Background Osteosclerotic metaphyseal dysplasia (OSMD) is a unique form of osteopetrosis characterised by severe osteosclerosis localised to the bone ends. The mode of inheritance is autosomal recessive. Its genetic basis is not known.
Objective To identify the disease gene for OSMD.
Methods and results By whole exome sequencing in a boy with OSMD, we identified a homozygous 7 bp deletion (c.5938_5944delGAGTGGT) in the LRRK1 gene. His skeletal phenotype recapitulated that seen in the Lrrk1-deficient mouse. The shared skeletal hallmarks included severe sclerosis in the undermodelled metaphyses and epiphyseal margins of the tubular bones, costal ends, vertebral endplates and margins of the flat bones. The deletion is predicted to result in an elongated LRRK1 protein (p.E1980Afs*66) that lacks a part of its WD40 domains. In vitro functional studies using osteoclasts from Lrrk1-deficient mice showed that the deletion was a loss of function mutation. Genetic analysis of LRRK1 in two unrelated patients with OSMD suggested that OSMD is a genetically heterogeneous condition.
Conclusions This is the first study to identify the causative gene of OSMD. Our study provides evidence that LRRK1 plays a critical role in the regulation of bone mass in humans.
- Osteosclerotic metaphyseal dysplasia
- Whole exome sequencing
- LRRK1 mutation
- Lrrk1 deficient mouse
Statistics from Altmetric.com
Osteosclerotic metaphyseal dysplasia (OSMD; OMIM %615198) is a unique form of osteopetrosis characterised by a distinctive pattern of osteosclerosis (localised increase of bone density).1 ,2 Only five patients in four families have been reported.2–5 The clinical features include developmental delay, hypotonia and seizures.2–5 Osteosclerosis localises predominantly to the metaphyses and epiphyseal margins of the appendicular bones and metaphyseal equivalents of the axial bones. The metaphyses show wide sclerotic bands and undermodelling, whereas the diaphyses are not sclerotic or even osteopenic. Osteosclerosis is also seen in the vertebral endplates, costal ends and margins of the flat bones. The skull is not affected. Laboratory abnormalities include raised levels of serum alkaline phosphatase (ALP), aspartate aminotransferase and creatine kinase and increased urinary pyridinoline and deoxypyridinoline excretion.2–5 An autosomal recessive inheritance has been suggested based on the occurrence in siblings born to first-cousin parents2; however, the disease gene has not yet been found.
To identify the disease gene for OSMD, we recruited three unrelated individuals diagnosed with OSMD and performed molecular genetic studies.
Materials and methods
Three patients (P1–3; see online supplementary table S1) diagnosed as having OSMD by experts of skeletal dysplasia (GN and GM) based on their clinical and radiographic findings were included in the study. P2 has been previously reported.2 P1 and P2 were consanguineous. The study protocol was approved by the ethical committee of RIKEN and participating institutions. Peripheral blood was obtained from the patients and their family members after informed consent and was used for the study.
Exome sequencing and variant calling
Genomic DNA was extracted by standard procedures from peripheral blood or Epstein–Barr virus immortalised lymphocytes. Concentration of DNAs for whole exome sequencing was measured using a Qubit V.2.0 Fluorometer (Life Technologies, Carlsbad, California, USA). Whole exome sequencing was performed as previously described.6 ,7 Briefly, DNA (3 μg) was sheared by Covaris S2 system (Covaris, Wobum, Massachusetts, USA) and processed by SureSelectXT Human All Exon V5 (Agilent Technologies, Santa Clara, California, USA). Captured DNA was sequenced using HiSeq 2000 (Illumina, San Diego, California, USA) with 101 bp pair-end reads with seven indices. Image analysis and base calling were performed using HiSeq Control Software/Real-time analysis and CASAVA 1.8.2 (Illumina). Reads were mapped to the reference human genome (hg19) by Novoalign-3.02.04. Aligned reads were processed by Picard to remove PCR duplicates. Variants were called by Genome Analysis Toolkit (GATK) v2.7-4 based on GATK Best Practice Workflow v3 and annotated by ANNOVAR (downloaded at 2013 June).
