Collagen genes and proteins in osteogenesis imperfecta · Clinical classification of osteogenesis imperfecta Osteogenesis imperfecta is a group of inherited collagen diseases in which

Journal of Medical Genetics, 1985, 22, 466-478

Collagen genes and proteins in osteogenesisimperfectaF M POPE, A C NICHOLLS, JANE McPHEAT, PHILIPPA TALMUD,AND R OWENFrom the MRC Clinical Research Centre, Watford Road, Harrow, Middlesex HAI 3UJ.

SUMMARY Type I collagen is a heteropolymer of al(I) and a2(I) chains, each of which is a

separate product of genes localised to chromosomes 17 and 7 respectively. Molecular defects oftype I collagen produce a group of inherited disorders of connective tissue primarily affectingbones, which are easily broken and collagen depleted (osteogenesis imperfecta). Sillenceclassifies these diseases into four groups, two of which are autosomal dominant and relativelymild, the others being either genetic lethals or responsible for very severe progressive disease.Here we describe two specific molecular abnormalities of type I collagen. One, a cysteinesubstitution in al(I) collagen, causes a mild Sillence type I disease, the other, a four base deletionin the C terminal extension of a2(I) collagen, causes progressive Sillence type III disease in thehomozygously affected patient and mild premature osteoporosis in his clinically symptomlessparents. We have briefly reviewed a variety of other similar mutations causing various 01

syndromes, which are tabulated, including various helical and non-helical deletions and a varietyof structural protein changes. Several restriction fragment length polymorphisms for a2(I) andcal(II) collagens have also been described, and 5' EcoRI and 3' MspI polymorphisms for a2(I)collagen segregate with Sillence type IV 01.

Modern recombinant DNA technology combinedwith protein chemistry provides a powerful means ofunderstanding single gene disorders. The bestknown application of this approach to date has beento the molecular pathology of the haemoglobino-pathies. The spectrum of genetic mechanisms re-sponsible for these diseases includes single basemutations, deletions and insertions, fusion genes,chain termination mutations, premature termina-tion, splice mutations, and various regulatorymutations. 1-3

Recent spectacular advances in the molecularbiology of interstitial collagen genes now allow asimilar approach to inherited collagen mutations in avariety of human diseases. Here we describe howthis approach has worked in the family of inheriteddiseases chiefly characterised by a hereditary prop-ensity to fracture bones from minimal injury-osteogenesis imperfecta.

Collagen genes and proteins

At least seven collagen proteins coded by 11 genesare already fully documented but 10 proteins with 16

Received for publication 3 July 1985.Accepted for publication 10 September 1985.

genes are likely. The genes are non-allelic and somecollagens are the product of more than one gene. Ingeneral there are classical and atypical collagenproteins. The former, better called the interstitialcollagens, include types I, II, III, and V.4 S Theseare widely distributed in a variety of scaffolding andsupporting tissues including skin, bone, and tendon(type I), cartilage (type II), arteries, veins,pleuroperitoneal linings, and skin (type III) col-lagens. In addition, type I collagen also occurs inskin and blood vessels and type V is also present inskin, blood vessels, and to a limited extent in bones.The other minority collagens include type IV,6specific for basement membranes, possibly withdifferent types in skin, lung, and glomerular base-ment membrane, other minority types includingtypes VI,7 VII,8 and VIII,9 which are blood vesselassociated, and type IX'° which includes the minor-ity cartilage components.

Structural composition of interstitial collagens

The interstitial collagens are the most abundantmembers of this protein family and are virtually theonly collagens contained in bone and thereforerelevant to osteogenesis imperfecta.

466

copyright. on 23 M

arch 2019 by guest. Protected by

http://jmg.bm

j.com/

J Med G

enet: first published as 10.1136/jmg.22.6.466 on 1 D

ecember 1985. D

ownloaded from

http://jmg.bmj.com/

Collagen genes and proteins in osteogenesis imperfecta

All have a regular order and gross structuralsimilarities sharing C and N terminal non-helicalextensions which have inter- and intra-chain disul-phide bonds. These regions are excised by specific Nand C terminal peptidases before the head to toeassembly of individual collagen triple helices to formthe final mature collagen fibrils." These then packin a characteristic one-quarter staggered pattern,which appears in longitudinal section as a regularbanded cross striation.

Intermolecular cross linking occurs between spe-cific overlapping peptide regions of the interstitialmolecules, and mutations of this region couldcertainly produce significant disease, as couldabnormalities within the procollagen extension pep-tides which contain globular non-helical and helicalsequences at the N terminus and non-helical exten-sions at the C terminus. The C propeptide alsocontains an asparagine residue which is later man-nosylated. The extension peptides are crucial tosecretion of the proteins and their correct assemblyinto the final mature fibrils.The amino sequence of these proteins is highly

conserved and is really a repeating polymer ofgeneral formula (Gly XY)303 where X and Y areoften proline and lysine respectively. These latterresidues are usually hydroxylated by specific prolineand lysine hydroxylases and this stabilises the triplehelix which would otherwise melt below normalbody temperatures. Other modifications include anexcess of hydrophobic side chains (a2(I), (al(IV),and a l(V) chains) and cysteine residues"1 which formand stabilise interchain disulphide bands in al(III)and al(IV) chains. Presumably these modificationsstabilise and strengthen the collagen triple helix andhave been highly conserved in evolution. Conver-sely, point mutations or other structural alterationsproduce instability of these regions and also veryprobably specific human diseases.

