Objectives Leri’s pleonosteosis (LP) is an autosomal dominant rheumatic condition characterised by flexion contractures of the interphalangeal joints, limited motion of multiple joints, and short broad metacarpals, metatarsals and phalanges. Scleroderma-like skin thickening can be seen in some individuals with LP. We undertook a study to characterise the phenotype of LP and identify its genetic basis.
Methods and results Whole-genome single-nucleotide polymorphism genotyping in two families with LP defined microduplications of chromosome 8q22.1 as the cause of this condition. Expression analysis of dermal fibroblasts from affected individuals showed overexpression of two genes, GDF6 and SDC2, within the duplicated region, leading to dysregulation of genes that encode proteins of the extracellular matrix and downstream players in the transforming growth factor (TGF)-β pathway. Western blot analysis revealed markedly decreased inhibitory SMAD6 levels in patients with LP. Furthermore, in a cohort of 330 systemic sclerosis cases, we show that the minor allele of a missense SDC2 variant, p.Ser71Thr, could confer protection against disease (p<1×10−5).
Conclusions Our work identifies the genetic cause of LP in these two families, demonstrates the phenotypic range of the condition, implicates dysregulation of extracellular matrix homoeostasis genes in its pathogenesis, and highlights the link between TGF-β/SMAD signalling, growth/differentiation factor 6 and syndecan-2. We propose that LP is an additional member of the growing ‘TGF-β-pathies’ group of musculoskeletal disorders, which includes Myhre syndrome, acromicric dysplasia, geleophysic dysplasias, Weill–Marchesani syndromes and stiff skin syndrome. Identification of a systemic sclerosis-protective SDC2 variant lays the foundation for exploration of the role of syndecan-2 in systemic sclerosis in the future.
- Gene Polymorphism
- Systemic Sclerosis
Statistics from Altmetric.com
Leri’s pleonosteosis (LP; MIM 151200) is an autosomal dominant condition characterised by flexion contractures of the interphalangeal joints, limited motion of multiple joints, and short broad metacarpals, metatarsals and phalanges.1 Additional features, including chronic joint pain, facial dysmorphism, bony overgrowths, short stature, spinal cord compression, carpal-tunnel syndrome and scleroderma-like thickening of skin, have also been described in some affected individuals.2–7 The clinical manifestations of LP overlap with a number of musculoskeletal conditions, including Myhre syndrome (MIM 139210), acromicric dysplasia (MIM 102370), geleophysic dysplasias (MIM 231050 and 614185), Weill–Marchesani syndromes (MIM 277600 and 608328) and stiff skin syndrome (MIM 184900). Recent genetic studies have implicated abnormalities of extracellular matrix (ECM) and SMAD/transforming growth factor (TGF)-β signalling in the pathogenesis of these conditions.8–13 The clinical overlap suggests that LP is an additional member of this ‘TGF-β-pathies’ group of disorders.14 Here we report phenotypic characterisation of LP and identification of its genetic and molecular basis.
There is an increasing catalogue of genes in which variants cause both rare single-gene disorders and predispose to more common disorders with shared phenotypic features. For example, variants in TREX1 cause Aicardi–Goutières syndrome (MIM 225750),15 familial chilblain lupus (MIM 610448)16 and systemic lupus erythematosus (SLE, MIM 152700).17 Similarly, here, we show that a common missense variant in a gene implicated in pathogenesis of LP could alter susceptibility to systemic sclerosis (SSc).
Methods and results
We ascertained a multigenerational white British family with LP (figure 1A; table 1). Some members of this family have been described in a previous publication from our centre.1 Ethics approval for this study was obtained from the NHS ethics committee (11/H1003/3) and University of Manchester. Informed consent was taken from all participants in the study.
Genetic basis of LP
We genotyped the two most distantly related affected individuals in family 1 (V:2 and III:6 in figure 1A and table 1) on whom samples were available, using Genome-Wide Human SNP Array 6.0 (Affymetrix, Santa Clara, California, USA) according to the manufacturer's protocol, with the aim to undertake linkage analysis. Genotypes and copy number data were generated within the Affymetrix Genotyping Console (V.188.8.131.520) using the Birdseed V2 algorithm (confidence threshold of 0.01) and SNP 6.0 CN/LOH algorithm, respectively. Copy number data were visualised at a 50 Kb threshold for genomic segment size with a minimum of 25 markers per segment. This revealed a 1 Mb microduplication of chromosome 8q22.1, flanked by single-nucleotide polymorphism (SNP) probe SNP_A-8284673 (rs6994748) and copy number probe CN_1318790, in both individuals (figure 2A).
