Osteocyte Protein Expression Is Altered in Low-Turnover Osteoporosis Caused by Mutations in WNT1 and PLS3

Abstract

Context:

Osteocytes express proteins that regulate bone remodeling and mineralization.

Objective:

To evaluate the relationship between osteocyte-specific protein expression and bone histology in patients with monogenic osteoporosis due to wingless integration site 1 (WNT1) or plastin 3 (PLS3) mutations.

Design and Setting:

Cross-sectional cohort study at a university hospital.

Participants:

Six patients (four males; ages: 14 to 72 years) with a heterozygous WNT1 mutation and five patients (four males; ages: 9 to 70 years) with a heterozygous/hemizygous PLS3 mutation.

Methods and Main Outcome Measures:

Immunohistochemistry was performed for fibroblast growth factor 23 (FGF23), dentin matrix protein 1 (DMP1), sclerostin, and phosphorylated (phospho–)β-catenin in iliac crest samples and compared with bone histomorphometry.

Results:

FGF23 expression in WNT1 patients was 243% that observed in PLS3 patients (P < 0.01). DMP1, sclerostin, and phospho–β-catenin expression did not differ between groups. Serum phosphate correlated inversely with FGF23 expression (r = −0.79, P = 0.01) and serum ionized calcium correlated inversely with sclerostin expression (r = −0.60, P = 0.05). Phospho–β-catenin expression correlated inversely with DMP1 expression (r = −0.88, P < 0.001), osteoid volume/bone volume (r = −0.68, P = 0.02), and bone formation rate (r = −0.78, P < 0.01). FGF23 expression did not correlate with DMP1 expression, sclerostin expression, or bone histomorphometry. Marrow adiposity was higher in WNT1 than in PLS3 patients (P = 0.04).

Conclusions:

Mutations that disrupt WNT signaling and osteocytic mechanosensing affect osteocyte protein expression. Abnormal osteocyte function may play a role in the pathogenesis of monogenetic forms of osteoporosis.

We studied bone from six WNT1 and five PLS patients. Bone FGF23 was higher in WNT1 than in PLS3 patients. Phospho–β-catenin correlated inversely with DMP1 and with bone turnover.

Bone health and strength are maintained by the coupled actions of bone-forming osteoblasts and bone-resorbing osteoclasts on the surface of trabecular bone, and there is increasing recognition that both bone formation and bone resorption are controlled by osteocytes buried within mineralized bone (1). Investigations into rare monogenic forms of osteoporosis have paved the way for better understanding of genetic determinants and molecular mechanisms affecting bone health and have provided new therapeutic opportunities for treating skeletal disorders (2). We have identified two large Finnish families with monogenic forms of early-onset osteoporosis, an autosomal dominant form caused by a missense mutation c.652T>G (p.C218G) in wingless integration site 1 (WNT1) (3) and an X-linked form resulting from a splice mutation c.73-24T>A in plastin 3 (PLS3) (4). Although the exact mechanisms whereby WNT1 and PLS3 mutations affect bone health are incompletely understood, WNT1 mutations result in impaired WNT/β-catenin signaling (5), whereas PLS3 appears to play a role in osteocyte dendrite function and skeletal mechanosensing (6). Affected members of each family presented with early-onset progressive osteoporosis defined by low bone mineral density (BMD), peripheral fractures, and multiple compression fractures, particularly in the thoracic spine. Although bone turnover markers in peripheral blood and urine were normal, iliac crest bone biopsies showed low-turnover osteoporosis in both pedigrees (3, 4).

