2 Med Genet 1995;32:284-289

Syndrome of the month

Perinatal lethal osteogenesis imperfecta

W G Cole, R Dalgleish

AbstractPerinatal lethal osteogenesis imperfecta isthe result ofheterozygous mutations oftheCOLlAl and COL1A2 genes that encodethe al(I) and a2(I) chains of type I colla-gen, respectively. Point mutations result-ing in the substitution of Gly residues inGly-X-Y amino acid triplets of the triplehelical domain of the al(I) or a2(I) chainsare the most frequent mutations. Theyinterrupt the repetitive Gly-X-Y structurethat is mandatory for the formation of astable triple helix. Most babies have theirown private de novo mutation. However,the recurrence rate is about 7% owing togermline mosaicism in one parent.The mutations act in a dominant neg-

ative manner as the mutant proa chainsare incorporated into type I procollagenmolecules that also contain normal proachains. The abnormal molecules arepoorly secreted, more susceptible to de-gradation, and impair the formation oftheextracellular matrix. The collagen fibres

are abnormally organised and min-eralisation is impaired.The severity of the clinical phenotype

appears to be related to the type of muta-tion, its location in the a chain, the sur-rounding amino acid sequences, and thelevel of expression of the mutant allele.

(J Med Genet 1995;32:284-289)

Osteogenesis imperfecta (OI) is a geneticallydetermined disorder of the connective tissues.The major phenotypic features include bonefragility, dentinogenesis imperfecta, bluesclerae, deafness, and ligamental laxity. It haslong been recognised to be clinically het-erogeneous with mild, moderate, severe, andlethal phenotypes. The Sillence classificationtakes these varying phenotypes into account.'

Department ofMedical Genetics,University of Torontoand The ResearchInstitute,The Hospital for SickChildren,555 University Avenue,Toronto, Ontario,MSG lX8,CanadaW G Cole

Department ofGenetics,University ofLeicester,Leicester LEI 7RH,UKR Dalgleish

Correspondence to:Professor Cole.

Figure 1 OI-IIA: radiograph showing severe osteopenia,continuously beaded ribs, platyspondyly, and broadcrumpled long bones.

Figure 2 OI-IIB: radiograph showing severe osteopeniaand broad crumpled femora. The ribs which are slenderwith expanded ends contain few fractures.

284

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from

Perinatal lethal osteogenesis imperfecta

Figure 3 OI-IIC: radiography showing minimalossification of the skeleton.

Type I is the common mild form, type II is theperinatal lethal form, type III is a severe form,and type IV is a moderately severe form. Wewill focus on OI type II (OI-II).

Clinical heterogeneityBabies are classified as having OI-Il if they diein the perinatal period. However, the babies

show clinical and radiological heterogeneitywhich appears to be established early in thepregnancy.2 The radiographic appearances areused to subclassify OI-II into groups A to Cor into groups I to V."3 The Sillence sub-classification will be used here.

OI-IIAThe babies are small for the period ofgestation.The face is hypoplastic with micrognathia, deepblue-grey sclerae, and a relatively large cal-varium, which is moulded and soft. The chestis small and the abdomen protuberant. Thelimbs are short and bowed. The hips are flexedand abducted. One or more limbs may be heldin a severely deformed position. The radio-graphs show generalised osteopenia. The di-aphyses of the long bones and ribs are as wideas their growth plates owing to the lack ofmetaphyseal remodelling (fig 1). The thin andcrumpled cortices give the long bones a con-tinuously beaded appearance. The pelvis andshoulder girdle have a similar appearance.There is generalised platyspondyly. The skulland face are severely porotic. Numerous worm-ian bones are present in the skull sutures. Thebabies are usually stillborn or die at birth.

OI-IIBThe ribs are less affected so that respiratorydistress is less and babies may survive formonths (fig 2). Otherwise the clinical andradiographic features are similar to those ofOI-IIA.

oi-iicThis is the least common subtype. The babiesare very small, osteopenia is severe, and thelong bones are slender, poorly modelled, andcontain numerous fractures (fig 3). The babiesare stillborn or die soon after birth.

