LRP5 has an essential role in the Wnt signaling pathway, since it acts as a co-receptor that binds Wnt proteins with Frizzled-receptors [5,6]. Mutations within the gene are known to lead to various bone disorders: gain-of-function mutations in the LRP5 gene can cause high-bone-mass (HBM) phenotypes in humans [7,8], whereas homozygous loss-of-function mutations cause osteoporosis-pseudoglioma syndrome (OPPG) characterized by early-onset osteoporosis and complications in eye development [9-11]. Similarly, transgenic mice with interrupted Lrp5 express a low bone mass phenotype, independent of Cbfa-1, including decreased osteoblast proliferation, osteopenia and persistent embryonic eye vascularization [12]. Furthermore, associations have also been reported between the LRP5 gene polymorphisms and bone mass and size [13-15].

LRP5 is widely expressed in most human tissues, with greater amounts in the liver and pancreas [16]. In bone, it is mainly expressed by the bone-forming cells, i.e. osteoblasts, in the endosteal and trabecular bone surfaces [7,9]. It is not known to be expressed by osteoclasts [9]. Recently, Lrp5 expressed in the murine duodenum was shown to affect the synthesis of gut-derived serotonin (5-hydroxytryptamine, i.e. 5-HT) by inhibiting expression of the serotonin rate-limiting enzyme tryptophan hydroxylase 1 (Tph1) [17]. Serotonin then affects bone formation, its effect being mediated by specific 5-HT transporters in the circulation and by binding to the 5-HT receptor 1 B (5-Htr1b) on osteoblasts [17,18]. However, other investigators have not observed a role for gut-expressed Lrp5 in regulating serotonin production or bone mass [19].

Patient M11 (Figure ​(Figure1A1A III:1) had recurrent fractures during childhood and adolescence and repeatedly subnormal BMD (the Z-score at lumbar spine at the time of diagnosis was -3.20), measured by dual-energy X-ray absorptiometry (DXA), and was diagnosed as having primary osteoporosis. M11 is currently 32 years and the most recent BMD measurements at the lumbar spine are normal (T-score 0.0) but osteopenic in the femoral neck (T-score -1.5). The patient has not received bisphosphonate treatment.

The mother (Figure ​(Figure1A1A II:2) and sister (Figure ​(Figure1A1A III:2) of patient M11 had the same mutation as the patient (c.3446 T > A; L1149Q), whereas the father (Figure ​(Figure1A1A II:1) did not carry the mutation. In addition, the DNA samples of maternal grandparents (Figure ​(Figure1A1A I:1 and I:2) and maternal aunt and uncle (Figure ​(Figure1A1A II:3 and II:4) were analyzed, but none was found to have the mutation. Since the mother and sister have been diagnosed as having osteoporosis whereas the father shows no signs of reduced BMD, the genetic findings are concordant with the clinical status of the patient and his family members. The mother's lumbar spine and femoral neck BMD T-scores at age 48 years, after some years of bisphosphonate treatment, were -2.3 and -1.3. The sister's lumbar spine and femoral neck BMD T-scores at 25 years were -1.4 and -2.5.

For the rest of the relatives, the grandmother (Figure ​(Figure1A1A I:1) had reduced BMD at age 83 years with lumbar spine and hip T-scores of -2.0 and -2.2, respectively. The grandfather's (Figure ​(Figure1A1A I:2) BMD measurements at 83 years were somewhat reduced (BMD T-scores of -0.9 and -2.7 for lumbar spine and hip). The aunt (Figure ​(Figure1A1A II:3) had osteopenia at 43 years (BMD T-scores 0.7 and -1.5 for lumbar spine and hip). No clinical data were available for the uncle of M11 (Figure ​(Figure1A1A II:4). The results suggest that M11 inherited a de novo mutation that had arisen in his mother.

Patient M13 (Figure ​(Figure1B1B II:1) was first presented at 13 years of age with multiple fractures in his shoulder, forearms and upper arms. The first fracture occurred at 7 years, and he has sustained altogether 8 fractures by age 14.5 years. When examined at the age of 13, the patient's alkaline phosphatase, parathyroid hormone, 25-hydroxyvitamin D₃, calcium, phosphate and creatinine levels were normal and renal function was normal. Radiographs of the spine revealed reduced mineralization but no vertebral compression fractures or structural abnormalities. Peripheral quantitative computed tomography (pQCT; Stratec QCT 900) of the distal radius showed reduced total and trabecular volumetric BMD Z-scores of -1.5 and -2.8. The lumbar spine BMD Z-score, measured by DXA, was -2.4. The diagnosis of osteoporosis was confirmed by bone biopsy, which showed reduced bone volume and loss of trabecular connectivity, consistent with osteoporosis.

The father (Figure ​(Figure1B1B I:2) of patient M13 (Figure ​(Figure1B1B II:1) shared the mutation (c.3553 G > A; G1185R) with his son, but the change was not detected in the patient's unaffected mother (Figure ​(Figure1B1B I:1) or sister (Figure ​(Figure1B1B II:2). The father has slightly reduced BMD on DXA (T-score at lumbar spine -0.3 and at femoral neck -1.7). Otherwise the family has a negative history of osteoporosis.

Neither one of the novel mutations (L1149Q, G1185R) was detected in controls. The evolutionary importance of the observed disease-causing missense mutations was assessed by aligning the human protein sequences with the corresponding sequences of other species. Both the newly found mutations encode for conserved amino acids (Additional file 3: Figure S1).

