Zehra Mustafa

Osteoarthritis-Susceptibility Locus on Chromosome 11q, Detected by Linkage

We present a two-stage genomewide scan for osteoarthritis-susceptibility loci, using 481 families that each contain at least one affected sibling pair. The first stage, with 272 microsatellite markers and 297 families, involved a sparse map covering 23 chromosomes at intervals of approximately 15 cM. Sixteen markers that showed evidence of linkage at nominal P</=.05 were then taken through to the second stage, with an additional 184 families. This second stage confirmed evidence of linkage for markers on chromosome 11q. Additional markers from this region were then typed to create a denser map. We obtained a maximum single-point LOD score, at D11S901, of 2.40 (P=.0004) and a maximum multipoint-LOD score of 3.15, between markers D11S1358 and D11S35. A subset of 196 of the 481 families, comprising affected female sibling pairs, generated a corrected LOD score of 2.54 (P=.0003) for marker D11S901, with evidence for linkage extending 12 cM proximal to this marker. When we stratified for affected male sibling pairs there was no evidence of linkage to chromosome 11. Our data suggest that a female-specific susceptibility gene for idiopathic osteoarthritis is located on chromosome 11q.

Stratification Analysis of an Osteoarthritis Genome Screen—Suggestive Linkage to Chromosomes 4, 6, and 16

To the Editor: We have previously carried out a two-stage genomewide linkage screen for osteoarthritis (MIM 165720) susceptibility loci, using an affected-sibling-pair approach (Chapman et al. 1999). In stage 1 of this screen, we tested 272 microsatellite markers in 297 families, each of which contained at least one pair of siblings who had undergone hip-, knee-, or hip and knee–replacement surgery for primary osteoarthritis. Loci that demonstrated evidence for linkage at nominal P=.05 were then taken through to stage 2, in which they were tested against a further 184 families. Sixteen markers within nine genomic regions from stage 1 had evidence of linkage, at P=.05. When the data for stages 1 and 2 were combined, the P value decreased for 3 of the 16 loci (D2S202, D11S907, and D11S903) and was constant for a 4th (D11S901). We subsequently concentrated our analysis on the chromosome regions to which these markers map. To test these linkages further, we genotyped additional markers and obtained maximum multipoint LOD scores (MLSs) of 1.2 for chromosome 2 and 3.1 for chromosome 11. Because there is evidence, from epidemiological, twin, and segregation studies, that the genetic contribution to osteoarthritis differs between the sexes and between different joint groups (Lindberg 1986; Cooper et al. 1994; Kaprio et al. 1996; Chitnavis et al. 1997; Felson et al. 1998), we stratified our chromosomes 2 and 11 linkage data according to sex and site of osteoarthritis (hip or knee). This stratification indicated that the suggestion of linkage to chromosome 2 was principally accounted for by affected sibling pairs with hip osteoarthritis (MLS 2.2), whereas the suggestion of linkage to chromosome 11 was restricted to affected female pairs (MLS 2.8). Because this analysis highlighted substantial differences between the strata tested, we have now reanalyzed stage 1 of our genome screen, for the remaining 20 autosomes, to determine whether any regions harbor susceptibility loci that are obscured in the unstratified data set. We stratified our stage 1 data into the same six strata tested in our analysis of chromosomes 2 and 11: affected females only (132 families), affected males only (60 families), hips only (194 families), knees only (34 families), female hip (85 families), and male hip (44 families). (A more detailed breakdown of these families can be found in the study by Chapman et al. [1999].) We did not stratify for female knee or male knee, because the number of families was too small (16 and 4, respectively) to allow reliable inference of linkage. Multipoint linkage analysis was performed on the stratified data by means of the ASPEX program. Ten of the 20 autosomes have one or more multipoint peaks with uncorrected MLS⩾1.0 for one or more of the six strata tested (table 1). The highest MLS is 3.9, for chromosome 4q in the female-hip strata, followed by 2.9, for chromosome 6 in the hip-only strata, and 2.1, for chromosome 16 in the female-hip strata. When we adjust MLS values to correct for the seven models tested (one unstratified analysis and six stratified analyses), by deducting from the original values (Kidd and Ott 1984), chromosome 4 has an MLS value of 3.1, chromosome 6 has an MLS value of 2.1, and chromosome 16 has an MLS value of 1.3. The uncorrected multipoint plots of these three chromosomes are shown in figure 1. Figure 1 Multipoint analysis. A, Chromosome 4, female hip (n=85 families) and female only (n=132 families). B, Chromosome 6, hip only (n=194 families). C, Chromosome 16, female hip (n=85 families) and female only (n=132 families). Table 1 Stratified MLSs The suggestion of linkage on chromosome 4 is centered on 4q12–4q21.2 and is restricted to female pairs with hip disease. Roby et al. (1999) have recently reported linkage of chromosome 4q to severe early-onset hip osteoarthritis in a large pedigree of Dutch origin. This locus maps to the telomeric end of 4q (4q35), placing it >50 cM distal to the linkage that we have observed. It is therefore unlikely that the two linkages have detected the same locus. More than 50 cM of chromosome 6 has an uncorrected MLS⩾2.0 in the hip-only stratum, between markers D6S257 and D6S262. This region of chromosome 6 contains a strong candidate gene for osteoarthritis, COL9A1 (6q12–6q13). This gene maps within the 11-cM interval between D6S257 and D6S286 and encodes the α1 chain of type IX collagen. This collagen is a quantitatively minor cartilage collagen that decorates the type II collagen fibril and that interacts with extrafibrillar macromolecules (Ayad et al. 1994). Two transgenic mouse models have demonstrated that mutations in the equivalent mouse gene can result in an osteoarthritis phenotype. In the first model, a truncated form of the gene resulted in mice with a mild osteochondrodysplasia phenotype and secondary osteoarthritis (Nakata et al. 1993). In the second model, a knockout mouse had no congenital abnormality but developed a severe osteoarthritis that was comparable, in timing and pathology, to human primary osteoarthritis (Fassler et al. 1994). A more detailed analysis of this second model revealed that the synthesis of the α1 polypeptide chain was necessary for type IX collagen assembly (Hagg et al. 1997). Chromosome 16 does not contain any known genes that can be considered as strong candidates for osteoarthritis susceptibility. As more genes are mapped, candidate loci on this chromosome may become apparent. Overall, the stratification of our genome screen has revealed additional chromosomal regions that may harbor susceptibility loci for osteoarthritis. Stratification increases the level of genetic homogeneity and can therefore assist in the mapping of loci for complex traits. Our analysis highlights the potential utility of this approach for osteoarthritis.