Genetic heterogeneityNumerous studies have shown that the majority

910 928

457446897 541 559 673 863

547560 625 700 805 907 976244 691 718 748 904 948 988

472 640 77

343247 352 415 565 59 631 862 913 964 1003

502 706 66S5nnA256 a5s 637 802 973 1100N

Figure 4 Map of glycine substitutions within the triple helical domains of the al (I) and a2(I) chains of OI-II. Each line represents the triple helicaldomain of an oa chain. The substituting residue is listed on the left. The residue numbers are indicated above each line for the cxl (I) chain and beloweach line for the ac2(I) chain. The residues are numbered relative to the first glycine residue of the triple helix.

Ala

Arg

Asp

Cys

Glu

Ser

Val I I I i -

285

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from

Cole, Dalgleish

Exon skipping mutations, deletions, and insertions of the ttiple helical domains of COLlAl and COLIA2 and thecarboxyl-terminal propeptide domain of COLlAI in OI-II

Domain Locus Mutation Reference

Helical COLlAl Skipping of exon 14 4Duplication of exons 14-17 causing insertion of 60 aa 584 aa deletion, 328-411 6, 7Skipping of exon 27 875 bp insertion derived from intron 35 9Skipping of exon 43 83 aa deletion (GlyAlaPro) in the region of 868-876 103 aa deletion (GlyAlaPro) 874-876 11Skipping of exon 44 8Skipping of exon 47 12

COL1A2 Skipping of exon 28 13Skipping of exon 33 14Deletion of exons 34-40 15

Carboxyl- COLlAl Asp 2099His 16propeptide Trp 1134Cys 17

Valll46Cys plus frameshift and truncation 182 aa deletion (Glu 1159-Tyr 1160) 16Leu 121OArg 16

of babies with OI-II are heterozygous for mut-ations of the COLlAl or COL1A2 genes thatalter the structure of the triple helical domainsor the carboxyl-terminal propeptides of theproocl(I) or proa2(I) chains of type I pro-collagen (fig 4, table). Each procollagenmolecule contains two proocl (I) chains and oneprocz2(I) chain. The propeptides are en-zymatically removed outside the cell to producetype I collagen that is laid down in the ex-tracellular matrix of the dermis, bone, dentine,sclera, ligaments, and fascia, the tissues affectedin OI.The first mutation defined in OI-II was a

heterozygous deletion within COLlAl that res-ulted in the loss of 84 amino acids from thetriple helical domain ofthe cxl (I) chain.67 How-ever, few deletions have subsequently beenidentified (table). Exon skipping mutations aremore common. They usually maintain thetranslational reading frame and the repetitiveGly-X-Y amino acid triplet structure as theexons encode complete triplets (table).The majority of mutations in 01-II involve

the substitution of Gly residues in Gly-X-Ytriplets within the triple helical domain of theoc chains. The ability of such mutations toproduce the OI-II phenotype was confirmedby reproducing it in transgenic mice bearing aCOLlAl mutation that substituted Cys forGly 859.'9 Numerous point mutations of theGGN codon for Gly have been identified (fig4). Substitutions by Ser are the most commonwhile substitutions by Ala and Glu are rare andsubstitutions by Trp have not been identifiedto date. Approximately two-thirds of the sub-stitutions involve the ocl (I) chain and a similarproportion involve the carboxyl-terminal halvesof the triple helical domains of the cxl (I) andoc2(I) chains. The non-uniform nature of thesubstitutions, cx chain involved, and sites withinthe cx chains are unlikely to be the result ofascertainment or technical biases. Similar res-ults were found when all babies born with OI-II in Victoria, Australia were studied.'7 Similarresults have also been obtained with modernmethods of mutation detection that screen fulllength cxl (I) and oc2(I) cDNAs.20Most unrelated babies with OI-II have their

own private de novo mutation (fig 4). Ex-ceptions include recurrent substitutions of Gly

352, 415, and 1003 by Ser in the ocl(I)chain.2>24 These mutations occurred at CpGdinucleotides, in a manner consistent with de-amination of a methylated cytosine residue.2'However, there do not appear to be any majormutational hot spots in the COLlAl orCOL1A2 genes in babies with OI-II.A few mutations in the carboxyl-propeptide

of procxl (I) chains have been identified in OI-II (table). They alter sequences that are im-portant in chain association, disulphide linkingof the chains, and interactions withchaperones.'6 18

In many instances, the mutations in OI-II are new dominant mutations, while in theremainder the mutation is inherited from amosaic parent.25 Isolated germline mosaicismhas not been identified to date.2526 The mosaicparent may appear to be clinically normal or tohave a mild or moderate form of OI. However,leucocytes, hair follicles, and dermal fibroblastscontain differing proportions ofthe normal andmutant alleles.2526 The proportion within theosteoblasts, which is likely to be the most im-portant determinant of the phenotype, isusually unknown. The recurrence risk of OI-II is approximately 7% because of the oc-currence ofgermline mosaicism in one parent.27

PathogenesisAll of the mutations that produce OI-II appearto act in a dominant negative manner. Themutant proo chains are incorporated into typeI procollagen molecules and produce majorabnormalities in collagen metabolism.

