Data from this study were presented at the 11th and Valedictory Workshop on Cell Biology of Bone and Cartilage in Health and Disease, March 18, 2006, Davos, Switzerland.
The authors state that they have no conflicts of interest.
Published online on February 12, 2007
We found a novel heterozygous missense mutation (M282V) in the LRP5 gene in a patient with a high bone mass phenotype. In vitro studies suggest that a reduced antagonistic effect of DKK1 on canonical Wnt signaling contributes to the molecular effect of this mutation and its pathogenic consequence.
Introduction: Gain-of-function mutations in the gene encoding LDL receptor–related protein 5 (LRP5) cause high bone mass. Recent studies revealed that a reduced inhibition of canonical Wnt signaling by Dickkopf 1 (DKK1) contributes to the pathophysiology of this disease phenotype.
Materials and Methods: We report on a 55-yr-old female patient with a high bone mass phenotype. Sequencing of exons 2–4 of the LRP5 gene was carried out to screen for disease-associated mutations in genomic DNA of the patient. The effect of the identified mutation on LRP5 membrane trafficking was studied by immunoblotting of a truncated form of LRP5. Additionally, Wnt signal activation in the absence and presence of DKK1 was assessed using a TCF4-based reporter gene assay in Saos-2 cells.
Results: Our patient presents with dense bones (Z-scores > +6), and radiographic examination showed a generalized thickening of the skeleton. BMD at the hip and lumbar spine significantly decreased through the passage to menopause, indicating no protection to bone loss. Further clinical evaluation revealed torus palatinus. Mutation analysis showed the presence of a novel heterozygous missense variant (844A→G; M282V) in LRP5, located in the first β-propeller domain of the extracellular portion. Although protein secretion seemed to be impaired, this mutant was able to transduce Wnt signals at levels comparable with wildtype LRP5. We additionally observed a less efficient inhibition of canonical Wnt signaling by DKK1.
Conclusions: Like all high BMD–associated gain-of-function LRP5 mutations described thus far, the M282V variant affects an amino acid located in the first β-propeller domain, underlining the functional importance of this region in the pathophysiology of these conditions. This mutation most likely alters a region important for LRP5 modulation by DKK.
Low-density lipoprotein receptor–related protein 5 (LRP5) is involved in canonical Wnt signaling, where it acts as a co-receptor for Wnt ligands to regulate intracellular signal transduction by β-catenin.(1) The protein is a key regulator in a wide range of developmental and physiological processes and is expressed in many embryonic and adult tissues, including osteoblasts of the endosteal and trabecular bone surfaces.(2) Recent advances in bone biology emphasize the importance of the canonical Wnt signaling pathway during osteoblast differentiation in embryonic development and for osteoblast maturation and activity during postnatal life.(3,4) Moreover, mutations in the LRP5 gene give rise to bone abnormalities. Whereas the autosomal recessive osteoporosis pseudoglioma (OPPG) syndrome, characterized by congenital blindness with osteoporosis, is associated with loss-of-function of LRP5,(5,6) gain-of-function mutations in LRP5 cause high BMD phenotypes.(7–13) Patients with activating LRP5 mutations share a similar skeletal phenotype including a dense skull base and cortical thickening of the tubular bones.(11) Occasionally, other clinical symptoms, including a wide jaw, torus palatinus, neurological complications, craniosynostosis, and oropharyngeal disease, were observed.(7,8,10–13)
Structurally, LRP5 consists of a large extracellular domain, a single transmembrane-spanning segment, and a cytoplasmic tail (www.stanford.edu/∼rnusse/wntwindow.html). The extracellular domain is composed of four epidermal growth factor (EGF)-like repeats, each separated by six YWTD spacer domains, that form a six-bladed β-propeller domain.(14) In addition, three LDL receptor domains flank the 23-amino acid membrane-spanning segment. Interestingly, the nine reported gain-of-function LRP5 mutations causing high BMD phenotypes are all located in the first β-propeller domain of LRP5.(7–13) Different proteins can bind to the extracellular portion of LRP5 and/or its homolog LRP6, thereby regulating the activity of the canonical Wnt signaling pathway.(15) Dickkopfs (DKKs) 1–4 comprise one family of proteins modulating this signaling cascade by direct binding to LRP5/6.(16,17) Antagonistic action of the DKKs is effected by the formation of a ternary complex with Kremen and LRP5/6, thereby triggering clearance of the Wnt co-receptor from the cell surface by endocytosis.(18) The third β-propeller domain of LRP5/6 is crucial for binding DKK, although other binding sites might also be involved.(16,17) Previous studies have shown that all high BMD LRP5 mutations studied thus far result in less efficient inhibition of canonical Wnt signaling by DKK1, most likely as a result of reduced affinity of mutant LRP5 to DKK1.(7,17,19)
In this report, we describe a novel LRP5 missense mutation, located in the first β-propeller domain of the LRP5 protein, in a female patient with increased BMD. We were able to show that this LRP5 sequence variant does not result in constitutive activation of canonical Wnt signaling and that, in the presence of Wnt1 ligand, the level of Wnt signal transduction was comparable with that observed with WT LRP5. Our results indicate that secretion of mutant LRP5 is impaired, although the protein is able to traffic to the cell surface at levels comparable with WT LRP5. Finally, we observed increased LPR5-mediated Wnt signal transduction in the presence of DKK1 that is most likely the result of reduced inhibition by DKK1.