Patients suffering from osteoporosis have increased numbers of adipocytes in their bone marrow (7, 8), and in vitro studies suggest that a reciprocal relationship exists between osteogenic and adipogenic pathways (8). Osteocytes are key regulators of bone modeling and remodeling, generating proteins that regulate osteoblast differentiation and skeletal mineralization (1); these proteins may likewise regulate marrow adipogenesis. In monogenetic diseases, disrupted expression and defective posttranslational modification of several osteocyte-specific proteins, including fibroblast growth factor 23 (FGF23) (9), dentin matrix protein 1 (DMP1) (10), and sclerostin, result in impaired bone mineralization and turnover and lead to various skeletal phenotypes ranging from rickets/osteomalacia to osteoporosis. Activation of the WNT/β-catenin pathway promotes osteogenic differentiation while inhibiting adipogenic differentiation of mesenchymal stem cells (11) and sclerostin, an osteocytic protein encoded by SOST and regulated by circulating levels of parathyroid hormone (PTH), antagonizes the canonical WNT/β-catenin pathway (12). FGF23, an osteocyte-specific protein that is regulated by circulating phosphate, PTH, and 1,25-dihyroxyvitamin D (13), acts as an endocrine factor by controlling renal phosphate reabsorption and renal 1,25-dihyroxyvitamin D synthesis (13, 14). The upstream regulator of FGF23, DMP1, is also expressed in osteocytes and regulates osteoblast/osteocyte maturation as well as skeletal mineralization (15); whether FGF23 and DMP1 affect adipogenesis is currently unknown.

Osteocyte-specific proteins are undeniably important in controlling adipogenic and osteogenic differentiation and in determining overall bone health (7, 8). However, the mechanism by which the signaling pathways of these hormones intersect and the roles that they play in human skeletal disease remain incompletely defined. Although PTH has been shown to increase WNT signaling by suppressing skeletal expression of sclerostin (16), we have previously reported that exogenous administration of PTH in the form of teriparatide paradoxically increases marrow adipocyte numbers in patients with both WNT1 and PLS3 mutations (17). This led us to hypothesize that disruptions in WNT1 and PLS3 might alter osteocyte-specific protein expression and disturb the balance between adipogenic and osteogenic differentiation. We thus used immunohistochemical and terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining to evaluate the role of WNT1 and PLS3 on osteocyte-specific expression of FGF23, DMP1, sclerostin, and phosphorylated (phospho–)β-catenin and on bone cell apoptosis. We also evaluated the relationship of this staining to histomorphometric parameters of trabecular bone volume, mineralization, turnover, and marrow adiposity in two large families with monogenic forms of osteoporosis.

Materials and Methods

Patients

We have previously described two Finnish families with monogenic osteoporosis; one caused by a missense mutation c.652T>G (p.C218G) in WNT1 (3) (hereafter referred to as WNT1 patients) and another by a splice mutation c.73-24T>A (p.Asp25Alafs*17) in PLS3 (4) (hereafter referred to as PLS3 patients) (Supplemental Fig. 1 ^{(166.8KB, tif)} ; Supplemental Table 1 ^{(32.4KB, docx)} ). Affected family members with a genetically verified mutation in WNT1 or PLS3 (3, 4) were invited to participate in a study characterizing skeletal and extraskeletal features of monogenic osteoporosis. Of the mutation-positive individuals in both pedigrees, 11 patients (six with WNT1 mutation and five with PLS3 mutation) with low BMD and/or spinal compression fractures underwent transiliac bone biopsy and thus comprised the study cohort for the present evaluation. All 11 patients were clinically examined at the Helsinki University Hospital. Five of the 11 had previously received bisphosphonate therapy. Serum biochemical values for ionized calcium, phosphate, and PTH were obtained at the time of bone biopsy, as previously described (3, 4) (Supplemental Table 2 ^{(31.2KB, docx)} ). The study was approved by the Research Ethics Committee of the Helsinki University Central Hospital and performed according to the ethical principles defined in the Declaration of Helsinki. All patients/guardians gave written informed consent prior to participation.

Histomorphometric analysis of iliac crest biopsies

Transiliac bone specimens were obtained from the anterior iliac crest using a manual drill with a trephine of 7.5 mm inner diameter (Rochester Bone Biopsy, Medical Innovations International) after double labeling with oral tetracycline (18). To ensure adequate tetracycline absorption, patients refrained from taking dairy products or supplemental calcium during tetracycline labeling periods. All biopsies were taken between 2006 and 2011 from the left iliac crest by an experienced orthopedic surgeon (V.-V.V. or T.L.) at the Helsinki University Hospital.

