The receptor activator of the NF-κB ligand (RANKL) is the central player in the regulation of osteoclastogenesis, and the quantity of RANKL presented to osteoclast precursors is an important factor determining the magnitude of osteoclast formation. Because osteoblastic cells are thought to be a major source of RANKL, the regulatory mechanisms of RANKL subcellular trafficking have been studied in osteoblastic cells. However, recent reports showed that osteocytes are a major source of RANKL presentation to osteoclast precursors, prompting a need to reinvestigate RANKL subcellular trafficking in osteocytes. Investigation of molecular mechanisms in detail needs well-designed in vitro experimental systems. Thus, we developed a novel co-culture system of osteoclast precursors and osteocytes embedded in collagen gel. Experiments using this model revealed that osteocytic RANKL is provided as a membrane-bound form to osteoclast precursors through osteocyte dendritic processes and that the contribution of soluble RANKL to the osteoclastogenesis supported by osteocytes is minor. Moreover, the regulation of RANKL subcellular trafficking, such as OPG-mediated transport of newly synthesized RANKL molecules to lysosomal storage compartments, and the release of RANKL to the cell surface upon stimulation with RANK are confirmed to be functional in osteocytes. These results provide a novel understanding of the regulation of osteoclastogenesis.
The receptor activator of the NF-κB ligand (RANKL) is a cytokine identified as the osteoclast differentiation factor. The essential role of RANKL in physiological osteoclastogenesis has been demonstrated by the phenotype of mice with a complete lack of osteoclasts upon genetic disruption of RANKL.[2, 3] Most studies have focused on the total cellular expression of RANKL;[4-7] however, only RANKL molecules presented on the cell surface can stimulate osteoclast precursors and determine the magnitude of RANKL signal input and the degree of osteoclastogenesis. Therefore, it is indispensable to investigate RANKL subcellular behavior to achieve a complete understanding of the osteoclastogenic process.
Recent studies showed that osteocytes serve as the major source of RANKL in the bone remodeling process.[8, 9] Before this discovery, osteoblastic cells had been generally believed to be the major source of RANKL supplied to osteoclast precursors,[10-12] so studies on the mechanisms regulating RANKL cell surface presentation were conducted in osteoblastic cells.[13-15] We have previously shown that most newly synthesized RANKL molecules are transferred from the Golgi apparatus to the lysosomal storage compartment in complex with osteoprotegerin (OPG) and that the cell surface presentation of RANKL molecules is tightly restricted.[13, 14] In primary osteoblastic cells derived from Opg-deficient mice, accumulation of RANKL was observed in the Golgi apparatus along with the increase in the amount of RANKL leaked to the cell surface. In addition, stimulation of osteoblastic cells with RANK extracellular domain-conjugated beads (RANK beads) induced the translocation of RANKL from lysosomal storage to the contact area with RANK beads.[14, 15] These observations were made in osteoblastic cells, and the RANKL subcellular behavior in osteocytes is totally unknown.
Osteocytes are derived from osteoblasts encased in bone matrix during the process of bone formation and undergo changes in cell shape and ultrastructure.[16, 17] During osteocyte differentiation, the cytoskeleton is reorganized, dendritic processes are extended, and the cell shape changes from round to stellate.[16, 17] Dendritic processes are considered to play essential roles in cellular communication; through their dendritic processes, osteocytes are in contact with each other, with osteoblasts, and with lining cells.[16, 17] Dendritic processes are known to be lost when osteocytes are isolated from bone matrix and placed in conventional 2D culture conditions, making the study of osteocyte functions in vitro difficult.
In the present study, a novel in vitro co-culture system of osteocytes and osteoclast precursors was developed to analyze how osteocytes support osteoclastogenesis. Assays using this co-culture system revealed that osteocytic RANKL is provided as a membrane-bound form to osteoclast precursors through osteocyte dendritic processes. Moreover, the mechanisms regulating RANKL subcellular trafficking, including OPG-mediated transport of newly synthesized RANKL molecules to lysosomal storage compartments, are confirmed to be functional in osteocytes. These results provide a novel understanding of the regulation of osteoclastogenesis.
