of type-I collagen are responsible for the mildest form of OI (OI type I according to Sillence et al. 1979). On the other hand, point mutations, splicing mutations, deletions, and insertions, which produce qualitatively abnormal collagen molecules, have been found to be the cause of a wider range of clinical forms. Point mutations that resulted in substitutions for glycine residues in the triple helical domain were found in several lethal cases, as well as in severe (OI type III) and also in a few mild/moderate cases (Byers et al. 1991; Kuivaniemi et al. 1991). Most of the authors have suggested a position-dependent effect of glycine substitutions within chains. So far this has been widely confirmed for glycine to cysteine substitutions in pro a l ( I ) chains (Starman et al. 1989). The most intriguing exceptions to the "phenotypic gradient" model are represented by the Gly-Ser substitutions mapped in the COL1A1 gene: serine substitutions at position 565 (Bateman et al. 1991) as well as 598 and 631 (Westerhausen et al. 1990) caused lethal phenotypes. A Ser-832 substitution was found in a moderate (OI type IV) case (Marini et al. 1989) while a Ser-844 substitution caused severe (type Ill) OI (Pack et' al. 1989). More Cterminal substitutions, Ser-913, 1003, and 1009, caused lethal, lethal, and extremely severe phenotypes, respectively (Cohn et al. 1990a). Here we present the molecular evidence for a new Gly- Ser substitution at position 901, which is caused by a G-to-A transition at nucleotide 3235 of the coding sequence of a pro a l ( I ) gene (COLIA1). This mutation was found in an 8-year-old boy affected with mild OI and, most strikingly, in his mother, who showed only mild features of the disease, but never suffered bone fractures. Our data indicate that a particular combination of nature and position of the substituting amino acid can play a crucial role in determining the phenotypic outcome.

481

Materials and methods

Clinical summary Clinical information on the family has been reported in a previous paper (Tenni et al. 1991). Briefly, the patient, now an 8-year-old boy, has been diagnosed as OI type IB. Due to repeated femoral bilateral fractures, he underwent intramedullar rodding at the age of 3. Since then the number of fractures has greatly reduced and he can walk without assistance. No hearing impairment has been detected so far. His mother, now 52, never reported bone fractures and, despite her very short stature (140 cm), she does not show any particular dysmorphism. Around age 40 she started showing mild hypoacusia and moderate osteoporosis. She was 44 at the time of the proband's pregnancy, she had never undergone any medical treatment, nor did she know she had a heritable connective tissue disorder. At the age of 48 she bore a second healthy son, who shows a bluish hue of the sclerae. This trait has been reported also in other maternal relatives, besides the proband and his mother, who do not show any other particular feature. A maternal uncle and his three sons were included in the molecular investigation with allele-specific oligonucleotides. Skin biopsies from the proband, parents, and controls were obtained after informed consent.

Detection of the mutation on R N A by chemical cleavage Cultures of skin fibroblasts were established and grown as reported (Tenni et al. 1990) and were used between the 3rd and the 15th passage. Total RNA was prepared from the patient's fibroblasts by the guanidinium isothiocyanate method. A normal eDNA probe corresponding to the C terminal region of proctl(I) chains (amino acid residues 822-1275; amino acid positions are conventionally numbered by assigning number l to the first glycine of the triple helical domain of ct chains) was obtained from a full-length al(I) cDNA (Tromp et al. 1988) after digestion with NcoI and EcoRI. The resulting 1365-bp fragment was 3' end-labeled with [a32P] dCTP and the Klenow fragment of DNA polymerase I (Boehringer). Approximately 10ng (3 x 105 cpm) of the double-stranded probe was mixed with 10 gg of total RNA. Denaturation was performed at 80~ for 10min in the buffer described previously (Lamande et al. 1989). Heteroduplex formation occurred during a subsequent 2-h incubation at 55~ Hydroxylamine and piperidine incubations were performed as previously reported (Valli et al. 1991).

