Annals of the Rheumatic Diseases 1994; 53: 789-797 789

REVIEW Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from

Genetic aspects of familial

Sergio A Jimenez, Rita M Dharmavaram

Human osteoarthritis (OA) is a heterogeneous may be responsible for the premature and multifactorial disease characterised by the and generalised degeneration of the tissue progressive deterioration of the cartilage of matrix. The abnormal genes could include the diarthrodial joints. Multiple aetiological and genes for cartilage matrix macromolecules, for pathogenetic mechanisms have been impli- enzymes involved in the biosynthesis of matrix, cated in its development and progression.' In for hormone and growth factor receptors in many instances OA is an acquired process chondrocytes, or for enzymes involved in the secondary to various metabolic, mechani- metabolic degradation of the tissue. Recent cal, or inflammatory-immunological events. evidence, however, suggests that the genes However, it has long been recognised that encoding the collagenous components of several distinct forms are inherited as dominant cartilage matrix are the most likely candidates. traits with a Mendelian pattern.2 3 The most The collagens represent the most abundant common form of inherited OA is characterised protein of articular cartilage matrix, com- by the presence of Heberden's and Bouchard's prising about 50% of the dry weight of the nodes and the concentric or uniform de- tissue. These molecules play a crucial role generation of the articular cartilage of several in the maintenance of the biomechanical joints, particularly the hips and knees.4 Many properties of cartilage, being responsible for studies have examined the genetic factors that the tensile strength and shear stiffness of the may be associated with either development or tissue. The remarkable complexity of the severity of this form of OA. Analysis of the organisation of the collagenous components in frequencies of HLA antigens in various popu- articular cartilage has recently become lations of patients affected have yielded apparent'7 18 and at least five different collagen conflicting results,5-9 although in one study an types representing the products encoded by at increased frequency of the HLA-A1B8 least 10 distinct genes have been recognised in phenotype was observed.9 This study also the tissue. Although the cartilage-specific type

examined the frequency of (x l-antitrypsin II collagen is the most abundant collagen http://ard.bmj.com/ phenotypes in these individuals and found a species in the tissue, it is very likely that the significant increase in the MZ phenotype,9 other collagen types (types VI, IX, X, and XI) although these results are unlikely to be related also play important structural and functional to the primary genetic defect in this form of roles. The normal supramolecular assembly of OA. A second type of inherited OA is familial the various cartilage collagens serves as a chondrocalcinosis-a disease in which calcium mechanical constraint to limit the expansion of

pyrophosphate dihydrate (CPPD) crystals are proteoglycans and their tightly bound water on September 24, 2021 by guest. Protected copyright. deposited in fibrous and hyaline cartilage.'0 molecules into the large hydrodynamic The observations that the degenerative domains characteristic of proteoglycans in free occasionally precedes or is not associated with solution. A failure of this collagenous assembly demonstrable deposition of CPPD crystals" would result in swelling of the proteoglycans, and that there is a physical association of increased tissue water, softening of the matrix CPPD crystals with components of cartilage and eventual degeneration of the cartilage. matrix'2 13 have led to the suggestion that There is, therefore, compelling evidence to cartilage matrix abnormalities may be a suggest that cartilage collagen genes may be primary common event leading to cartilage those at fault in heritable OA. Several studies degeneration, CPPD crystal deposition, or have identified mutations in the genes both. A third familial form of OA is known as encoding type I and type III procollagens in Stickler syndrome or hereditary arthro- patients with other heritable disorders that ophthalmopathy. 4 This syndrome is cause mechanical failure of the connective Jefferson Medical characterised by ophthalmological manifesta- tissues, such as .'9 College, as Thomas Jefferson tions such progressive vitreoretinal degen- These investigations have provided a number University, eration and severe myopia, peculiar facial of successful strategies for the identification of Philadelphia, features, and premature degenerative joint structural mutations in procollagen genes. PA 19107, USA disease. Other heritable S A Jimenez disorders accom- However, as discussed in more detail below, R M Dharmavaram panied by premature OA include hydroxy- many of these strategies have limitations when Correspondence to: apatite deposition disease'5 and certain forms applied to the study of mutations in the Dr S A Jimenez, of multiple epiphysial dysplasias.'6 cartilage collagen genes, such as the limited Bluemle Iife Sciences Building, The pattern of inheritance of these diseases availability of cartilage, the difficulty in 233 South 10th Street, is consistent with the hypothesis that mutations expanding chondrocyte populations in vitro, Rm 509, Philadelphia, in one or more of the genes encoding the the loss of chondrocyte-specific phenotype PA 19107-5541, USA. macromolecular components of articular during culture, and the difficulty in performing 79070imenez, Dharmavaram

extensive protein characterisation from small expression of hereditary arthro-ophthal- tissue samples. The following sections will mopathy or Stickler syndrome. In the study by Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from review some of the strategies that have been Knowlton et al,29 three large Stickler syndrome successfully utilised or that may be potentially families were analysed for co-inheritance of the useful to identify collagen gene mutations in clinical manifestations with the Hind-III and inherited diseases affecting articular cartilage. the variable number tandem repeat (VNTR) polymorphisms in COL2A1. Genetic linkage between the disease phenotype and COL2A1 Restriction fragment length was demonstrated in the largest family. The polymorphism (RFLP) analysis results from the second family also supported The development of recombinant probes that linkage to COL2A1, leading to the conclusion detect polymorphic sites in human DNA by that mutations in the COL2A1 gene are RFLP analysis has made available a vast responsible for the disease in these two resource of genetic markers to follow the families. In conrast, in the third family, re- inheritance of specific DNA sequences in combination between clinical expression and families.20'22 These polymorphic sequences COL2A1 was demonstrated suggesting, there- occur frequently in the flanking regions ofmost fore, that the syndrome may be heterogeneous genes, as well as randomly throughout genomic and that in certain families a gene other than DNA; their detection has made it possible to COL2A1 may be the defective gene. identify abnormal alleles of many genes and to Subsequently, Knowlton et al0 dem- trace their pattern of co-segregation with a onstrated co-inheritance of a phenotype of given disease phenotype in families. The premature OA and a mild chondrodystrophy premise for RFLP study of inherited forms of with polymorphisms in the type II procollagen OA is that, despite the heterogeneity and gene. Similar results were obtained in another variability of their phenotype, they are caused family with primary generalised OA.3" by a primary genetic defect that resides at However, a large study of 61 patients with a specific chromosomal locus. The disease primary generalised OA failed to show a phenotype and the defective gene must, significant difference in the frequencies of one therefore, map to the same chromosomal or more COL2A1 alleles between affected and location as reflected by their co-segregation in control individuals.32 These results thus families. Using appropriate restriction enzymes excluded the possibility that there is a common it becomes feasible to map the inherited trait mutation at COL2A1 that is responsible for the to a region in the chromosome bracketed by disease phenotype. In a more recent study, two specific markers. Weaver et al" analysed the inheritance pattern The potential benefits of utilising RFLP of Hind-III and Hinf-I RFLPs of COL2A1 in genetic linkage analysis in heritable forms of a large five generation family with multiple human OA is that it will permit the epiphysial dysplasia (MED). The data