The specimens were placed in 70% ethanol, then dehydrated and embedded in polymethylmethacrylate. For each individual patient, static histomorphometric parameters were evaluated in one entire undecalcified 5-µm section treated with toluidine blue; tetracycline labeling was assessed over the entire area of one unstained 10-µm section. Primary bone histomorphometric parameters were assessed in trabecular bone under ×200 magnification using the OsteoMeasure system (OsteoMetrics). Mineralized bone was defined by dark blue staining areas; pale-blue seams at least 1.5 µm in width were included in measurements of osteoid. Derived indices were calculated according to standard formulae (19). Nomenclature and abbreviations follow the recommendations of the American Society for Bone and Mineral Research. All histomorphometric analyses were performed by R.C.P. at the David Geffen School of Medicine at the University of California, Los Angeles (UCLA). Values were reported as z-scores relative to age-matched norms (20–22). Marrow adiposity was evaluated in an entire 5-µm section of bone with intact marrow and cancellous compartments. Sections were analyzed under ×100 magnification. The number of adipocytes was measured in all fields of each sample and the total number was divided by tissue area, including bone marrow and trabecular bone.

Immunohistochemistry of bone protein expression

The technique for immunohistochemical detection of protein in bone was adapted from a previously reported method (23). In brief, 5-μm sections of bone tissue were deplastified in xylene and chloroform, rehydrated in graded alcohol solutions, and partially decalcified in 1% acetic acid. Endogenous peroxidase activity was quenched in 3% hydrogen peroxide/methanol solution. Nonspecific binding was blocked in avidin-biotin solution and in 5% normal horse serum with 1% bovine serum albumin. Sections were incubated with affinity-purified polyclonal goat anti-human FGF23 (residues 225–244; Immutopics; dilution 1:500), monoclonal antihuman DMP1 (LFMb31, 62–513; Santa Cruz Biotechnology; dilution 1:50), monoclonal antihuman sclerostin (MAB1406; R&D Systems; dilution 1:500), or polyclonal antihuman phospho–β-catenin (ab47385; Abcam; dilution 1:500) primary antibody overnight at 4°C in a humidified chamber. Sections were then incubated with biotinylated anti-goat (Vector Laboratories), anti-rabbit (Bio-Rad Laboratories), or anti-mouse (Sigma-Aldrich) antibody, incubated for 30 minutes with a StreptABC complex/horseradish peroxidase kit (Vector Laboratories) followed by AEC substrate chromogen (Dako), and counterstained with Mayer’s hematoxylin (Sigma-Aldrich). Negative controls were performed for each bone section by omitting the primary antibody. Reproducibility was ensured by repeating the immunohistochemistical analysis on all specimens. FGF23 was assessed in the entire area of trabecular bone (which excluded the marrow space), and DMP1 was assessed in the entire trabecular tissue area (which included the bone marrow space). Some individual samples had some sclerostin and phospho–β-catenin expression within the trabeculae; however, these proteins were expressed primarily in the cortex, and thus their measurements were assessed exclusively in cortical bone. Immunoreactivity for FGF23 was quantified by counting the number of osteocytes expressing FGF23 in one entire 5-µm section of trabecular bone and normalizing this number by trabecular bone area. Immunoreactivity for DMP1, sclerostin, and phospho–β-catenin was quantified in one entire 5-µm section per protein using the Ariol SL-50 scanning system (24). Staining was normalized by trabecular (for DMP1) tissue area or cortical (for sclerostin and phospho–β-catenin) tissue area. All slides were scanned at ×20 magnification with a red filter and digitized (Applied Imaging). Analyzed fields were manually selected to avoid areas with tissue damage occurring during immunostaining. Staining was expressed as pixels/mm² (24). All analyses were performed with the MultiStain script. All protein expression analyses were performed at the David Geffen School of Medicine at UCLA (R.C.P.).