Materials and Methods
Preparation of primary cells and culture conditions
Primary osteocytes were isolated from the calvaria of 1- to 4-day-old C57BL6 mice according to a method previously described, with minor modifications.[8, 18] Briefly, minced calvarial pieces were digested for 20 minutes at 37°C in enzyme solution (0.1% collagenase type A (Oriental Yeast, Tokyo, Japan), 0.2% dispase II (Sanko Junyaku, Tokyo, Japan), 0.1% bovine serum albumin (BSA, Nacalai Tesque, Kyoto, Japan), 0.5% glucose, 25 mM HEPES, 10 mM NaHCO3, 3 mM K2HPO4, 70 mM NaCl, 30 mM KCl, 1 mM CaCl2, 60 mM sorbitol, pH 7.4), and subsequently washed with phosphate-buffered saline (PBS) to remove contaminating tissues, osteoblasts, and fibroblasts. After three washes, osteocytes were collected from the residual pieces, exposed to demineralizing solution (PBS, 0.1% BSA, 5 mM EDTA) for 20 minutes at 37°C, and then digested for 20 minutes at 37°C in enzyme solution. These collection steps were repeated twice. The cells collected were cultured for 12 hours on culture dishes coated with type I collagen (Nitta Gelatin, Osaka, Japan) in α-MEM containing 1% fetal bovine serum (FBS, Biowest, Nuaillé, France), 20 mM L-glutamine (Nacalai Tesque), and 1% penicillin-streptomycin (PCSM, Life Technologies, Carlsbad, CA, USA), and adherent cells were used for further experiments. Mouse bone marrow macrophages (BMMs) were collected from the tibia of 8- to 10-week-old C57BL6 mice using previously reported methods. Primary osteoblastic cells were collected from the calvaria of 1- to 4-day-old C57BL6 mice using previously reported methods. For the collagen-embedding 3D culture system, mouse primary osteocytes were suspended in ice-cold collagen sol (α-MEM containing 1% FBS, L-glutamine, PCSM, and 0.24% Cellmatrix Type I-A [Nitta Gelatin]), cast into plates, and cultured at 37°C to form collagen hydrogel (Fig. 1B). All animal procedures were approved by the Institutional Animal Care and Use Committee of Graduate School of Medicine, The University of Tokyo. Opg–/– mice were obtained from Japan Clea Co.[21, 22] C57/BL6 wild-type (WT) mice were purchased from Japan SLC (Shizuoka, Japan).
Total RNA was extracted using RNAiso Plus (TaKaRa, Shiga, Japan). As for osteocytes cultured under 3D conditions, collagen gel containing osteocytes was homogenized in the reagent and processed according to the manufacturer's standard protocol. As for the co-culture system, collagen gel together with porous filter was homogenized in the reagent and processed according to the manufacturer's standard protocol. The mRNA expression levels were determined by reverse transcription and real-time quantitative PCR with SYBR GreenER qPCR SuperMix Universal (Life Technologies) and the Eco Real-Time PCR system (Illumina, San Diego, CA, USA) and associated software. The expression levels of the marker genes were calculated using a standard curve prepared by serial dilution of the reference sample. The expression level of each marker gene was normalized to that of the housekeeping gene, mouse glyceraldehyde-3-phosphate dehydrogenase (Gapdh). Finally, the relative expression of the marker genes in primary osteocytes over time was compared with that on day 0 (freshly isolated cells). The sequence of each primer is as follows: 5′-CTT CAG GAA TGA TGC CAC AGA GGT-3′ and 5′-ATC TTT GGC GTC ATA GGG ATG GTG-3′ for mouse sclerostin (Sost), 5′-ACT TGT CGC AGA AGC ATC-3′ and 5′-GTG GGC GAA CAG TGT ACA A-3′ for mouse fibroblast growth factor 23 (Fgf23), 5′-GGC TGT CCT GTG CTC TCC CAG-3′ and 5′-GGT CAC TAT TTG CCT GTC CCT C-3′ for mouse dentin matrix acidic phosphoprotein 1 (Dmp1), 5′-CAG TGT TGT TCT GGG TTT TGG-3′ and 5′-TGG GGT CAC AAT ATC ATC TTC A-3′ for mouse podoplanin (Gp38), 5′-CCA AGC AGG AGG GCA ATA-3′ and 5′-AGG GCA GCA CAG GTC CTA A-3′ for mouse osteocalcin (Bglap), 5′-GGG CGT CTC CAC AGT AAG CG-3′ and 5′-ACT CCC ACT GTG CCC TCG TT-3′ for mouse alkaline phosphatase (Alpl), 5′-GTC TGT AGG TAC GCT TCC CG-3′ and 5′-CAT TTG CAC ACC TCA CCA TCA AT-3′ for mouse Rankl, 5′-ACC CAG AAA CTG GTC ATC AGC-3′ and 5′-CTG CAA TAC ACA CAC TCA TCA CT-3′ for mouse Opg, 5′-CCT TGC ACA AGC ACA TGT TC-3′ and 5′-CCA GAA GTA GTT GCA AGA CA-3′ for mouse calcitonin receptor (Calcr), 5′-CAG AAC GGA GGC ATT GAC TC-3′ and 5′-CCA CAG GAA TCT CTC TGT AC-3′ for mouse cathepsin K (Ctsk), 5′-GGT GAT GTC ACA GCA GAC GT-3′ and 5′-GGT CTC CCT GTC TTC TTT GC-3′ for mouse vacuolar-type H+-ATPase (Atp6v0d2), and 5′-ATG TGT CCG TCG TGG ATC TG-3′ and 5′-TGA AGT CGC AGG AGA CAA CC-3′ for mouse Gapdh.
