Age at onset–dependent presentations of premature hip osteoarthritis, avascular necrosis of the femoral head, or Legg-Calvé-Perthes disease in a single family, consequent upon a p.Gly1170Ser mutation of COL2A1

Authors


Abstract

Objective

To identify the genetic abnormality responsible for osteoarthritis (OA), avascular necrosis (AVN) of the femoral head, and Legg-Calvé-Perthes disease in a single family, and to determine factors responsible for the distinct phenotypes manifested by different family members.

Methods

Forty-two members of a 5-generation family were recruited and investigated. Diagnosis was made by independent orthopedic surgeons and radiologists. Histopathologic changes of the diseased tissue were examined. Linkage analysis was performed with markers spanning the COL2A1 locus. Haplotypes were constructed and mutation of the gene was detected. Structures of the wild-type and mutant proteins were modeled.

Results

Sixteen affected members were identified (5 with isolated precocious hip OA, 6 with AVN of the femoral head, and 5 with Legg-Calvé-Perthes disease). A p.Gly1170Ser mutation of COL2A1 cosegregated with the 3 diseases and was absent in controls. Of note, age at onset in relation to the closure status of the femoral head epiphysis was associated with the diseases, with Legg-Calvé-Perthes disease presenting prior to closure (at ages 6–14 years), AVN of the femoral head presenting during closure (at ages 15–18 years), and precocious OA of the hip presenting after closure (at ages 21–34 years). Molecular modeling predicted that the serine-to-glycine substitution loosens the helical structure of the protein.

Conclusion

The p.Gly1170Ser mutation of COL2A1 in the family described is responsible for pathology confined to the hip joint, which presents as isolated precocious hip OA, AVN of the femoral head, or Legg-Calvé-Perthes disease. Age at onset in relation to closure of the femoral head epiphysis appears to be a critical factor in determining disease pattern.

Osteoarthritis (OA), the most common adult rheumatic disease, is characterized by progressive cartilage degeneration (1). Avascular necrosis (AVN) of the femoral head is a consequence of impaired blood supply (2), while Legg-Calvé-Perthes disease is a form of AVN in growing children (3). These 3 pathologies are considered distinct entities. COL2A1 (Entrez gene ID: 1280) encodes the precursor of the type II collagen α1 chain, the most abundant cartilage component. A p.Gly1170Ser mutation of COL2A1 has been reported in 2 Taiwanese families with AVN of the femoral head (OMIM no. 608805) and in 1 Japanese family with Legg-Calvé-Perthes disease (OMIM no. 150600) (4, 5). We report our findings from 5 generations of a unique family in which the COL2A1 p.Gly1170Ser mutation resulted in 3 distinct disease phenotypes.

PATIENTS AND METHODS

The study was approved by the ethics committee of Sun Yat-Sen University and conducted in compliance with the Declaration of Helsinki. Written informed consent was obtained from all subjects. The proband (family member III-2), a 56-year-old woman, first reported groin pain and a limp at age 15 years. These symptoms gradually worsened until walking became very difficult. AVN of the femoral head was diagnosed at her local hospital, and a right total hip replacement was performed. Examination showed restricted movement of the left hip, and radiography confirmed femoral head collapse and subluxation. The family history identified similar symptoms in multiple members. The whole family was traced, and their medical records were reviewed. Diagnoses of Legg-Calvé-Perthes disease in member IV-4; of AVN of the femoral head in members I-2, II-6, III-2, III-6, III-14, III-18, III-27, and IV-2; and of hip OA in members II-13 and II-15 had been made at different hospitals, with bilateral hip replacement performed on members II-13, III-6, and III-18.

Forty-two family members were examined clinically. Samples of their blood were obtained for genotyping (Figure 1A) and detailed laboratory studies. Radiography and/or magnetic resonance imaging (MRI) of the hips were performed on 16 members, including 13 who had had groin pain and/or a limp and 3 who had requested the examination (Table 1). The remaining 4 members had previously undergone bilateral hip replacement (II-13, III-6, and III-18) or had refused consent for radiography/MRI (I-2) (Table 1).