Measurement of apoptosis in osteocytes and in bone marrow

Bone sections of 5 µm were mounted on poly-l-lysine–coated glass slides (Allegiance Healthcare). Sections were deplastified and then treated with 100 μL of 20 μg/μL proteinase K in 10 mM Tris (pH 8.0) for 20 minutes at room temperature, rinsed with 50 and 10 mM Tris-buffered sodium/potassium chloride, reincubated in 3% H₂O₂ in methanol, and rinsed with 50 mM Tris-buffered saline. DNA fragmentation was detected by the TUNEL reaction using Klenow terminal deoxynucleotidyl transferase per the manufacturer’s instructions (Oncogene Research Products). Sections were incubated in 0.15% CuSO₄ in 0.9% NaCl for 2 minutes and counterstained with 2% methyl green aqueous solution. Sections treated with 1 μg/μL DNase I in 50 mM Tris-buffered saline/1 mM MgSO₄ were used as positive controls. Sections not incubated with the transferase, but with vehicle alone, were used as negative controls. Osteocytes and bone marrow cells in which the nuclei were clearly dark brown, rather than blue-green, were considered apoptotic. TUNEL staining was evaluated in a subjective manner by a trained histomorphometrist (R.C.P.) who was blinded to patient genetic status. Its expression was assessed according to an ordinal scale (0, no expression; 0.5, minimal expression; 1, moderate expression; 2, significant expression). All apoptosis analyses were performed at the David Geffen School of Medicine at UCLA (R.C.P.).

Statistical analysis

Measurements for normally distributed variables are reported as mean ± standard error, and median values and interquartile range are used to describe nonnormally distributed variables. Ranges are presented where noted in the text. The Wilcoxon signed rank test and the χ² test were used to assess intergroup differences. Relationships between biochemical, bone histomorphometric, and immunohistochemical parameters were assessed by Spearman correlation coefficients. All statistical analyses were performed using SAS software (SAS Institute) and all tests were two-sided. A probability of a type I error <5% was considered statistically significant and ordinary P values are reported.

Results

Patients

Altogether 11 patients participated in the study (Supplemental Table 1 ^{(32.4KB, docx)} ; Supplemental Fig. 1 ^{(166.8KB, tif)} ). Six patients (two females and four males) had a heterozygous p.C218G (c.652T>G) mutation in WNT1 (3). Five patients (four males and one female) had a splice mutation in PLS3 (c.73-24T>A), which causes a frameshift and premature termination of messenger RNA translation (p.Asp25Alafs*17) (4). Of the five PLS3 patients, the one female was heterozygous and the four males were hemizygous for the frameshift mutation. No differences in gender or age were observed between the two patient groups. The patients’ ages ranged from 9 to 72 years (median age, 45 years).

Three of the 11 patients were children. All had normal growth and development. The 14-year-old boy with a WNT1 mutation had one asymptomatic spinal compression fracture and no peripheral fractures; his BMD Z-score was −2.0 for lumbar spine and −2.1 for total hip. The 9- and 13-year-old boys with a PLS3 mutation each had two to three previous forearm fractures, multiple thoracic vertebral compression fractures, and low BMD z-scores. The adult WNT1 and PLS3 patients also had multiple low-impact peripheral and vertebral fractures and low BMD values with otherwise normal stature and no extraskeletal features. Five of the eight adults had received osteoporosis treatment prior to the study (Supplemental Table 1 ^{(32.4KB, docx)} ) (3, 4). Biochemical values are displayed in Supplemental Table 2 ^{(31.2KB, docx)} .