identification of the genes at fault in these indicated that the segregation patterns of http://ard.bmj.com/ disorders. Even the finding of linked genetic COL2A1 and MED phenotype were dis- markers at some distance from the responsible cordant, with recombination observed in a gene would allow definition ofthe approximate minimum of two meioses in the portion of the chromosomal location of the abnormal gene family tested. From these results, it was and its cloning and sequencing. In addition, determined that the MED locus in this family informative RFLPs can be tested in families does not map within 2 centimorgans (cM) of

suffering from diseases with similar phenotypes COL2A1, therefore excluding this gene as the on September 24, 2021 by guest. Protected copyright. to determine if the same genetic defect is site of the mutation responsible for the disease present in all cases or if there is genetic phenotype. RFLPs in the type VI collagen gene heterogeneity among the various phenotypes. were also tested for linkage with MED in this RFLP analysis can also give valuable negative family. Recombination between COL6A2 and informnation, as the absence of co-segregation COL6A3 genes and MED was observed, also (i.e. recombination) with the disease pheno- ruling them out as candidate genes. Moreover, type can exclude the candidate gene as the site the data excluded MED from a range of 5 cM of the defect in a given family. on either side of the RFLP marker for Recently, several polymorphisms in the COL6A1, effectively excluding COL6A1 as genes for some of the collagens expressed in the candidate gene. articular cartilage and in surrounding DNA Despite the enormous potential of RFLP sequences have been identified. For example, analysis for the identification of the defective polymorphisms in the genes for type II genes in inherited diseases, the method has procollagen (COL2A1), the a2-chain of type certain limitations. For example, RFLP IX collagen (COL9A2), and type X collagen analysis depends on the ability to identify large (COLlOAl) have been described.2"27 The families in whom the disorder is clearly identification of these polymorphisms has per- inherited, and from whom blood or tissue mitted the application of RFLP analysis to test samples can be obtained from affected and the possibility that structural mutations in non-affected members of three or more various collagen genes expressed in cartilage generations. Also, RFLP analysis does not may be responsible for' the biomechanical allow the identification of the exact mutation failure of articular cartilage in familial OA. For responsible for the disease phenotype. example, Francomano et al28 and Knowlton Recently, several polymerase chain reaction et a129 demonstrated co-inheritance of certain (PCR) based techniques have been used to polymorphic sites in COL2A1 with the type DNA polymorphisms. One method relies Genetic aspects offamilial osteoarthritis 791

on the fact that in the human genome there are We have previously reported the ampli- 50 000-100 000 interspersed (CA)n blocks fication of the entire coding region of the Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from with n being 15-30. The function of such type X collagen gene, using PCR and SSCP dinucleotide repeats or microsatellites is analysis based on the previously reported unknown, but it is known that the number of sequence.26 We identified a Gly to Arg repeats within DNA varies among the individ- substitution of the carboxyl (C-) terminal of uals within a species.34 It was subsequently the protein that results in a loss of a restriction shown that blocks of human (CA)n repeats enzyme site. Restriction analysis of a large exhibit length polymorphisms and that these MED family ruled out this sequence change as length polymorphisms could be utilised to a cause of disease. Using a similar approach, follow allele co-segregation occurring during Sweetman et al27 have investigated if mutations meiosis.35 Subsequently, other short tandem in COLlOAl are responsible for diverse repeats such as tri- and tetranucleotide repeats, heritable cartilage diseases including achon- although less abundant than (CA)n markers, droplasia, , pseudoachon- have provided an additional source of highly droplasia and thanatotrophic dysplasia. They informative markers.36 The procedure for identified seven sequence changes in the typing the (CA)n block markers and other coding and flanking regions of COLlOAl in simple sequence repeats is now widely utilised, affected individuals. However, six of these mainly because of its speed-essentially it changes did not co-segregate with the involves only two steps: amplification of a expression of these diseases and were, there- DNA fragment by PCR in the presence of fore, polymorphisms rather than true muta- radioactive nucleotide and elecrophoresis on a tions. The sequence changes were also used to denaturing polyacrylamide gel. demonstrate discordant segregation between In the MED family studied by Weaver et al,33 the COLlOAl gene and and described above, in whom COL2A1 and pseudoachondroplasia. A seventh sequence COL6A1 were excluded as candidate genes, change that results in a Val to Met substitution such dinucleotide repeat and simple sequence at the C-terminal was only found in two indi- repeat polymorphisms have been utilised for viduals with hypochondroplasia from the same mapping of the disease locus to the family. Segregation analysis of the family was pericentromeric region ofchromosome 19.37 In inconclusive because of uncertainty in the another study, of nine multigenerational diagnosis of one of the family members. families with pseudoachondroplasia, a number Recently, SSCP analysis followed by of cartilage specific and non-cartilagenous genomic DNA sequencing has been used for extracellular matrix candidate genes including identification oftwo mutations in COLlOAl in aggrecan, proteoglycan link protein, small families with autosomal dominant Schmid cartilage matrix protein and various collagen metaphysial chondrodysplasia. One of them is genes have been excluded by RFLP analysis. a 13 bp deletion of 1857-1869 bp and the 1771 However, pseudoachondroplasia in these other is a T- C change at position bp http://ard.bmj.com/ families has been mapped to chromosome 19 that results in a Cys -Arg substitution at utilising chromosome 19 microsatellite amino acid 591 of the protein.404' markers on that chromosome.38 Another PCR based method that has been Another PCR based method that is being successfully utilised for the identification of utilised as a screening method for detection of mutations in the cartilage collagen genes is DNA sequence changes is the single strand known as 'heteroduplex analysis'. This