All LRP5 mutations associated with primary osteoporosis in our patient set (A29T, C913fs, R1036Q, L1149Q, G1185R) are located in the coding regions of the LRP5 gene. The L1149 and G1185 amino acids are conserved between species (Additional file 3: Figure S1), and are thereby likely to have structural and/or functional importance. Also, the site R1036 is quite well preserved as only one species out of eight differs from the human sequence at this position. The novel mutations (L1149Q and G1185R), as well as two mutations we have identified earlier (C913fs and R1036Q [4]), are located on the fourth propeller domain of LRP5 protein. Only one of the disease associated mutations (A29T) is situated in the first propeller domain of LRP5 [4]. However, primary osteoporosis and osteopenia have been confirmed in heterozygous carriers of OPPG-causing mutations located in other domains and on splice sites of LRP5 [10,25]. One of our patients (M13) with G1185R on the fourth propeller domain presented a graver phenotype than did his father who also had the mutation. The father had reduced BMD, but no osteoporosis. This finding is congruent with our previous and yet unpublished results ([4], Korvala et al. unpublished data) showing that phenotypes of affected offspring tend to be more severe than those of their parents. Reasons accounting for this may be variations in mutation penetrance or presence of other predisposing genetic factors [26,27] or the disorder may be multigenic in nature [28].

There is an interesting connection between the location of LRP5 mutations and resulting disorders (and presumably the disease causing mechanisms). HBM mutations are located in the first propeller domain of LRP5 whereas OPPG causing mutations are scattered mainly in the second and third propeller domains. Furthermore, different LRP5 domains bind to certain ligands in the Wnt signaling pathway: the first and second propeller domains of LRP5/6 participate in binding a certain class of the inducing ligands of the pathway e.g. Wnt1 and Wnt9b [29-31], but also the Wnt signaling inhibitors Wise, Sclerostin (SOST) and Dkk1 [31-33]. At the same time, the third and fourth propeller domains bind DKK1 [34,35] and another class of Wnt proteins e.g. Wnt3a (in LRP6) [30,31] while the cytoplasmic domain binds Axis inhibitor-1 (Axin) [36]. Moreover, the latest studies have indicated that different Wnts are able to bind to specific LRP6 propellers simultaneously [31], and compete with DKK1 binding [31,37]. These results may potentially be implicated also in LRP5. In conclusion, the site of mutation may be an important indicator for the resulting disorder, when assuming that the mutation affects the interaction between LRP5 and the ligand binding the mutation site.

Our in vitro studies with four LRP5 mutations causing primary osteoporosis showed that all the LRP5 constructs were able to mediate signaling and that the signaling activity was enhanced several-fold in all the constructs when Wnt3a was added (Figure ​(Figure3),3), supporting the role of Wnt3a as a ligand for LRP5. Wnt3a also enabled us to elucidate the differences in signaling response between the LRP5 constructs: mutants C913fs and L1149Q reduced activity significantly (by 47% and 29%, respectively) as compared with WT-LRP5, and activity was also reduced by R1036Q (by 24%), whereas G1185R had no effect on signaling activity. Although no clinical significance has been reported for R1036Q (GenBank rs61889560), it has been detected in four OPPG patients [38] since its identification in a patient with primary osteoporosis [4], supporting its role in bone development or maintenance. Further functional studies are necessary for G1185R as no effect was detected using the current methods and a different experimental approach may identify the underlying mechanism.

The fact that the signaling activity of HBM mutation G171V in our study was close to that of WT-LRP5 is consistent with earlier findings that HBM-LRP5's are not constitutively active but need a Wnt ligand to be activated [39-41]. In vitro LRP5 studies by others have focused on HBM mutations and the few studies addressing the impact of mutations causing osteoporosis have mainly been associated with OPPG. These have shown that mutations causing OPPG reduce Wnt and/or Norrin signaling [11,38,42], while some mutants are trafficked unequally to the cell membrane [23]. Crabbe et al. [43] concluded that mutations associated with idiopathic osteoporosis in adult men may change the expression of LRP5 protein and/or interfere with the interaction of LRP5 with Mesd or with the Wnt/Fzd complex. Saarinen et al. [38] found an association between three homozygous OPPG mutations (R570W, R925C, R1036Q) and glucose tolerance, and suggested a potential association with diabetes. Taken together, our findings are in line with the results of other in vitro LRP5/OPPG studies showing that mutations associated with low bone mass disorders reduce the ability of LRP5 to mediate Wnt-induced signaling and consequently result in a low bone mass phenotype.

Since Yadav et al. [17] have shown that Lrp5 produced in the intestine can inhibit Tph1 expression, and consequently also 5-HT synthesis and bone formation, we examined whether the LRP5 mutations causing primary osteoporosis influence Tph1 and/or 5-Htr1b expression in an in vitro system. Our results showed that only one of the mutations (L1149Q) reduced 5-Htr1b expression significantly in the presence of Wnt3a (p < 0.002; Figure ​Figure4B),4B), but neither HBM nor primary osteoporosis LRP5 mutations influenced Tph1 expression (Figure ​(Figure4A).4A). We cannot readily compare our in vitro results to the in vivo studies of Yadav et al. [17], and although the 5-Htr1b finding is of potential interest, it is still tentative and further investigation using alternative methods is needed to examine its biological significance. One restriction of the current study is the use of only one reference gene, β-actin, which has been commonly used as a reference gene in human and murine gynecological tissue studies [44-47] and has shown stable expression in human endometrium [48].