PCR and Sanger sequencing
We performed Sanger sequencing to confirm the deletion identified in the patient and parents by the exome sequencing. We amplified exon 34 by PCR using primers, LRRK1fsFP1 (forward primer): 5′-TTAGCCAGCTCCGCCTTCCT-3′ and LRRK1fsRP1 (reverse primer): 5′-TCCTTGGAGAACTAGAGGCCGAT-3′ and sequenced both strands. We also designed the primer sets to amplify all exons of LRRK1 and performed mutation screening for the subjects in whom LRRK1 mutation was not identified by the whole genome sequencing. Sanger sequencing was performed on a 3730 DNA analyzer (Life Technologies). Sequencher V.4.7 (Gene Codes, Michigan, USA) and Genetyx (Genetyx Inc, Tokyo, Japan) were used for aligning sequencing chromatographs to reference sequences and mutation detection.
Evaluation of the deletion found in LRRK1
We evaluated the deletion by databases, dbSNP (http://www.ncbi.nlm.nih.gov/projects/SNP/), 1000 genomes (http://www.1000genomes.org/), Human Gene Mutation Database (https://portal.biobase-international.com/hgmd/pro/start.php) and ESP6500 (http://evs.gs.washington.edu/EVS/).
Homozygosity mapping was performed using HomozygosityMapper as previously described8 on whole exome sequencing data of P1 and P2.
In vitro osteoclastogenesis and collection of osteoblast- and osteocyte-rich fractions
In vitro osteoclastogenesis was performed as described previously.9 Briefly, mouse bone marrow cells were cultured with 10 ng/mL macrophage colony stimulating factor (M-CSF) (R&D Systems, Minnesota, USA) for 2 days and bone marrow-derived monocyte/macrophage precursor cells (BMMs) were obtained. BMMs were cultured with 25 ng/mL receptor activator of nuclear factor κ-B ligand (RANKL) (PeproTech, Rocky Hill, New Jersey, USA) and M-CSF for 3 days. Osteoclastogenesis was evaluated by tartrate-resistant acid phosphatase (TRAP) staining using leucocyte acid phosphatase staining kit from Sigma (386-A). The staining mixture contained citrate, tartrate, acetate and naphthol AS-BI phosphoric acid solutions. The multinucleated cells (number of nuclei ≥3) were stained for 5–10 min at 37°C, then washed three times with phosphate-buffered saline. TRAP-positive multinucleated cells were counted. Osteoblast- and osteocyte-rich fractions were collected as described previously.10
Quantitative RT-PCR analysis
Total RNA was extracted using ISOGEN (Nippon Gene, Tokyo, Japan) following the manufacturer's protocol. Real-time quantitative RT-PCR analysis was performed as described previously.9 ,11 Briefly, first-strand cDNAs were synthesised from 0.5 μg of total RNA using Superscript III RT (Life Technologies). Quantitative RT-PCR analysis was performed with the CFX384 Touch real-time system (Bio-Rad Laboratories, Hercules, California, USA) using SYBR Green real-time PCR master mix (Toyobo, Osaka, Japan). The relative gene expression levels were calculated using the Ct method after the values were normalised against Gapdh expression. The primer sequences used in real-time RT-PCR analysis were: Gapdh, 5′-GGATGCAGGGATGATGTTCT-3′ and 5′-AACTTTGCCATTGTGGAAGG-3′; Ctsk, 5′-GGGAAGCAAGCACTGGATAA-3′ and 5′-CCGAGCCAAGAGAGCATATC-3′; Nfatc1, 5′-CAAGTCTCACCACAGGGCTCACTA-3′ and 5′-GCGTGAGAGGTTCATTCTCCAAGT-3′; Lrrk1, 5′-GAACTGAGCATGACCTGAGC-3′ and 5′-TCCTCCATGTCTAGGAACCC-3′; Dmp1, 5′-GGTTTTGACCTTGTGGGAAA-3′ and 5′-TTGGGATGCGATTCCTCTAC-3′.