conformation polymorphism (SSCP) analysis. procedure allows detection of mismatches on September 24, 2021 by guest. Protected copyright. The method involves amplification by PCR between double stranded DNA containing one of a discrete segment of genomic DNA in wild type strand and a complementary strand the presence of radiolabelled nucleotides, from a mutant gene. For this analysis, temperature denaturation of the PCR amplified PCR products of a test sample and products, and analysis of single strands on a wild type control are mixed, heated to denature non-denaturing polyacrylamide gel. Poly- the double stranded DNA and allowed to morphic differences in strand mobility result reanneal at a lower temperature. If the primary from the effects of primary sequence changes sequences of the sample and control differ, on the folded structure of a single DNA strand. 50% of the reannealed double stranded DNA The primary sequence differences alter the will be heteroduplex DNA. When separated on intramolecular interactions that generate a a non-denaturing polyacrylamide gel, the three dimensional folded structure. The reannealed products of the PCR reactions molecules may thus move at different rates migrate in a distinct pattern: homoduplex through a non-denaturing polyacrylamide gel. strands of mutant and control DNA migrate as Since the conformational variations are subtle, a function of their length; heteroduplexes the success of any particular SSCP analysis formed of one mutant strand and one wild type experiment depends on several factors strand migrate at a different rate through the including the sequence of the DNA fragment gel because the region of mismatch forms a being evaluated and optimisation of the 'kink' in the DNA. Thus a heteroduplex experimental conditions to maximise differ- frequently appears on the gel as a distinct band, ential migration among fragments. Under separate from the homoduplex DNA. Several optimal conditions the efficiency of SSCP to factors affect the resolution of single base detect DNA sequence differences such as poly- mismatches in heteroduplex DNA, including morphisms or mutations has been reported to DNA size (200-600 bp being optimal), be close to 100%.39 position of mismatch within the DNA 7927Jimenez, Dharmavaram

fragment (central position being most easily Identification oftype II procollagen gene Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from detected), and type and context of mismatch. mutations by analysis ofthe The advantage of heteroduplex analysis is that corresponding complementary DNA it is technically simple, does not require (cDNA) complex chemical or temperature gradients, Each of the mutations described above was and bands can generally be detected by identified by sequencing genomic DNA from ethidium bromide staining. Utilising this affected individuals. However, sequencing the method we have recently reported a mutation large genes encoding for the cartilage specific that results in a 10 bp deletion from collagens is time consuming, or is not possible 1867-1876 bp in a five generation family with because of the incomplete knowledge of their Schmid metaphysial chondrodysplasia.2 structure, therefore identification of mutations by sequencing the corresponding cDNAs has many practical advantages. This approach has Identification oftype II coliagen gene been very successful and has resulted in the mutations in affected individuals by identification of mutations in the genes genomic sequencing encoding for the structural connective tissue Determination of the sequence of the type II components in a large number of families with collagen gene has proved successful in certain various forms of osteogenesis imperfecta or the forms of heritable OA. For example, Ehlers-Danlos syndrome, and in some families sequencing exons 2-52 of the gene identified with vascular aneurysms.'9 The approach a single base mutation that results in the adopted in most ofthese instances has been the substitution of Cys for Arg at position 519 of determination of the nucleotide sequences of the ot 1(II) procollagen chain in affected cDNAs prepared by reverse transcription of members of a family with primary generalised total cellular RNA from cultured cells OA associated with mild chondrodysplasia.43 A (fibroblasts or amniotic cells) from affected mutation that introduced a stop codon in individuals. Many of these studies were greatly COL2A1 corresponding to position 732 of the facilitated by the previous identification and ox 1 (I) procollagen was found in affected characterisation ofprotein abnormalities which members of a family with Stickler syndrome," pointed to the likely nature of the mutations and two different mutations in exon 48 of and to the approximate regions within the gene COL2A1 have been found in two cases of containing the mutations. These protein spondyloepiphysial dysplasia.45 46 Further- abnormalities included the demonstration of more, a COL2A1 mutation that results in a Gly overmodification ofpost-translational reactions, to Ser substitution at residue 943 in the 'protein suicide' as a result of decreased C-terminal domain of the triple helix has been procollagen thermal stability, the finding of identified in a case of /hypo- disulphide bonded ct-chain dimers resulting chondrogenesis, a lethal perinatal form of from the presence of cysteine residues in the

short-limbed .47 The number ofnewly triple helical domains, or the demonstration of http://ard.bmj.com/ identified mutations in COL2A1 responsible shortened collagen ca-chains. for a variety of heritable diseases affecting Although this method allowed the cartilaginous structures has increased rapidly identification of a number of mutations as the methods for gene sequencing have affecting types I and III collagen genes, its improved in accuracy and speed. The success depends on the availability of large spectrum of mutations in COL2A1 that had numbers of cells expressing the transcripts of