MATERIALS AND METHODS
We report on a female patient, 55 yr old, of Belgian origin, presenting with a high bone mass phenotype. She is of normal height (168.9 cm) and weight (66.6 kg). Hyperostosis was found incidentally at the age of 37 yr. Clinical examination was not remarkable, except for a palpable torus palatinus the patient was not aware of (Figs. 1A and 1B). There are no indications of a family history for high BMD, although bone densitometry measurements have never been carried out in any of her relatives, as far as she knows. Unfortunately, none of them were willing to undergo densitometry.
Bone densitometry of the hip and lumbar spine (LS) was assessed at the age of 50 and 55 yr by DXA using Hologic QDR-4500A equipment (Hologic, Waltham, MA, USA), and values were compared with values for age- and sex-matched normal individuals. At 50 yr of age, these measurements revealed a BMD of 1.954 g/cm2 in the left total hip (Z-score = +8.75), a BMD of 1.791 g/cm2 in the femoral neck (FN; Z-score = +9.22), and a BMD of 1.699 g/cm2 in the L1–L4 spine (Z-score = +6.65). At age 55 yr, BMD at the left total hip, FN, and L1–L4 spine were 1.862 (Z-score = +8.23), 1.638 (Z-score = +8.18), and 1.613 g/cm2 (Z-score = +6.23), respectively. From these BMD values, it is interesting to note that through the passage to menopause (at 52 yr), the patient experienced a bone loss amounting to 4.7% at the hip and to 5.1% at the LS, implying capability of accelerating bone remodeling. Radiographic examination at several skeletal sites revealed a significant thickening of the skull and long bones. MRI and a standard radiograph of the skull showed a marked thickening of the calvaria, especially at the frontal and occipital bone (Figs. 1A and 1B). A postero-anterior radiograph of the hands showed cortical thickening of the metacarpal bones and phalanges (Fig. 1C). Additionally, cortical thickening and slight broadening of the diaphyses of the humerus, ulna, and radius (Fig. 1D) and a subtle increase in thickness of the femoral shaft were observed (Fig. 1E). Increased density of the vertebral bodies and posterior processes was also reported, without obvious osteoarthritis, except for incipient degenerative discopathy at the L5–S1 level. The patient did not suffer from obvious secondary clinical complications.
At the age of 52 yr, she was referred to us for mechanical pain in the right knee caused by gonalgia. Plain films of the knees revealed chondrocalcinosis and incipient osteoarthritis. Treatment consisted of paracetamol with effective improvement. At 54 yr of age, she suffered from cervical pain for which she underwent physiotherapy. Subsequent radiographic examination revealed arthrosis at the lumbar and cervical vertebrae. There was no evidence of spondylolysis. Only recently, the patient developed a breast cancer for which she is currently being treated.
Biochemical studies revealed normal values for serum calcium, phosphorus and magnesium, bone-specific alkaline phosphatase (ALP), PTH, 25(OH)D3, and C-terminal telopeptide (CTX). Slightly increased levels of 1,25(OH)2D3 were observed (Table 1).
Table Table 1.. Biochemical Survey for Bone Remodeling Parameters in the Patient
Mutation analysis of the LRP5 gene
Genomic DNA was isolated from peripheral blood leukocytes by standard techniques, and direct sequence analysis was carried out. Because all of the currently reported gain-of-function mutations in LRP5 involve exons 2, 3, and 4, only these exons, with inclusion of the adjacent splice sites, were PCR amplified and subjected to DNA sequencing. Sequencing reactions were performed using the ABI Big Dye Terminator v1.1 Cycle Sequencing kit (Applied Biosystems), and fragments were analyzed on an ABI 3130 Automated Sequencer. PCR primers and conditions and sequencing primers are available on request.
