The osteocyte product sclerostin is emerging as an important paracrine regulator of bone mass. It has recently been shown that osteocyte production of receptor activator of NF-κB ligand (RANKL) is important in osteoclastic bone resorption, and we reported that exogenous treatment of osteocytes with sclerostin can increase RANKL-mediated osteoclast activity. There is good evidence that osteocytes can themselves liberate mineral from bone in a process known as osteocytic osteolysis. In the current study, we investigated sclerostin-stimulated mineral dissolution by human primary osteocyte-like cells (hOCy) and mouse MLO-Y4 cells. We found that sclerostin upregulated osteocyte expression of carbonic anhydrase 2 (CA2/Car2), cathepsin K (CTSK/Ctsk), and tartrate-resistant acid phosphatase (ACP5/Acp5). Because acidification of the extracellular matrix is a critical step in the release of mineral from bone, we further examined the regulation by sclerostin of CA2. Sclerostin stimulated CA2 mRNA and protein expression in hOCy and in MLO-Y4 cells. Sclerostin induced a decrease in intracellular pH (pHi) in both cell types as well as a decrease in extracellular pH (pHo) and the release of calcium ions from mineralized substrate. These effects were reversed in the co-presence of the carbonic anhydrase inhibitor, acetozolamide. Car2-siRNA knockdown in MLO-Y4 cells significantly inhibited the ability of sclerostin to both reduce the pHo and release calcium from a mineralized substrate. Knockdown in MLO-Y4 cells of each of the putative sclerostin receptors, Lrp4, Lrp5 and Lrp6, using siRNA, inhibited the sclerostin induction of Car2, Catk and Acp5 mRNA, as well as pHo and calcium release. Consistent with this activity of sclerostin resulting in osteocytic osteolysis, human trabecular bone samples treated ex vivo with recombinant human sclerostin for 7 days exhibited an increased osteocyte lacunar area, an effect that was reversed by the co-addition of acetozolamide. These findings suggest a new role for sclerostin in the regulation of perilacunar mineral by osteocytes. © 2013 American Society for Bone and Mineral Research.
Sclerostin is the product of the SOST gene and is expressed in terminally differentiated ostocytes embedded within the mineralized bone matrix.[1, 2] Mutations of the SOST gene have been identified as causative in sclerosteosis, a high bone mass disease in humans.[3, 4] A 52-kb deletion of a regulatory region approximately 35 kb distal to the SOST gene is causative of the high bone mass phenotype seen in Van Buchem's disease.[5, 6] Deletion of Sost in mice also causes a striking, whole skeleton, high bone mass phenotype similar to that of sclerosteosis in humans. These findings indicate that sclerostin has a key role in the regulation of bone mass.
The mechanism of action of sclerostin remains to be fully elucidated. Sclerostin inhibits canonical Wnt and Wnt-induced BMP signaling by binding to the Wnt co-receptors LRP5/6.[8, 9] Evidence also suggests that LRP4[10, 11] is a vital receptor for sclerostin, presenting it to LRP5/6 and facilitating sclerostin inhibition of Wnt signaling. However, the relative importance of each of these receptor interactions, the circumstances under which each of these receptors plays a role in the sclerostin response, and the way in which Wnt signaling is inhibited as a result are not yet understood.
We have recently identified osteocytes as targets for sclerostin action and reported that treatment of mineralizing human osteoblast cultures with recombinant sclerostin resulted in the inhibition of osteoblast maturation to osteocytes and of mineralization in vitro. This effect was attributable in part to the increased synthesis of matrix extracellular phosphoglycoprotein (MEPE) and increased abundance of the MEPE-derived ASARM peptide. In addition to this anti-anabolic role, sclerostin was also found to stimulate osteocyte support of osteoclastogenesis in a RANKL-dependent manner, consistent with the recent demonstration that osteocytes are a major source of RANKL in post-developmental bone.[14, 15] It is therefore now apparent that osteocytes have an essential role in the regulation of bone mineral content, and it appears that sclerostin is an important negative regulator of bone mass and mineralization via several mechanisms.