Allele-specific oligonucleotide hybrid&ation Genomic DNA from peripheral blood samples and cultured dermal fibroblasts was prepared according to standard methods. A 500-ng sample was amplified by PCR using 20 pmol each of primers CDC22 (complementary to exon 44) and CDC26 (complementary to intron 45), the generous gift of C.D. Constantinou (Constantinou et al. 1990). Amplification was performed for 30 cycles consisting of denaturation (1 min at 94~ annealing (30 s at 52~ primer extension (45 s at 72~ with Perkin Elmer-Cetus Taq polymerase and buffer in the presence of 0.25 mM spermidine. Approximately 200 ng of the 571-bp product was ethanol precipitated, resuspended in 33% formaldehyde, boiled for 4 min, and immediately dot blotted on Hybond C (Amersham) duplicate filters. Allele-specific oligonucleotides carrying either the normal (N): 5'GGTCCTGTCGGCCCTGTTG 3', or the mutant (M): 5'GGTCCTGTCAGCCCTGTTG 3' sequence, were synthesized and end-labeled by [7-32P] ATP and T4 polynucleotide kinase (Pharmacia). Hybridization was carried out overnight at 52~ individual filters were rinsed briefly at room temperature with 6 x SSC, 0.5% sodium dodecyl sulfate (SDS) and then at 59~ for 10min. For the quantitation assay, serial twofold dilutions of amplified genomic DNA (from 100 ng to 12.5 ng) were dot blotted and hybridized as described above. The labeled N and M allelespecific oligonucleotides had comparable specific activities: 3.2 x 108 cpm/gg and 3.0 • 108 cpm/lag, respectively. After autoradiography, individual radioactive dots were excised from nitrocellulose filters and counted. Oligonucleotides and primers were prepared on a 391 Applied Biosystem DNA synthetizer.

Results A n 8-year-old b o y with r e p e a t e d f e m o r a l bilateral fractures a n d his m o t h e r , p r e s e n t i n g n o fractures, mild hypoacusia, a n d osteoporosis, were i n v e s t g a t e d in o r d e r to identify the genetic defect. Collagens type I f r o m fibroblast cultures d e r i v e d from skin biopsies were s t u d i e d ( T e n n i et al. 1991). O n the basis of the b i o c h e m i c a l data, the p r e s e n c e of a p o i n t m u t a t i o n l e a d i n g to a glycine s u b s t i t u t i o n in the

c D NA synthes& and amplification First strand eDNA was synthesized from 5 gg of total RNA with reverse transcriptase (BRL) and 100 pmol of the oligonucleotide primer: 5'CATCATCAGCCCGGTAGTAGCGGC 3', complementary to nucleotides 3638-3661 of the ct 1(I) eDNA sequence. One-tenth of the reaction mixture was used for polymerase chain reaction (PCR) amplification through 25 cycles using Taq polymerase (Perkin Elmer-Cetus). Each cycle consisted of denaturation at 94~ for 1 min, annealing at 60~ for I min, and primer extension at 72~ for 1 min. The second PCR primer: 5'GGTGAACCTGGCAAACAAGG 3' (corresponding to nucleotides 2940-2959) was deduced from the work of Lamande et al. 1989. The amplification mixture contained 40pmol of each primer and 4mM MgCI2. The expected PCR product of 721 bp was visualized on a 1% agarose gel.

Sequence determination of cloned cDNAs For sequence determination, PCR products were incubated with Sau3A, and a 468-bp fragment was recovered from 2% Nusieve (FMC) agarose and ligated to SmaI-linearized pUC19 plasmid vector. Several independent clones were isolated and sequenced using the Sequenase version 2.0 kit (US Biochemical Corp.)