been identified by late 1993 are listed in the interest, in order to obtain sufficient amounts on September 24, 2021 by guest. Protected copyright. table. of RNA for the abnormal protein, and on the information obtained from structural studies of the mutated protein. The availability of Mutations in the type IIprocollagen gene identified in heritable diseases ofcartilage (as cells that express the transcripts from the reported by late 1993) mutated genes is not a problem for Disease Mutation Reference investigation of mutations affecting types I and III collagen genes, because cultured SED Deletion AA 964 to 999 (exon 48) 45 Achondrogenesis II/ Gly943' Ser (exon 46) 46 dermal fibroblasts or amniotic cells can be successfully utilised. However, in the case of SED Duplication AA 970 to 984 (exon 48) 47 OA with mild Arg"9 Cys (exon 31) 43 mutations in articular cartilage collagen genes, chondrodysplasia this approach has been obstructed by the SED tGly997 - Ser (exon 48) 49 to Stickler syndrome Arg732 Stop (exon 40) 44 difficulty in obtaining sufficient tissue Hypochondrogenesis Gly 74 Ser (exon 33) 50 isolate RNA and the loss of cartilage Stickler syndrome Arg9 - Stop (exon 7) 51 on SED tAr.t' Cys (exon 41) 52 phenotype when chondrocytes are cultured SED Arg5' Cys (exon 11) 53 plastic substrata in vitro.59 SED Gly247 Ser (exon 19) Ritvaniemi et al (pers. comm.) SEMD Gly'54 Arg (exon 15) 54 Stickler syndrome Frame shift (exon 40) 55 SED G+5 IVS20 56 Amplification and sequencing ofhuman SED Gly4`' Ser (exon 30) Katzenstein et al (pers. comm.) Hypochondrogenesis Gly5833 Glu (exon 43) 57 type II collagen cDNA from freshly Kniest/Stickler syndrome Deletion IVS 12 58 isolated and from cultured human fetal ED/POA Gly'71 'Ser (exon 48) Williams et al (pers. comm.) Hypochondrogenesis Gly"' Arg (exon 48) Ganguly et al (pers. comm.) and adult chondrocytes Achondrogenesis II/ Gly"' Arg (exon 38) Williams et al (pers. comm.) The discovery of a specific DNA polymerase hypochondrogenesis that is capable of reiterative synthesis of DNA All these mutations were identified by genomic DNA sequencing except for t identified by sequencing copies fronT a given template at high of 'illegitimate' transcripts of the gene. SED = spondyloepiphysial dysplasia; SEMD = spondylo- epimetaphysial dysplasia; ED/POA = Epiphysial dysplasia and precocious OA. temperatures has resulted in the development Genetic aspects offamilial osteoarthritis 793

A Thus to sequence the entire coding region for 2 3 4 5 6 the mutant allele is feasible; single base Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from mutations or small insertions/deletions in coding or regulatory sequences can be identified; mutations in sequences important for splicing can also be detected utilising oligonucleotide primers specific for intronic sequences at the intron-exon boundaries. Thus the PCR amplified products will correspond to the coding sequences and to adjacent consensus sequences required for splicing. We have successfully used this approach and have amplified the entire type II collagen cDNA, a length of 4-4 kb, in only six fragments as shown in figure 1, in which the length and position of the PCR fragments shown in A are illustrated in B. Sequencing of these PCR fragments indicated that their sequence was identical to that of correspond- ing regions of the human type II collagen B cDNA reported previously.6' The RNA used in this experiment was obtained from freshly isolated 24 week old fetal chondrocytes. In *--- N --o -- Triple helix 'I- C order to determine the minimum amount of E E RNA required to amplify the entire type II I - collagen cDNA, we performed PCR amplifi- ATG X 500 bp X TA A cation of RNA in concentrations ranging from 1 ,ug to 0.1 ng. We found that 1 ng of RNA was the smallest amount required to obtain _ 2 3 4 5 6 amplification of each PCR fragment. In this Figure 1 Amplification ofthe entire type II collagen cDNA from humanfetal experiment, 1 ,ug of RNA was obtained from chondrocytes. cDNA was amplified as described in the text and 10% ofthe PCR product 3 X 105 freshly isolated human fetal chondro- was loaded on a 1% (w/v) agarose gel and stained with ethidium bromide. Lambda DNA cytes; therefore RNA equivalent to that digested with Hind-III was used as a marker. A: Amplification ofcDNA was carried out obtained from only 300 chondrocytes was using thefollowing primers: LANE 1 11-1/11-2; LANE 2 11-3/11-4; LANE 3 11-5/11-6; LANE 4 11-7/11-8; LANE 5 11-9/11-10; LANE 6 11- 1/11-12. B: Map oftype II collagen cDNA sufficient to obtain amplification of one indicating the length andposition ofthefragments shown in A. bp = Base pairs. fragment, or 1800 chondrocytes for amplifi- cation of the six fragments encompassing the

entire type II collagen cDNA. Amplification of http://ard.bmj.com/ human type II collagen cDNA was also of one of the most powerful techniques of obtained from chondrocytes from a small molecular biology.60 The PCR reaction hEas postmortem specimen of articular cartilage been successfully applied to cloning anid from a 75 year old woman immediately after sequencing of cDNAs and genes that were n(ot isolation (fig 2A), and after culture on porous amenable to the techniques previousl:Ly, hydroxyapatite ceramic crystals for 90 days