Figure 1.

Genetic analyses and structural modeling of the protein. A, Pedigree of the family. Markers used for the linkage analysis were D12S1663, D12S85, and D12S368. Solid bars denote the disease gene–bearing haplotype (4;4;5). The proband (family member III-2) is indicated by an arrow. The question mark in the box for family member IV-10 signifies that he carried the mutation but had not developed any clinical symptoms or radiographic abnormality. B, Partial sequencing results from the forward strand. Arrow denotes the 3665G→A mutation in exon 50. C, A spacefill model of the “wild-type” type II collagen α1 chain. Atoms of chain “a” are shown in green and glycine 21 is shown in magenta. The “b” chain is yellow and the “c” chain is olive. The location of the region affected by the mutation and within 3Å of the α carbon atom of glycine 21 is shown, prior to mutation, in the enlarged local spacefill model of the atoms. D, The spacefill model of the mutated protein type II collagen α1 chain. The hydroxymethyl group added as a consequence of the glycine-to-serine change at position 21 (red) is located in the center of the helix and may therefore loosen the structure. The enlarged local spacefill model shows atoms within 3Å of the α carbon atom of serine 21 after mutation. LCP = Legg-Calvé-Perthes disease; ANFH = avascular necrosis of the femoral head; OA = osteoarthritis; Gly21:a.c = carbon atom of the C=O bond in glycine 21 on chain “a” of the triple superhelix; Gly21:a.o = oxygen atom of the C=O bond in glycine 21 on chain “a”; Gly21:a.ca = α carbon atom of glycine 21 on chain “a.”

Table 1. Phenotype details*
Family member/age/sexAge at onset, yearsAge at diagnosis, yearsMajor symptomsRadiographic findingsMRI findingsPhenotype group
  • *

    ND = not determined.

  • Diagnosed at another hospital at age 25 years as having avascular necrosis of the femoral head (ANFH), but diagnosed by the authors at age 31 years as having Legg-Calvé-Perthes disease (LCP).

  • Diagnosed at another hospital at age 28 years as having avascular necrosis of the femoral head, but diagnosed by the authors at age 64 years as having osteoarthritis (OA).

  • §

    Family member III-18 underwent total hip replacement at another hospital in 2003. Member I-2 refused consent for radiography or magnetic resonance imaging (MRI). Member III-6 underwent total hip replacement at another hospital in 2005. Member II-13 underwent total hip replacement at another hospital in 2003. None of these patients underwent radiography or MRI administered by the authors.

Diagnosed by the authors      
 IV-4/30/M69Limping/groin painSubluxation/resorptionLCP
 V-2/7/F66Irritable hip/limpingNormalHydrops articuliLCP
 IV-17/11/M1111Irritable hip/limpingNormalEpiphysis destructionLCP
 IV-8/15/F1215Irritable hip/groin painNormalEpiphysis destructionLCP
 IV-2/32/M1425/31Groin pain/limpingCystic/collapseCystic/collapseLCP
 III-14/32/F1526Groin pain/limpingCystic/scleroticCystic/sclerotic/necrosisANFH
 III-2/56/F1535Groin pain/limpingSubluxation/collapseANFH
 III-27/28/M1822Groin pain/limpingCystic/scleroticCystic/sclerotic/necrosisANFH
 II-6/65/M2128/64Groin pain/limpingJoint space narrowing/marginal osteophytesJoint space narrowing/marginal osteophytesOA
 II-8/62/M2362Groin pain/limpingJoint space narrowing/marginal osteophytesOA
 II-15/52/F2638Groin pain/limpingJoint space narrowing/marginal osteophytesOA
 III-26/36/F3435Groin painRefused consent for radiographySeparation/destruction of articular cartilageOA
 IV-7/17/FTransient groin painNormalNormal
 IV-10/8/MNDNDNone (radiography/MRI requested)NormalNormalPresymptomatic stage
 II-5/68/FNone (radiography requested)Normal
 III-29/29/MNone (radiography requested)Normal
Not diagnosed by the authors§      
 III-18/46/F1626Groin pain/limpingANFH
 I-2/91/F1645Groin pain/limpingANFH
 III-6/46/F1721Groin pain/limpingANFH
 II-13/58/F2435Groin pain/limpingOA