Viral plasmid construction and lentivirus generation
Lentiviral plasmid of pSSFV-LRRK1 was generated by replacing green fluorescent protein (GFP) with a synthesised 3xFLAG/LRRK1 DNA (GenScript, Piscataway, New Jersey, USA) at the restriction sites of SgfI and PmeI of the pRRLsin-cPPT-SFFV-GFP-wpre vector. Lentiviral pSSFV-LRRK1-WDmt vector was generated by subcloning a synthesised mutated WD40 DNA fragment into the restriction sites of BsrGI and Pmel of the pSSFV-LRRK1. Lentiviral particles were generated by co-transfection of pSSFV-LRRK1 or pSSFV- LRRK1-WDmt with Pax2 and VSVG plasmids in 293T cells. Lentiviral productions were generated by transfection in 293T cells as described previously.12 Forty-eight hours after transfection, culture supernatants containing viral particles were collected, spun at 2000×g for 10 min and filtered through a 0.45 µm filter. Titres were determined by infecting 293T cells with serial dilutions and examining GFP expression of infected cells 24 h after infection. Primary monocytes derived from mouse bone marrow were seeded on bone slices in 48-well plates at a density of 10 000 cells/well and transduced by adding viral supernatant at a multiplicity of infection (MOI) of 5 in the presence of polybrene (8 µg/mL) and M-CSF (20 ng/mL) for 24 h, followed by replacement of a fresh osteoclast differentiation medium containing M-CSF (20 ng/mL) and RANKL (30 ng/mL).
In vitro osteoclast formation and bone resorption pit assays
Primary osteoclast precursors were isolated from the spleen or bone marrow of femur and tibia from 5-week-old mice as described previously.13 The isolated precursors were maintained in α-minimal essential medium supplemented with 10% fetal bovine serum, penicillin (100 units/mL), streptomycin (100 μg/mL) and M-CSF (20 ng/mL) at 37°C in 5% CO2 for 3 days to stimulate monocyte proliferation. To induce osteoclast differentiation, trypsinised precursors were seeded in 48-well plates (5000 cells/well) and incubated with M-CSF (20 ng/mL) and RANKL (30 ng/mL). The medium was changed every 2 days. Osteoclastogenesis was evaluated by counting TRAP staining-positive multinucleated cells having at least three nuclei. Osteoclast precursors were differentiated in vitro for 9 days until formation of large multinuclear cells for TRAP staining. Parallel cultures were maintained in osteoclast differentiation medium for another 4 days for pit staining. For bone resorption pit assays, slices from bovine cortical bone were placed in 48-well plates and cells were differentiated on top of the bone slices as described above. Cells on bone slices were digested with trypsin at 37oC overnight. Multinucleated cells were further removed by 5 min sonication in 1 M ammonia. Air-dried bone slices were stained with haematoxylin. The entire surface of each bone slice was examined and the total resorbed area per bone slice was quantified using ImageJ software (National Institutes of Health, USA).
Identification of a LRRK1 mutation in a patient with OSMD
We conducted whole exome sequencing in two probands (P1 and P2) (see online supplementary table S1) and obtained about 2.4–2.5 Gb sequences, which were successfully mapped to all human RefSeq. For both samples, at least 95.4% of all coding regions were covered in a depth of 10 reads (see online supplementary table S2). No common candidate genes that displayed mutations in the two subjects were found; however, we identified a homozygous 7 bp deletion (c.5938_5944delGAGTGGT) in the last exon of LRRK1 (leucine-rich repeat kinase 1) in P1. The deletion was not deposited in any databases, including dbSNP, 1000 genomes, Human Gene Mutation Database and ESP6500. The homozygosity mapping using the whole exome sequencing data in P1 and the parents showed that LRRK1 is residing in a 2.4 Mb homozygous region of the patient genome (see online supplementary table S3). The 7 bp deletion was confirmed by Sanger sequencing (figure 1A). The parents were heterozygous for the deletion.