because ofeither the development ofsecondalry (fig 2B). Thus even the small amounts of on September 24, 2021 by guest. Protected copyright. structures or the lack of available enzymne articular cartilage obtained from arthroplasty restriction sites for proper cloning. Thhe or from arthroscopic surgery can be possibilities of identifying cartilage collagen successfully utilised to identify mutations in gene mutations using small amounts (of COL2A1, by use of PCR. cartilage or few cultured chondrocytes hasve been improved dramatically by the PCR. Thhe amplification of specific DNA sequences can Identification ofmutations that do not overcome the limitations introduced by thie affect the coding sequence of COL2A1 lack of sufficient amounts of articular cartilagge In addition to the benefits of RFLP genetic necessary to obtain RNA for generation (of linkage analysis to identify the genes that may cDNA libraries. In this approach, minulte be at fault in heritable OA, RFLP analysis can amounts of RNA can be reverse transcribed tto give valuable negative information because the obtain cDNAs, and these cDNAs can then bre absence of co-segregation (i.e. recombination) PCR amplified using sequence specifiic with the disease phenotype can exclude a oligonucleotides based on the known sequencce candidate gene in a given family. Knowlton et of full length cDNAs. Since the entirre al9 showed the presence of recombination in sequence of the type II collagen cDNA haas one of the families with Stickler syndrome they been published, it has become feasible tto studied. This observation conclusively amplify the entire coding sequence from thie excluded a mutation in COL2A1 as the cause cDNAs. The PCR products can be sequenced of the disease phenotype in this family. Similar directly or can be cloned into bacterial vectolrs analysis of another family with Stickler for standard sequencing procedures. Given the syndrome and of a large family with severe advances in direct DNA sequencing, larg3e premature OA associated with CPPD stretches of DNA sequence can now b'e deposition failed to show co-segregation of obtained from a single sequencing reactiora. COL2A1 polymorphisms with the disease 7947Jimenez, Dharmavaram

phenotype29 (Jimenez and Knowlton, unpublished observations). These results Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from suggest that mutations in genes other than COL2A1 may be responsible for the disease phenotype in certain subsets of heritable OA. In addition, mutations that cannot be detected by sequencing of the gene may be responsible for the disease. These could involve, for example, mutations that result in abnormal splicing of the mature transcripts. It is, apparent, therefore, that a careful search for mutations that do not affect the coding regions of the COL2A1 or mutations in other genes expressed in articular cartilage matrix should be carried out in families with OA. An example of this has recently been reported by Vikkula et al.62 in a family in whom RFLP analysis indicated linkage to COL2A1 but sequencing of the entire coding sequence of the gene failed to show any mutation. In this family it was suggested that a mutation in the regulatory (promoter) regions of the gene that would not be detected by sequencing of the coding region of the gene may be responsible for the disease phenotype. In the past, attempts to identify mutations in collagen genes other than COL2A1, such as those coding for IX, X, or XI collagens, would have been nearly impossible with the small amounts of articular Figure 2 Amplification ofthe type II collagen cDNA from adult human chondrocytes. RNA was isolatedfrom cartilage available to extract mRNA. We chondrocytes obtainedfrom knee cartilage of a 75 year old decided, therefore, to examine if the methods woman. Lambda DNA digested with Hind-III was used as developed to amplify cDNA for type II a marker. A: Amplification ofcDNA fromfresh chondrocytes was carried out using thefollowingprimer collagen could be successfully utilised to pairs: LANE I 11-3/11-4; LANE 2 II-9/II-10; LANE 3 11-7/II- amplify cDNA corresponding to the minor 8. B: Amplification ofcDNA obtainedfrom RNA isolated collagens present in the tissue, and attempted from chondrocytes cultured in porous hydroxyapatite ceramic microcrystals using thefollowingprimer sets: LANE to amplify fragments of type IX and type XI 1 11-9/Il-10; LANE 2 11-3/11-4; LANE 3 11-7/71-8. collagen cDNA. As shown in figure 3A, we

amplified 780 bp of the t1 (IX) collagen chain: http://ard.bmj.com/ the length and position of the fragments shown in A are illustrated in B. We also amplified the entire published sequence of the ox2(XI) collagen cDNA which corresponds to 67% of the entire cDNA for this chain (fig 4); again, the figure shows the fragments, and their

length and position. These studies dem- on September 24, 2021 by guest. Protected copyright. onstrated that a combination of PCR with primer-extension/reverse transcription can be utilised to amplify cDNA for the various articular cartilage collagens from mRNA obtained from fresh and cultured human chondrocytes. From the results shown above, it is apparent that the use of cDNA for identification of mutations in type II and other collagens expressed in articular cartilage has several advantages over genomic sequence analysis. First, the target area to be analysed is much smaller; for example, the entire type II procollagen gene is about 30 kb, whereas the corresponding cDNA is only 4A4 kb. Second, it allows easy detection of mutations in collagen genes the structure of which has not 77QC; been completely established, such as the genes encoding for types IX, X, and XI collagens. Third, cDNA sequencing allows easy detection of splicing abnormalities such as those which occur when exons are skipped or when Figure 3 Amplification oftype IX collagen cDNA from fetal chondrocytes. PhiXl74 aberrant splice sites are used. Finally, the DNA digested with Hae-III was used as a marker. A: ThefoUowingprimer sets were used results demonstrate that cDNA analysis of for amplification: LANE I IX-1/IX-2; LANE 2 IX-3/IX-4. B: Map oftype IX collagen cDNA indicating the length and position offragments shown in A. bp = Base pairs. cartilage specific collagens can be successfully Genetic aspects offamilial osteoarthritis 795

1 2 3 4 5 Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from

Figure 5 Amplification oftype II collagen cDNA from Epstein-Barr virus transformed lymphocytes. PhiXl 74 DNA digested with Hae-III was used as a marker. The primerpairs usedfor amplification were: LANE 1II-1/II-13; LANE 2 I-9/II-10; LANE 3 II- 1/11I-14.

shown in figure 5, we obtained amplification of N Triple helix c cDNA fragments corresponding to the amino- terminal (lane 1), the triple helical (lane 2), and the C-terminal (lane 3) domains of the type II 500 bp collagen cDNA from RNA extracted from EBV-transformed lymphocytes. Sequencing of 3 5 1575 bp indicated that the illegitimate transcripts were correctly spliced. We also 2 4 demonstrated, for the first time, illegitimate Figure 4 Amplification oftype XI collagen cDNA from humanfetal chonidrocytes. transcription of the genes encoding for types PhiXI 74 DNA digested with Hae-III was used as a marker. Amplification ofcDNA was IX and XI collagens in EBV-transformed carried out using thefollowingprimer sets: LANE I XI-1/XI-2; LANE 2 XI-3/XI-4; LANE 3 XI-5/XI-6; LANE 4 XI-7/XI-8; LANE 5 XI-9/XI-10. The lower map of typxe XI collagen lymphocytes, as we were able to amplify cDNA cDNA indicates the length and position ofthefragments. bp = Base pairs. fragments corresponding to the C-termini of both of these collagens. However, we were unable to obtain any amplification of cartilage accomplished from small amounts of RNA, as specific collagen cDNAs from RNA obtained the entire type II cDNA can be annplified from either from untransformed lymphocytes or as little as 6 ng of chondrocyte RN{A. from lymphocytes induced to proliferate by exposure to phytohaemagglutinin (results not shown). Because of the very small amounts of