In vitro mutagenesis
The 844A→G mutation was introduced in the previously described full-length human wildtype LRP5 (WT-LRP5)(5) using the QuickChange Site-Directed Mutagenesis Kit (Stratagene). The complete insert sequence was verified for the presence of the mutation and absence of PCR errors. A truncated expression construct harboring the 844A→>G variant was made by replacing the 1.84-kb SacII-Bsu36I restriction fragment from a truncated form of human wildtype LRP5 (WT-LRP5N-myc)(5) with the SacII-Bsu36I restriction fragment from the 844A→G full-length LRP5 constructs. The truncated LRP5 expression constructs lack the transmembrane and cytoplasmic domains but have a c-myc epitope at the carboxy terminus.
Cell culture, transfection, and luciferase reporter assays
The human embryonic kidney (HEK) cell line 293T, obtained from D Huylebroeck (University of Leuven, Leuven, Belgium), and the Saos-2 human osteosarcoma cell line (ATCC) were grown in DMEM supplemented with FBS (10% vol/vol). Luciferase reporter assays were carried out as previously described.(20) Briefly, in Saos-2 cells, a mouse Wnt1-V5 (B Williams, Van Andel Research Institute),(19)mesd-c2 (B Holdener, State University of New York),(21) and a pGL3-OT reporter construct (B Vogelstein, The Johns Hopkins University School of Medicin)(22) were co-transfected with full-length LRP5 in the presence or absence of DKK1 (S Sokol, Mount Sinai School of Medicine).(23) Each transfection was carried out in triplicate and repeated independently in three separate experiments. Forty-eight hours after transfection, cells were lysed, and luciferase activity was measured on a Glomax 20/20n Luminometer (Turner Designs) using the dual-luciferase reporter assay system (Promega).
Isolation of cell membrane fractions and immunoblotting
To study the ability of truncated LRP5 (lacking the transmembrane and cytoplasmic domains but with a c-myc epitope at the carboxy terminus) to be secreted, we collected conditioned media (CM) and prepared total cell lysates and cytosolic and membrane fractions from HEK293T cells. Briefly, truncated LRP5 was co-transfected with a mesd-c2 expression construct in HEK293T cells. Seventy-two hours after transfection, CM was collected. To prepare total cell lysates, HEK293T cells were lysed in RIPA buffer (50 mM Tris-HCl pH = 8.0; 150 mM NaCl; 1% NP-40; 0.5% Na-deoxycholate; 0.1% SDS) for 20 minutes at 4°C. To fractionate the HEK293T cells into membrane and cytosolic fractions, cells were harvested after trypsinization and washed with ice-cold cell culture medium and then with membrane preparation buffer (20 mM Tris-HCl, pH 7.5, 300 mM sucrose, 60 mM KCl, 15 mM NaCl, 0.5 mM EDTA) supplemented with protease inhibitors (Complete mini EDTA-free; Roche). Cells were subsequently homogenized, and the nuclear pellet and cell debris were removed by low-speed centrifugation (800g for 5 minutes at 4°C). High-speed ultracentrifugation (100,000g for 1 h at 4°C) was performed to separate membranes and vesicles from soluble proteins. The soluble fraction was regarded as the cytosolic fraction. The pellet was resolved in membrane preparation buffer containing 0.5% Triton X-100 (membrane fraction). Immunoblotting assays were carried out as previously described.(20) The presence of truncated LRP5 in CM, total cell lysates, and cytosolic and membrane fractions was verified with the c-myc antibody (1:5000; Sigma-Aldrich). Equal protein loading was verified with the β-actin antibody for total cell lysate and cytosolic fractions (1:5000; Sigma-Aldrich) and with the integrin β1 antibody (1:333; R&D Systems) for membrane fractions. Chemiluminescence detection was carried out using the ECL Western Blotting Substrate (Pierce) according to the manufacturer's instructions after incubation of the blots with anti-mouse IgG-HRP (1:5000; Amersham Biosciences).
Statistical data analysis
All experimental data related to Wnt signaling were obtained from three independent experiments, each carried out in triplicate, and results are expressed as means ± SD. Comparisons were performed using a Student's t-test. Values of p < 0.05 were considered statistically significant.