Deletions, skipping mutations, and in-sertions in the triple helical domain produceabnormal alignment of the proct chains. Sub-stitutions of Gly residues interrupt the man-datory repetitive Gly-X-Y structure requiredfor normal formation of the triple helix. Theymay also produce a bulge or kink as a meansof accommodating the bulkier side chains. Forexample, kinks at the mutant site have beenobserved in type I collagen molecules con-taining mutant ocl (I) chains bearing sub-stitutions of Gly 718 or 748 by Cys.2829 Thelysine residues, particularly on the amino-ter-minal side of the substitution, are often over-modified by the prolonged action of lysyl

286

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from

Perinatal lethal osteogenesis imperfecta

hydroxylase and glycosyl transferases owing toa delay in the zipper-like folding of the collagenmolecules.30 The thermal denaturation tem-perature of the abnormal collagen is often re-duced but it is independent of the level ofovermodification ofthe oa chains.3' The variablelevels of overmodification and thermal stabilityof the mutant molecules are likely to reflectregional differences in helix stability and in thepropensity for renucleation of folding of thehelix beyond the mutation.32-34As type I collagen molecules contain two

oal (I) chains it is likely with heterozygous mut-ations of COLlAl that 25% of the moleculesare normal while 75% of the molecules containone or two mutant oal (I) chains. Conversely,as type I collagen molecules contain one cz2(I)chain, it is likely with heterozygous mutationsof COL1A2 that 50% of the molecules arenormal and 50% contain a mutant a2(I)chain.27 Procollagen containing one or more

mutant procx chains is poorly secreted and isexcessively degraded by cultured fibroblasts.22Once secreted, cleavage of the amino pro-peptides by N-proteinase may be reduced evenwith mutations distant from the cleavage site.28

Histological and biochemical anomalies havebeen observed in the dermis from babies withOI-II. The fibroblasts often contain dilated,rough endoplasmic reticulum because of im-paired secretion of the mutant type I pro-collagen molecules.35 The collagen fibrils are

smaller than normal. There is a reducedamount of type I collagen which consists of a

mixture of normal and mutant type I collagenmolecules. The mutant collagen chains are

more readily extracted which suggests that thetype I collagen molecules containing one or

more mutant chains may be degraded beforethey are incorporated into the more insolublecross linked extracellular matrix.36

Severe anomalies have also been observed inbone from babies with OI-II. The bone isseverely porotic. The normal cortical and tra-becular bone pattern is replaced by woven

bone.3537 There is an abundance of plumposteoblasts surrounded by little extracellularmatrix. The osteoblasts may also contain di-lated rough endoplasmic reticulum.2 Thegrowth plate is normal but cartilage cores per-sist in the trabeculae.38 The bone matrix con-

tains a reduced amount of type I collagen andincreased amounts of type III and Vcollagens.3639 The type I collagen includes bothnormal and mutant molecules.3640 The levelsof some non-collagenous proteins are also ab-normal.4' Osteocalcin, cx2-HS glucoprotein,and bone sialoprotein levels are increased, os-

teonectin is reduced, and decorin levels arenormal. There are dramatic changes in thecollagen fibrils and mineralisation. The col-lagen fibrils are thin and do not show the normalspatial arrangement of parallel bundles.242 Atthe mineralising front, apatite crystallites are

not oriented along the collagen fibrils as ob-served in normal bone. The crystallites are alsosmall and appear to grow by aggregation ofnew crystallites onto their surfaces.23843Many of these metabolic findings were also

observed in transgenic mouse models of

OI-II.'9" Expression of COLlAl constructsencoding a Gly 859 to Cys substitution or anin frame deletion in the triple helical domainof aol(I) chains resulted in the incorporationof mutant type I collagen molecules into thetissues.