Finer linkage mapping of a primary hip osteoarthritis susceptibility locus on chromosome 6

Primary osteoarthritis (OA) is a common late-onset disease that exhibits complex genetic transmittance. A previous genome-wide linkage scan of OA affected sibling pair families (ascertained by total joint replacement surgery) identified a region of suggestive linkage on chromosome 6, with a maximum multipoint-LOD score (MLS) of 2.9 in 194 families containing sibling pairs concordant for total hip replacement (THR-families). However, up to 50 cM of the chromosome had a multipoint-LOD score >2.0, vvindicating that the susceptibility locus was poorly mapped. We have now genotyped chromosome 6 to a higher density in an expanded cohort of 378 THR-families. We obtained an MLS of 2.8 to an 11.4 cM interval defined by markers D6S452 and 509-8B2, which map between 70.5 to 81.9 cM from the 6p-telomere. Stratification by gender revealed that this linkage was completely accounted for by female THR-families (n=146), with an MLS of 4.0 and with the highest two-point LOD score being 4.6 for marker D6S1573 (75.9 cM). The 11.4 cM interval just encompasses the candidate gene COL9A1 (81.9 cM). We identified and then genotyped twenty common single nucleotide polymorphisms (SNPs) from within COL9A1 in the 146 probands from our female THR-families and in 215 age-matched female controls. No SNP allele, genotype or haplotype demonstrated association to disease. Overall, we have narrowed the chromosome 6 OA susceptibility locus to a point at which linkage disequilibrium/association analysis is feasible, we have demonstrated that this locus is female specific, and found no evidence that COL9A1 encodes for the susceptibility.