Analysis of the purities of osteocyte and osteoblast population
Isolated osteocytes or osteoblasts were cultured for 8 hours on 96-well plates (8 × 104 cells/cm2) coated with type I collagen, and adherent cells were fixed with 4% paraformaldehyde at room temperature for 30 minutes. In the case of detecting the expression of SOST, BD GolgiPlug (BD Biosciences, San Jose, CA, USA), a protein transport inhibitor, was added to the media during the culture before fixation. After permeabilization treatment for 5 minutes with PBS containing 0.01% TritonX-100, samples were treated with blocking buffer (PBS containing 1% BSA) at room temperature for 1 hour. Then, the cells were incubated with primary antibodies: anti-Gp38 antibody (R&D Systems, Minneapolis, MN, USA), anti-DMP-1 antibody (Abcam Inc., Cambridge, MA, USA), anti-SOST antibody (R&D Systems), anti-bone-specific ALP antibody (TaKaRa), anti-fibroblast-specific marker antibody (ER-TR7) (Abcam Inc.) or corresponding control IgG derived from rabbit and goat (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Alexa Fluor-labeled secondary antibodies (Life Technologies) were used to detect primary antibodies. Finally, the cells were treated with 20 µg/ml RNase A (Sigma-Aldrich, St. Louis, MO, USA) and incubated with PBS containing Hoechst33342 (Sigma-Aldrich) at room temperature for 30 minutes. Fluorescence was detected using a cellular imaging and analysis system, IN Cell Analyzer 1000 (GE Healthcare, Waukesha, WI, USA) equipped with 10× Plan Apochromat objective lens and dichroic mirror 61002v2 (Chroma Technology Corp., Bellows Falls, VT, USA). Cell number and fluorescence intensity was analyzed with the associated software.
To collect osteocytes embedded in collagen gel, the gel was digested with collagenase type A at 37°C for 20 minutes. Cell suspension was centrifuged and the cell pellet was washed with PBS. Collected cells were lysed in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 1% NP-40, pH 7.4) and subjected to immunoblotting. The blots were probed with anti-Gp38 antibody (R&D Systems), anti-DMP-1 antibody (Abcam Inc.), anti-bone-specific ALP antibody (TaKaRa), or anti-GAPDH antibody (Cell Signaling Technology, Beverly, MA, USA) as the primary antibody. Then, proteins were detected with HRP-labeled secondary antibody (GE Healthcare), ECL Prime reagent (GE Healthcare Bioscience, Buckinghamshire, UK), and Chemidoc XRS (Bio-Rad, Hercules, CA).
Construction of protein expression vectors
To introduce proteins exogenously into osteocytes, lentiviral expression vectors were used in this study. Lentiviral vectors to introduce GFP, mouse RANKL fused with GFP at the N-terminus (GFP-RANKL), WT and mutant mouse OPG fused with His-tag at the C-terminus, mouse lysosome-associated membrane protein 1 (LAMP1) fused with Kusabira Orange at the C-terminus (LAMP1-KuOr), mouse formiminotransferase cyclodeaminase (FTCD) fused with Kusabira Orange at the C-terminus (FTCD-KuOr), and mouse calnexin fused with Kusabira Orange at the C-terminus (Calnexin-KuOr) were constructed using the ViraPower II Lentiviral Gateway Expression System (Life Technologies). Each gene construct was subcloned from a corresponding plasmid construct used in a previous report into the pLenti6.3/V5-DEST vector (Life Technologies). According to the manufacturer's protocol, lentiviral vectors and other packaging plasmids were transfected into 293FT cells using lipofectamine 2000. Twenty-four hours after transfection, sodium butyrate was added to the culture medium to a final concentration of 10 mM. Forty-eight hours after transfection, the culture medium was collected as a viral solution. In these constructs, protein expression is driven by CMV promoter.
Cell staining and confocal laser scanning microscopy analysis
For confocal cell imaging, cultured cells were treated with 4% paraformaldehyde and 0.02% Triton-X100 before staining. For actin fiber staining, phalloidin tetramethylrhodamine B isothiocyanate (TRITC) conjugate (Sigma-Aldrich) or Alexa Fluor 633 conjugated phalloidin (Life Technologies) was used according to the manufacturer's protocol. Fluorescence was detected using a confocal laser scanning microscope Fluoview FV1000 (Olympus, Tokyo, Japan) equipped with a UPLSAPO 100 × O objective lens (numerical aperture: 1.40, Olympus). Three-dimensional reconstructions of confocal sectioning images were performed using the IMARIS software (Carl Zeiss, Thornwood, NY, USA).
To examine the cell surface expression of GFP-RANKL in osteocytes, cell surface proteins were biotinylated using EZ-Link Sulfo-NHS-SS-biotin (Thermo Scientific, Bremen, Germany) following the manufacturer's protocol, and cells were harvested and lysed in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 1% NP-40, pH 7.4). Biotinylated proteins were collected using MagnaBind Streptavidin Beads (Thermo Scientific) following the manufacturer's protocol and subjected to immunoblotting. The blots were probed and proteins were detected with rabbit anti-GFP antibody (Life Technologies) or rabbit anti-Na+/K+ ATPase antibody (Santa Cruz Biotechnology) as the primary antibody, and HRP-labeled donkey anti-rabbit IgG antibody (GE Healthcare) as the secondary antibody.