Real-time PCR copy number analysis confirmed that the microduplication segregated with the phenotype in all the affected individuals (III:6, IV:2, V:1 and V:2 in figure 1A) and was absent in all the unaffected individuals (III:5 and IV:5) who were available for testing (see online supplementary figure S1).
A skin punch biopsy sample was obtained from two affected individuals, II:3 and IV:2 (aged 77 and 20 years, respectively, at biopsy), from family 1, and dermal fibroblasts were cultured using standard protocols. Metaphase fluorescent in situ hybridisation was performed on cultured fibroblast cell suspensions using standard protocols with Spectrum Green fluorophore prelabelled RP11-22I13 BAC probe (The Centre for Applied Genomics, Toronto, Canada) and Spectrum Orange fluorophore prelabelled 8q subtelomeric probe (Abbott Molecular, Maidenhead, UK) as control. The images were consistent with tandem microduplication at 8q22 (figure 2B).
The affected individual, also of white British origin, in family 2 was independently diagnosed with LP through the European Skeletal Dysplasia Network (ESDN) (figure 1B and table 1), and an SNP 6.0 array was genotyped. This showed a microduplication of 1.2 Mb at 8q22.1, flanked by copy number probes CN_377954 and CN_1320967, overlapping the duplicated segment in family 1 (figure 2A). It was not possible to determine segregation of the microduplication in this family. Neither microduplication is present in copy number polymorphism databases (Database of Genomic Variants, DECIPHER or Ensembl) or in our internal variation database generated from 190 microarray samples.
Segregation of the microduplication with the phenotype in family 1, confirmation of the microduplication by three independent methods, and finding of an overlapping microduplication in an unrelated individual with the same condition confirm the causal relationship between LP and 8q22.1 microduplication.
Gene expression profile in the critical region
The overlapping minimum critical region in the two families was ∼0.95 Mb (flanked by CN_377954 and CN_1318790) and consisted of six genes, GDF6, UQCRB, MTERFD1, PTDSS1, SDC2 and CPQ (figure 2A). We measured expression of the six genes within this shared region by real-time quantitative PCR (rtqPCR) in control adult human dermal fibroblasts (HDFs; Life Technologies, Paisley, UK) and cultured dermal fibroblasts from individuals II:3 and IV:2 of family 1. UQCRB, MTERFD1, PTDSS1 and PGCP were expressed at similar levels in fibroblasts from affected individuals in comparison with the control HDFs. However, GDF6, which encodes growth/differentiation factor 6 (GDF6), was found to be overexpressed significantly in fibroblasts from both individuals (40.2- and 13.2-fold increased in II:3 and IV:2, respectively, figure 2C). SDC2, which encodes syndecan-2, was overexpressed in II:3 (2.8-fold), but not in IV:2.
Expression profile of genes relevant to TGF-β-mediated ECM homoeostasis
Abnormalities of ECM underlie a number of related phenotypes.8–13 Both GDF6 and syndecan-2 modulate TGF-β signalling, a process that is of vital importance in normal ECM deposition.18 ,19 Syndecan-2 has known roles in deposition, assembly and cellular interactions of ECM.20–23 The expression profiles of downstream genes relevant to TGF-β-mediated ECM homoeostasis were therefore measured by rtqPCR (as described above).10 ,24 Expression of these genes was found to be altered compared with control fibroblasts in both II:3 and IV:2, but in an inconsistent manner (see online supplementary figure S2). In II:3, expression of FBLN4 (fibulin 4), TGFB1 (transforming growth factor β1) and TGFBR3 (betaglycan) was decreased, whereas expression of FBN1 (fibrillin) and FN (fibronectin) was increased compared with control fibroblasts. In IV:2, expression of FBLN4, FBN1 and FN was decreased, whereas there were no noticeable differences in TGFBR3 or TGFB1 levels (see online supplementary figure S2). The differences in expression patterns of ECM protein-encoding genes probably reflect age-related differences in the two cases (77 and 20 years, respectively).25 This difference was also reflected in the growth of the fibroblasts from II:3 and IV:2. After cells had been seeded at the same density on coverslips, fibroblasts from individual II:3 (aged 77 years) grew slowly, resulting in 3.0-fold lower cell numbers detected (88±11 nuclei/field) compared with the control cell line (264±13.8) after 1 week. This is compared with individual IV:2 (aged 20 years), where the cells grew rapidly and had 2.3-fold higher cell numbers detected (607±26) compared with the control cell line.