Mutation analysis of the LRP5 gene
Direct sequencing of exons 2–4 of the LRP5 gene revealed the presence of a heterozygous missense mutation in exon 4 resulting in the substitution of the A at nucleotide position 844 by a G (844A→G; Fig. 2A). This sequence variant is absent in a panel of 100 control DNA samples. It results in the alteration of a conserved amino acid at position 282 of LRP5 (M282V; Fig. 2B) and is located in the first β-propeller domain of the protein. A 3D structure mapping study of this domain using the Protein Explorer graphics visualization tool (http://proteinexplorer.org)(24) showed that the M282V mutation affects an amino acid residue located near the top surface and central region of the β-propeller structure. Interestingly, the high BMD LRP5 mutations studied by Ai et al.(19) are also clustered in this area.
Functional evaluation of the M282V missense mutation
The high BMD LRP5 mutants studied by Ai et al.(19) were able to transduce Wnt signals at rates comparable or slightly increased to that of wildtype (WT) LRP5, although the majority reached the cell surface less efficiently.(19) To evaluate the effect of the M282V mutation on protein synthesis, membrane trafficking, and secretion, WT and mutant LRP5 (M282V LRP5) protein lacking the transmembrane and intracellular domains tagged with a c-myc epitope at the carboxy terminus were expressed in HEK293T cells. When co-expressed with mesd-c2, truncated WT LRP5 was efficiently synthesized and secreted into the CM of HEK293T cells, as shown by immunoblotting (Fig. 3A). Although levels of M282V LRP5 in the total cell lysate were similar to WT LRP5, a significantly reduced secretion into the CM was observed (Fig. 3A). When trafficked normally, WT LRP5 protein that is secreted into the culture medium is glycosylated and migrates on SDS-PAGE at a slightly higher molecular weight than LRP5 recovered from the cell lysate.(21) M282V LRP5 secreted into the CM migrated at the same molecular weight as WT LRP5 (Fig. 3A), indicating that this mutant had undergone post-translational modification.
The reduced amount of M282V LRP5 in the CM led us to evaluate whether this mutation affects targeting of LRP5 to the cell membrane and/or cell secretion of LRP5. For this, cytosolic and membrane fractions of HEK293T cells expressing either WT or M282V LRP5 were analyzed by immunoblotting. Our data suggest that M282V LRP5 is efficiently trafficked to the cell membrane because the levels of WT and M282V LRP5 are similar in the membrane fraction (Fig. 3B). Considering the drastically reduced levels of M282V LRP5 in the CM, we believe that LRP5 secretion is significantly affected by the mutation. It should also be noted that in the cytosolic fraction of cells expressing WT LRP5 an additional fragment of lower molecular weight is observed that is not detectable in cells expressing M282V LRP5 (Fig. 3B).
We additionally studied whether the presence of M282V LRP5 decreases the secretion of WT LRP5 from HEK293T cells. Immunoblotting of total cell lysates and CM showed that co-transfecting equal amounts of truncated WT LRP5 tagged with a c-myc epitope at the carboxy terminus and full-length untagged M282V LRP5 does not result in a reduced amount of WT LRP5 in the CM compared with co-transfection with full-length untagged WT LRP5 (Fig. 3C).
All transfection efficiencies were similar as measured by β-galactosidase activity (data not shown), and equal protein loading was verified by immunoblotting of total cell lysates and cytosolic fractions using β-actin and of membrane fractions using anti-integrin β1.
Next, we assessed the effect of the M282V mutation on Wnt activation by LRP5 in the Saos-2 human osteosarcoma cell line by transiently transfecting pGL3-OT and pRL-TK luciferase reporter constructs along with untagged, full-length WT or M282V LRP5 and mesd-c2. In the absence of mWnt1-V5, neither WT nor M282V LRP5 were constitutively active (Fig. 4). Co-expression of mWnt1-V5 and WT LRP5 caused a >10-fold increase in luciferase activity. Potentiation of Wnt signal activation in cells expressing M282V LRP5 was not significantly different from cells expressing WT LRP5 in the presence of mWnt1-V5 (Fig. 4; p = 0.12). Thus, although M282V LRP5 protein secretion seems to be reduced, mWnt1 signal transduction levels were not negatively influenced by the mutation.