There is growing experimental support for the decades-old concept that osteocytes can liberate calcium and phosphate from bone mineral in a process termed osteocytic osteolysis.[16-21] It was reported recently that osteocyte perilacunar/canalicular remodeling is a means to provide calcium for milk production during lactation. The mechanisms, by which osteocytes remodel their perilacunar matrix, are unknown. However, Krempien and colleagues reported that PTH administration in rats results in an increased number of lysosomes in osteocytes, which were observed to fuse with the plasma membrane and empty their contents into the lacunar space. The induction of tartrate-resistant acid phosphatase expression in osteocytes treated with PTH was also observed. TRAP was also observed in osteocytes in the bones of lactating mice, and expression was rapidly lost after weaning. In addition, the expression of a number of genes usually associated with osteoclastic resorption was increased in an enriched osteocyte population from lactating mice compared with virgin or post-lactating mice, including cathepsin K, matrix metalloproteinase 13, a number of proton transporters and carbonic anhydrase 2. Interestingly, Atp6vob, encoding a subunit of the lysosomal proton pump V0 involved in acid extrusion in osteoclasts, has been identified in the MLO-Y4 cell transcriptome. These data suggest shared features between osteocytic osteolysis and osteoclast resorptive machinery.
In the current study, we examined the direct effect of recombinant human sclerostin (rhSCL) on the catabolic activity of osteocytes. We found that rhSCL stimulated the expression in osteocytes of a number of genes associated with bone resorption, including ACP5/Acp5, CTSK/Ctsk, and CA2/Car2, in both hOCy and mouse MLO-Y4 osteocyte models. With a focus on the involvement of CA2, treatment with rhSCL reduced the pH of the extracellular media and induced calcium release from mineralized matrix. Showing the potential relevance of these findings for human bone physiology, rhSCL treatment of human bone samples increased the osteocyte lacuna area. We conclude that sclerostin may have a catabolic action via induction of osteocytic osteolysis.
Materials and Methods
The use of all normal human donor-derived bone tissue was approved by the human ethics committees of the Royal Adelaide Hospital/University of Adelaide (Approval No. RAH090101). Human bone samples were obtained with informed written donor consent, as required and approved by the ethics committee.
Recombinant cytokines and antibodies
Recombinant human (rh) sclerostin (rhSCL) was purchased from R&D Systems (Minneapolis, MN, USA). Anti-mouse CA2 rabbit antibody and anti-human CA2 sheep antibodies were purchased from Abcam (Cambridge, UK) and R&D Systems, respectively. Alexa Fluor 488 goat anti-rabbit IgG and 647 donkey anti-sheep antibodies were purchased from Molecular Probes, Inc. (Eugene, OR, USA).
Cells and culture media
MLO-Y4 cells were passaged on type I collagen–coated plates, as described previously. Briefly, the cells were cultured in α-MEM containing 2.5% fetal calf serum (FCS) and 2.5% newborn bovine serum, L-glutamine (2 mM), HEPES (1 mM) at 37°C and 5% CO2 in a humidified incubator. Adult human primary osteoblasts were derived from intertrochanteric bone obtained at hip replacement surgery from a number of different donors. Sample processing and cell culture were as we have described previously. Mature osteocyte-like cultures were generated by culturing these cells for 35 days in mineralization medium, as previously described. Briefly, cells were cultured in α-MEM containing 5% FCS, L-glutamine (2 mM), HEPES (1 mM), ascorbate 2-phosphate (100 µM), KH2PO4 (1.8 mM), and dexamethasone (10 nM). Media containing all supplements were replaced every 4 days and cells were cultured for up to 5 weeks, generating human osteocyte-like cells (hOCy).[12, 27-29]
Preparation of total RNA and RT-PCR
Total RNA was extracted from cell lines and hOCy cultures, which were treated with rhSCL (1, 10, or 50 ng/mL) for 1, 3, or 7 days, and RNA was prepared as described above. Complementary DNA (cDNA) was prepared, and gene expression analyzed by real-time RT-PCR, as we have described previously. Relative gene expression between samples was calculated using the comparative cycle threshold (CT) method (ΔCT), using GAPDH or 18S rRNA as the reference gene, as indicated, and as we have published.[30, 31] Oligonucleotide primers were designed in-house to flank intron-exon boundaries and were purchased from Geneworks (Thebarton, Australia). Real-time oligonucleotide primers for the amplification of human and mouse CA2/Car2, ACP5/Acp5, CTSK/Ctsk, V-ATPase-a3 (OC116/Oc116), CLCN7/Clcn7, human GAPDH mRNA, and 18S rRNA were described previously.[30, 31] Sequences used for real-time RT-PCR amplification of mouse Gapdh mRNA were TGCACCACCAACTGCTTA (sense) and GCATGCAGGGATGATGTT (antisense).
