|Home | About | Journals | Submit | Contact Us | Français|
This report identifies human skeletal diseases associated with mutations in WNT1. In ten family members with dominantly inherited early-onset osteoporosis, a heterozygous missense variation c.652T>G (p.Cys218Gly) in WNT1 segregated with the disease, and a homozygous nonsense mutation (c.884C>A, p.Ser295*) was identified in two siblings with recessive osteogenesis imperfecta. In vitro, aberrant forms of WNT1 protein showed impaired capacity to induce canonical WNT signaling, their target genes, and mineralization. Wnt1 was clearly expressed in bone marrow, especially in B cell lineage and hematopoietic progenitors; lineage tracing identified expression in a subset of osteocytes, suggesting altered cross-talk of WNT signaling between hematopoietic and osteoblastic lineage cells in these diseases.
Osteoporosis is a common skeletal disorder characterized by low bone mineral density (BMD), impaired bone quality, and fragility fractures1. Multiple genetic loci have been defined from genome-wide association studies in patients with osteoporosis, including WNT ligands, yet the known loci are generally associated with fracture odds ratios below 1.1.2 Recently, novel bone metabolic pathways have been discovered in patients with osteogenesis imperfecta (OI), a Mendelian disease characterized by brittle bones.3 The role of the WNT pathway in bone formation and maintenance has been extensively studied since the identification of mutations in key signaling WNT mediators (LRP5 and sclerostin) in both diseases with high and low bone mass phenotypes.4,5 However, despite abundant studies in cell and mouse models, the identity of the key WNT ligand that signals via LRP5/6 in human bone formation has remained unknown.6
We ascertained a Finnish family (Fig. 1a) with severe early-onset and dominantly inherited osteoporosis. Clinical and radiologic evaluation of 16 family members confirmed osteoporosis with low BMD and low-impact vertebral and peripheral fractures in 10 family members (Table 1 and Fig. 1b). Affected persons had no extra-skeletal abnormalities. Serum and urine parameters of calcium homeostasis and bone turnover were normal (Supplementary Appendix, Table S1). Histomorphometry of transiliac bone biopsies in two adults showed severe low-turnover osteoporosis with low bone formation rate; a 14-year-old boy in the family had normal bone volume but low bone formation and remodeling for his age (Supplementary Appendix, Fig. S1, Table S2).
We also ascertained a Laotian Hmong family with two severely affected daughters suffering from a presumed recessive form of OI (Fig. 1c). In the first affected child, the first fracture was documented at one month of age. Radiographs of both affected children showed severe osteopenia with multiple fractures and sequelae over time, including vertebral compression fractures, resultant kyphoscoliosis, severe short stature, and long bone deformities (Fig. 1d). The older sister, now aged 26 years, is wheelchair-bound because of her bone disease but is able to perform most activities of daily living and is intellectually normal. Her head circumference is at the 25th percentile, as is her weight, but her height is more than 2 SD below the mean (less than 3 feet) with severe long bone deformities (Supplementary Appendix, Fig. S2).
In the second affected child, a third trimester prenatal ultrasound revealed multiple femoral and rib fractures. She has severe intellectual disability with absence of speech, and has been quadriplegic from toddlerhood. Her MRI revealed severe hypoplasia of the left cerebellar hemisphere with a short midbrain (Fig. 1e). She is now 23 years old, has no functional use of her hands, does not have language, and has achieved no motor milestones. She has unilateral ptosis and exotropia. Her head circumference is just below the 3rd percentile (51.5 cm), her weight is at the 25th percentile, and her height is more than 2 SD below the mean (less than 3 feet) with severe long bone deformities (Supplementary Appendix, Fig. S2).