Mutations of the carboxyl-terminus of theprool (I) chain can also produce dramaticchanges in type I collagen metabolism.'645 Inone heterozygous baby with OI-II, a codonframeshift mutation produced a truncatedproal(I) chain with other proct chains.'8 Theabnormal type I procollagen molecules werenot secreted and were degraded within the cell.The type I collagen content of the dermis andbone was reduced to about 20% of normal.This is the amount of normal collagen thatwould be expected if all of the molecules con-taining one or two mutant proozl (I) chainswere degraded. In contrast, mutations of thecarboxyl-terminus of the prooxl (I) chains thatprevent chain association and incorporation ofmutant chains into type I procollagenmolecules produce one form of mild OI-I.46

Genotype-phenotype relationshipsThere are two broad categories of genotype-phenotype relationships in osteogenesis im-perfecta. The first category includesphenotypes associated with haploinsufficiencyof COLlAl and COL1A2. The second cat-egory includes phenotypes associated withdominant negative mutations of COLlAl andCOL1A2. There are rare exceptions, such assome forms of autosomal recessive OI that donot appear to be the result of type I collagenmutations.47

Haploinsufficiency of COLlAl results in anapproximately 50% reduction oftype I collagensynthesis and the mild osteogenesis imperfectatype I phenotype with normal teeth, OI-IA.27Heterozygous Mov-13 mice, in which oneCOLlAl allele is inactivated by insertion of aretrovirus, also show the OI-IA phenotype.48 Incontrast, homozygous Mov-13 mice die aroundthe twelfth day of gestation as a result of theabsence of type I collagen from many tissues.49A human equivalent has not been identified asyet but would probably produce early death inutero and spontaneous abortion.A homozygous deficiency of at2(I) chains has

been reported in a child with moderately severeOI_III.51 The structure of the carboxyl-pro-peptide of the proa2(I) chains was altered sothat the chains would not associate. The meta-bolic consequences of the mutation were equi-valent to homozygous haploinsufficiency ofCOL1A2. The heterozygous parents were min-imally affected.The majority of deletions, insertions, and

splicing anomalies that alter the triple helicaldomain of the ocl(I) or oa2(I) chains producelethal phenotypes. However, the more com-mon glycine substitutions in the repetitiveGly-X-Y triplets of the triple helix producemild to lethal phenotypes. The site andtype of substitution have been proposed asimportant factors in determining the severityof the phenotype. For Gly substitutions by

287

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from

Cole, Dalgleish

Cys in the oc1(I) chain, carboxyl-terminal sitesproduce lethal phenotypes, mid-helical sitesproduce moderately severe phenotypes, andamino-terminal sites produce mild pheno-types.34 However, exceptions such as thesubstitution at Gly 244 by Cys in a babywith OI-II have been observed.5' Gradientsof phenotypic severity are less obvious withother Gly substitutions (fig 4). Lethal pheno-types are frequently observed with mid-helicalGly substitutions by Ser and Arg.There are a few rare instances of unrelated

subjects harbouring the same substitutions ofglycines in either the ocl (I) or oc2(I) chain.Somewhat unexpectedly, such people do not

always exhibit the same phenotype suggestingthe possible involvement of epistatic effects. Toillustrate this, an ocl (I)-Gly352Ser substitutionproduces OI-II in one person but OI-IV in twoothers.202223 A similar, though less dramatic,effect has also been noted in two subjects with

OI-III resulting from an cx2(I)-Gly859Ser sub-stitution.52 In contrast, three subjects har-

bouring an oc2(I)-Gly502Ser substitution

appear to exhibit exactly the same OI-IIpheno-type.53 Clearly evidence from more instances is

required to resolve the question of epistasis.Gly substitutions were proposed to be less

deleterious in the oc2(I) chain as type I collagenmolecules only contain one oc2(I) chain. How-

ever, a similar proportion of Gly substitutionsin the ocl(I) and oc2(I) chains produce lethal

phenotypes. The sequences around the Glysubstitution may be more important in de-

termining the consequences of the substitutionthan the oa chain involved.2332 For example,within the oc2(I) chain there are non-lethal

domains interspersed wth lethal domains.23The level of expression of the mutant allele

has been shown in transgenic mice also to be

an important determinant of the severity of the

phenotype. A direct relationship was found

between the severity of the phenotype and

steady state mutant mRNA levels in mice bear-

ing a Gly 859 substitution by Cys in COLlA1 .19These findings indicated that moderate levels

of expression of the mutant COLlAl were

sufficient to lead to dramatic dominant negativeeffects on the incorporation of type I collageninto the extracellular matrix. Similar findingswere also observed in transgenic mice that were

high expressors of a COLlAl gene construct

containing an in frame deletion.44 However,inbred mice expressing moderate levels of this

transgene showed marked phenotypic vari-

ability.54 The explanation for this variabilitywas not determined.