Western blot analysis
Immunoblot analyses were performed as described. In brief, cells were lysed in lysis buffer (10 mM Tris-HCl, pH 7.6, 0.1% sodium dodecylsulfate, 150 mM NaCl, 1% Triton, 0.1 mM Sodium melavanadate, one tablet/10 mL protease inhibitor (Complete Mini, Roche, Indianapolis, IN, USA). Whole-cell extracts were prepared by centrifugation at 10,000g for 15 minutes at 4°C after suspending cells in lysis buffer on ice for 30 minutes. Extracts (30 µg) were electrophoresed in a 12% SDS-polyacrylamide gel and blotted onto a polyvinylidene difluoride membrane. Immunoblot detection was performed with the relevant antibodies using an ECF substrate (Amersham Pharmacia Biotech, Buckinghamshire, UK) and blots then analyzed using a Fluorimager (Molecular Dynamics, Sunnyvale, CA, USA).
In situ immunofluorescence
MLO-Y4 cells and NHBC were seeded into type I collagen–coated and uncoated chamber slides (Lab-Tek, Nunc, Naperville, IL, USA), respectively, and cultured for the times and under the conditions indicated. Cells were then rinsed with PBS, fixed with 4% paraformaldehyde in PBS for 10 minutes on ice, rinsed with PBS, and permeabilized with 0.1% Triton X100 for 5 minutes. Nonspecific binding sites were blocked with PBS containing 10% goat serum for 30 minutes at room temperature. The cells were incubated for 1 hour with primary antibodies on ice. After three washes in PBS, cells were incubated with either anti-rabbit IgG or anti-sheep antibody, respectively, for 1 hour in a dark, humidified container. Cells were then washed in PBS and mounted (Prolong Gold with DAPI [4′,6-diamidine-2′-phenylindole dihydrochloride] anti-fade mounting media; Invitrogen, Carlsbad, CA, USA). Samples were examined by confocal microscopy on a Radiance 2100 confocal microscope (Bio-Rad Microscience Ltd., Hemel Hempstead, UK).
Measurement of intracellular and extracellular pH
Intracellular pH (pHi) changes in MLO-Y4 cells and NHBC treated with rhSCL were measured using the pH-sensitive dye, carboxy-SNARF-AM (Molecular Probes, Inc.).[33, 34] The cells under various conditions were washed with PBS and loaded with carboxy-SNARF-AM at 37°C for 30 minutes, at a final concentration of 10 µM. Dye loading in cells was monitored visually using a Radiance 2100 confocal laser scanning microscope (Bio-Rad): Cells were excited at 514 nm, and the emission fluorescence was measured at two different wavelengths, 580 and 640 nm. pHi was calculated from the ratio between the fluorescence at 580 and 640 nm. The calibration curve was derived using cells with their intracellular pH fixed at known values ranging from 6.2 to 7.8 in calibration solution containing 141 mM K+ and 20 µM nigericin. Extracellular pH (pHo) was measured directly in the well, using the Micro Combination pH Electrode (Lazar Research Laboratories, Inc., Los Angeles, CA, USA). For these experiments, the media volumes were reduced to 150 µL in the absence of HEPES buffer to detect the pHo change. The change of pH from the day 0 sample of each treatment was calculated.
Measurement of calcium levels
hOCy were treated with 50 ng/mL rhSCL in the presence or absence of 10−6 M acetozolamide for 3 days and the calcium level in the supernatants was measured. Cell layer-associated Ca2+ levels were also measured, as described previously. Briefly, calcium levels in the cell layer were determined from quadruplicate cultures. The cultures were first washed three times with Ca2+/Mg2+-free PBS, and calcium was then extracted using 0.6 N HCl. Calcium standards were made to create a standard curve. Calcium levels were measured in a colorimetric assay using the arsenazo III method, as per manufacturer's instructions (Themo Fisher Scientific Inc., Middletown, VA, USA), and the absorbance at 630 nm was measured on a MR7000 microplate reader (Dynatech Laboratories, Guernsey, Channel Islands). We also measured ionized calcium levels using a micro-calcium ion electrode (Lazar Research Laboratories Inc.), which was standardized using pure calcium chloride solutions, as per the manufacturer's instructions.
To visualize the calcium release from the matrix, we devised a novel method by using calcein-stained mineralized substrates. Osteologic slides were incubated with medium containing calcein (5 µg/mL, Sigma Chemical Co., St. Louis, MO, USA) overnight at 4°C. Labeled slides were then washed with 3 × 0.5 mL volumes of sterile PBS. MLO-Y4 cells were then seeded onto these surfaces and treated with rhSCL (50 ng/mL) in the presence or absence of acetozolamide (10−6 M) for 5 days. Cells were then washed in PBS and mounted with aqueous mounting media, and samples were examined by confocal microscopy (Bio-Rad Microscience Ltd.). The intensity of calcein staining (488 nm) was measured at four regions of each well, in quadruplicate.