Genetic variation including nonsynonymous polymorphisms of a major aggrecanase, ADAMTS‐5, in susceptibility to osteoarthritis

OBJECTIVE Given the recent characterization of ADAMTS-5 as the main aggrecanase of cartilage destruction in mouse models, we explored whether genetic variation and, in particular, putative damaging polymorphisms in the ADAMTS-5 gene modify susceptibility to osteoarthritis (OA). METHODS Two likely deleterious nonsynonymous single-nucleotide polymorphisms (SNPs) were identified in ADAMTS-5 by bioinformatics analysis, rs2830585 in exon 5 affecting a thrombospondin 1 motif, and rs226794 in exon 7. Exploration of their role was carried out in 3 steps, discovery, extension, and replication, on samples obtained from 4 European Caucasian collections, comprising a total of 2,715 patients with knee, hip, or hand OA and 1,185 OA-free controls. In addition, 6 tagSNPs were studied to fully evaluate genetic variation in the ADAMTS-5 locus. RESULTS Initial analyses of 2 sample collections (n = 277 and n = 159) showed a trend toward decreased frequency of the putative deleterious allele of rs226794 among patients with severe knee OA (P = 0.047 versus controls). However, results in patients with knee OA from 2 additional sample collections (n = 360 and n = 265) did not confirm this trend. No association was found with hip OA or hand OA. None of the other SNPs or haplotypes constructed with these SNPs showed a significant association with OA susceptibility. CONCLUSION Use of several collections of OA samples allowed us to obtain sound evidence against the participation of genetic variation in ADAMTS-5 in OA susceptibility. These results indicate the need to further explore the function of this aggrecanase in human OA to determine whether it is as critical as has been observed in mouse models.

Association of a functional microsatellite within intron 1 of the BMP5 gene with susceptibility to osteoarthritis

BackgroundIn a previous study carried out by our group, the genotyping of 36 microsatellite markers from within a narrow interval of chromosome 6p12.3-q13 generated evidence for linkage and for association to female hip osteoarthritis (OA), with the most compelling association found for a marker within intron 1 of the bone morphogenetic protein 5 gene (BMP5). In this study, we aimed to further categorize the association of variants within intron 1 of BMP5 with OA through an expanded genetic association study of the intron and subsequent functional analysis of associated polymorphisms.MethodsWe genotyped 18 common polymorphisms including 8 microsatellites and 9 single nucleotide polymorphisms (SNPs) and 1 insertion/deletion (INDEL) from within highly conserved regions between human and mouse within intron 1 of BMP5. These markers were then tested for association to OA by a two-stage approach in which the polymorphisms were initially genotyped in a case-control cohort comprising 361 individuals with associated polymorphisms (P ≤ 0.05) then genotyped in a second case-control cohort comprising 1185 individuals.ResultsTwo BMP5 intron 1 polymorphisms demonstrated association in the combined case-control cohort of 1546 individuals (765 cases and 781 controls): microsatellite D6S1276 (P = 0.018) and SNP rs921126 (P = 0.013). Functional analyses in osteoblastic, chondrocytic, and adipocytic cell lines indicated that allelic variants of D6S1276 have significant effects on the transcriptional activity of the BMP5 promoter in vitro.ConclusionVariability in gene expression of BMP5 may be an important contributor to OA genetic susceptibility.