Fig. 1. Mismatch analysis on mRNA :cDNA heteroduplexes. Total RNA from the proband's (P) and control (C) fibroblasts was hybridized to a 1365-bp fragment of al(I) cDNA. Heteroduplexes were reacted with hydroxylamine for the time indicated and subsequently treated with piperidine. Cleavage products were subjected to electrophoresis on a 5% polyacrylamide, 7 M urea gel and autoradiography. Sizes (in nt) of the fragments are indicated. The MW standard was a 32p-labeled 0X174 DNA digested with HaeIII (not shown)

482

Fig.2. Sequence of the ~1 Ser-901 mutation, cDNA sequence of mutant and normal COL1A1 alleles in the proband, and mutant COL1A1 allele in the mother. The sense strand sequence is shown: arrows point to the A-for-G substitution. The DNA sequence surrounding the mutation site is shown below

CB6 peptide of ~1(I) chains was suspected. The corresponding m R N A region was therefore subjected to molecular investigation by the chemical cleavage method. Heteroduplexes were formed between the patient's or control's m R N A and a wild-type c D N A probe. Subsequent treatment with hydroxylamine and piperidine produced a 240 + 5 bp fragment that was released from the 1365-bp p r o b e (Fig. 1). The same mismatch was found within the mother's m R N A (data not shown). The patient's and mother's m R N A s were reversetranscribed and amplified in vitro using a couple of oligonucleotide primers that had been designed to cover the whole CB6 peptide coding region, from amino acid 802 to amino acid 1042. On the basis of the chemical cleavage localization, sequence determination was restricted to only a portion of amplified D N A (468bp, nt 3193-3661) obtained after Sau3A digestion and cloning of the original P C R product (721bp). Several individual clones were isolated, and six clones for each subject were sequenced. For the proband, three of six clones showed a G - t o - A transition at nucleotide 3235 of the pro ~ l ( I ) gene sequence, which would lead triplet 901 in the triple helix to code for serine, instead of glycine. For his mother, two of six clones showed the same base substitution (Fig. 2). Other discrepancies with the published sequence (Bernard et al. 1983) were ascribed to errors in the original report. In all twelve clones sequenced, triplets 899 and 902 were C C T instead of CCC; and triplet 903 was G T T (Val), instead of G C T (Ala); as found also by L a m a n d e et al. (1989) and Constantinou et al. (1990).

Fig. 3. Pedigree and allele-specific oligonucleotide hybridization of the family members. Asterisks indicate healthy individuals with bluish sclerae. The proband is indicated by an arrow. Numbers below the dot-blot figure refer to pedigree numbers. 2L and 2F, amplified genomic DNA from leukocytes and fibroblasts, respectively, of the proband's mother (subject 2). c+, A positive control: the 721-bp PCR product obtained from the proband's cDNA utilized for the sequence determination shown in Fig. 3. C-, A negative control: amplified genomic DNA from a healthy individual unrelated to the family. M, Mutant; N, normal allele

Table 1. Ratio of mutant-to-normal allele in proband's and mother's genomic DNA. ASO, Allele-specific oligonucleotide; M, mutant, N, normal Amount ASO N (ng) (cpm)

ASO M (cpm)

Mean ratio M/N (_+ SD)

Proband's DNA (leukocyte)

100 50 25 12.5

6845 3 161 2 267 935

5707 3 567 2 097 992

0.986 _+0.2

Mother's DNA (leukocyte)

100 50 25 12.5

10769 4051 2 242 1144

11051 4294 2 049 1117

0.99 _+0.06

Mother's DNA (fibroblast)

100 50 25 12.5

10317 5196 2964 1304

11022 5838 2975 920

0.972 _+0.12

To investigate the presence of the Ser-901 mutation in the family, genomic D N A from leukocytes and fibroblasts of various subjects was amplified in vitro and tested with 32p-labelled allele-specific oligonucleotides. Only the proband and his mother were found to carry the Ser901 mutation, while the other family m e m b e r s examined were negative (Fig. 3).

Inheritance o f the mutation in the proband's family

In the clinical history of the family, the presence of blue sclerae had been reported, not only in the proband and in his mother, but also in family m e m b e r s who did not show any other OI sign.