To determine the involvement of specific gene products to the observed responses to sclerostin, we utilized the highly specific technique of RNA interference. We used siRNA to downregulate Car2 Lrp4, Lrp5, and Lrp6 expression, using a method that we described previously. For these experiments, MLO-Y4 cells were seeded at a density of 2 × 104 cell per well in type I collagen–coated 24-well plates with 0.5 mL of growth medium and incubated overnight. Cells were then transfected with either the test siRNA or nonsilencing control siRNA, using Lipofectamine 2000 transfection reagent (Invitrogen) as per the manufacturer's instructions. SiRNA sequences for mouse Car2, Lrp4, and the control, nonhomologous, scrambled sequence equivalent (siNEG) were purchased from Applied Biosystems (Carlsbad, CA, USA). For knockdown of mouse Lrp5 and Lrp6, corresponding EsiRNA sequences and a nonsilencing control siRNA (siEGFP) were purchased from Sigma. Twelve hours after transfection, cells were enzymically removed from dishes using trypsin and plated onto Osteologic slides coated with calcein or uncoated, and then treated with or without rhSCL (50 ng/mL) for the assessment of pHo and Ca2+ release, as indicated. Alternatively, transfected cells were plated into wells of a 24-well plate for gene expression analysis by RT-PCR, as described above.
Measurement of ex vivo osteocyte lacuna size and CA2 expression
Human trabecular bone obtained at surgery was cultured with 50 ng/mL rhSCL in the presence or absence of 10−8 M acetozolamide for 7 to 10 days. Samples were decalcified, embedded in paraffin wax, and 5-μm sections cut using a microtome (Leica Microsystems, Wetzlar, Germany). Sections were stained using toluidine blue, and osteocyte lacuna area (Lac.Ar) was quantified from images using a Quantimet imaging system (Leica) and Image J software.
Mice transgenic for human SOST (SOST-Tg) were generated as previously described. Femurs from 14-week-old female SOST-Tg or wild-type littermates were dissected and periosteum and adherent tissue removed. Femurs were then fixed in ethanol and paraformaldehyde, decalcified, and paraffin embedded. Sections (5 μm) were stained with toluidine blue and imaged as for human trabecular bone. The Lac.Ar of cortical bone osteocytes was measured in the diaphysis, in a 1500-μm section starting 1000 μm below the growth plate. The area of trabecular bone osteocyte lacunae was measured separately from the metaphyseal trabecular bone in the same sections.
Statistical differences between parametric data sets were assessed using Student's t test. For experiments with more than two parameters, one-way analysis of variance (ANOVA) followed by Tukey's post hoc analysis was used (GraphPad Instat software). A value for p < 0.05 was considered significant.
Effect of exogenous sclerostin on gene expression in osteocyte-like cells
Human osteocyte-like cells (hOCy) were obtained by long-term (35-day) culture of primary human osteoblasts maintained under mineralizing conditions, as we have described previously.[12, 13, 27, 29] hOCy and MLO-Y4 cells were treated with rhSCL for periods of 3 or 7 days, as were small pieces of human trabecular bone cultured ex vivo. hOCy expressed a gene profile consistent with cells of a mature osteocyte-like phenotype, including E11, DMP1, SOST, OCN, BSP1, and PHEX (data not shown). Treatment of hOCy with rhSCL stimulated the expression of CA2 mRNA in a dose-dependent manner, with levels returning to those of control cultures after 7 days (Fig. 1A). Similar induction of Car2 by sclerostin was also observed in MLO-Y4 cells, which were even more sensitive, as shown in Fig. 1B, where 1 and 10 ng/mL of rhSCL produced significant increases. rhSCL also augmented CA2 mRNA expression in human trabecular bone samples cultured ex vivo (Fig. 1C). MLO-Y4 and hOCy cells, and human bone samples, also expressed ACP5, CLCN7, OC116, and CTSK mRNA, which, with the exception of OC116, were also upregulated by rhSCL treatment (Table 1).