Both sisters had severe dental caries as young children necessitating extraction of all deciduous teeth in childhood. Permanent dentition is relatively normal in both with few caries and no signs of dentinogenesis imperfecta. Hearing is normal. Their sclerae are white. Both sisters have asynchronous eye blinking. Both have mild restrictive airway disease requiring inhaled bronchodilators. Their hands are markedly hyperextensible with marked laxity at the interphalangeal joints. Fibroblast collagen studies were normal in both (data not shown).
The other siblings in this family had no features of OI or any neurologic disease. The mother, at age 44, has normal BMD on DXA scan and normal spinal radiographs. The father, at age 43, has normal femoral BMD, but the lumbar spine (L1–L4) has a BMD with a Z-score of −1.8 (the father’s height is normal at 5 feet and 5 inches). His spinal radiographs show a mild compression deformity involving the superior endplate of the L5 vertebral body.
The families provided signed informed consent to participate in studies approved by the Ethics Committee of the Helsinki University Central Hospital for Family 1, and by the Institutional Review Board of the Baylor College of Medicine for Family 2. For Family 1, family members underwent questionnaires, examinations and DXA scans. Three subjects had transiliac bone biopsies after tetracycline double-labeling. We used a genome-wide 384 microsatellite marker scan, followed by fine-mapping for chromosome 12 using 29 additional markers, and finally, a targeted next-generation sequencing strategy described in the Supplementary Appendix to sequence the exons and flanking intron bases in the linkage region.
For Family 2, apart from the standard clinical investigation and care of patients with OI and with the neurological disease in the second sibling, the parents had DXA scan and spinal radiographs. Given the family history with two affected siblings and the negative fibroblast collagen studies, whole exome sequencing was performed as described previously7 to identify a potential recessive OI gene. Variants were called and analyzed with an in-house pipeline described in the Supplementary Appendix. The final variant filtering scheme focusing on rare recessive variants is detailed in Table S3 in the Supplementary Appendix.
A cDNA encoding WNT1 was cloned into mammalian expression plasmids and the mutations were introduced using standard techniques. The plasmids were transfected in HEK293T, MC3T3 and C57MG cells and the cells were tested for beta-catenin activation, WNT1 protein expression, target gene transcription and differentiation. To profile Wnt1 gene expression in vivo, we used wild type mice for quantitative real-time PCR assays, and lineage tracing in reporter Rosa-mT/mG miceintercrossed withWnt1-Cre transgenic mice as previously described.8,9 Details about these experiments are described in the Supplementary Appendix.
In Family 1, we performed a genome-wide microsatellite linkage analysis using DNA from 10 affected and 6 healthy family members. It revealed one putative linkage area of 25.5 Mb on chromosome 12 with a p-value of 0.01 (Supplementary Appendix, Fig. S3 and S4). Fine-mapping and targeted next-generation sequencing allowed the identification of a single novel variant in WNT1 (p.C218G) segregating with the phenotype. A c.652T>G mutation affects the first cysteine of the so-called WNT motif (C-[KR]-C-H-G-[LIVMT]-S-G-x-C), which is conserved across species and all 19 known human WNTs10. The mutation substitutes a polar sulfur-containing cysteine with a small non-polar amino acid glycine (Supplementary Appendix, Fig. S5a).
In Family 2, we first excluded mutations in the known recessive OI genes by Sanger sequencing, and we then performed whole-exome sequencing. Analysis of the rare or novel genetic variations revealed two potential candidates. The first was a variant in COL1A2 (NM_000089.3:c.3200G>A, p.Arg1067His). However, this variant was not considered as causal since the father, who did not show evidence of OI, harboured the mutation in his blood and fibroblasts, while both affected children were also heterozygous for this mutation. In addition, this COL1A2 variant is not predicted to cause the severe OI phenotype (with neurologic abnormalities) diagnosed in this family given that the amino acid change is within the X residue of the G-X-Y collagen triplet and that the collagen studies in fibroblasts were normal. The other variant found was a homozygous nonsense mutation in WNT1 (NM_005430.3:c.884C>A, p.Ser295*) (Supplementary Appendix, Fig. S5b,c). Both affected children were homozygous for the change and both parents were heterozygous. The mutation resides in the last exon of WNT1, and thus escapes nonsense-mediated decay and leads to the expression of a WNT1 protein truncated of its last 76 amino acids (Supplementary Appendix, Figs. S5d and S6).