Bone histology and histomorphometry

No differences in bone volume, trabecular thickness, or trabecular number were observed between the WNT1 and PLS3 patients. Histomorphometric parameters of osteoid volume, osteoid thickness, osteoid surface, mineral apposition rate, and bone formation rate were low and did not differ based on mutation. However, focal areas of osteoid accumulation were observed in four of five PLS3 (both adult and pediatric) patients but not in any WNT1 patients [Fig. 1(a) and 1(b)]. Marrow adiposity was higher in WNT1 than in PLS3 patients (149 ± 10/mm² vs 114 ± 9/mm², respectively; P = 0.04 between groups) (Supplemental Table 3 ^{(30.9KB, docx)} ). Osteocyte morphology was visibly abnormal in biopsy samples from two WNT1 patients, both of whom were adults (one a 52-year-old female and the other a 62-year-old male) and both of whom had been previously treated with bisphosphonates. In these samples, areas of woven bone were noted in which osteocytes had abnormally rounded shapes with a paucity of dendritic processes. Osteoid accumulation was also noted in the perilacunar area of some of these osteocytes [Fig. 1(c) and 1(d)]. These features were not noted in any biopsy samples from bisphosphonate-treated or untreated PLS3 patients.

Figure 1.

Histologic features of WNT1 and PLS3 patients. (a) Toluidine blue staining demonstrating areas of focal osteoid accumulation in PLS3 patients (arrows) but not in WNT1 patients (b). (c and d) In two WNT1 patients who had previously received bisphosphonate therapy, osteocytes with an abnormally rounded shape were noted in areas of woven bone (boxed region). Normal-appearing osteocytes are indicated by the arrow in image (c). Perilacunar osteoid accumulation was also noted in these regions [arrow, image (d)]. TB, trabecular bone.

Bone immunohistochemistry and TUNEL staining

In all samples, FGF23 was expressed in discrete clusters of osteocytes at the periphery of trabecular bone, and its expression was localized to osteocyte cell bodies [Fig. 2(a)]. DMP1 was expressed in osteocytes throughout trabecular bone, and its expression was observed in cell bodies and in dendritic processes [Fig. 2(b)]. In contrast to the trabecular expression of FGF23 and DMP1, sclerostin and phospho–β-catenin expression was observed primarily in cortical bone [Fig. 2(c) and 2(d)]. Staining for sclerostin could also be observed in deeper trabecular bone in all PLS3 patients, whereas in five of the six WNT1 patients its expression was localized to the portion of trabecular bone immediately adjacent to the cortex.

Figure 2.

Osteocyte-specific bone protein expression in representative WNT1 and PLS3 samples. (a) FGF23 was expressed in the cell bodies of osteocytes, which clustered together at the trabecular periphery. Greater numbers of FGF23-expressing osteocytes were observed in WNT1 than in PLS3 patients. (b) DMP1 was expressed in osteocyte cell bodies and in dendritic processes throughout trabecular bone. (c) Sclerostin and (d) phospho–β-catenin expression was observed primarily in cortical bone. Sclerostin was also observed in trabecular bone in PLS3 patients. Arrows point to areas of positive staining. BM, bone marrow; CB, cortical bone; TB, trabecular bone.

FGF23 expression normalized by trabecular bone area (FGF23/trabecular bone area) in WNT1 patients was 243% that observed in PLS3 patients (P < 0.01 between groups) (Table 1). No statistical difference in DMP1 expression normalized by total trabecular tissue area (including bone marrow and trabecular bone) (DMP1/tissue area) was observed between WNT1 and PLS3 patients (Table 1); however, the two patients with the highest trabecular bone DMP1 expression were both PLS3 patients. Cortical sclerostin expression, normalized by cortical tissue area, did not differ between WNT1 and PLS3 patients (Table 1). Phospho–β-catenin expression tended to be higher in WNT1 as compared with PLS3 patients (P = 0.22 between groups), and phospho–β-catenin was present in five of six WNT1 patients but was completely absent in three of five PLS3 patients (P = 0.14 between groups).

Table 1.

