The authors state that they have no conflicts of interest.
Mutations of SQSTM1 are an important cause of PDB, but other genes remain to be discovered. A major susceptibility locus for PDB was identified on chromosome 10p13 by a genome-wide linkage scan in families of British descent, which accounted for the vast majority of cases not caused by SQSTM1 mutations.
Introduction: Paget's disease of bone (PDB) has a strong genetic component, and several susceptibility loci have been identified by genome-wide linkage scans. We previously identified three susceptibility loci for PDB using this approach on chromosomes 5q35, 2q36, and 10p13 in 62 families of mainly British descent, but subsequently, mutations in the SQSTM1 gene were found to be the cause of PDB in 23 families from this cohort. Here we reanalyzed the results of our genome-wide search in families from this cohort who did not have SQSTM1 mutations.
Materials and Methods: The study population consisted of 210 individuals from 39 families of predominantly British descent with autosomal dominant inheritance of PDB in whom SQSTM1 mutations had been excluded by mutation screening. The average family size was 5.44 ± 3.98 (SD) individuals (range, 2-24 individuals). Genotyping was performed using standard techniques with 382 microsatellite markers spaced at an average distance of 9.06 cM throughout the autosomes. Multipoint linkage analysis was performed using the GENEHUNTER program under models of homogeneity and heterogeneity.
Results: Multipoint parametric linkage analysis under a model of homogeneity and nonparametric linkage analysis under a model of heterogeneity both showed strong evidence of linkage to a single locus on chromosome 10p13 (LOD score, +4.08) close to the marker D10S1653 at 41.43cM. No evidence of linkage was detected at the chromosome 2q36 locus previously identified in this population, and linkage to other candidate loci previously implicated in the pathogenesis of PDB was excluded.
Conclusions: We conclude that there is an important susceptibility gene for PDB on chromosome 10p13 in families of British descent and find no evidence to support the existence of a susceptibility locus on chromosome 2q36 or other previously identified candidate loci for PDB in this population. The gene that lies within the 10p13 locus seems to account for the development of PDB in the vast majority of families of British descent who do not carry SQSTM1 mutations.
Paget's disease of bone (PDB; MIM 167250, 602080) is a common disorder, affecting between 1% and 3% of individuals over the age of 55 yr in white populations.(1,2) The cause of PDB is not completely understood, but both genetic and environmental factors seem to play a role. Reductions in the prevalence of PDB have been observed in some,(3,4) but not all,(5) countries in the past 25 yr. This suggests that predisposing environmental influences on the disease have changed in prevalence in some populations over recent years. Several potential environmental triggers for PDB have been suggested, including dietary calcium deficiency,(6) mechanical loading of the skeleton,(7) zoonotic infections,(8) and occupational exposure to toxins,(9) but most of these associations are based on circumstantial evidence.(6) The only potential trigger for PDB that has been studied experimentally is paramyxovirus infection. There is evidence that paramyxoviruses can affect osteoclast function in vitro,(10,11) and targeted overexpression of the measles virus nucleocapsid protein in osteoclasts has been shown to increase bone turnover in vivo.(12) Extensive studies have been conducted to try and detect evidence of paramyxoviruses in PDB, but the data remain conflicting despite extensive research efforts over the past 30 yr.(13,14) Indeed, in a recent multicenter blinded study, no evidence of paramyxovirus infection was found in samples from patients with PDB using highly sensitive RT-PCR-based detection methods.(15) The possibility that heredity might play a role in the pathogenesis of PDB was first raised in the 1940s,(16) but over recent years, increasing interest has focused on the role of genetic factors in the pathogenesis of the disease. It has been estimated that between 15% and 40% of patients with PDB have at least one affected first-degree relative,(17-19) and many families have been described where the disease is inherited as an autosomal dominant trait.(19-21) Linkage analysis has identified several potential susceptibility loci for PDB, but the most important of these is the PDB3 locus on chromosome 5q35, which was independently identified in two different populations by genome-wide linkage scans.(20,21) Mutations affecting the ubiquitin associated (UBA) of SQSTM1 were subsequently found to account for 5q35-linked PDB in both populations,(22,23) and SQSTM1 mutations have now been identified as an important cause of PDB by a wide variety of investigators in several populations.(24-30) It has been suggested that SQSTM1 mutations may not be sufficient to cause PDB in isolation on the basis that penetrance is incomplete(22) and because overexpression of the P392L mutation of SQSTM1 in osteoclasts caused increased bone turnover and osteopenia in mice, rather than lesions typical of PDB. However, we have recently found that mice that carry a germline truncating mutation of SQSTM1 exhibit several features in common with PDB,(31) showing that mutations of SQSTM1 can cause a PDB-like disorder in the absence of another trigger factor. These observations show that SQSTM1 is an important susceptibility gene for PDB, but there remains a large number of families with autosomal dominant inheritance of classical PDB where involvement of SQSTM1 has been excluded, indicating that other genes remain to be discovered.(21,23,26,32) In a previous study, we reported the existence of potential loci for susceptibility to familial PDB on chromosomes 2q36 and 10p13 when we conducted a genome-wide linkage scan in 62 families with PDB under a model of heterogeneity.(20) Subsequently, however, mutations of the SQSTM1 gene were identified in 23 of the families included in that study. Because linkage analysis under a model of heterogeneity may be confounded by false-positive results because of errors in the disease model, or co-incidental allele sharing at more than one locus, we have now reassessed the contribution of these regions to PDB by conducting genome-wide linkage analysis in families from this cohort who did not have SQSTM1 mutations.