Quantitation of mutant-to-normal allele ratio in the mother"

We investigated the possibility that somatic mosaicism might account for the very mild maternal O I phenotype. For this purpose quantification of allele-specific oligonuc-

483 leotide hybridizations was performed on equal amounts of amplified genomic DNA from different tissues (as estimated by fluorescence on an ethidium bromide stained agarose gel). The results reported in Table 1 indicate a 1 : 1 ratio of mutant-to-normal allele in all the tissues examined; therefore, no cellular mosaicism was detected in two of the mother's tissues.

Discussion

The data presented here provide the molecular evidence for a new COLIA1 mutation, Ser-901, which causes mild OI in a young patient and subclinical OI in his mother. The clinical phenotype was originally classified as OI type IB (Tenni et al. 1991). The blue sclerae and mild hypoacusia meet Sillence's (Sillence et al. 1979) criteria for OI type I, while the presence of dentinogenesis imperfecta (DI) and short stature, although not accompanied by bone deformities, are compatible with OI type IV. Based on the recent literature (Byers et al. 1991), it appears that the most common OI phenotype with blue sclerae and normal stature is classifiable as type I and attributable to quantitative defects in COL1A1, while other mild/moderate phenotypes, due to structural defects, are more commonly classified as type IV. We believe that our patients may better belong to this second group. As reported for other glycine substitutions, in this case a G-to-A transition occurring at a CpG dinucleotide is responsible for the disease. Mutations at CpG dinucleotides are found commonly in a variety of human genetic diseases (Cooper and Youssoufian 1988), and they have been found in two COL1A1 recurrent substitutions. This notwithstanding, CpG dinucleotides do not appear as preferential mutation sites in COL1A1 (Pruchno et al. 1991). A more precise indication of the biological relevance of CpG dinucleotide mutations in the collagen genes, will be provided in the future by the delineation of a close-to-saturation map of mutations. This serine-for-glycine substitution falls within the CB6 peptide of the ~1(I) chain: the majority of glycine substitutions so far mapped in this region were incompatible with life. The only three viable exceptions are represented by the already mentioned Ser-832 substitution (OI type IV), the Ser-844, and Ser-1009 substitutions (OI type III). In our case, a Ser-901 substitution results in a much milder form of the disease. The idea that serine, due to its small and polar side chain and its ability to form hydrogen bonds, can somehow be accommodated in an axial position of the triple helix is an appealing one. Maybe serine is more easily tolerated in some (but not all) axial positions than other bulkier residues and, in these circumstances, it exerts a less destructive effect on collagen fibril performance. Another important issue about different amino acid substitutions within collagen is the effect on triple helix stability. The serine substitution at position 598 altered the thermal unfolding of abnormal molecules (20~ as compared to the normal melting temperature, Tm = 40~176 while Ser-631, Ser-832, and Ser-844 sub-

stitutions did not cause any substantial change in the Tm of abnormal collagen chains. The different thermal stabilities of the mutated collagens led to the conclusion that thermal unfolding of the protein is not fully cooperative, but is accomplished through the micro-unfolding of a series of independent "cooperative blocks" (Westerhausen et al. 1990). Previous biochemical studies on this family (Tenni et al. 1991) revealed a Tm of 37~ (4~176 lower than normal) of the defective molecules. Interestingly in the nearest COOH terminal glycine substitution mapped, Cys-904 (OI type II), the abnormal molecules had a similar Tm (37~ Constantinou et al. 1989). It is possible, according to the microunfolding model, that the region around the 901 and 904 positions contributes more than others to the general stability of the helix. According to the B~ichinger model (B~chinger et al. 1992), Gly-901 an Gly-904 lie in a region of high relative stability, flanked by regions of lower stability. The model could therefore explain the low melting temperature of both mutants, as well as justify the high degree of overmodification due to the difficult renucleation downstream to the mutation. Pulse labeling experiments indicated a slower secretion of overmodified procollagen molecules in the proband and his mother, and demonstrated that the secretion of normal chains was also decreased (70%-90% of controls). The mutant collagen molecules were able to copolymerize with normal ones and be deposited in the matrix formed by cultured dermal fibroblasts in the presence of dextran sulfate (M. Valli et al., in preparation). The ratio of mutant-to-normal primers in the matrix was about 30%, which is lower than the expected ratio of 75% ; therefore, normal molecules were preferentially laid down. The decreased procollagen secretion shown by our patients, as well as by the patient Cys-904, could depend on the location of the mutation. The integrity of the subdomain in which both mutations fall could play a crucial role in type-I procollagen secretion. The impairment of procollagen secretion and the decreased stability of the triple helix led to a diminished amount of type-I collagen available in the extracellular matrix for fibril formation. We can speculate that the disturbance caused by a serine substitution in the region around amino acid 900 does not impair, as other substitutions do, the performance of collagen fibrils that contain abnormal molecules in a percentage lower than that synthesized by the cell. While we demonstrated that the proband inherited the mutation from his mother, we considered the possibility of her being a mosaic due to a new mutation that occurred during early stages in her development, which might explain her subclinical phenotype. The data on leukocytes and fibroblasts do not reveal any difference between mother and son; therefore, no evidence of mosaicism was found. A limitation to this conclusion is that we could not determine the ratio of mutant-to-normal allele in other maternal tissues, such as osteoblasts. The clinical heterogeneity between the two relatives, anyhow, is not as striking as in the other documented cases of intrafamilial variability due to somatic mosaicism, where very mild and lethal phenotypes coexisted (Wallis