Osteocyte-Specific Protein Expression

Immunohistochemical Parameter	WNT1 Patients (n = 6)	PLS3 Patients (n = 5)
FGF23/trabecular bone area (number/mm2 of trabecular bone area)	3.4 ± 1.4	1.4 ± 0.4*
Sclerostin (pixels/mm2 of cortical tissue area)	0.075 ± 0.064	0.124 ± 0.083
Phospho–β-catenin (pixels/mm2 of cortical tissue area)	0.018 ± 0.011	0.008 ± 0.013
DMP1 (pixels/mm2 of trabecular tissue area)	0.075 ± 0.036	0.184 ± 0.212
Cortical osteocyte apoptosis (subjective scale)	1.0 (0.5, 2.0)	1.0 (1.0, 3.0)
Trabecular osteoblast apoptosis (subjective scale)	0 (0, 0.5)	0 (0, 0)
Trabecular osteoclast apoptosis (subjective scale)	0.5 (0, 1.0)	0.5 (0.5, 0.5)
Bone marrow apoptosis (subjective scale)	0.5 (0, 1.0)	0.5 (0.5, 1)

Bone protein expression and bone apoptosis in six patients with WNT1 mutation c.652T>G (p.C218G) and five patients with PLS3 splice mutation c.73-24T>A. *P < 0.01 between patient groups.

Given the abnormal osteocyte morphology observed on histology in two WNT1 patients, TUNEL staining was performed in all biopsies to evaluate osteocyte apoptosis. Apoptotic osteocytes were observed in cortical osteocytes in five of six WNT1 and in four of five PLS3 patients (Fig. 3); apoptosis did not differ in patients who were naive to bisphosphonates as compared with those who were receiving ongoing or had received prior bisphosphate treatment. Cortical osteocyte apoptosis was very high in two PLS3 patients, both of whom were children. This high level of osteocyte apoptosis was not observed in any WNT1 patients. Bone marrow apoptosis also appeared to be stronger in PLS3 patients; TUNEL staining was present in bone marrow of all but one PLS3 patient but was present in only three of six WNT1 patients. Neither osteocyte nor bone marrow apoptosis was related to patient age.

Figure 3.

Marrow apoptosis is increased in PLS3 patients. (a) TUNEL staining demonstrated increased apopotosis in cells in the bone marrow of PLS3 patients. (b) Osteocyte apoptosis, in contrast, was similar between groups. Arrows point to areas of positive staining.

When both PLS3 and WNT1 patients were considered together as one group, serum phosphate concentrations correlated inversely with bone FGF23 expression (r = −0.79, P = 0.01) whereas serum ionized calcium levels, but not PTH, correlated inversely with cortical sclerostin expression (r = −0.60, P = 0.05). An inverse correlation was observed between phospho–β-catenin expression and trabecular DMP1 expression [Fig. 4(a)] (r = −0.88, P < 0.001). Osteoid volume/bone volume also correlated inversely with phospho−β-catenin expression (r = −0.68, P = 0.02) and directly with trabecular DMP1 expression (r = 0.75, P < 0.01) [Fig. 4(b)] whereas phospho–β-catenin expression correlated inversely with bone formation rate (r = −0.88, P < 0.01) [Fig. 4(c)]. Trabecular FGF23 expression did not correlate with trabecular DMP1 expression, cortical sclerostin expression, or any parameter on bone histomorphometry. There was no correlation between cortical sclerostin and phospho–β-catenin expression or between adipocyte number and sclerostin expression.

Figure 4.

Correlations between bone protein expression and histomorphometric parameters of bone turnover. (a) Scatter plot of cortical bone phospho–β-catenin expression vs trabecular bone DMP1 expression; trabecular DMP1 expression correlates inversely with cortical phospho–β-catenin expression. (b) Scatter plot of trabecular DMP1 expression vs trabecular osteoid volume/bone volume (OV/BV); trabecular osteoid volume correlates inversely with trabecular DMP1 expression. (c) Scatter plot of cortical phospho–β-catenin expression vs trabecular bone formation rate/bone surface (BFR/BS); trabecular bone formation rate correlates inversely with cortical phospho–β-catenin expression in the transiliac bone biopsy samples from six patients with a heterozygous WNT1 mutation c.652T>G (p.C218G) and five patients with a PLS3 mutation c.73-24T>A. The r value reflects the Spearman correlation coefficient. The open circles represent WNT1 patients, and the closed squares represent PLS3 patients.