Initial investigations included radiography as well as clinical and laboratory tests to exclude pathology secondary to other causes, such as systemic lupus erythematosus, sickle cell disease, Gaucher's disease, alcohol abuse, or steroid therapy. The blood tests performed included tests for anti–double-stranded DNA and antinuclear antibodies for systemic lupus erythematosus, hemoglobin electrophoresis or high-performance liquid chromatography to detect hemoglobin S in sickle cell disease, and measurement of β-glucocerebrosidase activity for Gaucher's disease. Family members were subsequently re-examined by 2 independent orthopedic surgeons, and imaging results were reviewed by 2 independent radiologists (neither the surgeons nor the radiologists had any knowledge of the pedigree). The 3 phenotypic groups of hip OA, AVN of the femoral head, and Legg-Calvé-Perthes disease were diagnosed using internationally recognized criteria (6–10). Age at onset was defined as age at the first appearance of persistent or recurrent limping and/or persistent and progressive groin pain, not explained by another cause. During this time the proband underwent left total hip replacement, and the femoral head bone and cartilage were examined by light microscopy.

The pedigree demonstrated autosomal-dominant inheritance (Figure 1A). Linkage analysis was performed with the MLINK program under the autosomal-dominant model, with the parameters of disease allele frequency 10–4, 100% or age-dependent penetrance, and equal recombination rates in males and females (11). Since mutation of the COL2A1 gene had been implicated previously in AVN of the femoral head and Legg-Calvé-Perthes disease in separate families, the microsatellite markers D12S1663, D12S85, and D12S368 (Applied Biosystems, Foster City, CA) spanning an 8-Mb region across the COL2A1 locus were selected. Haplotypes were also constructed by Cyrillic 2.1 (http://www.cyrillicsoftware.com/). Mutation was detected by sequencing the polymerase chain reaction products of all exons and splice sites of the gene on both DNA strands on an ABI 3730XL instrument (Applied Biosystems). We also sequenced DNA from 50 healthy unrelated age- and sex-matched Chinese Han control subjects.

The secondary structures of collagen isoforms 1 and 2 were predicted using nnPredict (http://www.cmpharm.ucsf.edu/∼nomi/nnpredict.html) and Predict Protein (http://www.predictprotein.org/). Domain information was acquired from InterPro (http://www.ebi.ac.uk/interpro/ISpy?mode=single&ac=P02458). The homology modeling programs SWISS MODEL (http://swissmodel.expasy.org//SWISS-MODEL.html) and CPHmodels (http://www.cbs.dtu.dk/services/CPHmodels/) were used to develop an appropriate model to mimic the effects of the mutated region. An approximate crystal structure of the collagen triple helix [(Pro-Pro-Gly)10]3 “1K6F” was chosen from the Protein Data Bank to illustrate the outcome of the p.Gly1170Ser mutation (12). The glycine at position 21 in the “a” chain of the triple superhelix (Gly21:a) was replaced by a serine (Ser21:a).

RESULTS

Clinical characteristics.

Sixteen affected patients were identified (Figures 2A–C and Table 1), including 11 patients with previously diagnosed disease and 5 with new diagnoses (II-8, III-26, IV-8, IV-17, and V-2). Among the 16 patients who underwent hip radiography/MRI at our hospital, 12 were diagnosed by the 2 independent orthopedic surgeons and 2 independent radiologists (Table 1). Previous diagnoses made at different hospitals were confirmed, except for patients II-6 and IV-2, who, although they had been diagnosed elsewhere as having AVN of the femoral head, were reclassified as having OA and Legg-Calvé-Perthes disease, respectively. The remaining 4 patients had undergone bilateral hip arthroplasty or had refused consent for imaging examinations (Table 1). Although family member IV-7 reported groin pain, radiography and MRI showed no abnormality and the symptoms resolved spontaneously. Family member IV-10, although a carrier, was asymptomatic with normal radiography and MRI findings. Family members II-5 and III-29, also asymptomatic, requested radiography studies, the findings of which were normal (Table 1).