Brachydactyly Type B: Linkage to Chromosome 9q22 and Evidence for Genetic Heterogeneity

Brachydactyly type B (BDB), an autosomal dominant disorder, is the most severe of the brachydactylies and is characterized by hypoplasia or absence of the terminal portions of the index to little fingers, usually with absence of the nails. The thumbs may be of normal length but are often flattened and occasionally are bifid. The feet are similarly but less severely affected. We have performed a genomewide linkage analysis of three families with BDB, two English and one Portugese. The two English families show linkage to the same region on chromosome 9 (combined multipoint maximum LOD score 8.69 with marker D9S257). The 16-cM disease interval is defined by recombinations with markers D9S1680 and D9S1786. These two families share an identical disease haplotype over 18 markers, inclusive of D9S278-D9S280. This provides strong evidence that the English families have the same ancestral mutation, which reduces the disease interval to <12.7 cM between markers D9S257 and D9S1851 in chromosome band 9q22. In the Portuguese family, we excluded linkage to this region, a result indicating that BDB is genetically heterogeneous. Reflecting this, there were atypical clinical features in this family, with shortening of the thumbs and absence or hypoplasia of the nails of the thumb and hallux. These results enable a refined classification of BDB and identify a novel locus for digit morphogenesis in 9q22.

Genetic association analysis of RHOB and TXNDC3 in osteoarthritis.

To the Editor: In the May 2006 issue of The American Journal of Human Genetics, Mahr et al.1 reported an association with osteoarthritis (OA [MIM 165720]) for a SNP (rs49846015) located immediately 5′ of the coding region of RHOB (on chromosome 2p24.1 [MIM 165370]) and for a SNP (rs4720262) located immediately 5′ of the coding region of TXNDC3 (on chromosome 7p14.1 [MIM 607421]). RHOB codes for a GTP-binding protein whereas TXNDC3 codes for a thioredoxin protein. The association study by Mahr et al. was performed with 171 patients with OA (74% females) who had undergone joint-replacement surgery (68% knee and 32% hip) and with 182 healthy control subjects (66% females), all of European white ethnicity. Possession of a copy of the G allele of rs49846015 was an OA risk factor (P=.0007), as was possession of the T allele of rs4720262 (P=.0007). To assess the robustness of these associations, we have genotyped the SNPs in our collection of >1,500 case patients with OA (mean age 65 years; age range 56–85 years) and >700 age-matched control subjects (mean age 69 years; age range 55–89 years). As in the study by Mahr et al.,1 our case patients were ascertained by joint-replacement surgery (hip, knee, or hip and knee) due to severe end-stage OA. Our control subjects had no signs or symptoms of arthritis or joint disease (pain, swelling, tenderness, or restriction of movement). All case patients and control subjects were individuals from the United Kingdom who are of white European ethnicity. Further details about the ascertainment of our case patients and control subjects have been published elsewhere.2 Ethical approval for our study was obtained from the appropriate ethics committees, and informed consent was obtained from each individual studied. When planning our investigation, we noted that rs49846015 was absent from dbSNP. A correspondence with Sandra Mahr (personal communication) revealed that rs585017 was the correct accession number for this SNP. rs585017 and rs4720262 were genotyped by mass spectrometry (homogeneous MassARRAY system [Sequenom]), and the genotype and allele distributions in case and control groups were compared using standard χ2 analysis-of-contingency tables. Tables ​Tables11 and ​and22 list the results for rs585017 and rs4720262, respectively. Both SNPs were in Hardy-Weinberg equilibrium in the case and control groups. There were no significant differences (all P values >.05) in genotype or allele frequencies for either SNP between the case and control groups. This was also the case when the data were stratified by sex, with male case patients compared with male control subjects and female case patients compared with female control subjects. Table 1.  Association of RHOB SNP rs585017 between Our Case Patients with OA and Control Subjects Table 2.  Association of TXNDC3 SNP rs4720262 between Our Case Patients with OA and Control Subjects The frequency of the G allele of rs585017 in our study is comparable to that in the study by Mahr et al.,1 with a frequency of 27.4% in our control group and 23.9% in Mahr et al.s control group (P=.20). However, the frequency of the T allele of rs4720262 shows a highly significant difference between the two studies, with a frequency of 28.8% in our control group and 13.4% in Mahr et al.s control group (P<.0005). In dbSNP, the T allele of rs4720262 is listed as having a frequency of 31.2% in the AFD-EUR panel (23 unrelated American individuals of European descent and one sample from a human variation panel of 50 whites) and a frequency of 21.7% in a HapMap-CEU panel (30 mother-father-child trios from the CEPH collection of Utah residents with northern and western European ancestry). Our T-allele frequency of 28.8% is between these two dbSNP-reported frequencies, whereas the T-allele frequency of 13.4% reported by Mahr et al. is substantially lower than both database frequencies. This implies that the control frequencies reported for this SNP by Mahr et al. may not accurately reflect the true frequency of this SNP in Europeans of white ancestry. To assess the power of our study, we conducted power calculations by using Quanto, version 1.5,3,4 with the following options: an unmatched case-control study design, a population risk of severe OA of 5%, a significance level of 0.05, a G-allele frequency of 23.9% for rs585017, a T-allele frequency of 13.4% for rs4720262, and a log-additive inheritance mode. The allele frequencies and the inheritance mode were selected to agree with the results of Mahr et al. Table 3 lists, for each comparison, the minimum odds ratio (OR) detectable with 80% power for our study. We calculated an OR of 2.05 for the association with rs585017 and an OR of 2.26 for the association with rs4720262 from the results reported by Mahr et al. All the ORs given in table 3 are lower than these values, indicating that the sample sizes used in our study are more than adequate to detect the ORs previously observed by Mahr et al. In fact, the power to detect an OR of 2.05 for the association with rs585017 was ⩾99.7% for all comparisons, and the power to detect an OR of 2.26 for the association with rs4720262 was ⩾99.6% for all comparisons. Table 3.  The Minimum Detectable ORs for Our Analysis under the Log-Additive Model with Power ⩾80% and a Significance Level of 5% Overall, our study does not replicate the previous findings of an association between OA and the RHOB SNP rs585017 and the TXNDC3 SNP rs4720262. Our study was adequately powered to detect an association comparable to that reported by Mahr et al., and we avoided potential confounding factors by using the same disease ascertainment (joint replacement of the hip or knee) and the same ethnic group (Europeans of white ethnicity) used in the original study. As more studies are reported, an accurate estimation of the effect of these two SNPs on OA susceptibility will become apparent.