Discussion

Despite their well-recognized importance to bone health, the exact role of osteocytes in osteogenic and adipogenic differentiation and in pathogenesis of skeletal disease remains incompletely defined. Our study demonstrates alterations in osteocyte protein expression that may mediate the osteoporosis resulting from specific genetic mutations. In short, FGF23 and phospho–β-catenin expression was higher, while DMP1 expression tended to be lower, in WNT1 as compared with PLS3 patients. These differences coincided with higher bone marrow adipocyte numbers in WNT1 patients. Osteocyte apoptosis was more prominent in PLS3, as compared with WNT1, patients. This assessment of bone protein expression in subjects with WNT1 and PLS3 mutations brings novel insights to osteocyte function in these monogenic forms of primary osteoporosis.

Subtle differences in bone histology that were not captured by bone histomorphometric measurements were noted between WNT1 and PLS3 patients. It is interesting that although all patients in the present study were diagnosed with osteoporosis based on bone density and clinical history, histomorphometric parameters of bone volume, trabecular thickness, and trabecular separation were on the lower end of the normal range in both groups of patients. Importantly, however, note that these traditional parameters of bone volume do not reflect trabecular connectivity, an important component of bone strength. Moreover, rounded osteocytes with a paucity of dendritic processes were noted in the biopsies from two WNT1 patients who had previously received therapy with bisphosphonates. These abnormal features were found in areas of woven bone, suggesting that alterations in WNT1 may result in a loss of directional control in bone remodeling at certain sites in bone during treatment with bisphosphonates. Additionally, although overall histomorphometric parameters of osteoid accumulation did not differ between groups, prominent focal areas of osteoid accumulation were noted in PLS3, but not in WNT1, patients. This finding suggests a specific role for PLS3 in the local regulation of bone mineralization. A decreased prominence of apoptosis in both osteocytes and in bone marrow of WNT1 compared with PLS3 patients suggests that differences in programmed cell death may account for local differences in bone remodeling and mineralization in patients with these specific mutations.

A unique spatial expression of osteocyte-specific proteins in bone was confirmed in the present study. As we have previously shown in iliac crest biopsies from children with kidney disease (24), FGF23 is found primarily at the trabecular periphery, in discrete osteocyte clusters, a location suggesting that it is expressed only in relatively young osteocytes. By contrast, sclerostin and phospho–β-catenin are both found primarily in more mature osteocytes in cortical bone (24). The location of DMP1 expression throughout trabecular bone suggests that this protein, in contrast to FGF23 and sclerostin, is expressed at various stages of osteocyte maturation.

We observed differences in expression patterns of osteocyte-specific proteins between WNT1 and PLS3 patients, which may help elucidate the mechanisms through which these mutations lead to skeletal fragility. FGF23 expression was higher in WNT1 than in PLS3 patients. DMP1 has been shown to regulate FGF23 expression through an FGFR1-dependent signaling pathway (15), and increased bone FGF23 expression has been linked to altered DMP1 expression in skeletal mineralization defects in children with chronic kidney disease (24). However, bone FGF23 expression did not correlate with any parameters of skeletal mineralization or with bone DMP1 expression in the present cohort of osteoporotic patients with normal kidney function; thus, increased FGF23 expression may not be due to altered DMP1 expression or function in WNT1 patients. Interestingly, note that PTH has been shown to increase FGF23 expression in osteoblast cell lines by activating the nuclear receptor–associated protein 1 (25); whether decreased canonical WNT signaling in bone may lead to increased FGF23 through altered nuclear receptor–associated protein 1 expression remains unknown and warrants further investigation.