|Gene name||Gene symbol||MLO-Y4||hOCy||Bone|
|Carbonic anhydrase 2||CA2/Car2||3.5 ± 1.5*||16.2 ± 5.5**||1.8 ± 0.6*|
|V-ATPase a3||OC116/Oc116||1.1 ± 0.1||1.7 ± 0.6||1.0 ± 0.6|
|Chloride channel 7||CLCN7/Clcn7||3.8 ± 1.3*||1.1 ± 0.3*||1.1 ± 0.8*|
|TRAP||ACP5/Acp5||4.4 ± 1.4*||2.6 ± 0.9**||1.4 ± 0.3**|
|Cathepsin K||CTSK/Ctsk||2.2 ± 0.9*||1.7 ± 0.7*||1.3 ± 0.4*|
Effect of sclerostin on CA2 protein expression in osteocyte like cells
Western blotting analysis revealed that CA2 protein expression in hOCy cells was increased after 3 days of treatment with rhSCL (Fig. 2A). Immunostaining of hOCy cells revealed that rhSCL enhanced the cytoplasmic CA2 expression (Fig. 2B). Similarly, rhSCL clearly increased the CA2 protein expression in MLO-Y4 cells, as determined by both Western analysis (Fig. 2C) and immunostaining (Fig. 2D).
Intracellular and extracellular pH change in MLO-Y4 cells treated with rhSCL
To explore whether the pHi changed after treatment of rhSCL, we utilized the pH-sensitive dye, carboxy-SNARF-AM. pHi was measured in the absence of HEPES to avoid any buffering effects on pHi. Treatment of hOCy cells with rhSCL significantly lowered the pHi (Fig. 3A), consistent with increased carbonic anhydrase activity. Similar results were obtained for MLO-Y4 cells (Fig. 3C). Extracellular pH (pHo), measured concurrently using a micro-pH meter, showed that treatment with rhSCL resulted in reduced pHo in both cell types compared with untreated samples (Fig. 3B, D).
Effect of sclerostin on extracellular mineral release
We next investigated whether the actions of sclerostin on pH resulted in mineral breakdown and calcium release. To do this, mineralized cultures of hOCy cells were treated with rhSCL, and the media calcium concentration was measured. Medium from cells treated with rhSCL had a higher total calcium concentration compared with that from untreated cells, and the removal of calcium from the mineralized cell layer into the medium was blocked in the presence of the carbonic anhydrase inhibitor, acetozolamide (Fig. 4A). Conversely, the total calcium content of the remaining mineralized matrix was decreased with rhSCL treatment, and this was also inhibited by acetozolamide (Fig. 4B). Interestingly, the level of calcium in cultures treated with acetozolamide alone was higher than in untreated cultures (Fig. 4B), implying that the incorporation of calcium into osteocyte layers is dynamic and that this is regulated in part by the constitutive activity of carbonic anhydrase. Because MLO-Y4 cells do not lay down a mineralized matrix, we devised a novel method to measure calcium release from around these cells. First, we incorporated the fluorescent calcium mimetic, calcein, into hydroxyapatite layers. We then seeded MLO-Y4 cells onto these now fluorescent surfaces and tested for effects of rhSCL on bound calcein. Although we could not detect calcein released into the media, the remaining calcein intensity in culture layers treated with rhSCL was significantly lower in comparison with untreated cells. Co-addition of acetozolamide completely blocked this effect (Fig. 4C). Similar to the effect seen in hOCy cultures, addition of acetozolamide alone resulted in increased remaining calcein, suggesting that osteocytes are involved in regulating mineral deposition through their expression of carbonic anhydrase.
Effect of Car2 siRNA knockdown on MLO-Y4 cells
To further investigate the importance of CA2 expression for the effect of sclerostin on osteocyte-like cell resorption, we employed RNA interference to knock down Car2 expression in MLO-Y4 cells. Transient transfection of Car2-siRNA into MLO-Y4 cells reduced Car2 mRNA levels by up to 81% after 48 hours, whereas no effect on Car2 mRNA levels was seen using a scrambled siRNA sequence (Fig. 5A). Consistent with effects seen in nontransfected cells (Fig. 3), rhSCL resulted in a reduction in pHo in control siRNA-transfected MLO-Y4 cells. However, rhSCL had no effect on pHo in Car2 siRNA-transfected cells (Fig. 5B). The addition of acetozolamide had no effect on these parameters when Car2 expression was knocked down. Treatment with rhSCL resulted in a significant increase in ionized calcium levels in control siRNA cultures, an effect that was blocked using acetozolamide (Fig. 5C). However, ionized calcium did not change in response to rhSCL in Car2 siRNA-transfected cells (Fig. 5C). We also tested the effects of siRNA-mediated knockdown of Car2 mRNA on the ability of MLO-Y4 cells to release calcein from labeled mineralized substrates. As shown in Fig. 5D, Car2 siRNA transfection prevented the removal of calcein and decrease in fluorescence intensity in response to rhSCL. Together, these data demonstrate that Car2 expression in MLO-Y4 is induced by rhSCL and that CA2 plays a direct role in the resulting loss of calcium from mineralized substrates, strongly implying that sclerostin-induced CA2 plays a role in osteocytic osteolysis.