Western blot analysis of SMAD proteins
Notably, SMAD4 expression was modestly increased in fibroblasts from both individuals (1.15- and 1.26-fold, respectively, see online supplementary figure S2). The downstream effects of the bone morphogenetic protein (BMP) and the canonical TGF-β signalling pathways are mediated by phosphorylation of SMAD1/5/8 and SMAD2/3, respectively.26 The pSMADs bind with SMAD4, are translocated to the nucleus, and induce or repress BMP and TGF-β target genes. We therefore explored expression of the SMAD family of proteins in more detail in fibroblasts. Western blot analysis demonstrated no significant changes in pSMAD1/5/8 and pSMAD2 levels (figure 3). As expected from gene expression studies, SMAD4 protein was significantly upregulated in IV:2, but was found to be decreased in II:3. SMAD6, on the other hand, was markedly decreased in both samples.
SDC2 variants in SSc
The thickened skin phenotype of LP shows similarity to the scleroderma of SSc. We therefore hypothesised that genomic changes in one or more genes within the causal 8q22.1 duplication of LP could affect susceptibility to scleroderma of SSc. We examined this possibility in a cohort of 330 individuals, all diagnosed as having SSc by a consultant rheumatologist in a specialist centre (see online supplementary table S1 for phenotypic details). Ideally in this cohort, we would have liked to undertake copy number analysis and sequencing of coding and regulatory regions of all the six genes within the overlapping 8q22.1 minimum critical region identified in LP. However, the amount of DNA available from this cohort was limited, which forced us to prioritise our candidates. Owing to the gene and protein expression results in patients with LP, variants in GDF6 and SDC2 were also our prime candidates for SSc susceptibility. SDC2 is overexpressed in the lesional fibroblasts from individuals with scleroderma.27 For this and other reasons given in the Discussion section, we decided to only examine SDC2 in the SSc cohort.
In genomic DNA samples of individuals from this cohort, we undertook copy number analysis of SDC2 by TaqMan-based qPCR (as described above) and Sanger sequencing of all five exons of SDC2 (NM_002998.3) using standard protocols. The primer sequences and reaction conditions are available on request. Sequence traces were analysed using Staden package V.2.0. No rare pathogenic copy number or sequence variants were detected in any individuals with SSc. To examine if more common SDC2 variants alter susceptibility to SSc, we calculated the frequencies of known SDC2 SNPs in our SSc cohort and compared them with data from a local ethnically matched healthy control population (n=308) and the European cohort from the Exome Variant Server (EVS, n=4300). For a missense SDC2 variant, rs1042381 (c.211T>A; p.Ser71Thr), we observed a minor allele frequency of 0.122 (T=593; A=83) in individuals with SSc, 0.167 (T=513; A=103) in local controls, and 0.182 (T=7032; A=1568) in the EVS cohort. The difference in the rs1042381 minor allele frequency of SSc individuals and local controls (p=0.0009, OR=0.58 (95% CI 0.41 to 0.80)) and the EVS cohort (p<1×10−5, OR=0.52 (95% CI 0.40 to 0.67)) was highly significant (see online supplementary table S2).
In this study, all individuals affected with LP had limited motion of multiple joints with a variable degree of contracture of the interphalangeal joints and chronic joint pain. All the affected individuals for whom radiographs were available had abnormal segmentation of the cervical vertebrae (figure 1A,B) and had short, narrow palpebral fissures and similar facial features (figure 1A,B). There were also notable differences in the phenotypes of the two families. For example, unlike in members of family 1, the metacarpals, metatarsals and phalanges were strikingly short and broad in our proband from family 2. All the affected individuals from family 1 were of average stature, whereas the height of the proband from family 2 was 3 cm below the 0.4th centile. Thickening of skin, especially on plantar and palmar aspects, was remarkably prominent in the proband from family 2.