Characterisation of the LRRK1 mutation
Human LRRK1 was first identified as a gene encoding a related protein of MFH (malignant fibrous histiocytoma)-amplified sequences with leucine-rich tandem repeats, an oncogene promoting tumorigenesis in nude mice.14 The LRRK1 gene (NM_024652) consists of 34 exons, encompassing ∼150 kb on chromosome 15q26.3. LRRK1 encodes a 2015 amino acid multidomain protein that contains ankyrin repeats, leucine-rich repeats, Ras of complex proteins (Roc), C-terminal of Roc (COR) and serine threonine kinase domain as well as seven tryptophan-aspartic acid dipeptide (WD) 40 domains at the C-terminus (figure 1B).15–17 The deletion in P1 is located within the last exon (exon 34) of LRRK1. It is predicted to result in a frame shift that destroys the seventh WD40 domain and adds a unique stretch of 66 amino acids to the C-terminal of the LRRK1 protein (p.E1980Afs*66). The WD40 domain is a ring-like structure that folds into a seven-bladed β propeller and also mediates protein–protein interaction.18 Lrrk1 is a mediator in the c-Src signalling pathway in mice and there is evidence that it negatively regulates bone mass.19 Hence, the partial deletion of WD40 domain of LRRK1 may abolish the interaction with additional unknown proteins in the c-Src signalling pathway.
Clinical features of the patient with the LRRK1 mutation
P1 was a 14-month-old boy who presented with failure to thrive, hypotonia and a delay in psychomotor development. Parents were of Moroccan (Berbers) origin and first cousins. He was born at 40 weeks with a birth weight of 3700 g. Physical examination at the age of 14 months did not reveal dysmorphic features but neurological evaluation showed hypotonia with weak deep tendon reflexes. The abdomen was slightly distended, but hepatosplenomegaly was clinically absent. His weight at 14 months was 8400 g (5th centile), his height 72 cm (2nd centile is 73 cm) and his fronto-occipital circumference 48 cm (98th centile). A skeletal survey (figure 2) showed a normal skull and vertebral height. The vertebral endplates, costal ends and iliac margins were sclerotic. Broad sclerotic bands were also seen in the metaphyses of the long and short tubular bones. Bone density was not increased in the diaphyses of the tubular bones. Metaphyseal undermodelling was evident in the proximal femora, distal radii and distal ulnae. Because of the failure to thrive and hypotonia, a metabolic investigation was initiated. This investigation showed normal values for serum amino acids and urinary organic acids but high levels of creatine kinase (31 300 U/L; normal 24–170 U/L), aspartate aminotransferase (649 U/L; normal 20–60 U/L), alanine aminotransferase (295 U/L; normal 19–61 U/L) and lactate dehydrogenase (5075 U/L; normal 425–975 U/L) in blood. ALP was decreased (103 U/L; normal 129–291 U/L). Lead poisoning was excluded by urine analysis. Additionally, a SNP array analysis was performed and showed a deletion (size: 102–113 kb) in the dystrophin gene on Xp21.1. A multiplex ligation-dependent probe amplification analysis confirmed the deletion involving exons 48–50 of the dystrophin gene, which was compatible with the diagnosis of Duchenne muscular dystrophy. The last follow-up at the age of 25 months showed a height of 79.8 cm (P2=82 cm), a weight of 10.8 kg (10th–25th centile) and a head circumference of 50 cm (98th centile). No dysmorphic features were present and the limbs were unremarkable (see online supplementary figure S1A). He had very weak deep tendon reflexes (due to the muscular dystrophy). Radiological examination showed that the metaphyseal sclerotic bands and marginal sclerosis of the flat bones had become less conspicuous (see online supplementary figures S1B–D).