Effect of sclerostin receptor knockdown
To investigate which of the reported receptors for sclerostin were important for the response to sclerostin, we used siRNA to knock down individually the expression of Lrp4, Lrp5, and Lrp6 mRNA. The extent of knockdown in their expression was 76%, 52%, and 65%, respectively, after 24 hours and was confirmed to be specific for each of the Lrps (Supplemental Fig. S1). Knockdown of each of these receptors inhibited the induction of Car2 mRNA expression by sclerostin (Fig. 6A). Examination of other genes potentially related to an osteocytic osteolysis effect revealed that induction by sclerostin of both Catk and Acp5 mRNA was also abrogated by knockdown of all receptors (Fig. 6B, C). Interestingly, the induction by sclerostin of Clcn7 mRNA expression was unaffected by Lrp4 knockdown but was abrogated in the case of Lrp5 and Lrp6 knockdown (Fig. 6D). Consistent with these effects, knockdown of each receptor resulted in the abolition of an effect of sclerostin on both pHo change (Fig. 6E) and calcium release (Fig. 6F), both implicating these receptors in the sclerostin response in a nonredundant fashion and the interruption of the β-catenin signaling pathway in this response.
Effect of sclerostin on osteocyte lacunar size
To test whether sclerostin induction of CA2 resulted in an effect on osteocyte activity in situ, human trabecular bone obtained from patients undergoing hip replacement surgery was cultured ex vivo for up to 7 days in the presence or absence of rhSCL. Resulting histological assessment of decalcified sections of rhSCL-treated bone revealed a significant increase in the mean osteocyte lacunar area (Lac.Ar) compared with bone incubated in control medium (Fig. 7). The increase in lacunar size was reversed in the co-presence of acetozolamide (Fig. 7), suggesting that the sclerostin-induced increase in CA2 levels may be a component of osteocytic osteolysis in human bone.
Analysis of SOST-Tg mice bone revealed a significant increase in mean Lac.Ar in cortical bone compared with that in age-matched WT littermates (WT = 7.5 ± 0.1 μm2 versus Sost-Tg = 8.0 ± 0.1 μm2, p < 0.001). Interestingly, however, there was no difference in this parameter in trabecular bone.
The available evidence points to sclerostin being a local regulator of bone metabolism. The expression of sclerostin in bone appears to reflect local strains perceived by osteocytes, with expression low under load and high in the unloaded state.[39-41] We have reported that osteocytes and pre-osteocytes are likely cellular targets for the activity of sclerostin.[12, 13] Sclerostin in these cell types exerted anti-anabolic effects, such as increasing the expression of the mineralization inhibitory MEPE-ASARM peptides, as well as pro-catabolic effects through stimulating osteocyte support of osteoclast function. Many groups have reported that osteocytes are capable of removing perilacunar mineral and matrix, although the mechanisms by which they do so have not been fully characterized. At least some features of osteocytic osteolysis appear to resemble the machinery used by osteoclasts to resorb bone mineral. Here, we sought to determine whether sclerostin can also regulate osteocytic removal of bone mineral. Recombinant sclerostin treatment, in two osteocyte models, resulted in upregulation of the mRNA species CTSK, ACP5, CLCN7, and CA2, confirming that osteocytes can express a repertoire of genes associated with bone resorption. The a3 subunit of the V-ATPase (OC116) was also detected, but its expression was not changed by rhSCL treatment. Demineralization by protons is a critical first step in the degradation of the bone matrix, and we observed that the addition of rhSCL increased the level of CA2 mRNA and protein in both human primary osteoblastic cells differentiated to osteocyte-like stage (hOCy) and in MLO-Y4 cells, resulting in upregulation of their resorptive activity. Thus, sclerostin treatment reduced both the pHi and pHo, and promoted the release of calcium from the endogenous mineralized matrix, in the case of hOCy cells, and from the mineralized substrate, on which MLO-Y4 cells were plated. These activities were reversed with the addition of the carbonic anhydrase inhibitor, acetozolamide, and with the siRNA-mediated knockdown of Car2 mRNA expression in MLO-Y4 cells. Importantly, rhSCL treatment of