All high BMD LRP5 mutations studied by Ai et al.(19) resulted in a decreased inhibition of LRP5-mediated Wnt signal transduction by DKK1, most likely as a result of reduced affinity of DKK1 to LRP5. To evaluate whether the M282V mutation acts through a similar mechanism, we set up a Wnt reporter assay in Saos-2 cells in which we co-transfected DKK1 together with mWnt1 and WT or M282V LRP5. As can be seen in Fig. 4, WT LRP5-mediated Wnt signaling was significantly reduced by DKK1 (p < 0.001). Wnt signaling mediated by M282V LRP5 was significantly inhibited only at the highest concentration of DKK1 (p < 0.001) Whereas the induction of luciferase activity by WT LRP5 was inhibited >46% and 65% by DKK1 at low and high concentration, respectively, a low concentration of DKK1 did not affect M282V LRP5-mediated Wnt signal transduction, whereas a higher DKK1 concentration only caused a 30% inhibition. Taken together, these data show that the M282V mutation reduces DKK1-mediated antagonism of Wnt signaling.
This study reports on a female patient with a high bone mass phenotype caused by a heterozygous missense mutation (844A→G) in the LRP5 gene that results in the substitution of a highly conserved amino acid (M282V). This sequence variant can be added to the list of gain-of-function mutations in LRP5 resulting in high BMD abnormalities, increasing the total number to 10 heterozygous missense mutations.(7–13) Patients harboring high BMD LRP5 variants share a similar skeletal picture with a predominant involvement of the skull and cortices of the long bones. Clinical complications associated with these mutations are variable. Frequently, torus palatinus, a wide and deep mandible, and neurological complications occur.(7,10–13) In addition to the bone phenotype, these features might be considered strong predictors of activating LRP5 mutations. Conversely, craniosynostosis, developmental delay, and exostoses of the oropharynx were only described in single studies.(8,10) Our patient presented with extremely high BMD Z-scores at the LS and left hip, and radiographic examination revealed sclerosis at all skeletal sites studied, including skull, hands, vertebrae, pelvis, and tubular bones. Clinical complications caused by cranial nerve compression, as well as an enlarged mandible, were not present. Detailed examination did, however, reveal a palpable torus palatinus, of which the patient was not aware of. Recently, she developed a breast cancer. An interesting finding was the significant bone loss (4.7% at the hip and 5.1% at the LS) through passage to menopause. In Belgian women, menopause takes place around the age of 52 yr, and an average bone loss of 5% over a period of 2 yr is observed. A study of transgenic mice overexpressing human G171V LRP5, a model that mimics the high bone mass phenotype, also revealed an age-related decline in a number of key bone parameters, including the bone volume fraction, trabecular thickness, and connectivity density, of distal femur in women. Although these parameters were significantly higher compared with age-matched control littermates, a consistent decrease was observed from the age of 17 wk onward.(25) The decrease in BMD observed in our patient and in several bone parameters in female G171V transgenic mice suggest that high BMD–associated LRP5 mutations do not protect against age-related bone loss.
The high BMD LRP5 mutations are all located in the first YWTD β-propeller domain of LRP5. Moreover, a homology model for this β-propeller domain revealed that the M282V variant is in close proximity with the other high BMD mutations, more precisely at one location on the protein surface, suggesting a common molecular mechanism.(19) It is difficult to predict the functional significance of this mutation, whereas little is known about the structure/function relationships of LRP5. However, by analogy with the LDL-R, it is possible that the surface region of the YWTD-EGF domain is involved in protein–protein interactions, and that mutations in amino acid residues at this site disrupt functionally important protein binding sites.(26)
DKK proteins are potent secreted modulators of Wnt signaling that bind directly to the extracellular portion of LRP5/6.(16,17) Although DKK binding to LRP5 requires the third β-propeller domain of LRP5, other binding sites, possibly located within the first and/or second YWTD repeat domains, are likely to be involved.(16,17) Previous studies suggest that a reduced affinity to DKK1 and a consequent decreased inhibition of Wnt signal transduction is a common mechanism by which the high BMD LRP5 mutations affect canonical Wnt signaling.(17,19) In this study, we were able to show that the novel M282V LRP5 mutation found in our patient similarly reduces DKK1-mediated inhibition of Wnt signaling. Studies that unravel the exact roles of the different DKKs in the regulation of osteogenesis are sparse. However, recent findings in Dkk1 and Dkk2 mouse models and in vitro studies emphasize the importance of these proteins in bone formation.(27–30)