Little is known about the levels of expressionof the mutant alleles and the levels of in-

corporation of mutant molecules into the ex-

tracellular matrix in human OI-II. However,the collagen content of the dermis and bone is

reduced and mutant collagen is detectable in

babies with Gly substitutions in the triple hel-

ical domain of the cx chains.36 In one baby with

a substitution of Gly 580 by Asp in the oc2(I)chain, the bone matrix was shown to contain

a ratio of mutant to normal oc2 (I) chains of

0.7:1.40The genotypes associated with subtypes of

OI-II have only been studied in a few babies.The substitution of Gly by Arg at Gly 391produced 01-IIA, at Gly 667 produced OI-IIA/IIB, and at Gly 976 produced 01-IIB.35The substitution of Gly by Val at Gly 1006produced OI-IIC, Gly 973 produced 01-IIA,and Gly 256 produced IIA.55 A carboxyl-pro-peptide mutation of the proctl(I) chain pro-duced OI-IIB.45

Prenatal diagnosisOI-II is often detected on routine ultrasoundscans of pregnant women with a history ofosteogenesis imperfecta.5657 The skull is poorlyechogenic and the limb bones are char-acteristically broad, shortened, and fractured.The chest is also abnormal. OI-IIA is the mostcommon subtype.565' These scans are oftenundertaken at about 16 weeks but can be doneearlier in families with a history of previousbabies with 01-II. In the latter families, in-trauterine detection of the COLlAl or

COL1A2 mutations can be used ifthe mutationhas been determined before the pregnancy.

Some of this work was supported with grants from the MedicalResearch Council of Canada (WGC) and the Medical ResearchCouncil (RD).

1 Sillence DO, Senn A, Danks DM. Genetic heterogeneity inosteogenesis imperfecta. _7 Med Genet 1979;16:101-16.

2 Cohen-Solal L, Zylberberg L, Sangalli A, Gomez Lira M,Mottes M. Substitution of an aspartic acid for glycine 700in the a2(I) chain of type I collagen in a recurrent lethaltype II osteogenesis imperfecta dramatically affects themineralization of bone. _7 Biol Chem 1994;269:14751-8.

3 Byers PH, Tsipouras P, Bonadio JF, Starman BJ, SchwartzRC. Perinatal lethal osteogenesis imperfecta (OI type II):a biochemically heterogeneous disorder usually due tonew mutations in the genes for type I collagen. Am _7 HumGenet 1988;42:237-48.

4 Bonadio J, Ramirez F, Barr M. An intron mutation in thehuman oal (I) collagen gene alters the efficiency of pre-mRNA splicing and is associated with osteogenesis im-perfecta type II. .TBiol Chem 1990;265:2262-8.

5 Cohn DH, Zhang X, Byers PH. Homology-mediated re-

combination between type I collagen gene exons resultsin an internal tandem duplication and lethal osteogenesisimperfecta. Hum Mutat 1993;2:21-7.

6 Chu M-L, Gargiulo V, Williams CJ, Ramirez F. Multiexondeletion in an osteogenesis imperfecta variant with in-creased type III collagen mRNA. _7 Biol Chem 1985;260:691-4.

7 Barsh GS, Roush CL, Bonadio J, Byers PH, Gelinas RE.Intron-mediated recombination may cause a deletion ofan oal type I collagen chain in a lethal form of osteogenesisimperfecta. Proc Natl Acad Sci USA 1985;82:2870-4.

8 Byers PH. Brittle bones fragile molecules: disorders ofcollagen gene structure and expression. Trends Genet 1990;6:293-300.

9 Genovese C, Brufsky A, Shapiro J, Rowe D. Detection ofmutations in human type I collagen mRNA in osteogenesisimperfecta by indirect RNase protection. _7 Biol Chem1 989;264:9632-7.

10 Hawkins JR, Superti-Furga A, Steinmann B, Dalgleish R.A 9-base pair deletion in COLlAl in a lethal variant ofosteogenesis imperfecta. _7 Biol Chem 199 1;266:22370-4.