New approaches to identify typ)e II correctly spliced mRNA in lymphoblasts (less http://ard.bmj.com/ coliagen gene mutations when articular than 1 mRNA molecule per 500-1000 cells), cartilage tissue is not available large amounts of starting material (at least 500 Despite the extraordinary sensiitivity of the ng RNA) and increased number of PCR cycles PCR, the method requires the avaLilability of at were required for amplification of cDNA from least minute amounts of cartiilage, or of low abundance transcripts such as those for cultured chondrocytes that moaintain their types II, IX, and XI collagens. The increased

differentiated phenotype. Furthermore, the number of PCR cycles could increase the error on September 24, 2021 by guest. Protected copyright. success of this approach depends (on the ability in amplified DNA sequences, as Taq DNA to generate cDNAs that cover the entire coding polymerase has a relatively high rate of single sequence, or on obtaining shorter cDNAs that base misincorporation. However, misincorpor- cover the region where the mutati4on is located. ated bases can be easily detected by sequencing Because these problems can oftein be difficult at least five separate clones from each of two to overcome, approaches that do not depend independent amplifications. Furthermore, this on the availability of cartilage many be required problem can be overcome by the use of direct in the majority of cases. Orie of these sequencing of PCR products with which the approaches was used recently tLo identify a authentic sequence is in vast excess over the mutation that results in a 4Gly to Ser sequences with misincorporated bases. substitution at residue 997 ofF the cx I (II) procollagen chain in a patient with spondyloepiphysial dysplasia con,genita, using Conclusion PCR amplification ofthe correspoinding cDNA The explosive advances in the application of from 'illegitimate transcripts' ol )tained from molecular biological methods have allowed the fibroblasts and Epstein-Barr virus (EBV)- identification of a large number of gene transformed lymphocytes.49 However, from mutations in various heritable diseases. The these studies it was not clear if tkie transcripts well known occurrence of certain subsets of contained the full length type II collagen OA displaying an autosomal pattern of mRNA, as only 621 bp of the cDNA were inheritance has rendered these diseases sequenced. We recently used this,approach to amenable to similar studies. amplify cartilage specific cDNAs from mRNA The methods reviewed here have already obtained from non-cartilaginous sources.26 As allowed the identification of a large number of 796 7Jimenez, Dharniavarati