Co-culture assay and RANK-bead stimulation assay
A 3D co-culture model of osteocytes and BMMs was developed. First, osteocytes were cultured on membranes with 3-µm-diameter pores (Millipore, Bedford, MA, USA) with collagen coating for 8 hours. Subsequently, collagen sol was cast into culture plates, and the porous membranes with osteocytes attached on one side were placed on the surface to embed osteocytes in collagen sol (Fig. 2A). After 1 hour of incubation at 37°C to gelatinize collagen sol, BMMs suspended in α-MEM containing 10% FBS, L-glutamine, PCSM, and 50 ng/mL mouse macrophage colony-stimulating factor (M-CSF, R&D Systems) were seeded. The culture medium was replenished every 3 days with fresh medium. Tartrate-resistant acid phosphatase (TRAP) enzymatic activity, which was evaluated by measuring p-nitrophenol production from p-nitrophenyl phosphate in tartaric acid, was assessed after 7 days of co-culture using the TRACP & ALP Assay Kit (TaKaRa) following the manufacturer's protocol. In the case of soluble RANKL (sRANKL) production inhibition experiments, recombinant mouse tissue inhibitor of metalloproteinase-2 (TIMP2, R&D Systems) was added to the medium at 2, 6.6, or 20 nM. The concentration of RANKL in culture supernatants or cell lysates was measured using Quantikine Immunoassay kits (R&D Systems) following the manufacturer's protocols. To examine the release of RANKL from lysosomal compartments to the osteocytic cell surface, RANK beads were added to the culture medium instead of BMMs and incubated for 12 hours (Fig. 4B). RANK beads were prepared with 6-µm-diameter polystyrene beads coated with protein G (Spherotech, Lake Forest, IL, USA) and recombinant RANK-Fc fusion protein (R&D Systems), as described previously. Proteins captured by the beads were collected and analyzed using previously reported methods. The activity of the lysosomal enzyme N-acetyl-β-glucosaminidase (NAGA) that leaked into the cultured medium upon stimulation with RANK beads was assessed by previously reported methods.
All data are expressed as the mean ± SD. from three independent determinations. Statistical analysis was performed using Student's t test or analysis of variance (ANOVA) followed by Dunnett's test where applicable.
Osteocyte characteristics can be maintained by collagen-embedding culture system
To analyze the osteoclastogenic ability of osteocytes, primary osteocytes were obtained from 1- to 4-day-old mouse calvaria and used for the analyses. First, we confirmed the purity of the isolated osteocyte population using IN Cell Analyzer 1000 (Fig. 1A). Co-staining of osteocyte marker Gp38 and osteoblast marker ALP showed that the fraction of osteocytes (Gp38-positive and ALP-negative) in the isolated cell population was about 65% and that contamination with osteoblastic cells (Gp38-negative and ALP-positive) was about 10%. Judging from the cell morphology, the remaining 25% were fibroblasts. Co-staining with anti-Gp38 and an anti-fibroblast-specific marker antibody (ER-TR7) showed that the percentage of fibroblasts (Gp38-negative and fibroblast marker-positive) in the cell population was about 25% (Supplemental Fig. 1A). Co-staining with anti-ALP and anti-fibroblast marker antibodies yielded similar results. Based on these results, we inferred that the remaining fraction is mostly fibroblasts. We also analyzed the osteoblastic cell population as a reference. The results showed that the fraction of osteoblastic cells (Gp38-negative and ALP-positive) within the cell population was about 95%. The results of DMP-1/ALP co-staining were similar. In addition, we also tried immunodetection of SOST, which is a marker for mature osteocytes. Because SOST is a secreted protein, the cells were cultured in the presence of the BD GolgiPlug reagent, which is a protein secretion inhibitor, before fixation. The results showed that the fraction of mature osteocytes (SOST-positive and ALP-negative) in the isolated osteocyte population was about 50%. These results indicated that the isolated osteocyte populations were sufficiently pure and that the level of contamination with osteoblastic cells was relatively low. Physiologically, osteocytes are buried in bone matrix and are known to dedifferentiate when cultured in conventional 2D conditions.[17, 23, 24] Therefore, the present study's design adopted a collagen-embedding 3D culture condition to culture osteocytes in vitro. As depicted in Fig. 1B, primary osteocytes were embedded in collagen, and the effect of collagen embedding on osteocyte characteristics was examined. Total RNA was extracted over time, and the mRNA expression level of markers of osteocyte differentiation was assessed by quantitative PCR. Results show that the mRNA expression levels of Sost and Fgf23, which are mature osteocyte markers,[16, 17] tended to be downregulated with time but were relatively higher under the 3D culture condition compared with those under a conventional 2D culture condition (Fig. 1C). In addition, the expression levels of Dmp1 and Gp38, an early osteocyte marker, were not affected in 3D culture, whereas those tended to increase in 2D culture (Fig. 1C). We also measured the expression levels of Bglap and Alpl, which are osteoblast markers. The expression levels of Bglap and Alpl were also not affected in 3D culture, whereas those tended to increase in 2D culture (Fig. 1C). Additionally, we measured the expression levels of Rankl and Opg mRNA. The results showed that the expression level of Rankl tended to be downregulated with time but was higher under the 3D culture condition compared with that under a 2D culture condition (Fig. 1C). The results also showed that the expression level of Opg tended to increase in 2D culture (Fig. 1C). Subsequently, we confirmed the protein expression levels of these markers by immunodetection: The protein levels of Gp38 and DMP1 were similar under the 3D culture condition but were slightly increased over time under a 2D culture condition (Fig. 1D). ALP protein expression was not detectable in the osteocyte under the 3D culture condition but increased and became detectable under a 2D culture condition (Fig. 1D). We also measured the secretion of SOST and FGF23 into the culture medium using ELISA and showed that the amount of secreted SOST and FGF23 was kept higher under the 3D culture condition compared with those under a 2D culture condition (Fig. 1E). In addition, we also measured the secretion of sRANKL and OPG into the culture medium using ELISA and showed that the secretion of sRANKL was kept higher in 3D culture and the secretion of OPG was kept lower in 3D culture (Fig. 1E). These results suggest that 3D culture condition is adequate for studying osteocyte function in vitro. Subsequently, the isolated osteocytes were transduced with GFP using a lentivirus expression system, and the cellular morphology was analyzed by confocal laser scanning microscopy (Fig. 1F). A stellate shape with long cell processes was observed in 3D culture. In addition, phalloidin staining confirmed the actin fiber orientation in the dendritic processes (Fig. 1F). These results support the physiological relevance of the 3D culture system.