Collagen genes

The rapid advances in human recombinant DNAtechnology in the past five or six years has allowedthe cloning and sequencing of numerous collagenand other structural genes. cDNA clones fromcollagen mRNA enabled al(I) and a2(I) collagengenes from sheep and chick to be quicklyisolated.'2 13 As expected, these have non-codingintrons separating the coding exons and the chickena2 gene proves to have more than 50 introns withexon sizes which are always multiples of nine basepairs, the most common being 54 base pairs. 14 Thisreflects a fundamental three amino acid Gly XY unitcoded by nine base pairs with an ancestral unit ofsix such repeats to give a 54 base pair 18 amino acid

sequence. Subsequently, various other human andanimal interstitial collagen genes have been isolatedand sequenced, and usually cDNA clones have beenused for screening human or other mammalianlibraries for the relevant genomic sequences.l517There is always nearly 5 kb of coding sequences ingenes ranging from 18 to 35 kb. The organisation ofthe 3' ends of the various interstitial genes is alsohighly conserved. Exon 1 codes for the C terminal48 amino acids (and the 3' untranslated region).Exon 2 contains sequences coding for the carbohy-drate attachment site and is closely conservedbetween particular collagen types like I, II, and III.Specific collagen types are also conserved betweenvarious animal species. Similarly, exon 3 codes forthe intramolecular cysteines and exon 4 the junctionof the C propeptide telopeptide and the beginning ofthe helix. l The chromosomal locations of thevarious interstitial genes have also been identified.Thus, al(I) is on chromosome 17, pro a2(I) is on

4 'W

M-a..,)

()

FIG 1 Close up oflongitudinal sections ofnormal(a) and OI (b) bones. The bony trabeculae (arrowed) arereduced in size in 01 bones compared with normal.

467

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/

F M Pope, A C Nicholls, Jane McPheat, Philippa Talmud, and R Owen

/);

FIG 2 Epiphyseal plates ofnormal (a) and 01 (b) longbones. By comparison with normal the OI cartilage cellpalisades are sparser and more poorly organised. Theaccompanying bony trabeculae are less well developed inOI bone.

chromosome 7, types III and V collagens are onchromosome 2, al(II) collagen is on chromosome12, and al(IV) is on chromosome 13. 1821 Thereason for such dispersion is completely unclear buthas precedents in other proteins, such as the globinclusters in which the human a gene cluster consistsof embryonic and duplicated al and a2 genes onchromosome 16 and the I, a sequence on the shortarm of chromosome 11.1

Clinical classification of osteogenesis imperfecta

Osteogenesis imperfecta is a group of inheritedcollagen diseases in which bones are abnormallyfragile, collagen depleted, and osteoporotic.22 23Van Gieson staining shows a marked depletion ofcollagenous material which is also distorted andbadly organised compared with normal bones (figla, b). In the worst forms of the disease there are

sometimes minor disturbances at the cartilage-boneinterface of the epiphyseal plates, which often showirregularity and relative depletion of the normalcartilage palisades (fig 2). Additionally there aredeficiencies of both peripheral cortical bone and themore spongy perimedullary trabecular bone which issparse and deficient in osteoid matrix, althoughproperly calcified. Since the severer forms of diseaseaffect long bones, lacunar cartilage cells of theepiphyseal plate may also sometimes be faulty aswell as the nearby osteoblasts, which are the princi-pal bone forming cells.

Clinically, osteogenesis imperfecta is not a singleentity, but is a family of similar disorders sharing atendency to brittle, easily fractured, collagen de-pleted bones. Although a disease of considerableantiquity, recognised in Egyptian Mummies and inpre-Norman Scandinavian Britons, the disease hasseveral different clinical patterns. This was recog-nised early from the descriptions of Looser, who in1906 described differences in the early onset (con-genital) and late onset (tarda) forms.24 Sillence hasestablished beyond any doubt the clinical andgenetic heterogeneity of the disease25 26 (table 1). Inthe main he recognised four types, two of which arerelatively mild and autosomal dominant, and two ofwhich are often autosomal recessive and severe oreven lethal. Sillence type I disease with at least threesubgroups is clearly autosomal dominant with multi-ple affected generations, blue sclerae, deafness, andshort stature. In the main, bones break when thechild begins to walk, deafness is common, andthere may or may not be dentinogenesis imperfectawhich breeds true in any given family. Although thedisease does not usually occur at birth there areexceptions. Additionally, although the disease isfairly uniform within families, occasional very muchmore severely affected persons occur, possiblybecause of a second (different) mutation. Sillencetype IV disease is similar to type I except that thesclerae are white rather than blue. Again, truebreeding varieties with and without dentinogenesisoccur and, like type I disease, the frequency offractures drops dramatically at puberty although itmay increase again in menopausal women.

Sillence type II disease is a lethal, short limbeddwarfism which may be fatal in utero, at birth, orperinatally.26 The commonest cause of perinataldeath is respiratory infection from lung restrictionsecondary to fractured ribs (fig 3a, b). Even ifinfancy is survived, it is rare for affected children tolive beyond the age of 5 years. Although Sillencerecognises several types, a reasonable classificationis into broad and thin boned types. Broad boneddisease may be confined to the limbs but can alsoaffect the ribs. Different patterns of rib fractures

468

A.F

4-41

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


TABLE 1 Osteogenesis imperfecta: classification modified after Sillence.

Clinical features Inheritance Type

Mild disease: blue sclerae; deafness: improves at adolescence. Autosomal dominant Ishort stature: variable penetrance; (a) normal teeth.(b) dentinogenesis. (c) scanty fractures

Lethal short limbed dwarfism: fatal in utero or perinatal life: Autosomal recessive. may be 11several types: (a) broad bones, broad ribs, (b) broad bones. autosomal dominant; geneticthin or beaded ribs. (c) thin or normal bones compounds with double heterozygositv

Severe fractures in utero or early childhood; blue sclerae; Autosomal recessive IIInormal teeth; progressive course; popcorn deformities oflower femur and upper tibia

Similar to type I except sclerae white: dentinogenesis or Autosomal dominant IVnormal teeth; thin boned

Loose jointedness and features of Ehlers-Danlos syndrome Variable Vassociated with OI; usually normal stature: minor fractures orosteoporosis

distinguish various types (fig 3a, b), but biochemicalinvestigation will ultimately test the validity of thisclassification.