484 et al. 1990; C o h n et al. 1990b; E d w a r d s et al. 1990; C o n s t a n t i n o u et al. 1990). I n this case, as in m a n y o t h e r O I families, the variability of clinical expression r e m a i n s une x p l a i n e d , a n d we can only speculate o n the possible effects of genetic b a c k g r o u n d differences b e t w e e n individuals, which could m o d u l a t e the p h e n o t y p e .

Acknowledgements. We

thank Drs. C. D. Constantinou and D.J. Prockop for the gift of oligonucleotide primers CDC22 and CDC26: the other oligonucleotides described were synthesized by P. Lorenzi. This work was supported by grants from the Italian National Research Council Target Projects "Genetic Engineering" and "Biotechnologies and Bioinstrumentation," ECCA on HCTD, and by a Veneto Region Medical Research Grant. The support and collaboration of the Associazione Italiana Osteogenesi Imperfetta is gratefully acknowledged.

References B~.chinger HP, Morris NP, Davies JM (1992) Thermal stability and folding of triple helices of interstitial collagens. Am J Med Genet (in press) Bateman JF, Hannagan M, Lamande S, Moeller I, Chan D, Cole WG (1992) Collagen I mutations in perinatal lethal osteogenesis imperfecta. Am J Med Genet (in press) Bernard MP, Chu ML, Myers JC, Ramirez F, Eikenberry EF, Prockop DJ (1983) Nucleotide sequences of complementary deoxyribonucleic acids for the pro al chain of human type I procollagen: statistical evaluation of structures that are conserved during evolution. Biochemistry 22 : 5213-5222 Byers PH, Wallis GA, Willing MC (1991) Osteogenesis imperrecta: translation of mutation to phenotype. J Med Genet 28 : 433-442 Cohn DH, Wallis G, Zhang X, Byers PH (1990a) Serine for glycine substitutions in the al(I) chain of type I collagen: biological plasticity in the Gly-Pro-Hyp clamp at the carboxyl-terminal end of the triple helical domain. Matrix 10 : 236 Cohn DH, Starman BJ, Blumberg B, Byers PH (1990b) Recurrence of lethal osteogenesis imperfecta due to parental mosaicism for a dominant mutation in a human type I collagen gene (COL1A1) Am J Hum Genet 46:591-601 Constantinou CD, Nielsen KB, Prockop DJ (1989) A lethal variant of osteogenesis imperfecta has a single base mutation that substitutes cysteine for glycine 904 of the cd(I) chain of tpye I procollagen: the asymptomatic mother has an unidentified mutation producing an over-modified and unstable type I procollagen. J Clin Invest 83 : 574-584 Constantinou CD, Pack M, Young SB, Prockop DJ (1990) Phenotypic heterogeneity in osteogenesis imperfecta: the mildly affected mother of a proband with a lethal variant has the same mutation substituting cysteine for ~I-glycine 904 in a type I procollagen gene (COLIA1). Am J Hum Genet 47:670-679 Cooper DN, Youssoufian H (1988) The CpG dinucleotide and human genetic disease. Hum Genet 78 : 151-155