11 Wallis GA, Kadler KE, Starman BJ, Byers PH. A tripeptidedeletion in the triple-helical domain of the prooal (I) chainof type I procollagen in a patient with lethal osteogenesisimperfecta does not alter cleavage of the molecule by N-proteinase. I Biol Chem 1992;267:25529-34.

12 Wallis GA, Starman BJ, Chessler SD, Willing MC, ByersPH. Mutations in the CB6 and carboxyl-terminal regionsof COLlAl causing lethal OI have different effects on

the stability and secretion of the type I collagen molecule.IVInternational Conference on Osteogenesis Imperfecta, Pavia,Italy, 9-12 September 1990; abstract 55.

13 Tromp G, Prockop DJ. Single base mutation in the proca2(I)collagen gene that causes efficient splicing of RNA fromexon 27 to exon 29 and synthesis of a shortened but inframe proot2(I) chain. Proc Natl Acad Sci USA 1988;85:5254-8.

14 Ganguly A, Baldwin CT, Strobel D, Conway D, Horton W,Prockop DJ. Heterozygous mutation in the G+5 position ofintron 33 of the pro-oa2(I) gene (COL1A2) that causes

aberrant RNA splicing and lethal osteogenesis imperfecta:use of carbodiimide methods that decrease the extent ofDNA sequencing necessary to define an unusual mutation._7 Biol Chem 199 1;266:12035-40.

15 Willing MC, Cohn DH, Starman B, Holbrook KA, Green-

288

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from

Perinatal lethal osteogenesis imperfecta

berg CR, Byers PH. Heterozygosity for a large deletion inthe Y2(I) collagen gene has a dramatic effect on type Icollagen secretion and produces perinatal lethal os-teogenesis imperfecta. J. Biol Chem 1988;263:8398-404.

16 Chessler SD, Wallis GA, Byers PH. Mutations in the carb-oxyl-terminal propeptide of the proo 1(I) chain of type Icollagen result in defective chain association and producelethal osteogenesis imperfecta. _7 Biol Chem 1993;268:18218-25.

17 Bateman JF, Lamande SR, Hannagan M. Moeller I, DahlHHM, Cole WG. Chemical cleavage method for thedetection of RNA bases changes: experience in the ap-plication to collagen mutations in osteogenesis imperfecta.Am JMed Genet 1993;45:233-40.

18 Bateman JF, Lamande SR, Dahl HHM, Chan D, MascaraT, Cole WG. A frameshift mutation results in a truncatednonfunctional carboxyl-terminal prootl (I) propeptide oftype I collagen in osteogenesis imperfecta. _7 Biol Chem1989;264: 10960-4.

19 Stacey A, Bateman J, Choi T, Mascara T, Cole W, JaenischR. Perinatal lethal osteogenesis imperfecta in transgenicmice bearing an engineered mutant pro-oal (I) collagengene. Nature 1988;332:131-6.

20 Mackay K, Byers PH, Dalgleish R. An RT-PCR-SSCPscreening strategy for detection of mutations in the geneencoding the al chain of type I collagen: application tofour patients with osteogenesis imperfecta. Hum Mol Genet1993;2: 1155-60.

21 Pruchno CJ, Cohn DH, Wallis GA, et al. Osteogenesisimperfecta due to recurrent point mutations at CpG di-nucleotides in the COLlAl gene of type I collagen. HumGenet 1991;87:33-40.

22 Bateman JF, Moeller I, Hannagan M, Chan D, Cole WG.Characterization of three osteogenesis imperfecta collagena 1 (I) glycine to serine mutations demonstrating a position-dependent gradient ofphenotypic severity. Biochem3t 1992;288:131-5.

23 Marini C, Lewis MB, Wang Q, Chen KJ, Orrison BM.Serine for glycine substitutions in type I collagen in twocases of type IV osteogenesis imperfecta (OI). Additionalevidence for a regional model of OI pathophysiology. 7Biol Chem 1993;268:2667-73.

24 Mottes M, Gomez Lira MM, Valli M, et al. Paternal mo-saicism for a COLlAl dominant mutation (oal Ser-415)causes recurrent osteogenesis imperfecta. Hum Mutat1993;2:196-204.