To evaluate the functional consequence of the WNT1 mutations, the mutant and wild type WNT1 were expressed in C57MG cells, which do not express endogenous Wnt111. All proteins showed similar cellular distribution, indicating that the mutant proteins were stable and that the mutations did not alter intracellular targeting of WNT1 (Supplementary Appendix, Fig. S7). We next assessed induction of canonical WNT signaling. Activation of this pathway results in accumulation of non-phosphorylated active β-catenin in the nucleus, where it induces target gene expression in co-operation with TCF/LEF transcription factors.6 In contrast to wild type WNT1, WNT1C218G and WNT1S295* did not induce significant accumulation of non-phosphorylated or total β-catenin in either cytosolic or nuclear fractions (Fig. 2a,b). We also found that in a superTOPFlash-luciferase reporter assay,12 both mutant proteins had significantly reduced capacity to activate canonical WNT signaling compared to wild type. On co-transfection, WNT1C218G did not interfere with the induction of superTOPFlash reporter by wild type WNT1, implying that WNT1C218G does not function in a dominant negative manner; however, the WNT1S295* protein seemed to have mild, context-dependent dominant negative activity (Supplementary Appendix, Fig. S8). To assess the effect of the mutations on endogenous targets of WNT signaling, we expressed the mutant proteins in MC3T3 osteoblastic cells. The expression of the downstream β-catenin targets (Axin2 and Lef1) were significantly induced by the normal WNT1, but not by the two mutants (Supplementary Appendix, Fig. S9). In addition, stable expression of the abnormal proteins stimulated less mineralization than did wild type WNT1 (Fig 2c). These assays demonstrate the markedly diminished capacity of the abnormal proteins to induce canonical WNT signaling and associated osteoblast differentiation.
To gain insight into how WNT1 modulates bone mass, we analyzed its expression pattern. We did not detect significant Wnt1 mRNA in mouse calvarial osteoblasts, osteoclasts or human mesenchymal stromal cells (MSCs) (data not shown). In a panel of 19 mouse tissues we consistently detected Wnt1 expression by RT-PCR in the brain, femur and spleen; the brain showed the highest relative mRNA expression (Supplementary Appendix, Fig. S10). Interestingly, we found clear Wnt1 expression in the hematopoietic bone marrow (Fig. 2d). We used Fluorescence Activated Cell Sorting (FACS) to isolate major hematopoietic lineages present in the bone marrow - B-cell, monocyte/macrophage and erythrocyte lineages - and analyzed Wnt1 expression by RT-PCR (Fig. 2d, Supplementary Appendix, Fig. S11). We observed that Wnt1 was expressed in B220+ cells of the B-cell lineage and to a lesser extent in the lineage-negative cells that represent hematopoietic progenitor cells, while the myeloid and erythroid lineages were Wnt1-negative. We further performed lineage tracing experiments using Wnt1-Cre transgenic mice intercrossed with the Rosa-mT/mG reporter mice which express green fluorescent protein upon Cre-mediated activation in a cell-type specific fashion.8,9 Fluorescence analysis showed strong expression in a subset of osteocytic cells in the subchondral bone and less so in cortical bone (Fig 2e, Supplementary Appendix, Fig. 12).