Sillence type III disease is also called the progres-sively deforming disease; fractures occur at orshortly after birth and continue throughout child-hood and adolescence if the child survives. Grossly )oidistorted limbs seem to be the rule and types withnormal and distorted facies can be recognised(fig 4a, b). There may be overlap with Sillence typeII 01 and some thin and broad boned survivors lookvery like Sillence type II disease (in early child-hood). There is considerable controversy as to thegenetics of this variant. Although several sibs insome families and proven heterozygote parents inother families are certain indications of autosomalrecessive inheritance in some instances at least, inthe main population surveys do not show a sufficientnumber of affected subjects within the families ofproven index patients. Possible explanations include ba sizeable proportion of new dominant mutations, ahigh frequency of (undetected) miscarriages shortlyafter conception, and the reluctance (conscious orotherwise) of parents with one severely crippledchild to run the risk, however small, of anotheraffected child.

Molecular mechanisms of disease in osteogenesisimperfecta

As illustrated above, collagen is very obviouslydepleted in histological sections of long bones inosteogenesis imperfecta. Since type I collagen(al(I)2a2) is the predominant protein of bones, itseems a logical conclusion that molecular abnorma-

FIG 3 (a, b, c) Chest x-rays from various Sillence type H lvariants showing thickened, beaded, and normal ribbedpatterns. Beaded and thickened ribs can both beaccompanied by widened, crumpled upper and lower limbbones.

Xak

-

.5..}. 3i* 5

.§

.;X

469

WE"

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


FIG 4 (a, b) Sillence type III OI showing normal andabnormal facies. Both types are associated with severelydistorted and progressive limb fractures.

lities of genes coding for this substance might befaulty in this disease. In this section we illustrate thispoint by describing two specific abnormalities oftype I collagen associated with OI. The first, a milddisease, has an amino acid substitution in the al(l)protein, and the second, a severe form of 01, hascompletely absent collagen a2 chains from 01 bonesbecause of a rather subtle mutation in the a2(I)procollagen gene.

AN a(I) SUBSTITUTION CAUSES MILDAUTOSOMAL DOMINANT SILLENCE TYPE I 01This disease presented as a series of badly healedfractures to the left tibia of a 9 year old male (fig 5).He fractured his right tibia quite justifiably from afall from a climbing frame but subsequently hadbroken his left tibia which then repeatedly fracturedthrough the union. Eventually, healing was achievedonly by rodding of the affected bone.27 At thisoperation the surgeon noted that the abnormal bonewas unusually hard but also brittle. We wereprovided with bone and skin biopsies from which wesuccessfully cultured skin and periosteal fibroblasts.Using radiolabelled 14C glycine and proline we thenexamined the cultured collagens by column chroma-tography and polyacrylamide gel electrophoresis(figs 6 and 7a, b). His mother, who had prematurelumbar spinal osteoporosis (fig Sc) and a history ofchildhood fractures of small hand bones, was alsostudied. Column chromatography of pepsinisedradiolabelled collagens showed an unexpectedshoulder in the position of al(I) collagen. Polyacry-lamide slab gel electrophoresis showed an extradisulphide linked dimer in addition to normal al(I),which could be reduced to monomeric form bymercaptoethanol (fig 7a). This was clearly separatefrom al(III) collagen, which is normally disulphidelinked, in contrast to normal al(I) collagen, whichnever is. Then we fingerprinted the mutant molecule

(1!

FIG 5 Clinicalfeatures of Sillence type 101 child with coll(itg, cU(1)clv.steine su.bstitution. Note (a) normalface and teeth,(b) tibial sclerosis with a nicked fracturie through abnorssnallvl united callus, and (c) premature lumbar osteoporosis of hisclinically affected mother.

470

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


03-

0 2

al (1)

g2 (I)

E0.1 \\ to

33

O al(1) x0-4 Bone 3.5 0

3-0

.2.0

02 - X 1.5

.1.0

0.1 . .0.5

0 10 20 30 40 50 60Fraction number

FIG 6 CMC chromatography of 4C radiolabelledpepsinised collagens from skin and bone fibroblasts. Bothrecordings show an unexpected shoulder in the al (I) peakedlabelled alX.

using a variety of double two dimensional elec-trophoresis. Mercaptoethanol reduction in bothdimensions showed the new band on a diagonalwhich was lacking in normal tissues. When theradiolabelled proteins were run in the first dimen-sion by molecular weight after conversion into theirconstituent cyanogen bromide peptides, and thenrun in the second dimension, either mercaptoethanolreduced or unreduced (fig 7c), the unreduceddisulphide linked dimer in the extra band wasconverted to a peptide in the position of collagenpeptide al(CB6) in the second dimension. Thisconclusively localises the mutation to the C terminalend of the collagen helix.28 Three possible mutationswill create a cysteine residue in a collagen genewhere none existed before. Glycine, arginine, orserine can all mutate by a single base substitution.For experimental and clinical reasons we do notexpect a glycine to cysteine substitution in ourpatient. Firstly, Steinmann et a129 have conclusively

shown just such a mutation in a patient withclassical, lethal, broad boned Sillence type II OI. Inthis patient the collagen melted abnormally easilyand they have shown (in collaboration) a glycine tocysteine substitution by molecular cloning of themutant gene.30 In contrast, our patient's meltingcurves are normal3l and, unless Steinmann's patientcontains a second mutation which alters the meltingprofile, we do not expect this to be the substitutionin our patient who has a mild, non-lethal disease.Cloning and sequencing of the mutant proteinshould therefore show an arginine or serine substitu-tion in this instance. Presumably, the appearance ofcysteine in a protein helix normally lacking it subtlyalters the properties of the molecule in such a way asto allow the extra hardness but normal brittleness torecur.