25 Wallis GA, Starman BJ, Zinn AB, Byers PH. Variable ex-pression of osteogenesis imperfecta in a nuclear family isexplained by somatic mosaicism for a lethal point mutationin the a1 (I) gene (COLlA1) of type I collagen in a parent.Am3tHum Genet 1990;46:1034-40.

26 Cohn DH, Starman BJ, Blumberg B, Byers PH. Recurrenceoflethal osteogenesis imperfecta due to parental mosaicismfor a dominant mutation in a human type I collagen gene(COLlAI). AmJHum Genet 1990;46:591-601.

27 Byers PH, Wallis GA, Willing MC. Osteogenesis imperfecta:translation of mutation to phenotype. 7 Med Genet 1991;28:433-42.

28 Lightfoot SJ, Holmes DF, Brass A, Grant ME, Byers PH,Kadler KE. Type I procollagens containing substitutions ofaspartate, arginine and cysteine for glycine in the procal (I)chain are cleaved slowly by N-proteinase, but only thecysteine substitution introduces a kink into the molecule._7 Biol Chem 1992;267:25521-8.

29 Vogel BE, Doelz R, Kadler KE, Hojima Y, Engel J, ProckopDJ. A substitution of systeine for glycine 748 of the oalchain produces a kink at this site in the procollagen Imolecule and an altered N-proteinase cleavage site over225nm away. J Biol Chem 1988;263:19249-55.

30 Raghunath M, Bruckner P, Steinmann B. Delayed triplehelix formation of mutant collagen from patients withosteogenesis imperfecta. .7 Mol Biol 1994;236:940-9.

31 Rao VH, Steinmann B, de Wet W, Hollister DW. Decreasedthermal denaturation temperature of osteogenesis im-perfecta mutant collagen is independent of post-trans-lational overmodifications of lysine and hydroxylysine. _7Biol Chem 1989;264:1793-8.

32 Bachinger HP, Morris NP, Davis JM. Thermal stability andfolding of the collagen triple helix and the effects ofmutations in osteogenesis imperfecta on type I collagen.AmJMed Genet 1993;45:152-62.

33 Westerhausen A, Kishi J, Prockop DJ. Mutations that sub-stitute serine for glycine al-598 and glycine oal-631 intype I procollagen. The effects on thermal unfolding ofthe triple helix are position-specific and demonstrate thatthe protein unfolds through a series of cooperative blocks.J Biol Chem 1990;265:13995-4000.

34 Wenstrup RJ, Shrago-Howe AW, Lever LW, Phillips CL,Byers PH, Cohn DH. The effects of different cysteine forglycine substitutions within a2(I) chains. Evidence ofdistinct structural domains within the type I collagen triplehelix. _7 Biol Chem 199 1;266:2590-4.

35 Cole WG, Chow CW, Rogers JG, Bateman JF. The clinical

features of three babies with osteogenesis imperfecta re-sulting from the substitution of glycine by arginine in thepro a1 (I) chain of type I procollagen. _7 Med Genet 1990;27:228-35.

36 Bateman JF, Chan D, Mascara T, Rogers JG, Cole WG.Collagen defects in lethal perinatal osteogenesis im-perfecta. Biochem. 1986;240:699-708.

37 Sztrolovics R, Glorieux FH, Travers R, van der Rest M,Roughly PJ. Osteogenesis imperfecta: comparison of mo-lecular defects with bone histological changes. Bone 1994;15:321-8.

38 Omoy A, Kim OJ. Scanning electron microscopy findingsin osteogenesis imperfecta lethalis. Isrr7Med Sci 1 977;13:26-32.

39 Brenner RE, Vetter U, Nerlich A, Worsdorfer 0, TellerWM, Muller PK. Osteogenesis imperfecta: insufficientcollagen synthesis in early childhood as evidenced byanalysis of compact bone and fibroblast cultures. Eur _7Clin Invest 1989;19:159-66.

40 Niyibizi C, Bonadio J, Byers PH, Eyre DR. Incorporationof type I collagen molecules that contain a mutant a2(I)chain (Gly580-+Asp) into bone matrix in a lethal case ofosteogenesis imperfecta. _7 Biol Chem 1992;267:23108-12.

41 Vetter U, Fisher LW, Mintz KP, et al. Osteogenesis im-perfecta: changes in noncollagenous proteins in bone. 7Bone Mineral Res 199 1;6:501-5.

