These authors contributed equally to this work
Apoptotic Bodies Convey Activity Capable of Initiating Osteoclastogenesis and Localized Bone Destruction†
Version of Record online: 4 FEB 2008
Copyright © 2008 ASBMR
Journal of Bone and Mineral Research
Volume 23, Issue 6, pages 915–927, June 2008
How to Cite
Kogianni, G., Mann, V. and Noble, B. S. (2008), Apoptotic Bodies Convey Activity Capable of Initiating Osteoclastogenesis and Localized Bone Destruction. J Bone Miner Res, 23: 915–927. doi: 10.1359/jbmr.080207
The authors state that they have no conflicts of interest.
- Issue online: 4 DEC 2009
- Version of Record online: 4 FEB 2008
- Manuscript Accepted: 31 JAN 2008
- Manuscript Revised: 23 JAN 2008
- Manuscript Received: 20 APR 2007
- bone resorption;
Introduction: Osteocyte apoptosis co-localizes with sites of osteoclastic bone resorption in vivo, but to date, no causal molecular or signaling link has been identified between these two processes.
Materials and Methods: Osteocyte apoptotic bodies (OABs) derived from the MLO-Y4 osteocyte-like cell line and primary murine osteocytes and apoptotic bodies (ABs) derived from primary murine osteoblasts were introduced onto the right parietal bone of murine calvariae, and osteoclastic bone resorption was examined 5 days after treatment. In addition, the ability of primary murine and cell line–derived OABs to support osteoclastogenesis was examined in vitro in co-culture with murine bone marrow hematopoietic progenitors in the absence of RANKL or macrophage-colony stimulating factor.
Results: For the first time, we show that OABs are capable of initiating de novo osteoclastic bone resorption on quiescent bone surfaces in vivo. Furthermore, the addition of OABs to mononuclear osteoclast precursors (OPs) in vitro resulted in the maintenance of OP cell numbers and an increase in the proportion and activity of TRACP+ cells. In contrast, application of ABs from osteoblasts showed no osteoclastogenic activity either in vivo or in vitro. The osteoclastogenic capacity of OABs was shown to be independent of the known osteoclastogenic factor RANKL but dependent on the induction of TNF-α production by OP.
Conclusions: These data point to a mechanism by which dying osteocytes might target bone destruction through the distribution of OAB-associated signals and give further physiological meaning to the apoptotic process in bone.
Inappropriate bone destruction by mineralized matrix-resorbing osteoclasts underlies a number of bone diseases including osteoporosis. Osteoclasts are produced as a result of the fusion of mononuclear macrophage/monocyte lineage precursors and, whereas a number of key molecules are known to show osteoclastogenic activity both in vitro and in vivo,(1) the mechanisms by which the initiation of bone resorption at particular anatomical sites in a bone is achieved are unknown.
Osteocytes are located within lacunae in the bone matrix and form an extensive communication network between each other and cells on the bone surface through long cytoplasmic processes and gap junctions.(2) It has been proposed that the osteocyte directs the location of bone destruction(3) and formation(4,5) in response to a range of environmental inputs including mechanical stimulation, such that a bones shape, mass, and hence, strength is appropriate for its current or recent use.
Studies in vivo have shown an association between osteocyte apoptosis and regions of osteoclastic bone resorption.(6–10) The demonstration that the targeted destruction of damaged bone was preceded by osteocyte apoptosis in the region of damage has led to the hypothesis that signals released as a result of osteocyte apoptosis might initiate the process of targeted bone resorption in health and disease.(3) Despite literature showing an association between the two processes, no causal or molecular link has been shown between osteocyte apoptosis and the formation of bone-resorbing osteoclasts from their mononuclear precursor cells.
Potential osteocyte-related molecular mechanisms capable of controlling targeted bone resorption include either the loss of antiresorptive signals(11–13) or the gain of proresorptive signals.(14) Because bone containing low numbers of viable osteocytes is not readily remodeled(15–19) and osteocytes are, in vitro, known to be associated with pro-osteoclastogenic activity,(20) it seems likely that osteocytes produce pro- rather than antiresorptive signals.