Osteocytes have the ability to stimulate osteoclast formation efficiently from the opposite side of porous membranes
Recent reports indicated that osteocytes mainly provide RANKL signals to osteoclast precursors during the physiological process of osteoclast formation; however, the mechanism of signal stimulation remains unclear.[8, 9] Under physiological conditions, osteocytes are embedded in bone matrix, and the extremities of dendritic processes are exposed to the bone marrow cavity. In contrast, in conventional 2D co-cultures, BMMs can contact osteocytes at the osteocyte cellular body. Therefore, in the present study, a novel co-culture system of osteocytes and BMMs was developed to prevent the interaction of BMMs with osteocyte cellular bodies (Fig. 2A). This novel co-culture system is thought to mimic the physiological states of osteocytes supporting osteoclastogenesis. As a result, the elevation of TRAP activity was efficiently stimulated when osteocytes were embedded in collagen on the opposite side of a porous membrane (Fig. 2B). In addition, we confirmed the time course of osteoclast marker gene mRNA expression. Significant increase in Calcr, Ctsk, and Atp6v0d2 mRNA expression was observed over time (Fig. 2C). We also visualized TRAP, actin fibers, and nuclei in the osteoclasts by fluorescence staining. The results confirmed the formation of multinucleated cells showing TRAP expression and actin rings (Fig. 2D). The number of multinucleated (at least three nuclei) TRAP-positive cells formed in the BMM/osteocyte co-culture system was comparable with that formed in the reference culture (BMMs stimulated with 100 ng/mL sRANKL) (Fig. 2E). These results indicated that osteocytes are able to support osteoclast formation through a porous membrane, however; results shown in Fig. 1 indicated that osteocytes tended to dedifferentiate to a more immature osteocyte phenotype with time in culture, even under 3D culture conditions. Thus, it is necessary to confirm that the osteoclastogenic ability observed in the above experiments was indeed that of osteocytes. To address this issue, we examined the effect of adding recombinant OPG protein to the co-culture system at different time points (Fig. 2F). When recombinant OPG (100 ng/mL) was added to the medium for the duration of the co-culture, TRAP activity at day 7 decreased to about 15% of the control value (co-culture in the absence of recombinant OPG) (Fig. 2F). By contrast, when OPG protein was washed away on day 1, TRAP activity on day 7 decreased to about 50% of that in the control (Fig. 2F). In the same manner, TRAP activity on day 7 decreased to about 30% and 25% of that in the control when OPG was removed on day 3 and day 5, respectively (Fig. 2F). These results indicated that the contribution of RANKL signaling originating from osteocytes during the earlier stage (1 to 3 days) to TRAP activity on day 7 is considerable. Furthermore, we also examined the effect of the timing of co-culture initiation. When BMMs were added to osteocytes on day 3, TRAP activity on day 10 decreased to about 20% of that in the control (TRAP activity on day 7 when BMMs were added on day 0) (Fig. 2G). Similarly, when BMMs were added to osteocytes on day 6, TRAP activity on day 13 decreased to less than 10% of that in the control (Fig. 2G). These results indicate that the dedifferentiation of osteocytes with time in culture suppresses their ability to stimulate TRAP activity. In addition, osteoblasts strongly stimulated TRAP activity when cultured with BMMs on the same side of a porous membrane, whereas osteoblasts could not stimulate TRAP activity efficiently when cultured on the opposite side (Fig. 2H). Based on these results, we concluded that osteocytes have the ability to stimulate osteoclast formation efficiently from the opposite side of porous membranes.