Figure 2.

Radiographic and histopathologic assessments. A, Radiography of the hip in patient IV-4. The patient was limping at age 6 years and was diagnosed as having Legg-Calvé-Perthes disease at age 9 years. The radiograph shows collapse and resorption of the femoral head (arrows), lateral subluxation, and decreased neck-shaft angle. B, Radiography of the hip in patient III-14. The patient started having groin pain at age 15 years. The radiograph shows cystic and sclerotic changes as well as partial collapse of the femoral head (arrows). The patient was diagnosed as having avascular necrosis of the femoral head. C, Radiography of the hip in patient II-6. The patient's age at onset of disease was 21 years. The radiograph shows joint space narrowing, marginal osteophytes and sclerosis, and cystic changes in the femoral head and acetabulum, without collapse (arrows). The femoral neck-shaft angle is normal. The patient was diagnosed as having precocious osteoarthritis. D, Histopathology of the femoral head bone of the proband (patient III-2), showing empty lacunae without osteocytes, fragmentation and dissolution of the bone matrix, and no local inflammatory infiltration. The patient was diagnosed as having avascular necrosis of the femoral head (hematoxylin and eosin stained; original magnification × 400).

Following further analysis, it became apparent that the different phenotypic groups in the 16 affected family members were associated with age at onset. Those ages 6–14 years presented with Legg-Calvé-Perthes disease, those ages 15–18 years presented with AVN of the femoral head, and those ages 21–34 years presented with precocious hip OA (Table 1). This relationship also existed among the 4 patients previously diagnosed elsewhere (Table 1). Clinical, radiographic, and laboratory investigations excluded secondary causes for the 3 diseases and found no evidence for skeletal abnormalities outside the hips.

Histopathology demonstrated empty lacunae without osteocytes, fragmentation and dissolution of the bone matrix, and no local inflammatory infiltration. The bone marrow and blood vessels between the trabeculae and related bone were necrotic (Figure 2D).

Linkage analysis and mutation detection.

Linkage analysis produced logarithm of odds scores of 4.21 at D12S1663 and 5.45 at D12S85 (both at θ = 0) and of 6.78 at D12S368 at θ = 0.05 with 100% penetrance, and of 3.91, 4.71, and 5.89, respectively, at age-dependent penetrance. All affected family members and family member IV-10 shared the same disease gene–bearing haplotype (4, 4, 5) (Figure 1A). Sequencing of both DNA strands revealed a 3665G→A transition (GenBank accession no. NM_001844) at exon 50, resulting in a p.Gly1170Ser change at the protein level (GenBank accession no. NP_001835) (Figure 1B). The mutation was detected in, and cosegregated with, all the affected individuals who exhibited localized hip OA, AVN of the femoral head, or Legg-Calvé-Perthes disease. It was absent in the controls and unaffected family members, except for family member IV-10, who at age 8 years carried the mutation but had not developed any clinical symptoms or radiographic abnormality (Figure 1A).

Structural modeling of the mutant protein.

The protein COL2A1_HUMAN has 3 isoforms (http://www.expasy.org/uniprot/P02458). The mutation at amino acid position 1170, resulting in a change from glycine to serine in isoform 2, also occurred in isoform 1 at position 1101. Our domain search and secondary structural modeling did not reveal significant change in the mutant protein. However, it suggested that the serine-to-glycine substitution disrupts local structures, generating a loose superhelix (Figures 1C and D).