MATERIALS AND METHODS
The participants in this study were comprised of 210 individuals from 39 families of British descent in whom mutations of SQSTM1 had been excluded by mutation analysis. Details of the recruitment strategy and procedures for disease ascertainment were as described previously.(20) Within the study cohort, 90 subjects were diagnosed as having PDB (42.8%), whereas PDB was excluded in 40 subjects (19.0%), and in the remaining 80 subjects (38.1%), affection status was unknown. Seventeen of the families were resident in the United Kingdom (44%), 12 in Australia (26%), 9 in New Zealand, (23%), and 1 in Canada (3%). The average family size was 5.44 ± 3.98 (SD) individuals, with a range of 2-24 individuals. Pedigree diagrams showing the size and structure of the individual families included in this study are shown on the University of Edinburgh website (http://www.mcm.ed.ac.uk/rheum). All subjects gave informed consent to being included in this study, which was approved by the relevant research ethics committees in participating centers.
Genotyping and linkage analysis
Microsatellite genotyping was carried out on leukocyte DNA as previously described(20) using 382 microsatellite markers from the ABI Prism Linkage Mapping Set MD-10, which has an average intermarker distance of 9.06 cM. Positions of the microsatellites were obtained from the Marshfield database, and genotype data were checked for Mendelian inheritance inconsistencies using the PEDCHECK program.(33) Linkage analysis was performed using the GENEHUNTER program, version 1.3,(34) under models of homogeneity and heterogeneity. We also performed two-point linkage analysis for selected loci using MLINK. For the linkage analysis, we assumed an autosomal-dominant mode of transmission; a disease allele frequency of 0.002325; and a phenocopy rate at 0.028675 to account for the estimated population frequency of PDB in elderly subjects from the UK population(3) and the assumption that ∼15% of cases will be familial in nature. The principal analysis was conducted assuming a disease penetrance of 95% by the age of 65 yr, because previous studies in this cohort of families indicated that 56% of subjects had developed PDB by age 65.(20) However because recent reports have indicated that penetrance of PDB mediated by SQSTM1 mutations can vary from 79%(22) to 100%(23,26,35) by the seventh decade, we also conducted the linkage analysis using penetrance thresholds of 75%, 80%, 85%, and 90%. Analysis using the SLINK program showed that we had >80% power to detect significant linkage (Z = 3.3) using this model, assuming a penetrance of 95% and that the proportion of linked families (α) was 0.75 or greater.
The results of parametric linkage analysis under a model of homogeneity are shown in Fig. 1A. This analysis showed significant evidence of linkage to the PDB6 region on chromosome 10p13, where the multipoint LOD score was +4.08 at 41.43 cM, close to the marker D10S1653. This analysis excluded involvement of several regions that have previously been implicated in regulating genetic susceptibility to PDB, including the PDB1 locus on chromosome 6p21(36); the PDB2 locus on chromosome 18q21(37); the PDB4 locus on chromosome 5q31(21); the PDB7 locus on chromosome 18q23(38); and the PDB5 locus on chromosome 2q36, which was previously identified as showing possible evidence of linkage to PDB in the cohort.(20) Corresponding results under a model of heterogeneity are shown in Fig. 1B. This analysis also showed evidence of significant linkage to chromosome 10p13 where the HLOD score was identical to the LOD score and also maximized at 41.43 cM. No evidence of linkage to chromosome 2q36 was detected, and the only other region in the genome that showed any evidence of linkage was on chromosome 12p13, where the maximum parametric LOD score was +1.76 and the HLOD score was +1.82 at 2.52 cM.
We repeated the linkage analysis using different models of penetrance ranging between 75% and 90%, and this yielded very similar results with the exception that the multipoint LOD score was slightly lower in the region of strongest linkage. For example, when a penetrance of 75% was used, the maximum LOD score was +3.4, at 41.43 cM on chromosome 10p13, compared with +3.6 for a penetrance of 80%; +3.7 for a penetrance of 85%, and + 3.9 for a penetrance of 90%.