mutations in the genes encoding articular DNA identification of a polymorphism that changes Gly to Arg at position 545 by single-strand conformation Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from cartilage collagens in patients with various polymorphism analysis.Biocheni Biophys ResCUonni 1992; inherited disorders of articular cartilage (listed 187: 420-4. 27 Sweetman W A, Rash B, Sykes B,ctal. SSCP and in the table) and will permit the rapid segregation analysis of the human type X collagen gene characterisation and sequencing of cDNA for (COIlOAl) in heritable forms of chondrodvsplasia. Ant ifHittn Geniet 1992; 51: 841-9. the various cartilage specific collagens from 28 Francomano C A, Liberfarb R M, Hirose T, ct a!. The small amounts of chondrocyte RNA, or Stickler syndrome: evidence for close linkage to the struc- tural gene for typeII collagen. Geniontiics 1987; 1: 293-6. from RNA obtained from non-cartilaginous 29 Knowlton R G, WeaverE J, Struyk A F,etal. Genetic sources. linkage analysis of hereditary arthro-ophthalmopathv (Stickler syndrome) and the type II procollagen gene. An; _i Hunt Geniet 1989; 45: 681-8. The expert assistance of R Tate and 0 Ma in the preparation 30 Knowlton R G, Katzenstein P L, Moskowritz R W, ctal. of the manuscript is gratefully acknowledged. Supported by Demonstration of genetic linkage of the typeI1 NIH Program Project Grant1-AR39740.PO procollagen gene (COL2AI) to primary osteoarthritis associated with a mild chondrodvsplasia.N\' Eng!7 Ales! 1990;322: 526-30. 31 Palotie A, Vaisarnen P, Ott J, ctal. Predisposition to familial I Moskowitz R W. Clinical and laboratory findings in osteoarthrosis linkedto typeII collagen gene.I aicct osteoarthritis. In: McCarty DJ, Koopman W J,cds. 1989; 1: 924-7. ArthnitLs and allied conditiOns, 12th ed. Philadelphia: Lea 32 Priestlev L, Fergusson C, Ogilvie D, ct ci!. A limited asso- &Febiger, 1993: 1735-60. ciation of generalized osteoarthritis with alleles at the type 2 Stecher R M, Hersh A H, Hauser H. Hebcrden's nodes. II collagen locus: COI 2A1. BrifRheituni 1991; 30: 272-5. The family history and radiographic appearance of large 33 WeaverE J, Summerville G P, Yeh G,et al. Exclusionoftvpc family. A ni _7Hlon Geniet 1953; 5: 46-69. II and type VI procollagen gene mutations in afive 3 KellgrenJ H, Lawrence JS, Bier F. Genetic factors in gencration familyNvith multiple epiphvseal dvsplasia. Ant generalized osteoarthritis. Ann Rheunpii Dis 1963; 22: 7 A/cdGcn,ct 1993;45: 345-52. 237-55. 34 Tvler-Smith C, Brown W R A. Structure of the major block 4 Marks S,J Stewart I M, Hardinge 1K. Primary osteoarthritis of alphoid satellite DNA on the human Y chromosome. of the hip and Heberden's nodes. Ann RhliniDIis 1979; 7 Al t/Bitt 1987; 195: 457-70. 38: 107-11. 35 Weber J L, May PE. Abundant classof DNA poly- 5 LawrenceJ S, Gelsthorpc K, Morell G. Heberden's nodes morphismsswhich can be typed using the polvmerase and HLA markers in generalized osteoarthritis. i chain reaction. AniifY HutipiGen7et 1989; 44: 388-96. Rheuniiatol 1983; 10 (suppl 9): 32-3. 36 Edsvards A, Civitello A, Hammond H A, Caskev C T. DNA 6 Brodskv T, Appelboam A, Govaerts Famev J P. Antigens typing and genetic mapping with trimeric and tetrameric HLA et nodosites d'Heberden. Acta Rheninatologica 1979; tandem repeats. AifiYHunt Gettet 1991; 49: 746-56. 3: 95-105. 37 Oehlmann R, Summerville GP, Yeh G, Weaver E J, Jimenez 7 Ercilla M G, Brancos M A, Breysse Y, etal. HLA antigens S A, Knowlton R G. Genetic linkage mapping of multiplc in Forestier's disease, ankylosing spondylitis, and polv- epiphyseal dvsplasia to the pericentromeric region of arthrosis of the hands. i7 Rheioniiatol 1977; 4 (suppl3): chromosome 19. Atis7 Huni Geziet 1994; 54: 3- 1 0. 89-93. 38 HechtJ T, Francmano C A, Briggs M D, et al. Linkagc 8 Benavides G, Cervantes A, Silva B, Katona G, Lardizabal of tvpical pseudoachondroplasia to chromosome 19. J. HLA and Heberden's nodes in Mexican mestizos. Clin GcetitnicCs 1993; 18: 661-6. Rheuniiatol 1985; 4: 97-8. 39 Sheffield V C, Beck S,J Kwitek A E, Sandstrom D W, Stonc 9 Pattrick M, Manhire A, Ward A M, Doherty M. HLA-A, E M. The sensitivity of single-strand confsormation B antigens and cx 1 -antitrypsin phenotypes in nodal polymorphism analysis for the detection of single base generalized osteoarthritis and erosive osteoarthritis. Ann substitutions. Gec1is,1ics 1993; 16: 325-32. Rheion Dis 1989; 48: 470-5. 40 Warman M L, Abbot M, ApteS S,ct al. Atype X collagen 10 Ryan L M, McCarty D J. Calcium pyrophosphate crvstal mutation causes Schmid metaphvseal chondrodvsplasia. deposition disease: pseudo-gout; articular chondro- .Natur7c Gecnct 1993; 4: 79-82. calcinosis. In: McCarty D J, Koopman W J, eds. Arthritis 41 McIntoshI, Hefferon T W, Warman M L, Olsen B R, anzd allied conditions, 12th ed. Philadelphia: Lea & Febiger, Francomano C A. Mutation of type X collagen 1993: 1835-55. (COLOAl) resulting in Schmid metaphvseal chondro- 11 Bjelle A 0. Morphological study of articular cartilage in dysplasia. Anii_7fHiti Geniet 1993; 52 (suppl): A1201. http://ard.bmj.com/ pyrophosphate arthropathy. Ann Rheionii Dis 1972; 31: 42 Dharmavaram R M, Elberson M A, Peng M, Kirson L A, 449-56. Kelly TE, Jimenez S A. Identification of a mutation in 12 Bjelle A 0. Cartilage matrix in hereditary pyrophosphate tyIpe X collagen in a familv with Schmid metaphNyseal arthropathy. _f Rheumiatol 1981; 8: 959-64. chondrodysplasia. Hunt MoI Genzet 1994; 3: 507 9. 13 Beutler A, RothfussS, Clayburne G, Sieck M, Schumacher 43 Ala-KokkoL, Baldwin C T, Moskowitz R W, Prockop D H R. Calcium pyrophosphate dihvdrate crystal deposition J. Single base mutation in the type II procollagen gene in synovium. Relationship to collagen fibers and (COL2AI) as a cause of primarv osteoarthritis associated chondrometaplasia. Arth/tins Rheuni 1993; 36: 704- 15. sith a mild chondrodysplasia. Prtoc Natl Accais Sc'i U'SA 14 Stickler G B, Belau P G, Farrell FJ, ct al. Hereditarv 1990;87: 6565 8.