Here we studied the role of osteocyte apoptosis in the initiation of osteoclastogenesis and subsequent bone resorption by introducing osteocyte apoptotic bodies (OABs) to the bone-forming surface of murine calvariae in vivo. In addition, we studied the potential for OABs to promote osteoclastogenesis in vitro after their introduction to bone marrow–derived osteoclast precursors (OPs) in the absence of the known pro-osteoclastogenic factors RANKL and macrophage-colony stimulating factor (M-CSF). Our results show that apoptotic bodies derived from osteocytes, independent of additional pro-osteoclastogenic factors, are associated with potent osteoclastogenic activity capable of influencing the bone environment in terms of local resorptive activity.
MATERIALS AND METHODS
All materials were purchased from Sigma unless otherwise stated.
Cells and cell culture
Marrow OP cells were isolated from femurs of 3- to 6-wk-old mice. Mature osteoclasts and stromal cells were removed from cultures using differential plating of bone marrow for 1 and 24 h, respectively.(21) Nonadherent OP cells were plated at 40,000 cells/ml and shown to produce bone-resorbing osteoclasts after treatment with M-CSF at 50 ng/ml and RANKL (R&D Systems) at 100 ng/ml. Time zero (T0) cultures were incubated for 24 h after isolation from the bone marrow. Osteocytes and osteoblasts were isolated from the calvariae of 3- to 6-wk-old mice.(22) The calvariae were stripped of periosteum and incubated with trypsin for 10 min at 37°C. Calvariae were subjected to 10 sequential digestions with collagenase (0.75 mg/ml) for 25 min at 37°C, and cell were populations collected for each fraction and maintained in αMEM supplemented with 5% FCS, 5% newborn cell serum (NCS), penicillin (50 IU/ml), streptomycin (50 μg/ml), and l-glutamine. Osteoblast cultures were characterized by positive staining for alkaline phosphatase and the ability to form mineralized nodules after culture in osteogenic (OS) mix (αMEM supplemented with 10% FCS, penicillin [50 IU/ml], streptomycin [50 μg/ml], l-glutamine, 1,25-VitD3, 10 nM dexamethasone, 100 nM ascorbic acid, and 50 μg/ml β-glycerol phosphate), which were stained with von Kossa (5% wt/vol silver nitrate for 30 min under strong light). Osteocyte populations were identified by distinctive dendritic morphology and lack of proliferative capacity after incubation with 50 μM 5-bromo-2 deoxyuridine (BrdU) and 50 μM deoxycytidine. BrdU-labeled nuclei were visualized using immunocytochemical staining. Osteoblastic cells where found in fractions 2–5, and osteocytes were found in fractions >6. The MLO-Y4 osteocyte cell line was cultured as described previously.(23)
Production of apoptotic and necrotic cells
Osteocytes or osteoblasts were maintained in medium containing 0.1% FCS for 7 days to promote apoptosis.(24) OABs and osteoblast apoptotic bodies (ABs) were sorted with Annexin V-biotin (Roche) and streptavidin-conjugated magnetic beads.(25) To produce apoptotic osteocyte conditioned media (aOCM), the medium was retained after OAB were purified and centrifuged at 20,000g to remove debris. Healthy osteocyte conditioned media (hOCM) was produced by collecting medium from healthy osteocyte cultures and centrifuging at 20,000g to remove debris. Necrotic osteocytes were produced by incubation in medium containing 0.8 mM H2O2 for 3–5 h and characterized as described previously(26) and were separated into membrane and soluble fractions by centrifugation at 20,000g. To produce necrotic cells at densities equivalent to those of apoptotic bodies, necrosis was induced in cultures containing the same number of cells as those used to generate the apoptotic body densities used in the previous experiments (i.e., original cell number equivalents).
Estimation of protein content of ABs
ABs at 100,000/ml obtained from primary osteocytes, MLO-Y4 osteocyte cell line, and primary osteoblast origin were lysed after a 5-min incubation at 4°C in lysis buffer (25 mM Tris HCl, pH 7.6, 150 mM NaCl, 1% [vol/vol] NP-40, 6 mM sodium deoxycholate, 1 mM PMSF, 1 mM Na3VO4, and 1 mM NaF) containing protease inhibitor cocktail (Roche), and protein content was estimated using the Bradford assay.(26)
OP cells grown in the absence of RANKL and M-CSF were subjected to experimental treatments for 72 h. Treatments included healthy osteocytes at 2000 cells/ml; primary OABs (pr.OABs) at 50,000–200,000 OAB/ml, MLO-Y4 OABs at 6,250–200,000 OAB/ml; osteoblast ABs (prim. osteoblast ABs) at 100,000 AB/ml; necrotic osteocyte cells at 100,000 cells/ml; 500 μl media conditioned by apoptotic osteocytes both cell line (aOCM) and primary (primary aOCM); and healthy osteocytes (hOCM) or necrotic cells (necrotic OCM). In studies using specific blocking agents, cells were treated with either the RANKL inhibitor osteoprotegerin (OPG) at 50–1000 ng/ml for 30 min before addition of treatments, after which time it remained in the culture for the 72-h period or with a TNF-α neutralizing antibody (R&D Systems) at 0.02–0.1 μg/ml for 1 h before addition of experimental treatments for 72 h. Supernatants were analyzed for the presence of TNF-α using a LiquiChip Mouse Cytokine detection kit (Qiagen) and measurement on a Luminex xMAP array system.