A SMALL MUTATION OF THE a2(i) COLLAGENGENE CAUSES SILLENCE TYPE III OIIn this example, the disease presented at 5 weeks ofage with recurrent fracturing of various long bones.Although at first the child was suspected to havebeen battered, the fractures continued while inhospital and subsequently fracturing has occurredseveral times a month. When first examined by us at21 months of age there was already quite obviousbowing and deformity of the long bones, althoughthe face and skull were relatively normal. The teethappeared normal at this stage. Subsequently therewere persistent and frequent fractures so that by theage of 5 years the upper and lower limbs wereseverely distorted and deformed. Fracturing hasbeen so frequent that orthopaedic setting andplastering has never been a practical possibility andthe fractured regions are merely bandaged firmly.The x-rays exclude the broad boned, lethal form ofOI and early ones showed osteoporotic long boneswith multiple healed and recent fractures. Subse-quent x-rays have shown severe osteoporosis andextreme alteration of the long bones with pseudoar-throses of the humerus and an expanded 'popcorn'deformity of the lower femur and upper tibia (fig 8).The trunk and face are relatively normal, althoughthe skin is soft and hyperextensible and the jointsmarkedly hypermobile (fig 9).

Family historyThere is parental consanguinity as the parents, wholive in the Hartz mountains of northern Germany,are third cousins. Clinically they look normal buthave early premature osteoporosis of the lumbarspine.

Biochemical investigationsRadiolabelled profiles of cultured skin fibroblast

471

- . I

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


.Mw;.5° *-0.

< ~~~~Molecular weight--

Molecularweight

1 2 3 4

ReducedI

*MSH-

-:a.e.m . . O

.B.S '1(111)*~~~~ ~ ~~~~~~~~~~~~~~~~~~~~~~~ Patient

-MSH.4:s W.:

*0ir

:~1

+IMSH

.4

A.6 *.F. lw~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

!.-\

0

tl(I) X x t1 (1

medium showed a surprising finding. Collagena2 chains were absent from the medium proteins ofthe affected patient (fig 10). Furthermore, hisparents were intermediate in this respect andproduced between one-third and one-half of theexpected a2 chains.33 We also proved by cyanogenbromide peptide mapping that the a2(I) collagen

FIG 7 (a) Radiolabelled "C collagen proteinsshowing a disulphide linked dimer 'X' beforereduction which after treatment withmercaptoethanol becomes a monomer 'X'running below al(III) monomer. Track4-control 1, 2, and 3 respectively containedcollagens from skin and bone fibroblasts.(b) Skin fibroblasts from mother's "4C labelledcollagen proteins run in the first dimension bymolecular weight and in the second dimensionafter mercaptoethanol reduction. The arrowshows the disulphide linked dimer.(c) Cyanogen bromide cleaved peptides of in situa chains separated by molecular weight in thetransverse dimension and by peptide mappingin the vertical dimension. Two gels have beenrun back to back and one has beenmercaptoethanol reduced. The arrowed peptidein the disulphide linked peptide map does notmove in the position of al(I) CB6 unlessmercaptoethanol reduced. Note that the arrowson the left and right are homologous.

product was absent and not migrating anomalously,and showed that tissues from the patient were

deficient (depleted) in this substance, so that an

artefact of culture could be confidently excluded.Next we collaborated with Dr Darwin Prockop

and Dr Jeanne Myers of the Rutgers UniversityMedical School, New Jersey, to identify the specific

(aIUnreduced

0D

cntroi

472

--*.- X

A&- iL1-iisi OM - ol (1)0 .0- ."O 2

A.4p

.fl 1)x

A. A...... A.--

-.*-- t 1 ( I )ON"* --W- 1 2

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


FIG 8 X-ray oflower limb showing widely expanded lowerfemur and upper tibial epiphyses with widespreadosteoporosis.

molecular defect responsible for this disease. First,we showed that sufficiently rapid and heavilylabelled cells produced small amounts of collagen a2chains which occurred only within the cell layer anddid not become exported into the surroundingmedium.34 This explained why a2 collagen wasabsent from normal tissues. Secondly, Northern andSouthern blotting experiments showed no majordeletion of the a2 gene. We then suspected that theC terminal propeptide might be altered so as toinhibit the incorporation of a2 chains with themature type I collagen triple helix. We already hadevidence that instead of the heteropolymerical(I)2a2 chains, the affected child produced signifi-cant quantities of the usually rare al(I)3, so calleda(I) trimer. To detect a small mutation of the 3' endof the gene coding for this region, Dr Leon Dicksonfrom Dr Prockop's laboratory constructed a 687base pair gene fragment which he and Dr Myersthen used in S1 blotting experiments with cDNA(gene) RNA hybrid experiments.35 By using S1nuclease digestion of 3' and 5' labelled RNA/cDNAhybrids, a 12 base deletion was detected in the 3'propeptide coding region (fig 11). Subsequent clon-ing and sequencing of this mutation have shown afour base pair deletion (fig 12) which completelyalters the phase of the last 34 amino acids of the a2 Cpropeptide region.36 Because of this, assembly intothe normal type I helix is impossible and the a2 gene

FIG 9 Clinical appearance of child aged 5 yearsshowing progressive deformity of upper andlower limbs with relatively normalface and trunk.

1 2 3

- 1 (111) .....

,.1, o,(I) |

FIG 10 Comparison of type I collagen a chains secretedinto medium "/C radiolabelled skin fibroblasts afterprecipitation with 2 6 mmolll NaCI. Normal controls (track1) contain a] (III), al (I), and a2(I) chains. The latter areabsentfrom the supernatant and precipitate of the O0mutant.

473

A.WNWASA.