human trabecular bone derived from the proximal femur cultured ex vivo also resulted in increased CA2 mRNA expression, as well as increased osteocyte lacunar size in a manner consistent with the hypothesis that an acute increase in sclerostin levels can potentially result in osteocytic osteolysis. Consistent with this, analysis of SOST-Tg mice revealed a significant, albeit modest, increase in Lac.Ar in cortical but, interestingly, not trabecular bone. Contrary to our acute treatment of human bone samples with rhSCL, the SOST-Tg model is one of chronic overexpression and may be complicated by negative feedback compensatory mechanisms, which may have arisen during development to limit the extent of the effect observed. Intuitively, such a compensatory mechanism would be necessary for the purposes of maintaining skeletal integrity, as well as calcium and phosphate homeostasis. Even with acute exposure to sclerostin, the effects appear to be transient. For example, in the cell models examined, only a transient increase in CA2/Car2 mRNA levels was observed in response to SCL, and we have previously reported that sclerostin treatment causes the downregulation of expression of at least LRP4 at the mRNA level. One of the unexplained features of osteocytic osteolysis in general is that it appears to be self-limiting. Conceivably, the physiological process involves the dynamic release and recapture of mineral around the osteocyte, which may be difficult to image. Together, our findings are consistent with an accumulating body of evidence suggesting that sclerostin exerts a catabolic function as well as an anti-anabolic effect on bone metabolism.[2, 43-47]
CA2 is almost ubiquitously expressed but is considered the major generator of protons during osteoclastic resorption. CA2 deficiency in humans causes osteopetrosis, and CA2-deficient mice display metaphyseal widening and increased trabecular bone volume consistent with osteopetrosis. This mechanism for bone resorption appears to be at least partially preserved in osteocytes. The primary osteoclastic cellular mechanism responsible for extracellular acidification is proton secretion by the multi-subunit V-ATPase. Although MLO-Y4 cells have been reported to express the ATP6V0 subunit, and we have found expression of the a3 subunit of this complex enzyme in MLO-Y4 cells and hOCy cells, it remains to be seen if osteocytes express a functional V-ATPase.
LRP5 deficiency and sclerostin overexpression have been shown to have similar effects on bone in terms of reducing bone mass,[37, 50] supporting the notion that at least some actions of sclerostin are mediated via binding to LRP5. As stated, LRP6 and LRP4 are also implicated in sclerostin action.[8-11] This provided the rationale for using the MLO-Y4 osteocyte model to investigate the receptors and pathway through which sclerostin might regulate the expression of CA2 and the other identified potential mediators of osteocytic osteolysis. Knockdown of Lrp4 mRNA expression resulted in abrogation of induction of Car2, Catk, and Acp5 mRNA induction, pHo decrease, and calcium release by sclerostin, consistent with a putative role for LRP4 as an essential chaperone receptor.[10, 11] Interestingly, the induction of Clcn7 was unaffected by Lrp4 knockdown. The effects of sclerostin on all of the above genes examined were also abrogated when either Lrp5 or Lrp6 mRNA expression was inhibited. Sclerostin has been shown to bind directly to the first β-propeller domain of both LRP5 and LRP6 via loop 2 of its core domain, and this was not involved in binding to LRP4.[51, 52] Leupin and colleagues identified that propeller 3 of the LRP4 protein was the most likely to bind sclerostin but did not analyze which region of the sclerostin molecule was involved. Interestingly, loss of the ability of either LRP5[8, 53, 54] or LRP4 to bind sclerostin results in bone overgrowth in humans. In our experiments, knockdown of each of these receptors in MLO-Y4 cells appeared to abrogate the effects of sclerostin in a nonredundant fashion, except in the case of Clcn7. Although it is not clear how the interdependence of these receptors arises, our findings are consistent with sclerostin mediating effects through each of these receptors and with in vivo data that demonstrate receptor-specific knockdown effects on bone mass in the case of both Lrp5 and Lrp6.