To explore the reasons for the discrepancy between our previous study and this study with regard to the chromosome 2q36 linkage, we reviewed the results from our previous genome-wide scan.(20) This revealed that, of the 10 families that were estimated by GENEHUNTER as being most likely to have linkage to chromosome 2q36 under heterogeneity analysis (probability > 0.64), 7 families also had evidence of linkage to chromosome 5q35, and the affected members of these families turned out to have mutations in the SQSTM1 gene.
Details of the linked region on chromosome 10p13 are shown in more detail in Fig. 2. The multipoint LOD score plot and HLOD plot were superimposed over the region of strongest linkage between markers DS10547 and DS10548. Moreover, two-point linkage analysis using five microsatellite markers spread across the region gave results consistent with those observed in the multipoint analysis and the highest two-point LOD score was 3.06 (θ = 0) at D10S1653, close to the region of greatest linkage at 41.43 cM, identified by multipoint analysis.
To try and refine the critical interval, we conducted haplotype analysis in affected families for markers that encompassed the region of linkage. Two of the families analyzed were informative, and details are shown in Fig. 3. In family 13, there was evidence of recombination between D10S548 and D10S1653 in subject 3.3, and in family 61, there was recombination between D10S189 and D10S547 in subjects 2.3 and 2.4. These events narrow the likely critical region to an interval of ∼26.7 cM between the markers D10S189 (19 cM) and D10S548 (45.7 cM).
Genetic factors play an important role in the pathogenesis of PDB, but current evidence suggests that the disease is genetically heterogeneous and can result from mutations in one of several disease genes. The most important gene for classical PDB is SQSTM1, which lies within the PDB3 locus on chromosome 5q35. Laurin et al.(22) were first to identify a proline-leucine substitution at codon 392 in SQSTM1 (P392L) as a cause of familial PDB in French-Canadian families, and we also identified the P392L mutation and other UBA domain mutations of SQSTM1 in 5q35-linked PDB patients of British descent.(23,32) Subsequently, a large number of investigators identified UBA domain mutations of SQSTM1 in patients with PDB.(24-27,29,30) Functional studies indicate that all of these mutations impair ubiquitin binding,(39) although the mechanism by which this leads to the disease phenotype remains incompletely understood.(40)
Although SQSTM1 is an important cause of classical PDB, it is also clear that other genes remain to be discovered. For example, Laurin et al.(21) identified a candidate locus for PDB on chromosome 5q31, which accounted for ∼50% of cases of familial PDB not caused by SQSTM1 mutations in the French-Canadian population. Similarly, studies in Holland and the United Kingdom have indicated that ∼60% of patients with familial PDB do not have SQSTM1 mutations.(26,32)
The genome-wide linkage scan previously performed in this population suggested that candidate loci on chromosomes 2q36 and 10p13 may contribute to the pathogenesis of PDB, although neither locus reached the threshold for genome-wide significance.(20) In this study, however, we repeated analysis of data from the genome-wide linkage scan in 39 families where SQSTM1 mutations had been excluded with markedly different results to those previously reported.(20) The new analysis provided strong evidence for involvement of the 10p13 locus in PDB where the LOD score was +4.08 but showed no evidence of linkage to the previously identified region on chromosome 2q36 and excluded involvement of other loci previously implicated in the pathogenesis of PDB on chromosomes 5q31, 6p21, 18q21, and 18q23.(21,36,38,41) This indicates that familial PDB may be less genetically heterogeneous than previously thought and indeed, on the basis of the present evidence, it would seem that the vast majority of cases of familial PDB in patients of British descent are caused either by SQSTM1 mutations or the 10p13 locus identified in this study. Indeed, our results suggest that the gene that lies within the 10p13 locus may be an even more important cause of PDB than SQSTM1 in our population, because all of the families included in this study showed evidence of linkage to this region. Although the 10p13 locus did not emerge as a candidate region in the genome scan performed by Laurin et al.(21) in the French-Canadian population, it would be of great interest to determine if families with PDB not caused by SQSTM1 mutations from other populations show evidence of linkage to 10p13.(26) The candidate locus we have identified is large and contains many genes that could potentially be involved in bone metabolism, although none of these is known to be directly involved in RANK-NF-κB signaling or the ubiquitin proteasome pathways that are known to be involved in the pathogenesis of PDB and related disorders.(42)
Positional cloning studies are now in progress to try to identify the gene that is responsible for the occurrence of Paget's disease within the 10p13 chromosomal region. Identification of this gene will give important insights the mechanisms that underlie abnormal bone remodeling, which characterizes PDB, and this in turn should lead to better strategies for prevention and treatment.
This study was supported in part by grants from the Arthritis Research Campaign (R0589, R0616, R0544); the National Association for Relief of Paget's Disease (UK); the Medical Research Council; and the Paget's Disease Charitable Trust (Auckland, New Zealand).