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/

FM Pope, A C Nicholls, Jane McPheat, Philippa Talmud, and R Owen

A B L

0X -Si 1 C1- P M F I C1 P M F Cl P

1353 M1078 _

872

am* _ 687

(nt )(nt603

_

/.80

FIG 11 Sl nuclease mapping ofmRNA from patient (P) and parents (M+F) compared with control (Cl) using 3' and 5'

labelled cDNA to 687 base pair C terminal fragment. Here the affected child shows only a 480 base pair protected fragmentwhen labelled 3' and a 195 base pair fragment when labelled 5' (not shown). Overall the child's mRNA is protected by 675

base pairs instead of 687 (12 bases short). The parents are heterozygous for 480 and 687 base pair protected fragments.

Control

AC

Pro C

31

Patient

3 T

A

C

G

A

A

C

A

AA

5 A

A

Lys A

Asn A

A

C

Thr A

A~~~~~A

Lys A 5'

_..u

ff(h' P 1Itf 1(i tU / '1 l

product is intracellularly degraded (and never

appears extracellularly). Prenatal diagnosis of thisdisease would be possible by fibroblast culture of a

fetoscopic biopsy and examination for a2 proteinchains. Better would be Southern blotting of poly-acrylamide Southern blots using the 687 bp fragmentas a probe.

Other type I collagen mutations

HiS Comparatively rapid progress has been made inother forms of osteogenesis imperfecta and a varietyof type I collagen mutations has been identified byseveral different workers. We have correlated theclinical pattern of disease (Sillence type) with the

Ser biochemical and genetic abnormalities in table 2. Ingeneral, specific abnormalities include small dele-tions, amino acid substitutions, an apparent aminoacid insertion, protein over-hydroxylation of lysyl

Thr residues, and a variety of slowly migrating al(I) anda2(I) collagen chains. Other defects are certain andwill probably include many of the full range ofmutations already mentioned earlier as applying tothe haemoglobinopathies1-3 (table 2).

Alternative (complementary) approaches

The evidence so far available (and summarisedabove) suggests that the causes of the differenttypes of OI form no coherent pattern but are a

series of individual mutations. As such they do nothave general applications to the disease as a whole,

M F C2

474

... -d,M, Aft ON 4mqp %dow, mo

*W

..mom". ago&,. -&Udk MIAla. Abou., qwl* *V* mffdwlm.. m 4im4w. mmpw 4M WNW qw gm

mombi.-FMFP- 4p

46*..Nmmk

-.4plop-0414

moo..._amma

WWaaw

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


TABLE 2 OI type correlated with collagen protein and gene defects.

Protein defect Gene defect Clinical type

Increased III/I collagen ratios3739 IaCysteine substitution in al(I) CB6 (normal ? arginine or serine -. cysteine in al(I) collagen Ia

melting curve)27 28 31Diminished pro al(l) synthesis"?Cysteine substitution in al(I) CB6; lowered Glycine -. cysteine (al(I) collagen) Ila

melting curve; overhydroxylated collagen29 3"Increased IIVI ratios; one pro al(I) allele shortened 0-5 kb helical pro al(I) deletion Ila

by 50+ amino acids; reduced collagen synthesis4144Overhydroxylated skin and bone tissues45 One gene normal IIaAbnormal pro ct2(I) genes46 Structural mutation pro a2(I) IIDiminished pro al(I) production;40 overhydroxylated In second pro al gene? IINH2 terminals 3/4 portion4 48

Overhydroxylated pro al(I) chains57 300 base pair deletion in al(II) gene; Ila? polymorphic marker

No a2(I) collagen in tissues or secreted in tissue Four base pair deletion pro al(I) C propeptide IIIculture; rapidly degraded intracellular pro a2 chains323'

Poorly secreted type I collagen; excessive ? amino acids substitution in C propeptide IIImannosylation of C propeptide49

Short a2(I) deletion near amino terminal end of ? gene deletion at 5' end pro a2(I) gene IVtriple helix5"

Ialthough of course they are highly relevant to theindividual families concerned. The alternativemethod of identifying causes of mutated collagengenes is to search for random restriction fragmentlength polymorphisms near or within the gene,which can then be shown to be in marked linkagedisequilibrium with the relevant collagen gene. Thisapproach has been very successful in globin genepathology, especially in the P globin gene relatedcluster on chromosome 11, of which the HpaI 13 kbassociated sickle gene polymorphism is an excellentexample.' 51 52 Another good example is the poly-morphism at the haemophilia B locus. As far asOI is concerned, al(I), a2(I), and al(II) linkedRFLPs would all be useful for the diagnosis ofautosomal dominant disease. In principle, a similarapproach could be used (in much smaller pedigreesand therefore with lesser accuracy) for autosomalrecessive families.

COLLAGEN a2(I) RFLPTwo intragenic RFLPs have been identified for thehuman collagen a2 gene by Tsipouras and hiscolleagues.53 4 One is an EcoRI fragment in themiddle of the 5' end of the gene and the other is anMspI polymorphism localised to intron 6 at theopposite 3' end. The EcoRI polymorphism gener-ates an extra site in a fragment which is normally 13kb which, with an added site, generates 9-5 and 3.5kb fragments (fig 13a, b). The MspI polymorphismat the opposite end of the gene normally generatesa 2*1 kb fragment, which in the presence of anadded site becomes 1*6 and 0*5, giving 2-1, 1-6, and0-5 kb fragments in heterozygotes; this has also beendescribed by others.55 The allelic frequencies areEcoRI (no site) 68%, MspI (no site) 14%. Both

EV 702

I: 1 11_

12~~~~~~~~~

I2 _3

II 2

III112 S

III 2IIT 3 _

III 4IV 3. *

a

IV34IV 5IV 6IV- *IV

7 WM

IV-9IV 10IV 11

3-8 9 5 13Kilobase pairs

FIG 13 (a) Hypothetical pedigree of typical autosomaldominant Sillence type IV O. (b) The restriction patterncorresponds to the pedigree numbers showing that thedisease segregates with the extra restriction site which gives9-5 and 3-5 kb fragments instead of 13 kb when cut withEcoRI and probed with the ax2(I) specific probe NJ3.