The dependence of the effects of sclerostin, at least in MLO-Y4 cells, on LRP5/6 expression suggests that Wnt signaling inhibition may be at least partly responsible for the effects of sclerostin reported here because these are the Wnt co-receptors that facilitate canonical β-catenin signaling. A critical question arising from this study is the nature of the physiological Wnt signal inhibited by sclerostin in the models utilized. Thus far, the activity and/or the binding of Wnt1, Wnt3a, and Wnt9b have been shown to be inhibited by sclerostin, but the activities of the remaining 17 or so family members remain to be determined. A recent study by Ettenberg and colleagues suggested that Wnt ligands that activate β-catenin signaling can be grouped into two classes, depending on which propeller region of LRP5/6 they bind. They showed that whereas Wnt1 bound to propeller 1, Wnt3a bound to propeller 3. Interestingly, sclerostin also bound to propeller 1 and so may be assumed to inhibit the activity of Wnt ligands that also occupy binding sites at this region. However, Holdsworth and colleagues demonstrated that LRP6 could adopt at least three conformational states and in at least one of these, propellers 1 and 3 are juxtaposed, potentially explaining why sclerostin under certain conditions can also inhibit Wnt3a signaling.
The conditions under which sclerostin might effect a change in bone mineralization remain to be elucidated. We have recently reported that pro-catabolic stimuli increase sclerostin expression in bone; for example, exposure of osteoblast-lineage cells to the proinflammatory cytokines tumor necrosis factor-alpha (TNFα) and TNF-related weak inducer of apoptosis (TWEAK).[57, 58] We have also identified that sclerostin regulates physiological bone matrix mineralization through regulation of the PHEX/MEPE axis. It is possible that the regulation of CA2 and that of MEPE-ASARM are functionally linked, with CA2-mediated acidification reducing the propensity of perilacunar mineral deposition and MEPE-ASARM peptides physically blocking mineral crystal formation. Sclerostin has also been reported to be upregulated under other conditions, such as mechanical unloading,[39, 41] glucocorticoid treatment, and in response to 1,25(OH)2-vitamin D3 alone or in combination with retinoic acid.[60, 61] In addition, BMPs-2, -4, and -6 induced the mRNA levels of SOST in a time- and dose-dependent manner; BMP stimulatory effects on SOST were further enhanced by retinoic acid or 1,25(OH)2-vitamin D3. These results suggest that sclerostin might play a role not only in physiological remodeling but also in the case of immobilization and in pathological conditions, such as autoimmune disease and inflammatory disorders.
Despite the progress in our understanding of the function of sclerostin, important aspects remain to be addressed in future research. These include the identification of factors that regulate sclerostin expression and determine its highly restricted expression pattern in bone. How sclerostin mediates its effects in osteocyte-lineage cells is also an open question. Sclerostin is widely described as an inhibitor of the canonical Wnt-signaling pathway, but the precise way in which it does this and through which receptors or interaction partners it acts remains to be elucidated, as discussed above. Our previous data suggest that sclerostin is an active signaling molecule in its own right beyond that of an inhibitor of cell signaling, capable of inducing mitogen-activated protein kinase signaling and de novo gene expression in the absence of an exogenous Wnt or BMP signal.[12, 13] In addition, it is possible that sclerostin exerts its various effects through binding to receptors other than those discussed above. Sclerostin may have roles beyond the regulation of bone mass; for example, sclerostin also appears to play a role in B-cell development because mice lacking Sost expression have defective marrow B-lymphopoiesis.
In summary, we have identified for the first time to our knowledge that sclerostin can target osteocytes to stimulate their resorptive activity, at least in part by stimulating CA2 expression. That sclerostin may be involved in the direct catabolic activity of osteocytes represents a new paradigm and suggests that osteocyte production of sclerostin may play intrinsic roles in bone remodeling and possibly calcium homeostasis. These novel observations expand our view of the mechanism of bone remodeling controlled by sclerostin, combining inhibiting osteoblastic/osteocyte-mediated bone formation and mineralization and stimulation of both osteoclastic and osteocytic bone resorption. The circumstances, under which these various processes occur, should be the subject of ongoing research.
All authors state that they have no conflicts of interest.
This work was funded by Project Grants (565372 and 1004871) from the National Health and Medical Research Council of Australia (NHMRC). The authors thank Hui Peng Lim for technical assistance and are grateful to the surgeons and nursing staff of the Department of Orthopaedics & Trauma, Royal Adelaide Hospital, for the provision of human bone samples.
Authors' roles: Study design: GA and MK. Study conduct: GA, MK, and AW. Data collection: MK, AW, RO, PA, and GT. Data analysis: MK, AW, and GA. Data interpretation: GA, MK, AW, DF, and LB. Drafting manuscript: GA, MK, AW, and DF. Revising manuscript content: GA, DF, LB, MK, AW, RO, PA, and GT. Approving final version of manuscript: GA, MK, DF, LB, AW, RO, PA, and GT. MK, AW, and GA take responsibility for the integrity of the data analysis.