Phenotype of the Lrrk1 knockout mice
Lrrk1 knockout (KO) mice have been reported to exhibit severe osteopetrosis.19 We conducted a detailed skeletal survey and histologic analysis of the Lrrk1 KO mice and concluded that the bone changes were very similar to those seen in P1 (figure 3). Briefly, the vertebral bodies showed endplate sclerosis, but the skull was spared. The ribs were somewhat broad and sclerotic. The metaphyses of the long and short tubular bones of the mice were all severely sclerotic and undermodelled, while the bone density of the diaphyses was decreased (see online supplementary table S4). Histology of the growth plate of the Lrrk1 KO mice showed an elongated hypertrophic zone and delayed replacement of cartilage by bone (see online supplementary figure S2), corresponding to the metaphyseal changes seen on the radiographs. The close phenotypic resemblances between P1 and the Lrrk1 KO mouse suggested that the deletion in P1 is a hypomorphic mutation that causes OSMD in humans.
Functional analysis of the LRRK1 mutation
To test the lentiviral titre and expression levels of the FLAG-tagged LRRK1 proteins, we transduced 293T cells with lentiviral particles at 5 MOI and extracted the total cellular protein. 293T cells can be easily transduced. We typically obtain 95% or greater transduction efficiency at 5 MOI. We tried different dosages of MOI in primary osteoclast precursors and found that at >5 MOI, some cells began to die after transduction. Therefore, we maintained 5 MOI for primary osteoclast precursors for osteoclast differentiation assays. An aliquot of the cellular protein was used for a western blot with an antibody against FLAG tag. The cells infected with lentivirus expressed comparable levels of wild-type and mutant LRRK1 proteins (figure 4A). The transduction efficiency of lentiviral particles in primary cultures of monocytes derived from Lrrk1 KO mice was >70% (figure 4B). To examine the function of the mutant LRRK1, osteoclast precursors from Lrrk1 KO mice were infected with lentiviral particles expressing GFP (control), wild-type LRRK1 or mutant LRRK1. Normal numbers of osteoclasts were generated in the KO and wild-type mice. The areas of mature TRAP-positive osteoclasts on bone slices were comparable for Lrrk1-deficient osteoclasts expressing GFP, wild-type LRRK1 and mutant LRRK1. However, while the wild-type LRRK1 in Lrrk1-deficient cells completely rescued the defect in osteoclast resorption activity, the mutant LRRK1 failed to restore the osteoclast activity (figure 4C). There was no significant change in pit areas in the cells expressing mutant LRRK1 as compared with that expressing GFP. The area of resorption pits was reduced by 90% in Lrrk1-deficient GFP expressing cells as compared with that of the wild-type cells (figure 4D). The individual resorption pits remained smaller and the pits were shallower in the mutant LRRK1-expressing osteoclasts. These data strongly suggest that the mutation has a hypomorphic effect on the function of LRRK1.
Screening of LRRK1 mutation in P2 and P3
We checked the coverage of the exome sequence data for P2 and examined all exons that had low coverage (<96%) by Sanger sequencing; however, we found no variants likely to be disease causing. In P2 who was a child of consanguineous parents, we found some heterozygous variants in LRRK1, and homozygosity mapping using the exome data showed no long homozygous stretch containing LRRK1. These results suggest that P2 is unlikely to have a LRRK1 mutation. We also Sanger sequenced P3 for all LRRK1 exons and flanking introns, but found no LRRK1 mutation.