Genetics of osteoarthritis

Epidemiological studies have demonstrated a major genetic component to osteoarthritis (OA), with heritability estimates of over 50% for most joint sites. These studies have also highlighted differences in the degree of OA heritability between joint sites and between the sexes, implying a high level of heterogeneity. We published a genome-wide scan in 1999. We had focused on families containing siblings with severe large-joint OA ascertained by joint-replacement surgery. Our data showed that OA genetic susceptibility did exhibit joint specificity and that this susceptibility had a greater role in female disease. During the past 5 years we have been investigating our linkage regions and we have so far identified FRZB (chromosome 2q32.1), COL9A1 (6q12-q13), BMP5 (6p12.1) and IL4R (16p12.1-p11.2) as encoding for OA susceptibility. Common variants at these genes affect either the structural properties of the protein (FRZB and IL4R) or the transcription of the gene (BMP5 and COL9A1). The variants are particularly relevant to the development of hip OA in females. What is particularly interesting about our recent discoveries is that the proteins encoded for by IL4R, BMP5 and FRZB are involved in chondrocyte cell signalling and signal transduction pathways. It appears probable, therefore, that OA genetic risk for the hip is principally accounted for by aberrant cell signalling. This was not anticipated. In this presentation I will focus on our latest genetic findings. I will also discuss the results from other studies. A number of OA genome-wide scans have been performed, some on large joints and others investigating hand disease. Many loci appear unique to one particular study, with only some loci being positive in multiple studies. The reasons for this will be discussed. A concern often expressed at orthopaedic and rheumatology meetings is that genetic linkages and associations are not consistently reproduced. These are reasonable criticisms but it needs to be remembered that the genetic component of a complex trait such as OA will not be mediated by fully penetrant risk alleles. Instead, any one allele will contribute only a fraction of the overall risk and this allele will, by chance, have varying frequencies in different cohorts. It is unreasonable, therefore, to expect a linkage or an association in one cohort to be replicated in all cohorts.