Several lines of evidence indicate that canonical WNT signaling is essential for normal skeletal development and homeostasis.6,13 It induces osteoblast differentiation and bone formation in early osteoblast progenitors and regulates osteoblast-dependent osteoclastogenesis in mature osteoblasts/osteocytes.14–17 Moreover, mice lacking the WNT receptor FZD9 display a cell autonomous defect of bone formation.18
The patients presented here have a form of autosomal dominant osteoporosis caused by a missense mutation in WNT1 and a severe form of OI (that could be considered as prenatal-onset osteoporosis) caused by a homozygous truncation mutation in WNT1. These mutations lead to a loss-of-function of WNT1 signaling and impair bone formation, and the truncating mutation seems to have a mild, context-dependent dominant negative activity in vitro. In Family 2, the significant intellectual disability with cerebellar malformation in one of the affected sisters and a severe clinical OI phenotype in both suggests that homozygous mutations in WNT1 may variably affect other signaling pathways critical for central nervous system development consistent with the mouse Wnt1 knockout phenotype.19
Communication between osteoblasts/stromal cells and hematopoietic cells within the bone marrow hematopoietic stem cell niche is essential for normal hematopoiesis, and WNT signaling plays a role in this crosstalk.20 Conversely, hematopoietic cells and especially B cells have been suggested to regulate bone formation.21 Together, our data suggest that WNT1 exhibits a complex expression pattern that is likely dynamic both temporally and spatially. The net effect of the identified mutations is likely complex, interfering with WNT signaling between different cellular compartments including the hematopoietic and osteocytic lineages and potentially different Wnt-FZD-Lrp5/6 interactions in a context-dependent fashion. These data support a role for hematopoietic cells in regulating bone formation and implicate WNT1 as a key signaling molecule mediating these effects. This work also identifies WNT1 as an important WNT ligand for regulating bone mass in humans and, as such, it may serve as an important biomarker of skeletal health and/or target for therapy in OI and osteoporosis.
We thank the FIMM personnel (Pekka Ellonen, Anna-Maija Sulonen, Maija Järvinen, and Sari Hannula) for their assistance in the next-generation sequencing and Dario Greco for his help with the blasting of the Sure Select library. We thank Professor Anthony Brown for providing the C57MG-A5-cell line. Inari Tamminen is acknowledged for help with bone histomorphometry, Sofia Oja and Ariel Noro for help with mesenchymal stromal cell cultures, and Merja Lakkisto for expert technical assistance. We thank Cell Imaging Core at Turku Centre for Biotechnology for their help in FACS sorting. This study was financially supported by the Folkhälsan Research Foundation (C.L., O.M., M.W. and A-E.L.), the Academy of Finland (R.K. #139165, and O.M. #132894 and #250780), the Sigrid Juselius Foundation (R.K. and O.M.), the Foundation for Pediatric Research (O.M.), the Waldemar von Frenckell Foundation (C.L.), the Helsinki University Research Funds (O.M.), Helsinki, Finland, and European Calcified Tissue Society Career Establishment Award (R.K.). Research funding includes NIH grants PO1 HD22657 (B.L., D.K., D.C.), PO1 HD070394 (B.L., D.K., D.C.), and The Rolanette and Berdon Lawrence Bone Disease Program of Texas (BL). P.M.C. is supported by a CIHR clinician-scientist training award and the O’Malley Foundation. Kyu Sang Joeng is supported by NRSA fellowship F32 AR063616. We thank Shalini N. Jhangiani for coordination of exome sequencing and Terry Bertin for real-time PCR analyses. We thank Alyssa Tran for clinical research support. This work was also supported by the BCM Intellectual and Developmental Disabilities Research Center (HD024064) from the Eunice Kennedy Shriver National Institute of Child Health & Human Development.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
AUTHOR CONTRIBUTIONSCML, KSJ, PMC, KT, MG, MP, MW, TJH, VNP, MA, TL, HK, AEL, LN performed experiments. CML, OM, DK, CJRC were involved in the clinical assessment. JTL and PMC performed exome analysis. CML, RK, DC, RG, WGC, BHL and OM helped plan experiments. RK, BHL, and OM take responsibility for the integrity of the data. All authors reviewed the manuscript.