The identification of a biallelic and hypomorphic LRRK1 mutation in a boy with radiographic features of OSMD indicates that this gene plays a role in bone homoeostasis in humans. This assumption is reinforced by the observation that Lrrk1 KO mice exhibit a similar radiographic phenotype with dense and broad osteosclerotic bands in the metaphyses of tubular bones (figures 2 and 3, see online supplementary table S4). In addition, the growth plate abnormalities in these mice with expanded hypertrophic zone and delayed endochondral ossification (see online supplementary figure S2) may explain the shortness in the affected boy and KO mice. The osteopetrotic changes in the mouse result from markedly delayed resorption of cartilage matrices owing to dysfunction of mature osteoclasts. While Lrrk1 expression is very low in murine osteoblasts and osteocytes compared with osteoclasts (see online supplementary figure S3), its expression is increased in osteoclasts in the late stage of differentiation (see online supplementary figure S4), suggesting a role of Lrrk1 in the maturation of multinucleated osteoclasts. Consistently, Lrrk1 KO mice show osteoclast-rich osteopetrosis due to the dysfunction of multinucleated osteoclasts.19 Osteoclast precursors derived from Lrrk1 KO mice differentiate into multinucleated cells in response to M-CSF and RANKL treatment, but these cells cannot form peripheral sealing zones and ruffled borders and hence experience difficulties in resorbing bone.19 This defect of osteoclast function is partially rescued by overexpression of the constitutively active form of c-Src (Y527F). Immunoprecipitation assays in osteoclasts showed a physical interaction of Lrrk1 with the C-terminal end of Src kinase (Csk).19
We could not identify a LRRK1 mutation in P2 and P3. P2 is one of the original patients reported with OSMD.2 We examined both patients for the entire coding exons of LRRK1 and their surrounding introns by, respectively, exome sequencing and Sanger sequencing, but found no pathogenic nucleotide change. Although P2 is a child of consanguineous parents, homozygosity mapping using the exome data showed that LRRK1 is not residing in a homozygous stretch. In addition, Sanger sequencing of LRRK1 revealed several heterozygous SNPs in P2. These data suggest that OSMD is a genetically heterogeneous condition and that other genes must also be responsible for this rare disorder. The heterogeneity of OSMD is also reflected by the observation that P1 did not have raised ALP, although this has been described in the original patients. Interestingly, the Lrrk1 KO mice have decreased ALP levels. The differences in ALP levels in OSMD may indicate that the causal genes interfere at different stages in osteoclastogenesis and/or osteoclast function.
Our data provide new insights into the pathophysiology of increased bone mass phenotypes and also may provide us with new opportunities for the treatment of osteoporosis, one of the most serious health problems in the elderly.20 Understanding the genetic defects causing pycnodysostosis (OMIM: 265800) and sclerosteosis (OMIM: 269500) has resulted in the development of good drug targets for osteoporosis, exemplified by the use of inhibitors for cathepsin K and ClC-7 and of anti-sclerostin antibodies.21–23 It is of note that bone mineral density of Lrrk1 KO mouse is the highest among 3629 distinct KO mouse lines, including Sost and c-Src KO mice.19 LRRK1 would therefore be a good target molecule for the treatment of osteoporosis. The KO mice have lifelong accumulation of bone; the age-related bone loss does not seem to occur in these mice, as trabecular bone mineral density measurements remain stable across ages. The mice also respond normally to the anabolic actions of teriparatide treatment and are resistant to ovariectomy-induced bone loss, the model of postmenopausal osteoporosis.19 These findings suggest that LRRK1 may be a fascinating new drug target that is specific to osteoporosis and has few or no side effects during long-term treatment.
We thank the patients and their families for their help with the study. We also thank N Atsumi for checking the English. The Department of Veterans Affairs in Loma Linda, California provided facilities to carry out the Lrrk1 knockout mouse work.
AI and WX contributed equally.
Contributors SI designed the project and provided overall project management. AI, GM and SI wrote the paper. AI performed the sequence experiments and analysed the data. WX and SM generated LRRK1-deficient mice and analysed the data. TN and ZW performed the experiments and analysed the data. MKFD, MK, WVH, DR, JS, HO and GM collected and managed the DNA samples. NoM and NaM performed the exome experiments and analysed the data. GN analysed the clinical data.
Funding This study was supported by research grants from the Japan Agency For Medical Research and Development (AMED) (contract No 14525125) and by the European Commission Seventh Framework Programme (SYBIL project; grant number 602300). WX was supported by grant 1R21 AR066831-01 from the National Institutes of Health and a GAP (Grants in Aid Program) grant from the American Society for Bone and Mineral Research. ZW was supported by Takeda Science Foundation, Japan.
Competing interests None declared.
Patient consent Obtained.
Ethics approval This study was performed with the approval of the ethical committees of RIKEN and participating institutions.
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.