475

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


these polymorphisms have proved informative inSillence type IV 01. A typical example of thesegregation of the EcoRI site is shown in fig 12, inwhich the presence of the extra restriction sitesegregates with the gene for type IV 01. Two newpolymorphic markers using EcoRI and BglII withthis gene have also just been published.56 Noexamples of linkage to Sillence type I 01 have so farbeen established, suggesting that perhaps non-a2chain mutations are responsible for this disease andimplicating the a2 chain in the white eyed Sillencetype IV variant.

COLLAGEN al(II) POLYMORPHISMSCartilage collagen is an interstitial collagen closelyrelated to types I and III collagens. Although ithas never previously been implicated in OI, recentevidence suggests that it might be. We recentlydescribed a 300 bp deletion in a type I like gene intwo Caucasian and two Asian patients with lethal 01congenita (fig 13a).57 Subsequently, Cheah et allocalised the deletion to the 3' non-coding region ofthe gene and also recognised that the gene islocalised to chromosome 1258 and is the gene forc(l(II)58 (in contrast to the al(I) and a2(I) genes on

chromosomes 17 and 7 respectively). Gene sequenc-ing and Northern hybridisation confirmed the geneto be al(II) collagen which was simultaneouslyconfirmed by Sangiorgi et al.59 Since then, we haveobserved several variations of this deletion withsmall and large deletions and also probably insertionsin this region (fig 13b).60 Sykes and his colleagues6' 62in studies following our original observations haveshown a frequency of these deletions close to 20% incertain Asian and African populations but haveobserved it in very few Caucasians (none in our

case). The presence of the deletion in broad bonedlethal 01 Caucasian families is unexplained butreliably detects the OI gene if present in this racialgroup. In Asian or West Indian families the deletiondoes in any case have a 20% frequency. Time willprove whether this deletion segregates with OI inthat population and also whether the increasedfrequency of the deletion is accompanied by a

markedly increased frequency of lethal congenitalOl in that population. (These data are not at presentavailable.) Alternatively, only some of this family ofdeletions are associated with disease states or thedeletions may in all cases be irrelevant to thedisease. Cloning, sequencing, and transientexpression experiments will ultimately resolve thesespeculations but there is a strong implicationthat cartilage collagen genes and proteins may havea significant role in the generation of the more

severe and lethal forms of hereditary brittle bonedisease.

FIG 14 (a) EcoRI restriction pattern of three 01 congenitapatients' tracks (1, 2, 4) showing an apparent 300 bpdeletion when probed with nick translated 32P labelledcosH col 1. (b) Varied restriction pattern of BamHIIEcoRldouble digested human DNA when probed with 4-3 kbEcoRI fragment of cosH col 1. Tracks 2, 3, and 4 showpatterns encountered in O congenita patients of Asian andCaucasian extraction. Track 5 is from a West Indian patientwith an apparent insertion of 300 base pairs.

Other polymorphisms of the same gene have alsobeen described, namely a HindIII and a PvuIIpolymorphism, within the 9-3 EcoRI fragment 3-8kb 5' to the Eco7 4.3 kb fragment described above.Using the 9*2 kb Eco fragment as a probe, PvuIIdigestion detects a 3*3 kb fragment in those lackingan extra site and a 1-7 and 1 6 kb fragment when an

extra site is present. The HindIII variable site lies1.5 kb 3' of the PvuII variable site. Sykes and hiscolleagues63 have recently failed to show segregationof this marker with Sillence type I 01, suggestingthat the cartilage gene cannot therefore be involvedin the aetiology of this 01 variant. Provided there isno intragenic recombination within the remaining 18kb of the gene, this conclusion seems valid.

References

Weatherall DJ. The new genetics and clinical practice. London:The Nuffield Provincial Hospital Trust, 1983:45-89.

2 Collins FS, Weissman SM. Molecular genetics of humanhaemoglobin. Prog Nucleic Acid Res Mol Biol 1984;31:315-462.

3 Kazazian HH. Molecular pathology of the ,B-globin gene cluster.Perspective on disease trends. Trends Neurosciences 1985;8:192-200.

4 Bornstein P, Sage H. Structurally distinct collagen types. AnnuRev Biochem 1980;49:957-1003.Burgeson RE, Hollister DW. Collagen heterogeneity in humancartilage. Identification of several new collagen chains. BiochemBiophys Res Commun 1979;87:1124-31.

6 Timpl R, Wiedemann H, Van Delden V, Furthmayr H, KuhnK. A network model for the organisation of type IV collagenmolecules in basement membranes. Eur J Biochem1981 120:203-1 1.

U)

2 9 --

476

/4 :..'r14

.. 0 "'.i

-57

0p-L.,

i-

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


7Furthmayr H, Wiedemann H, Timpl R, Odermatt E, Engel J.Electron-microscopical approach to a structural model ofinternal collagen. Biochem J 1983;211:303-13.Bentz H, Hiorns P, Murray LW, Sakin LY, Hollister DW.Burgeson RE. Isolation and prenatal characterisation of a newhuman collagen with an extended triple helical domain. ProcNatl Acad Sci USA 1983 ;80:3168-72.

9Sage H, Trueb B, Bornstein P. Biosynthetic and structuralproperties of endothelial cell type VIII collagen. J Biol Che,n1983;258:13391-401.Gibson GJ, Kielty CM, Garner C, Schor SL, Grant ME.Identification and partial characterisation of three lowmolecular-weight collagenous polypeptides synthesised by chon-drocytes cultured within collagen gels in the absence and in thepresence of fibronectin. Biochem J 1983:211:417-26.Prockop JD. Kivirikko KI. Heritable diseases of collagen. NEngl J Med 1984:311:376-86.