Direct interaction between osteocytes and BMMs is needed to stimulate osteoclast formation
Next, to determine the contribution to osteoclastogenesis of sRANKL produced by osteocytes, the concentration of sRANKL in the supernatant of the co-cultures was measured by ELISA. The concentration of sRANKL in the culture medium was approximately 100 pg/mL (Fig. 3A), yet upregulation of TRAP activity was not observed when recombinant sRANKL was added to the medium at the same concentration in the absence of osteocytes on the opposite side of the porous membrane (Fig. 3B). These results indicate that sRANKL production does not solely contribute to the osteoclastogenic ability of osteocytes. To confirm this, co-culture assays were performed in the presence of recombinant TIMP2 added to the medium to inhibit sRANKL production from osteocytes. The inhibition of sRANKL production from osteocytes by TIMP2 was confirmed by ELISA (Fig. 3C). Subsequently, the effect of TIMP2 on TRAP activity in co-culture was examined, and the results showed that the inhibition of sRANKL production does not affect the elevation of TRAP activity stimulated by osteocytes (Fig. 3D). These results suggest that osteocytes provide RANKL signaling through cell-to-cell interaction with BMMs. So, co-cultures of osteocytes and BMMs were analyzed by confocal laser scanning microscopy. Fig. 3E shows a representative image of osteocytes interacting with BMMs via a dendritic process extended through a pore of the membrane. To determine whether the direct interactions between osteocyte dendritic processes and BMMs are necessary for the upregulation of TRAP activity, the effect of doubling the porous membrane to reduce interactions between osteocytes and BMMs was examined. Results showed that TRAP activity in co-culture was significantly reduced in the presence of the double membrane (Fig. 3F), indicating that the direct interaction between osteocyte dendritic processes and BMMs is needed for osteoclastogenesis. To confirm this point further, the effect of pore size on osteoclast formation was also examined. Along with a membrane pore size getting smaller, TRAP activity in co-culture tended to be reduced (Fig. 3G). These results indicate that osteocytes provide RANKL signaling through the direct interaction between osteocyte dendritic processes and osteoclast precursors.
Osteocytic RANKL is predominantly localized in lysosomes and released to the dendrite surface
As we previously reported, osteoblastic RANKL is predominantly localized in lysosomes and released to the cell surface in response to stimulation with RANK beads, which mimic the interaction with RANK expressed on the cell surface of osteoclast precursors;[14, 15] however, the regulation of RANKL subcellular localization in osteocytes is unknown. A series of experiments showed that RANKL is provided to osteoclast precursors through the dendritic processes of osteocytes, suggesting that the regulation of RANKL subcellular localization in osteocytes may be important to understand the osteoclastogenic process. Therefore, RANKL subcellular localization was analyzed in osteocytes by confocal laser scanning microscopy. GFP-RANKL introduced into osteocytes was predominantly co-localized with the lysosome marker LAMP-1-KuOr, and cell surface expression was minimal (Fig. 4A). Subsequently, stimulation-dependent release of RANKL to the cell surface was also investigated. To measure the amount of RANKL released in the stimulation area, collagen-embedded osteocytes were cultured on a porous membrane as described for the co-culture system, and RANK beads were added to the medium instead of BMMs (Fig. 4B). Interactions between osteocyte dendritic processes and RANK beads were confirmed by confocal laser scanning microscopy (Fig. 4C), indicating that RANK-bead stimulation can mimic the situation of co-culture with BMMs. After 12 hours of incubation, proteins captured on the RANK beads were collected. Immunoblot analysis of the samples collected showed that GFP-RANKL was released to the surface of osteocyte dendritic processes where RANK beads contact (Fig. 4D). Leakage of NAGA, which is known to be a lysosomal enzyme, into the medium upon stimulation with RANK beads was also measured (Fig. 4E). As a result, NAGA was leaked into the medium significantly after RANK-bead stimulation, indicating the fusion of lysosomes to the plasma membrane. These results suggested that lysosomal RANKL was released to the cell surface upon stimulation of cells with RANK beads.
OPG regulates RANKL subcellular localization in osteocytes
We previously reported that OPG functions as a RANKL trafficking regulator in osteoblasts; however, the effect of OPG co-expression on the subcellular localization of RANKL in osteocytes is unknown. Therefore, RANKL localization in osteocytes derived from Opg–/– mouse calvaria was investigated. GFP-RANKL tended to accumulate in the Golgi apparatus, which was confirmed by co-transduction of FTCD-KuOr, a marker for the Golgi apparatus (Fig. 5A). Rescue experiments were also performed by introducing OPG in Opg–/– osteocytes: The lysosomal localization of RANKL was recovered when GFP-RANKL was introduced with WT OPG (Fig. 5B). We also reported that the heparin-binding domain (HBD) of OPG was indispensable for OPG function as a trafficking regulator in osteoblastic cells, and we confirmed this point in the present study. When an HBD deletion mutant (OPG-ΔHBD) was co-introduced in Opg–/– osteocytes, GFP-RANKL tended to accumulate in the Golgi apparatus and the lysosomal localization of RANKL was not recovered, indicating that the OPG HBD is necessary for the trafficking regulatory activity (Fig. 5C).
We had also shown that RANKL accumulation in the Golgi apparatus resulted in the upregulation of RANKL transfer to the plasma membrane in osteoblastic cells.[13, 14] Therefore, to determine the amount of RANKL on the osteocytic cell surface in Opg-deficient cells, a biotinylation assay was performed to capture cell surface proteins selectively. The amount of exogenously introduced GFP-RANKL on the surface of Opg–/– osteocytes was greater than that in WT osteocytes (Fig. 5D). Detection of endogenous RANKL localized on the osteocyte surface using biotinylation was not successful, probably because of the low biotinylation efficiency under 3D culture conditions and the low sensitivity of anti-RANKL immunodetection. Therefore, to further investigate this question, we quantitated the amount of endogenous sRANKL released into the culture supernatant. We assumed that the amount of secreted sRANKL would be greater if the amount of endogenous membrane-bound RANKL expressed on the surface of Opg–/– osteocytes was greater than that expressed on WT osteocytes. In Opg–/– conditions, the amount of sRANKL was greatly increased compared with WT conditions (Fig. 5E). These results indicated that the large increase in the amount of RANKL localized on the osteocyte surface leads to the uncontrollable activation of osteoclasts. This mechanism might account for the previously reported osteoporotic phenotype of Opg–/– mice.