Immunohistochemical staining for RANKL
MLO-Y4 OAB and healthy MLO-Y4 were stained for the presence of RANKL using anti-mouse RANKL affinity purified Goat IgG (R&D Systems) with chromogenic detection using the horseradish peroxidase (HRP)-DAB system (CTS Series; R&D Systems). Osteocyte ABs were cytospun onto glass slides before fixation in 4% (wt/vol) paraformaldehyde solution for 10 min followed by 3 × 5-min washes in PBS. Healthy MLO-Y4 were grown in culture, media were removed, and the cell layer was washed in PBS before fixation in 4% paraformaldehyde. Before incubation of the samples with primary anti-murine-RANKL antibody, OABs and healthy MLO-Y4 were taken through blocking stages for peroxidase, serum, avidin, and biotin following the manufacturer's instructions within the HRP-DAB Cell and Tissue Staining Kit (R&D Systems). Samples were incubated in mRANKL affinity-purified Goat IgG (5 μg/ml) for 16 h at 4°C followed by 3 × 15-min wash in PBS before incubation for 1 h at room temperature in anti-goat IgG biotinylated secondary antibody. Samples were washed (3 × 15 min in PBS) before a 30-min incubation at room temperature in high-sensitivity streptavidin-HRP conjugate. Samples were washed (3 × 5 min in PBS) and incubated for 120 s with DAB chromogen.
OP cells were permeabilized in acetone and incubated in sodium tartrate to determine acid phosphatase using naphthol AS-BI phosphate(27) and counterstained with DAPI to determine total cell number in cultures or tissue sections. The proportion of positive cells per microscope field was determined using image capture and image analysis software (NIH Scion Image).
Assay of bone resorptive activity
OP cells were placed on OsteoSite (IDS) dentine discs(28) and subjected to treatments at 37°C for 72 h. At the end of the experimental period, the discs were sonicated in 0.5 ammonium hydroxide, and the resorption pits were visualized under light microscopy after staining with tolidine blue (1% wt/vol solution). The percentage of dentin slice surface resorbed was determined using image analysis software.
In vivo examination of the osteoclastogenic activity of OABs in parietal bones
To determine the potential osteoclastogenic activity of OABs in vivo, they were injected in a calvarial model of osteoclastogenesis both in the presence and absence of specific inhibitors. Twenty-eight female 5-wk-old C57Bl/6 mice were divided into seven treatment groups (n = 4 animals per group) and given one subcutaneous injection comprising either vehicle (PBS) or treatment (50 μl total volume) onto the periosteum on the right side of the parietal bones. Treatment groups included 5 × 104 MLO-Y4 OAB/animal; 5 × 104 MLO-Y4 OAB + 100 ng/ml OPG/animal; 5 × 104 MLO-Y4 OAB + 0.1 μg/ml anti-TNF-α/animal; 5 × 104 primary osteocyte AB/animal; 5 × 104 primary osteoblast AB/animal; 50 μl anti-TNFα control (0.1 μg/ml); or 50 μl PBS (vehicle control). After 5 days, the animals were killed by CO2 inhalation, and the parietal bones were removed and flash-frozen in hexane at −80°C. Seven-micrometer cryosections sections were prepared over the whole distance across the parietal bone from each treatment group. Four sections were taken for analysis at 0.5-mm intervals from each treatment group. The sections were stained for TRACP and counterstained with DAPI nuclear stain. Six images were collected for each section, and the resorbed surface was calculated on the bone formation surface against the total length of that surface using NIH Scion Image Analysis software.