12 Frischauf AM, Lehrach H, Rosner C, Boedtker H. Procollagencomplementary DNA a probe for messenger RNA purificationand the number of type I collagen genes. Biochemistry1978;17:3243-9.Boyd CD, Tolstoshev P, Schafer MP, et al. Isolation andcharacterisation of a 15 kilobase genomic sequence coding forpart of the pro at2 chain of sheep type I collagen. J Biol Chem1980;255:3212-20.

14 Wozney J, Hanahan D. Tate V, Boedkter H. Doty P. Structureof the pro a2(1) collagen gene. Nature 1981:294:129-35.

5 Myers JC, Dickson LA, de Wet WJ, et al. Analysis of thie 3' endof the human pro ct2(I) collagen gene. J Biol Chemn1983;258:10128-35.

16 Chu M-L, de Wet W, Bernard M, et al. Human pro al(I)collagen gene structure reveals evolutionary conservation of a

pattern of introns and exons. Nature 1984:310:337-40.17 Yamada Y, Liau GM, Mudryj M, Obici S. de Crombrugghe B.

Conservation of the sizes for one but not another class of exonsin two chick collagen genes. Nature 1984;310:333-7.

18 Junlen C, Weil D, Myers JC, et al. Assignment of the human proa2(I) collagen structural gene (COL1A2) to chromosome 7 bymolecular hybridization. Am J Hum Genet 1982;34:381-7.

19 Huerre C, Junien C, Weil D, et al. Human type I procollagengenes are located on different chromosomes. Proc NatlAcad SciUSA 1982;79:6627-30.

2(0 Solomon E, Hiorns LR, Spurr N, et al. Chromosomal assign-ment of the gene coding for human type Il. III and IV collagen:a dispersed gene family. Proc Natl Acad Sci USA 1985;82:3330-4.

21 Emanuel BS, Cannizzaro LA, Seyer JM, Myers JC. Humanal(III) and a2(V) procollagen genes are located on the long armof chromosome 2. J Biol Chem (in press).

22 McKusick VA. Heritable disorders of connective tissue. 4th ed.St Louis: Mosby, 1972:390-454.

23 Hollister DW, Byers PH, Holbrook KA. Genetic disorders ofcollagen metabolism. In: Advances in human genetics. Vol 12.New York: Plenum Press, 1982:1-87.

24 Looser E. Zur Kenntnis der Osteogenesis imperfecta congenitaund tarda (sogenannte idiopathische Osteopsathyrosis). Mit-teilungen aus dem Grenzgebieten der Medizin und Chirurgie1906;161-207.

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

oh Sillence DO, Barlow KK, Garber AP, Hall JG, Rimoin DL.Osteogenesis imperfecta type II. Delineation of the phenotypewith reference to genetic heterogeneity. Am J Med Genet1984;17:407-23.

27 Nicholls AC, Craig D, Pope FM. An abnormal collagen a chaincontaining cysteine in autosomal dominant osteogenesis imper-fecta. Br Med J 1984;288:112-3.

2 Nicholls AC, Pope FM, Steinmann B. Cysteine in the triple-helical domain of the rtl(I) collagen chain produces a mild,

dominantly inherited form of osteogenesis imperfecta. lInutliolBiol Chem. Submitted for publication.

29 Steinmann B. Rao VH. Vogel A. Bruckner P. Kitzelmann R,Byers PH. Cysteine in the triplc-helical donmain of one allclicproduct of the al(l) gene of type I collagen produces a Iethalform of osteogenesis imperfecta. J Biol Chem 1984;259:11129-38.

30 Steinmann B, Bvers PH. Personal communication. 1985.31 Steinmann B, Nicholls AC, Pope FM. Two different cysteine

substitutions in the same al(I) CB6 peptide of the (il(1) collagcnchain produce a lethal and a mild form of osteogenesisimperfecta. Pediatr Res Soc Eur 1985.

32 Nicholls AC, Osse G. Schloon HG, et al. The clinical featur-es ofhomozygous al(l) collagen deficient osteogenesis imperfecta. JMed Genet 1984:21:257-62.

33 Nicholls AC, Pope FM, Schloon H. Biochemical hcterogeneityof osteogenesis imperfecta. Lancet 1979;i:1193-4.

34 Deak SB, Nicholls AC, Pope FM. Prockop DJ. The molcculardefect in a nonlethal variant of osteogcncsis imperfecta. J BiotChem 1983;258:15192-7.

35 Dickson LA, Pihlajaniemi T. Deak S, et al. Nuclease S1 map-ping of a homozygous mutation in the carboxylpropeptide-coding region of the pro a2(I) collagen gene in a patient withosteogenesis imperfecta. Proc Natl Acad Sci USA 1984;81:4524-8.

36 Pihlajaniemi T, Dickson LA. Pope FM. et al. Osteogenesisimperfecta: cloning of a pro-a2(I) collagen gene with aframeshift mutation. J Biol Chem 1984;259:23941-4.

37 Byers PH, Barsh GS, Peterson KE, Holbrook DA. Rowe DW.Molecular mechanisms of abnormal bone matrix formation in

osteogenesis imperfecta. In: Veis A. ed. Chemistr'v atuZ biologyof minieralized connective tissues. Amsterdam: Elsevier/NorthHolland, 1981:213-22.

38 Sykes B, Francis MJO. Smith R. Altered relation of twocollagen types in osteogenesis imperfecta. N Erugl J Med1977:296:1200-3.

39 Pope FM, Nicholls AC. In: Veis A, ed. Chemtlistrs' anid biology ofmineralized connective tissues. Amsterdam: Elsevier/North Hol-land, 1981:223-7.