DISCUSSION

The cosegregation of the p.Gly1170Ser mutation with affected individuals and its absence in unaffected family members (except for family member IV-10) and controls demonstrates its causative role in the disease phenotypes seen. The cosegregation of the mutation with 3 different phenotypic groups suggests that the abnormality in type II collagen is an important factor in the development of the 3 diseases, and it even suggests that they may represent one entity by etiology and underlying mechanism in this family. The fact that family member IV-10 carries the mutation without abnormality suggests that he may be in the presymptomatic stage. The phenotypic variations between individuals cannot be explained by environmental factors, since family members had been living in separate provinces with different environmental exposures. However, the association between the age at onset and different phenotypic groups is apparent and coincides with the growing stages of the femoral head epiphysis.

The data show that the epiphysis closes at ages 15–17 years in girls and at ages 17–19 years in boys in the Chinese population (13). The ages at onset of our patients with Legg-Calvé-Perthes disease are consistent with those reported for a Japanese family with the disease (5). Thus, Legg-Calvé-Perthes disease emerged prior to the closure of the femoral head epiphysis, AVN of the femoral head emerged during the closure, and precocious hip OA emerged after the closure (Table 1). This finding suggests that the patient's age, and the associated developmental stage of the femoral head epiphysis at the time that the COL2AI mutation becomes significant, has a direct bearing on the disease phenotype. This in turn will inform the choice of treatment and the prognosis.

COL2A1 mutations have been reported to be associated with a wide range of diseases including spondyloepiphyseal dysplasia, achondrogenesis-hypochondrogenesis type II, and Stickler syndrome type I (www.genecards.org/cgi-bin/carddisp.pl?gene=COL2A1). A further paradox surrounding the p.Gly1170Ser mutation is that it results in isolated hip pathology, while most type II collagenopathies are associated with more diverse disease. To date, the reported COL2A1 mutations causing isolated bilateral hip disease are p.Gly1170Ser and p.Gly717Ser (4). Both mutations replace the glycine in the Gly-X-Y repeats of the protein with a serine residue. Since glycine and serine are polar, uncharged amino acids, it is not surprising that the serine-to-glycine substitution fails to impart secondary structural changes. However, the mutation adds a large hydroxymethyl group (CH2OH) into the center of the superhelix, which is predicted to disrupt the local structure and thereby loosen the superhelix, resulting in the pathologies. The fact that the mutation leads to disease confined to the hip may reflect the predominant weight-bearing function of the joint. However, the mechanism(s) by which the same mutation results in 3 separate phenotypic groups during different stages of epiphyseal development, and impairs the femoral head blood supply, remains to be determined.

The data presented here from a single family suggest that a detailed family history is required in the investigation of patients presenting with early-onset isolated hip OA, AVN of the femoral head, or Legg-Calvé-Perthes disease, and that COL2A1 mutations should be sought. Since de novo mutations may occur, this should also apply to sporadic cases. Hence, identification of a mutation, in addition to providing evidence of carrier status, also allows evaluation of risk. Moreover, the fact that the same p.Gly1170Ser mutation has been identified in families from different populations implies that the 3665G of the gene is a mutational hot spot, and that it should be given priority in mutation detection.

Our findings not only confirm the causal role of the mutation in AVN of the femoral head and Legg-Calvé-Perthes disease previously reported in separate families, but they also (together with the findings of previous investigators) suggest that the p.Gly1170Ser mutation represents a new type II collagenopathy, with pathology confined to the hip and manifesting as 3 distinct phenotypic groups, dependent upon age at onset. This revelation and the demonstration of the structural changes of the mutant protein also add further insight into the possible mechanisms of the 3 diseases.

AUTHOR CONTRIBUTIONS

Drs. Huang and Y. Wang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Su, Li, Zhou, X. Wang, Mason, Huang, Y. Wang.

Acquisition of data. Su, Li, Liu, Mow, Huang, Y. Wang.

Analysis and interpretation of data. Su, Li, Liu, Zhou, Patil, Mow, Mason, Huang, Y. Wang.

Manuscript preparation. Su, Li, Liu, Zhou, X. Wang, Patil, Mow, Mason, Huang, Y. Wang.

Statistical analysis. Li, X. Wang, Huang, Y. Wang.

Ancillary