The function of OPG as a RANKL trafficking regulator is crucial for controlled osteoclastogenesis
Finally, the impact of RANKL subcellular trafficking on the osteoclastogenic ability of osteocytes was examined. The above results with osteocytes are similar to the previous observations in osteoblastic cells suggesting that the regulation of RANKL subcellular trafficking by OPG is crucial for controlling the osteoclastogenic ability of osteocytes. The co-culture assay showed that Opg–/– osteocytes exhibit potent stimulatory effect on TRAP activity compared with WT osteocytes (Fig. 6A). The contribution of OPG, as a regulator of RANKL trafficking, to the control of osteoclastogenesis was also examined using a RANK-CRD-OPG-ΔCRD chimeric construct previously shown to retain decoy receptor activity but lack RANKL trafficking regulatory activity in osteoblastic cells. As expected, the introduction of RANK-CRD-OPG-ΔCRD in Opg–/– osteocytes did not rescue RANKL lysosomal localization (Fig. 6B). Quantitation of sRANKL in the culture supernatant showed that the reduction in sRANKL secretion upon co-expression of RANK-CRD-OPG-ΔCRD was small (Fig. 6C). This indicates that the effect of RANK-CRD-OPG-ΔCRD on the amount of endogenous RANKL presented on the osteocytic surface was small. Co-culture assays showed that the introduction of RANK-CRD-OPG-ΔCRD had only a slight effect on TRAP activity in Opg–/– osteocytes' co-culture (Fig. 6D), whereas WT OPG introduction reduced the TRAP activity to a level comparable with that of WT osteocytes (Fig. 6A, H). In addition, time profiles of mRNA expression levels of osteoclast markers, Calcr, Ctsk, and Atp6v0d2, were analyzed (Fig. 6E). Results showed that the effect of RANK-CRD-OPG-ΔCRD introduction on osteocyte marker expression was not significant, whereas WT OPG introduction significantly reduced the induction of Calcr and Ctsk mRNA expression. We also confirmed that the secreted amount of exogenously introduced OPG or its chimeric protein to the culture media are comparable to the level secreted from WT osteocytes (Fig. 6F, J). These results indicate that RANKL trafficking regulation by OPG plays an important role in the control of osteoclastogenesis.
In addition, RANK-CRD-OPG-ΔCRD was shown to compete with endogenous OPG for binding to RANKL in the Golgi apparatus and to inhibit selectively the function of endogenous OPG as a trafficking regulator when introduced into osteoblastic cells. Therefore, the effect of RANK-CRD-OPG-ΔCRD introduction in WT osteocytes was also investigated. Quantitation of sRANKL in the culture supernatant suggested that the amount of endogenous RANKL presented at the cell surface increased with co-expression of RANK-CRD-OPG-ΔCRD in WT osteocytes (Fig. 6G). Co-culture assays also showed that the TRAP activity in the co-culture with WT osteocytes introduced with the RANK-CRD-OPG-ΔCRD construct was significantly increased, while that in the co-culture with OPG-introduced osteocytes was slightly reduced (Fig. 6H). In addition, RANK-CRD-OPG-ΔCRD introduction led to the increase in induction ratio of Calcr and Ctsk mRNA expression compared with the control co-culture, whereas WT OPG introduction reduced the induction of Atp6v0d2 mRNA expression (Fig. 6I). We also confirmed that the secreted amount of exogenously introduced OPG or RANK-CRD-OPG-ΔCRD protein to the culture media is comparable to the level secreted from WT osteocytes (Fig. 6J). We already confirmed that the RANK-CRD-OPG-ΔCRD recombinant protein retains the ability to inhibit RANKL signaling as a decoy receptor for RANKL. Considering these data, the role played by OPG in regulating RANKL trafficking is important for controlling the osteoclastogenic ability of osteocytes.
In conclusion, our findings suggest that osteocytes provide RANKL to osteoclast precursors at least through direct cell-to-cell interaction at the extremities of dendritic processes, and the function of OPG as a RANKL trafficking regulator is crucial for controlling the osteoclastogenic ability of osteocytes (Fig. 7).