Edwards MJ, Byers PH, Cohn DH (1990) Mild osteogenesis imperfecta produced by somatic mosaicism for a lethal mutation in a type I collagen gene. Am J Hum Genet 47 : A215 Kuivaniemi H, Tromp G, Prockop DJ (1991) Mutations in collagen genes: causes of rare and some common diseases in humans. FASEB J 5 : 2052-2060 Lamande SR, Dahl HHM, Cole WG, Bateman JF (1989) Characterization of point mutations in the collagen COLIAI and COL1A2 genes causing lethal perinatal osteogenesis imperfecta. J Biol Chem 264:15809-15812 Marini JC, Grange DK. Gottesman GS, Lewis MB, Koeplin DA (1989) Osteogenesis imperfecta type IV. J Biol Chem 264: 11893-11900 Pack M, Constantinou CD, Kalia K, Nielsen KB, Prockop DJ (1989) Substitution of serine for cd(I)-glycine 844 in a severe variant of osteogenesis imperfecta minimally destabilizes the triple helix of type I procollagen. J Biol Chem 264: 1969419699 Pruchno CJ, Cohn DH, Wallis GA, Willing MA, Starman BJ, Zhang X, Byers PH (1991) Osteogenesis imperfecta due to recurrent point mutations at CpG dinucleotides in the COL1A1 gene of type I collagen. Hum Genet 87 : 33-40 Sillence DO, Senn A, Danks DM (1979) Genetic heterogeneity in osteogenesis imperfecta. J Med Genet 16 : 101-116 Starman BJ, Eyre D, Charbonneau H, Harrylock M, Weis MA, Weiss L, Graham JM, Byers PH (1989) Osteogenesis imperfecta: the position of the substitution for glycine by cysteine in the triple helical domain of the pro ul(I) chain of type I collagen determines the clinical phenotype. J Clin Invest 84: 12061214 Tenni R, Rossi A, Valli M, Mottes M, Pignatti PF, Cetta G (1990) Anomalous cysteine in type I collagen: localisation by chemical cleavage of the protein using 2-nitro-5-thiocyanobenzoic acid and by mismatch analysis of cDNA heteroduplexes. Matrix 10 : 20-26 Tenni R, Biglino P, Dyne K, Rossi A, Filocamo M, Pendola F, Brunelli P, Buttitta P, Borrone C, Cetta G (1991) Phenotypic variability and abnormal type I collagen unstable at body temperature in a family with mild dominant osteogenesis imperfecta. J Inherited Metab Dis 14 : 189-201 Tromp G, Kuivaniemi H, Stacey A, Shikata H, Baldwin CT, Jaenisch R, Prockop DJ (1988) Structure of a full-length cDNA clone for the prepro ul(I) chain of human type I procollagen. Biochem J 253:919-922 Valti M, Mottes M, Tenni R, Sangalli A, Gomez Lira M, Rossi A, Antoniazzi F, Cetta G, Pignatti PF (1991) A de novo G to T transversion in a pro cd(1) collagen gene for a moderate case of osteogenesis imperfecta. J Biol Chem 266 : 1872-1878 Wallis GA, Starman BJ, Zinn AB, Byers PH (1990) Variable expression of osteogenesis imperfecta in a nuclear family is explained by somatic mosaicism for a lethal point mutation in the cd(I) gene (COL1A1). Am J Hum Genet 46 : 1034-1040 Westerhausen A, Kishi J, Prockop DJ (1990) Mutations that substitute serine for glycine al-598 and glycine c~1-631 in type I procollagen. J Biol Chem 265 : 13995-14000