40 Barsh GS, Byers PH. Abnormal secretion of type I procollagenin a variety of osteogenesis imperfccta. Proc Natl Acuad Sci USA1981;78:5142-6.

41 Penttinen RP, Lichtenstein JR, Martin GR, McKusick VA.Abnormal collagen metabolism in cultured cells in osteogcncsisimperfecta. Proc Natl Acad Sci USA 1975:72:588-9.

42 Barsh GS, David K, Byers PH. Type I osteogenesis impertecta,a nonfunctional allele for proa(l) chains of type I collagen. ProcNatl Acad Sci USA 1982;79:3838-42.

43 Chu M-L, Williams CJ. Pepe G, Hirsch JL, Prockop DJ,Ramirez F. Internal deletion in a collagen gene in a perinatallethal form of osteogenesis imperfecta. Nature 1983:304:78-80.

44 Williams CJ, Prockop DJ. Synthesis and processing of a typc Iprocollagen containing shortened procsl(l) chains by fibroblastsfrom a patient with osteogenesis impcrfecta. J Biol Che,n1983:258:5915-21.

45 Trelstad RL, Rubin D, Gross J. Osteogenesis imperfectacongenita: evidence for a generalized molecular disorder ofcollagen. Lab Invest 1977:36:501-8.

46 de Wet WJ, Pihlajaniemi T, Myers J. Kelly TE, Prockop DJ.Synthesis of a shortened proa2(I) chain and decreased synthesisof proa2(I) chains in a pro band with osteogenesis imperfecta. JBiol Chem 1983;258:7721-8.

47 Bonadio J. Holbrook KA. Gelinas RE, Jacob J, Byes P1H.Altered triple helical structure of type I procollagen in lethalperinatal osteogenesis imperfecta. J Biol Chemn 1985;260:1734-42.

48 Byers PH, Bonadio JF. Steinmann B. Invited editorial com-

ment. Osteogenesis imperfecta: update and perspective. Amn JMed Genet 1984;17:429-35.

477

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/


49 Peltonin L, Palotie A, Prockop DJ. A defect in the structure oftype I procollagen in a patient who had osteogenesis imperfecta:excess mannose in the COOH-terminal propeptide. Proc NatlAcad Sci USA 1980;77:6179-83.

50 Byers PH, Shapiro JR, Rowe DW, David KE, Holbrook KA.Abnormal a2 chain in type I collagen from a patient with a formof osteogenesis imperfecta. J Clin Invest 1983;71:689-97.

51 Kan YW, Dozy AM. Polymorphism of DNA sequence adjacentto human fi globin structural gene: relationship to sicklemutation. Proc Natl Acad Sci USA 1978;75:5631-5.

52 Kan YW, Dozy AM. Antenatal diagnosis of sickle cell anaemiaby DNA analysis of amniotic-fluid cells. Lancet 1978;ii:910.

5 Tsipouras P, Myers JC, Ramirez F, Prockop DJ. Restrictionfragment length polymorphism ascribed with the proa2(I) geneof human type I procollagen. Application to a family with an

autosomal dominant form of osteogenesis imperfecta. J ClinInvest 1983;72:1262-7.

54 Tsipouras P, Borresen A-L, Dickson LA, Berg K, Prockop DJ,Ramirez F. Molecular heterogeneity in the mild autosomaldominant forms of osteogenesis imperfecta. Am J Hum Genet1984;36:1172-9.

5 Grobler-Rabie AF, Brebner DK, Vandenplas S, et al. Poly-morphism of DNA sequence in the human proa2(I) collagengene. J Med Genet 1985;22:182-6.

56 Brebner K, Grobler-Rabie AF, Bester AJ, Mathew CG, BoydCD. Two new polymorphic markers in the human proa2(I)collagen gene. Hum Genet 1985;70:25-7.

57 Pope FM, Cheah KSE, Nicholls AC, Price AB, Grosveld FG.Lethal osteogenesis imperfecta congenita and a 300 base pairgene deletion for an al(I)-like collagen. Br Med J 1984;288:431-4.

58 Cheah KSE, Stoker NG, Griffin JR, Grosveld FG, Solomon E.Identification and characterisation of the human type II collagengene (COL2A1). Proc Natl Acad Sci USA 1985;82:2555-9.

59 Sangiorgi FO, Benson-Chanda V, de Wet WJ, Sobel ME,Tsipouras P, Ramirez F. Isolation and partial characterization ofthe entire human proal(II) collagen gene. Nucleic Acids Res1985;13:2207-25.

6 Pope FM, Nicholls AC. Lethal osteogenesis imperfecta and a

gene deletion. Br Med J 1984;288:138.61 Sykes B, Ogilvie D. Lethal osteogenesis imperfecta and a gene

deletion. Br Med J 1984;288:1380-1.62 Sykes B, Ogilvie D. Lethal osteogenesis imperfecta. Is the

collagen gene deletion a normal variant rather than a causalmutation? J Med Genet 1985;22:138A.

63 Sykes B, Smith R, Vipond S, Paterson C, Cheah K, Solomon E.Exclusion of the al(II) cartilage collagen gene as the mutantlocus in type IA osteogenesis imperfecta. J Med Genet1985 ;22: 187-97.

Correspondence and requests for reprints to Dr F MPope, Dermatology Research Group, Clinical Re-search Centre, Watford Road, Harrow, MiddlesexHAl 3UJ.

478

copyright. on 23 M


http://jmg.bm

j.com/

J Med G


ecember 1985. D

ownloaded from

http://jmg.bmj.com/

Documents

Collagen genes and proteins in osteogenesis imperfecta · Clinical classification of osteogenesis imperfecta Osteogenesis imperfecta is a group of inherited collagen diseases in which