Genomic microduplications resulting in aberrant gene expression underlie a number of disorders of skeletal development.28–31 Copy number gains can result in aberrant expression of genes via a number of mechanisms including increased expression due to dosage effect and disruption of regulatory elements resulting in unusual spatial or temporal expression patterns. The phenotypic variability of LP may reflect differences in the expression of critical genes due to the span or orientation of the duplicated region, even though the microduplications in the two families overlap significantly. Alternatively, the structural alterations resulting in cis- or trans-acting long-range effects on other genomic regions may influence the different phenotype.32 Our results demonstrate the range of clinical features of the condition, and identification of more individuals with LP or overlapping duplications will further define the phenotypic characteristics of the disorder.
Interestingly, the phenotype of LP overlaps significantly with the TGF-β-pathies group of disorders.14 The articular and facial phenotype of LP is most reminiscent of Mhyre syndrome. However, lack of intellectual disability may help in distinguishing the two conditions. In addition, as demonstrated, unlike those with Mhyre syndrome, all individuals with LP may not have short stature. The brachydactyly strongly resembles the hands and feet in acromicric and geleophysic dysplasias. The skin thickening, when present, is similar to, although less dramatic than, that of stiff skin syndrome.
On the basis of expression levels in fibroblasts and the known functions of genes in the common duplicated interval in two families, we hypothesised that aberrant expression of GDF6 and/or SDC2 underlies the LP phenotype. GDF6 encodes GDF6, which belongs to a subgroup of BMP signalling molecules, which includes GDF5 and GDF7.18 ,33 GDF6 is also termed BMP-13 and cartilage-derived morphogenetic protein-2 (CDMP2). Loss-of-function GDF6 mutations result in Klippel–Feil syndrome (KFS; MIM 118100).34 Interestingly, the abnormal cervical vertebral segmentation in individuals described here resembles the phenotype of KFS. Both haploinsufficiency and duplication of certain genes can result in the same or a similar phenotype.35 ,36 In addition to the similarities to KFS, the skeletal phenotype of LP has other features (table 1). In this context, it is notable that loss-of-function and gain-of-function mutations of GDF5 result in distinct phenotypes.37 Regulated expression of GDF5, GDF6 and GDF7 is critical for normal joint development.38 Administration of GDF6 in rats induces development of neotendon and ligament tissue in ectopic sites39 and enhances tendon healing.40 Interestingly, one individual with LP has been noted to have greatly thickened ligaments.6 Hence, it is likely that aberrant or overexpression of GDF6 is associated with the skeletal and articular features of LP. SDC2 encodes syndecan-2, which is a transmembrane heparan sulfate proteoglycan.41 In mice, sdc2 is predominantly expressed in mesenchymal cells, which are precursors of connective tissue cells that form rudiments of the skeletal system. Its expression persists in the perichondrium, periosteum, periarticular, tendinous and dermal tissues,42 overlapping with the tissues altered in individuals with LP. SDC2 is upregulated in affected tissue in keloid scars43 and scleroderma.27 It is likely that aberrant/increased expression of SDC2 contributes to the non-skeletal phenotype of LP.
SMAD6 is an inhibitor of the TGF-β superfamily,44 and its downregulation probably results in overall activation of the SMAD signalling pathway. Interestingly, this corresponds to activation of the SMAD proteins in Myhre syndrome, which bears significant phenotypic overlap with LP and results from increased stabilisation of the mutant SMAD4.8 In Myhre syndrome, enhanced levels of phosphorylation of SMAD2/3 and SMAD1/5/8 have been demonstrated in fibroblasts. In contrast, there were no alterations of these SMAD proteins in LP (figure 3), indicating a different pathogenic mechanism. These results must be viewed in the context that fibroblasts for the expression studies were available from only two individuals with LP, and dermal cells may not be the most fitting model to test molecular mechanisms for some aspects of the phenotype observed in LP. However, greater understanding of SMAD signalling regulation has implications for a range of disorders such as Marfan syndrome, rheumatic diseases and cancer.