The first and second YWTD β-propeller domains are also involved in binding MESD, a specialized chaperone protein for members of the LDL-R family that ensures proper protein folding and facilitates the transit through the secretory pathway to reach the plasma membrane.(21) We were able to show that a truncated form of M282V mutant LRP5 was less capable of being secreted into the CM of HEK293T cells. Subsequent studies revealed the presence of M282V LRP5 in the membrane fractions of HEK293T cells, suggesting efficacious protein trafficking to the cell membrane. Our data therefore suggest that the secretion of mutant LRP5 from the cells is drastically impaired, resulting in reduced amount of mutant LRP5 presented in the CM. We were additionally able to show that co-expression of M282V LRP5 does not seem to affect the secretion of WT LRP5. It is also interesting to note that an additional low molecular weight LRP5 fragment was observed in the cytosolic fraction of HEK293T cells that was not detectable in cells expressing M282V LRP5. Similarly, Hsieh et al.(21) reported two fragments of different molecular weight for LRP5/6 in the presence of mesd, with the low molecular weight protein representing LRP5/6 molecules retained in the endoplasmic reticulum (ER), whereas LRP5/6 shifts to a high molecular weight form when it is transited through the secretory pathway. Our data therefore suggest that, in case of mutation M282V, the fraction of LRP5 protein in the ER is significantly reduced. Whether the M282V mutation affects MESD binding needs further study. Despite the reduced amount of M282V LRP5 being secreted, Wnt signal transduction levels induced by this mutant were comparable with the levels observed for WT LRP5. It has already been shown that a small amount of LRP5 at the cell surface is sufficient to transduce canonical Wnt signals, suggesting that other Wnt signaling participants are rate limiting.(19)
We showed in Saos-2 human osteosarcoma cells that M282V LRP5 does not result in constitutive activation of Wnt signaling in the absence of exogenously added Wnt ligand. When mWnt1 was co-expressed, this LRP5 mutant was able to transduce Wnt signals at levels comparable with WT LRP5. More interestingly, our data strongly indicate that the presence of the M282V mutation results in a decreased inhibition of canonical Wnt signaling by DKK1. In line with previous results, a reduced DKK1-mediated inhibition of canonical Wnt signaling can be put forward as a common mechanism resulting in LRP5-associated high BMD conditions. Disruption of DKK1 inhibition most likely increases the ability of Wnt proteins to stabilize β-catenin, consequently leading to stimulation of osteogenesis. However, at present, we can not exclude that other molecular mechanisms additionally contribute to the pathophysiology of disease. It has been documented that YWTD β-propeller domains 1 and 2 of LRP5/6 are also involved in binding Wnt ligands and sclerostin, an antagonist of canonical Wnt signaling.(16,31,32) Human genetic studies revealed an important function of sclerostin as a negative regulator of bone formation as loss-of-function results in sclerosteosis and van Buchem disease, two similar bone conditions characterized by a generalized hyperostosis mainly involving the skull and mandible.(33–36) This protein is secreted from osteocytes after they come embedded in a mineralized matrix, thereby arresting further osteoblastic bone formation.(37) Recent findings strongly indicate that sclerostin antagonism is affected by the presence of HBM LRP5 mutations and is most likely caused by impaired binding between both proteins.(38,39) Further studies are needed to evaluate whether the M282V LRP5 mutation similarly affects sclerostin antagonism of canonical Wnt signaling.
To conclude, we identified a novel heterozygous LRP5 missense variant that can be added to the list of gain-of-function mutations, all located in the first β-propeller domain of LRP5, resulting in conditions with a high bone mass phenotype. Our findings additionally provide further evidence that a reduced inhibition of canonical Wnt signaling by DKK1 contributes to the molecular mechanism by which these LRP5 mutants exert their effect.
We thank Dr Frédéric Lecouvet and Dr Filip Vanhoenacker, radiologists at Université Catholique de Louvain (Brussels, Belgium) and the University Hospital of Antwerp (Antwerp, Belgium), respectively, for critical evaluation of the patient's X-rays and MRI scans. WB holds a postdoctoral fellowship obtained from the Flemish Fund for Scientific Research (F.W.O.). This work was supported by the F.W.O. (Grant G.0117.06) and the EU FP6 project ANABONOS (LSHM-CT-2003–503020) to WVH, and by the Special Research Fund (B.O.F.) of the University of Antwerp to WB.