In the present study, phospho–β-catenin, the inactivated product of the downstream effector of WNT signaling, was present in the bone of patients with both mutations, including those with heterozygous WNT1 mutations; this is consistent with the partial loss of function in canonical WNT signaling that has been demonstrated in these patients (3). Phospho–β-catenin expression was absent in three of the five PLS3 patients but absent in only one of the six WNT1 patients. Although we were unable to measure the unphosphorylated, active form of β-catenin in these samples, these findings suggest that disruptions in WNT1 increase deactivation and degradation of the downstream effectors of canonical WNT signaling. The inverse relationship between the phosphorylated, inactive form of β-catenin and bone formation rate is consistent with the concept that active WNT signaling is anabolic to bone (26), whereas the inverse correlation between DMP1 and phospho–β-catenin is consistent with animal data suggesting that DMP1 may regulate osteocyte maturation and mineralization through a canonical WNT signaling pathway (27). Although the correlations between cortical phospho–β-catenin expression and trabecular DMP1 and bone formation rates were unexpected, given the lack of proximity of phospho–β-catenin to trabecular bone, WNT inhibitors and DMP1 are both known to be secreted into circulation, and thus it may be that canonical WNT signaling, DMP1, and bone formation are related by endocrine, rather than by paracrine, mechanisms.

A varied expression pattern of sclerostin was also observed between patient groups. Cortical sclerostin expression was observed both in WNT1 and PLS3 patients; however, trabecular expression of sclerostin was only observed in PLS3 patients. The absence of trabecular staining for sclerostin in all WNT1 patients suggests that impaired canonical WNT signaling may result in altered local expression of WNT inhibitors in these patients. Furthermore, the lack of correlation between cortical sclerostin and phospho–β-catenin expression, as well as the lack of correlation between sclerostin and adipocyte number, suggests a disconnect between expression of WNT antagonists, canonical WNT signaling, and adipogenesis in the bone of these osteoporotic patients.

We acknowledge this study to have certain limitations, mainly concerning the limited cohort size and the lack of healthy control comparators. Specifically, age-matched controls were not available for immunohistochemical analysis. We have previously reported normal values for FGF23 per bone area (1.34 + 0.3 osteocytes/mm²) and bone DMP1 (0.02 ± 0.01 pixels/mm²) expression in adolescents (24); these FGF23 values are similar to those observed in PLS3 patients and lower than our present observations in WNT1 patients. DMP1 values appear to be similar between WNT1 and PLS3 patients and adolescent controls. However, the effect of age on these parameters is unknown; similarly, control values for sclerostin, phospho–β-catenin, and apoptosis are not available. Additionally, we exclusively analyzed bone from the iliac crest; because bone loading plays an important role in WNT signaling (26), osteocyte-specific protein expression could differ in other parts of the skeleton. However, considering the rarity of these forms of monogenic forms of osteoporosis, the lack of knowledge in this specific subject, and the invasive nature of obtaining a bone biopsy, we think that our unique data offer valuable insights into human bone biology. Future studies are needed to determine how these findings differ in other forms of osteoporosis, such as juvenile osteoporosis, senile osteoporosis, and steroid-induced osteoporosis, all of which have different underlying pathogeneses. Future studies are also needed to determine what effects, if any, bisphosphonate therapy has on osteocyte-specific protein expression in these other osteoporotic conditions.

In conclusion, FGF23 expression is increased in patients with WNT1 mutations as compared with patients with PLS3 mutations, suggesting that FGF23 may be regulated by canonical WNT signaling in bone. In contrast to bone disease in children with chronic kidney disease (24), bone FGF23 expression in patients with osteoporosis due to WNT1 and PLS3 mutations did not correlate with either DMP1 expression or with indices of skeletal mineralization. By contrast, DMP1 and phospho–β-catenin expression correlated strongly with histomorphometric parameters of bone turnover, supporting previous animal data suggesting that DMP1 regulates bone cell maturation through a WNT-dependent mechanism (27). Immunohistochemical evaluation of bone protein expression may, in the future, improve diagnosis and treatment of osteoporotic conditions.