It remains unclear why most FBN1 mutations cause Marfan syndrome,45 but those in the TB5 domain cause autosomal dominant Weill–Marchesani syndrome, acromicric or geleophysic dysplasias,10 ,13 and those in the RGD-containing TB4 domain of FBN1 cause stiff skin syndrome.11 Similar to LP, individuals with Weill–Marchesani syndrome, acromicric and geleophysic dysplasias have brachydactyly, short stature and joint contractures, whereas the opposite phenotype of arachnodactyly, tall stature and joint hypermobility is observed in Marfan syndrome. We have recently shown that the FBN1 mutations causing Weill–Marchesani syndrome and acromicric and geleophysic dysplasias disrupt the interaction of heparan sulfate with the TB domain of fibrillin-1.46 However, the downstream effects of the altered interactions remain to be discovered. The striking similarity between the phenotype of LP, Weill–Marchesani syndrome, acromicric and geleophysic dysplasias suggests that the effect of FBN1 mutations on expression of GDF5/6/7 should be explored in these conditions. Another interesting observation is that individuals with trisomy 8 mosaicism have joint contractures and abnormal vertebral segmentation.47 Given that the microduplication is present at 8q22, it is possible that aberrant expression of GDF6 and/or SDC2 underlies this phenotype. Rare heterozygous deletions at 8q22.1 have been associated with Nablus mask-like syndrome, a condition that shares a few characteristics with LP, such as blepharophimosis, tight facial skin, camptodactyly and joint contractures.48 ,49 The agreed critical chromosomal region for Nablus mask-like syndrome50 ,51 is located centromerically adjacent to the duplicated region for LP. Notably, the molecular mechanism underlying Nablus mask-like syndrome is not clear. Close proximity to the LP region and overlap of cutaneous phenotype may suggest a shared mechanism for the two disorders mediated via altered genomic architecture.
The lower frequency of the missense SDC2 variant in individuals with SSc suggests that the minor rs1042381 allele may be protective against the disease. The expression of SDC2 is increased in scleroderma.27 The protective effect of the minor allele therefore could be hypothesised to result from partial loss of function of the encoded protein. Of note, the minor allele leads to substitution of serine with threonine at position 71 (p.Ser71Thr) in the ectodomain of syndecan-2, which is proposed to function in proteoglycan attachment. The available structural information on this domain of syndecan-2 is limited, but the serine residue at position 71 is fully conserved to Takifugu rubripes (fugu) (see online supplementary figure S3). The Grantham score (available from EVS), which categorises substitutions depending on chemical dissimilarity, is 58 (scores between 50 and 100 suggest moderate conservation) for this substitution. Of note, the missense variant is not represented on the Illumina CNV 370K and 550K Beadchip arrays used for genotyping in the large genome-wide association study (GWAS) of SSc, which may explain why this association has not been reported previously.52 Further, the SNP is not tagged to a haplotype captured by any of the genotyped SNPs in previous GWASs and therefore imputation is not possible. It is not unprecedented for variants relevant to a common disease not to be detected by GWASs but subsequently by a more targeted candidate approach.53 To test if the observed association is truly protective against SSc, validation in an independent cohort and functional characterisation of the amino acid substitution is required.
In summary, we have identified chromosome 8q22.1 microduplications as the cause of LP. Our work demonstrates the phenotypic range of the condition. We hypothesise that the phenotype of the condition is due to aberrant expression of GDF6 and SDC2. We show dysregulation of genes relevant to ECM homoeostasis in LP, highlighting the link between TGF-β signalling, GDF6 and syndecan-2. Further, we present evidence of decreased expression of SMAD6 in LP dermal fibroblasts. We report a missense SDC2 variant, which is protective against SSc.
The URLs for data presented herein are as follows:
Database of Genomic Variants—http://projects.tcag.ca/variation/
European Skeletal Dysplasia Network (ESDN)—http://www.esdn.org/eug/
EVS, NHLBI GO Exome Sequencing Project (ESP), Seattle, WA—http://evs.gs.washington.edu/EVS/) (accessed Jun 2013).
We thank Dr Stephen Roberts and Dr Stephen Eyre, University of Manchester for advice.
Review history and Supplementary material
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
SB and SAC contributed equally.
Contributors SB, SAC, CMK and WGN designed the study. MB, DD, BK, HK, AS-F, SU, HE, JW and ALH provided clinical details or patient samples. SC, SAC, SBD, JEU, GE and JH performed experiments. SB, SAC, CMK, WGN, CLRM and WWY analysed data. SB and WGN wrote the manuscript. All authors read and approved the manuscript.
Funding We acknowledge the support of Manchester Biomedical Research Centre, Wellcome Trust Institutional Strategic Support Fund award (ISSF fund—097820) and the Skeletal Dysplasia Group for funding. CMK and SAC are funded by the Medical Research Council grant G0801787.
Competing interests None.
Patient consent Obtained.
Ethics approval NHS ethics committee and University of Manchester.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.