progressive arthro-ophthalmopathy.Mavo Clinl Pr-oc 1965; 44 Ahmad N N, Ala-Kokko L, Knowlton R G, ct al. Stop on September 24, 2021 by guest. Protected copyright. 40: 433-55. codon in the procollagen II gene (COL2AI) in a family 15 Marcos J C, DeBenyacar M A, Garcia-Morteo 0, sith the Stickler svndrome (arthro-ophtalmopathv). I)ist Maldonado-Cocco J A, Morales V H, Laguena R P. .NatlAcad Sci USA 1991; 88: 6624-7. Idiopathic familial chondrocalcinosis due to apatite 45 Lee B, Vissing H, Ramirez F, Rogers D, Rimoin 10. crystal deposition. Ani7Med 1981; 71: 557-64. Identification of the molecular defect in a family svith 16 Spranger J. The epiphyseal dysplasias. Clin Orthop Rel Res spondyloepiphyseal dvsplasia. Science 1989; 244:978-80. 1975; 114: 46-59. 46 Tiller G, Rimoin D L, Murray L W, Cohn D H. Tandem 17 Mayne R. Cartilage collagens. Arthritis Rheinn 1989; 32: duplication within a type II collagen gene (COL2A1) 241-6. exon in an individual with spondyloepiphvseal dysplasia. 18 Eyre D R. Structure and function of cartilage collagens. In: Prsoc Natl Acad Sci USA 1990; 87: 3889-93. Brandt K D, ed. Cartilage changes in osteoarthliitis. 47 Vissing H, D'Alessio M, Lee B, Ramirez F, Godfrey M, Indianapolis: Indiana University Press, 1990: 12-6. Hollister D W. Glvcine to serine substitution in the triple 19 Prockop DJ. Mutations that alter the primary structure of helical domain of pro-ot I(11) collagen results in a lethal type I collagen.YBiolChe?n 1990; 265: 15349-52. perinatal form of short-limbed dwarfism. i Bit1t Clici 20 Botstein D, White R L, Skolnick M, Davis R W. 1989; 264: 18625-7. Construction of a genetic linkage map in man using 48 Pun Y L, Moskowitz R W, LieS,ce al. Clinical correlations restriction fragment length polymorphisms. Ant i Hunt of osteoarthritis associated with a single-base mutation Genet 1980; 32: 314-31. (Arginine5"' Cysteine) in type II procollagen gene. 21 Willard H F, Skolnick M H, Pearson P L, Mandel J-L. Arthritis Rheuni 1994; 37: 264-9. Report on the committee on human gene mapping by 49 Chan D, Cole W G. Low basal transcription of genes for recombinant DNA techniques. Crogenet Cell Gen1et 1985; tissue-specific collagens by fibroblasts and lympho- 40: 360-489. blastoid cells. Application to the characterization of a 22 Antonarakis S. Diagnosis of genetic disorders at the DNA glycine 997 to serine substitution in alpha 1(11) collagen level. NEnglif Med 1989; 320: 153-63. chains of a patient with spondyloepiphvseal dysplasia. 7 23 Sangiorgi F 0, Benson-Chanda V, de Wet W J, Sobel M E, Biol Chemi 1991; 266: 12487-94. Tsipouras P, Ramirez F. Isolation and partial 50 Horton W A, Machado M A, Ellard J, ec al. Characterization characterization of the entire human pro-l1(II) collagen of a type II collagen gene (COL2AI) mutation identified gene. Nucleic Acids Res 1985; 13: 2207 25. in cultured chondrocytes from human hypochondro- 24 Vaisanen P, Elima K, Pelotie A, Peltonen L, Vuorio E. genesis. Proc NatlAcad Sc USA 1992; 89: 4583-7. Polymorphic restriction sites of type II collagen gene: their 51 Ahmad N N, McDonald-McGinn D M, Zackai E H, et al. location and frequencies in the Finnish population. Hunit A second mutation in type II procollagen gene (COL2A1) Hered 1988; 38: 65-71. causing Stickler syndrome (arthro-opthalmopathy) is also 25 Weaver E J, Knowlton R G. A PvuII polymorphism near the a premature termination codon. Antif Hiont Ge?iet 1993; 5' end of the type II procollagen gene (COL2AI). Nucleic 52: 39-45. Acids Res 1989; 17: 6429. 52 Chan D, Taylor T K, Cole W A. Characterization of an 26 Dhamavaram R M, Peng M, Strawbridge R R, Jimenez S arginine 789 to cysteine substitution in alpha 1(II) A. Polymerase chain reaction amplification of the coding collagen chains of a patient with spondyloepiphyseal sequences of the type X collagen gene from genomic dvsplasia. 7 Bitt! Chliet 1993; 268: 15238-45. Genetic aspects offamilial osteoarthritis 797

53 Williams C J, Considine E L, Knowlton R G, et al. 267: 22522-6. Spondyloephyseal dysplasia and precocious osteoarthritis 58 Winterpacht A, Hilbert M, Schwaze U, Mundles S, in a family with an Arg75-Cys mutation in the pro- Spranger J, Zabel B U. Kniest and Stickler dysplasia Ann Rheum Dis: first published as 10.1136/ard.53.12.789 on 1 December 1994. Downloaded from collagen type II gene (COL2A1). Hum Genet 1993; 92: phenotypes caused by collagen II gene (COL2A1) defect. 499-505. Nature Genet 1993; 3: 323-6. 54 Vikkula M, Ritvaniemi P, Vuorio A F, Kaitila I, Ala-Kokko 59 Benya P D. Introduction and survey of techniques for L, Peltonen L. A mutation in the amino terminal end of chondrocyte culture. In: Maroudas A, Kuettner K, eds. the triple helix of type II collagen causing severe osteo- Methods in cartilage research. London: Academic Press, chondrodysplasia. Genomics 1993; 16: 282-5. 1990: 85-9. 55 Brown D M, Nichols B E, Weingeist T A, Kimura A E, 60 Saiki R K, Bugawau T L, Horn G T, Mullis R B, Erlich Sheffield V C, Stone E M. Type II procollagen gene H A. Analysis of enzymatically amplified beta-globin and mutation in Stickler syndrome. Am Jf Hum Genet 1992; HLA-DQ alpha DNA with allele-specific oligonucleotide 51 (suppl): A301. probes. Nature 1986; 324: 163-6. 56 Tiller G E, Weis M A, Eyre D R, Rimoin D C, Cohn D 61 Baldwin C T, Reginato A M, Smith C, Jimenez S A, H. An RNA splicing mutation (G + 51VS20) in the gene Prockop D J. Structure ofcDNA clones coding for human for type II collagen (COL2A1) produces spondylo- type II procollagen. The a1(II) chain is more similar to epiphyseal dysplasia congenita. AmJ Hum Genet 1992; 51 the al(I) chain than two other a chains of fibrillar (suppl): A37. collagens. BiochemJt 1989; 262: 521-8. 57 Bogaert R, Tiller G E, Weis M A, et al. An amino acid 62 Vikkula M, Palotie A, Ritvaniemi P, et al. Early-onset substitution (Gly853-Glu) in the collagen alpha 1(II) osteoarthritis linked to the type II procollagen gene. chain produces hypochondrogenesis. Biol Chem 1992; Arthritis Rheum 1993; 36: 401-9. http://ard.bmj.com/ on September 24, 2021 by guest. Protected copyright.