42 Stoss H, Freisinger P. Collagen fibrils of osteoid in os-teogenesis imperfecta: morphometrical analysis ofthe fibrildiameter. Am _7 Med Genet 1993;45:257.

43 Vetter U, Eanes ED, Kopp JB, Termine JD, Robey PG.Changes in apatite crystal size in bones of patients withosteogenesis imperfecta. Calc Tiss Int 1991;49:248-50.

44 Khillan JS, Olsen AS, Kontusaari S, Sokolov B, ProckopDJ. Transgenic mice that express a mini-gene version ofthe human gene for type I procollagen (COLlA1) developa phenotype resembling a lethal form of osteogenesisimperfecta. 7 Biol Chem 199 1;266:23373-9.

45 Cole WG, Campbell PE, Rogers JG, Bateman JF. Theclinical features of osteogenesis imperfecta resulting froma non-functional carboxyterminal procal(I) propeptide oftype I procollagen and a severe deficiency of normal typeI collagen in tissues. 7 Med Genet 1990;27:545-51.

46 Willing MC, Cohn DH, Byers PH. Frameshift mutationnear the 3' end of the COLlAl gene of type I collagenpredicts an elongated prooal(I) chain and results in os-teogenesis imperfecta type I. _7 Clin Invest 1990;85:282-90.

47 Williams EM, Nicholls AC, Daw SCM, et al. Phenotypicalfeatures of a unique Irish family with severe autosomalrecessive osteogenesis imperfecta. Clin Genet 1989;35:181-90.

48 Bonadio J, Saunders TL, Tsai E, et al. Transgenic mousemodel of the mild dominant form of osteogenesis im-perfecta. Proc Natl Acad Sci USA 1990;87:7145-9.

49 Schnieke A, Harbers K, Jaenisch R. Embryonic lethal muta-tion in mice induced by retrovirus insertion into the a 1 (I)collagen gene. Nature 1983;304:315-20.

50 Nicholls AC, Osse G, Schloon HG, et al. The clinicalfeatures of homozygous a2(I) collagen deficient os-teogenesis imperfecta. _7 Med Genet 1984;21:257-62.

51 Fertala A, Westerhausen A, Morris G, Rooney JE, ProckopDJ. Two cysteine substitutions in procollagen I: a glycinereplacement near the N-terminus of a1(I) chain causeslethal osteogenesis imperfecta and a glycine replacementin the ca2(I) chain markedly destabilizes the triple helix.Biochem J 1993;289:195-9.

52 Rose NJ, Mackay K, Byers PH, Dalgleish R. A Gly859Sersubstitution in the triple helical domain of the ca2 chainof type I collagen resulting in osteogenesis imperfecta typeIII in two unrelated individuals. Hum Mutat 1994;3:391-4.

53 Rose NJ, Mackay K, De Paepe A, Steinmann B, PunnettHH, Dalgleish R. Three unrelated individuals with peri-natally lethal osteogenesis imperfecta resulting from ident-ical Gly5O2Ser substitutions in the a2-chain of type Icollagen. Hum Genet 1994;94:497-503.

54 Pereira R, Khillan JS, Helminen HJ, Hume EL, ProckopDJ. Transgenic mice expressing a partially deleted genefor type I procollagen (COLlAl). A breeding line with aphenotype of spontaneous fractures and decreased bonecollagen and mineral. _7 Clin Invest 1993;91:709-16.

55 Cole WG, Patterson E, Bonadio J, Campbell PE, FortuneDW. The clinicopathological features of three babies withosteogenesis imperfecta resulting from the substitution ofglycine by valine in the procal (I) chain of type I pro-collagen. iMed Genet 1992;29:112-8.

56 Constantine G, McCormack J, McHugo J, Fowlie A. Pre-natal diagnosis of severe osteogenesis imperfecta. PrenatDiagn 1991;11:103-10.

57 Sharony R, Browne C, Lachman RS, Rimoin DL. Prenataldiagnosis of the skeletal dysplasias. Am i Obstet Gynecol1993;169:668-75.

58 Morin LRM, Herlicoviez M, Loisel JC, Jacob B, FeuillyC, Stanescu V. Prenatal diagnosis of lethal osteogenesisimperfecta in twin pregnancy. Clin Genet 199 1;39:467-70.

289

on Novem

ber 5, 2020 by guest. Protected by copyright.

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.32.4.284 on 1 A

pril 1995. Dow

nloaded from