Osteoblastic cells were believed for years to be the major source of RANKL presented to osteoclast precursors in physiological osteoclastogenesis;[10-12] however, two recent reports showed that the osteocyte is a central player in regulating physiological osteoclastogenesis.[8, 9] Hence, we faced the urgent need to reinvestigate the molecular mechanisms involved in the regulation of osteoclast formation by osteocytes. Although the earlier study showed that osteocytic MLO-Y4 cells could support osteoclast formation efficiently, the way of RANKL signal delivery to BMMs was yet to be clarified. To investigate molecular mechanisms in detail, in vitro assay systems reflecting physiological situations are needed; however, the in vitro experimental systems generally used to analyze the osteoclastogenic process are designed to assess interplays between osteoblastic cells and osteoclast precursors. Therefore, appropriate in vitro assay methods for evaluating the osteoclastogenic ability of osteocytes needed to be established in this study. As mentioned earlier, previous reports pointed out that isolated osteocyte-like cells tend to dedifferentiate and lose their characteristic morphology over time when cultured in conventional 2D conditions and that the osteocytic differentiation of osteoblasts does not take place in 2D culture conditions. To analyze the function of osteocytes in the osteoclastogenic process, it is desirable that the properties of the isolated osteocytes be maintained for a prolonged time. Therefore, a collagen-embedding 3D culture system was adopted for culturing osteocytes because type I collagen is a major constituent of bone matrix and forms a hydrogel at 37°C. In addition, reports show that osteoblastic cells tend to acquire osteocyte-like features when embedded in collagen hydrogel.[23, 29] The results of the present study showed that the physiological properties of osteocytes can be maintained for about a week when osteocytes are embedded in collagen. The molecular mechanisms involved in the maintenance of osteocytic features are not yet understood, but attachment to the extracellular collagen matrix might trigger intracellular signals in osteocytes. A previous report showed that various α/β integrin subunit isoforms, including α5, αV, β1, and β3, are expressed in osteocytes,[30-32] and receptors such as integrin might play a role in maintaining osteocytic features. Next, a co-culture system of osteocytes and BMMs was established to enable the analysis of the osteoclastogenic process. The combination of collagen embedding for osteocytes and a porous membrane between osteocytes and BMMs enables the evaluation of osteocyte function in osteoclastogenesis. The results of the co-culture assay showed that osteocytes possess the ability to stimulate osteoclast formation, even from the opposite side of the porous membrane.
These results raised the question of how RANKL signaling is transduced, ie, the question of the contribution of direct cell-to-cell interactions and soluble factors. It is well established that the extracellular portion of RANKL can be shed by matrix metalloproteases to form sRANKL.[33, 34] Although previous studies using osteoblastic cells as a RANKL source indicated that direct interactions between osteoclast precursors and osteoblastic cells are needed for in vitro osteoclastogenesis, it remains unclear whether direct interactions are necessary in the case of osteocytes. Results of a series of experiments showed that direct interactions between osteocytes and BMMs are necessary for efficient osteoclastogenesis and that the contribution of sRANKL is minor. As we showed in Supplemental Fig. S1D, RANK beads added to osteoblasts from the opposite side of the membrane did not capture RANKL, whereas beads added to osteocytes from the opposite side of the membrane captured RANKL. This result is consistent with the observation that osteoblasts co-cultured on the opposite side of the separation membrane could not stimulate TRAP activity. Confocal laser scanning microscopy analyses confirmed that the dendritic processes of osteocytes extended through the pores of the separation membrane and made direct contact with BMMs. We also analyzed osteoblasts cultured on the opposite side from BMMs using confocal laser scanning microscopy. The results showed that osteoblasts did not extend the dendritic process through membrane pores. These data suggest that the mechanisms regulating RANKL signal output at the extremities of osteocyte dendritic processes may be crucial for the regulation of the degree of osteoclastogenesis.
We previously reported that the presentation of RANKL on the osteoblastic cell surface is tightly regulated through complex subcellular trafficking mechanisms.[13-15] In the present study, OPG function was reinvestigated in osteocytes. Results showed that RANKL was predominantly localized in lysosomes in osteocytes as observed in osteoblastic cells. The property of OPG to regulate RANKL trafficking and restrict cell surface RANKL is also functional in osteocytes as in osteoblastic cells. In addition, the release of RANKL to the extremities of osteocyte dendritic processes was triggered upon stimulation with RANK beads, which is also similar to the release of RANKL observed in osteoblastic cells. Moreover, experiments using the RANK-CRD-OPG-ΔCRD chimera, which retains function as a decoy receptor but not as a trafficking regulator, showed that the OPG function of RANKL trafficking regulator is more crucial to the control of osteoclastogenesis than the decoy receptor function. These results indicate that the regulation of RANKL subcellular trafficking by OPG also plays a crucial role in the control of osteoclastogenesis. Although further in vivo validation of this concept, including a genetic approach, is needed, findings in the present study provide a novel understanding of the regulatory mechanisms of osteoclastogenesis.
All authors state that they have no conflicts of interest.
This work was supported in part by a Grant-in-Aid for Scientific Research (B) 24390349 and Grant-in-Aid for Young Scientists (Start-up) 24890048 from the Japan Society for the Promotion of Science and a Grant-in-Aid for Scientific Research on Innovative Areas “HD-physiology” 22136015 from the Ministry of Education, Culture, Sports, Science, and Technology.
Authors' roles: MH, YI, and YK designed the study, performed experiments, analyzed data, and wrote the manuscript. MH, NH, and SA performed experiments and wrote the manuscript. HS designed the study, analyzed data, and revised the manuscript. MH, YI, YK, MH, NH, SA, and HS approved the final version of the manuscript. MH, YI, and YK take responsibility for the integrity of the data analysis.