Images of histological sections of calvariae (Fig. 2A) and surface imaging of resorption lacunae on dentin slices (Fig. 3G) were captured using an upright microscope (Nikon Eclipse E800) with a 20×/0.50 NA objective and a DXM1200 camera (Nikon). Images of TRACP+-stained cells (Fig. 3E) were acquired using an inverted microscope (Nikon Eclipse TS100) with a 10×/0.30 NA objective and a DXM1200 camera (Nikon). SEM images (Fig. 1) were acquired using a Zeiss DSM962 with a Kontron IBAS external control computer, which was operated at 20 kV using a 0.5 NA and an annular solid state BSE detector (KE Electronics), as previously described.(29)
For the in vivo experiment, statistical differences between treatment groups was determined using a one-way ANOVA followed by Tukey-Kramer test posthoc; the error bars represent the mean ± SE from n = 4 animals per treatment group.
For the in vitro experiments, the results presented are representative of n ≥ 3 experiments, each with n = 4 replicates per treatment group with the error bars representing mean ± SD. Statistical differences between treatment groups was determined using one-way ANOVA followed by the Tukey-Kramer test posthoc.
Characterization of primary murine osteoblasts and osteocytes
Cell fractions obtained from calvarial digestions 2–5 had the morphological characteristics of osteoblasts, stained positive for alkaline phosphatase, and formed mineralized nodules as identified by von Kossa staining after 21 days in culture with osteogenic media (data not shown). Calvarial digestions above fraction 6 contained cells that showed the morphological features of osteocytes and lacked proliferative capacity as determined by BrdU incorporation (data not shown).
Characterization of apoptotic bodies
The primary murine and MLO-Y4 osteocyte cell line and primary murine osteoblasts were induced to undergo apoptosis in vitro when cultured in the absence of serum. The ABs were collected and characterized based on the exposure of phosphatidyl serine on the plasma membrane detected with fluorescently labeled Annexin V. The staining was combined with propidium iodide (PI) to allow differentiation between apoptotic, secondary necrotic, and necrotic cells. Although >80% of ABs were Annexin V positive and <5% were positive for PI alone at this stage, the ABs were further purified based on Annexin V binding to ensure the absence of necrotic cells. The size of the ABs varied in diameter, having an approximate range of between 20 nm to 1 μm (Figs. 1E and 1F). In addition, we observed that, for an equal number of ABs from each cellular source, we obtained an equal amount of protein. ABs derived from primary osteocytes contained 0.21 ± 0.01 μg/ml protein; MLO-Y4 osteocyte ABs contained 0.22 ± 0.01 μg/ml protein; and primary osteoblast ABs contained 0.23 ± 0.01 μg/ml protein.
OABs induce bone resorption in vivo
To determine whether OABs can directly induce bone resorption in vivo, purified OABs derived from both primary murine osteocytes and the MLO-Y4 osteocyte cell line were injected subcutaneously over mouse calvarium. This in vivo calvarial injection model is an established technique that has been used previously to show the local effects of interleukin 1 (IL-1) on bone resorption(30) and TGF-β on appositional bone growth.(31)
OABs were injected directly onto the formation (external) side of the right parietal bone of the mouse calvarium that, under normal circumstances, lacks resorptive activity. Within 5 days of application, primary (pr.OAB) and MLO-Y4 cell line–derived OABs were shown to induce bone resorption on the previously bone-forming surface as evidenced by the presence of TRACP+ osteoclasts and associated bone resorption lacunae (Figs. 2C–2E).
In contrast, calvarial surfaces injected either with PBS vehicle control or primary osteoblast ABs (Ob.ABs) appeared smooth and covered by lining cells with no evidence of TRACP+ osteoclasts (Figs. 2A and 2B).
Quantification of the extent of the resorption surface on pr.OAB and OAB-treated calvariae showed a significant increase in resorption across the right side of the parietal bone (Fig. 2F). No resorption foci where seen on the noninjected (left) side of the pr.OAB- and OAB-treated parietal bones (data not shown), indicating that OABs induced stimulation of osteoclast activity. Our experiments showed the ability of OABs but not ABs from a related cell source, primary osteoblasts, to induce the activity of osteoclastic bone resorption in a targeted manner in vivo.
Apoptotic osteocytes support the population of bone marrow resident OPs and engender the formation of active osteoclasts in vitro
The survival of bone marrow OPs and their subsequent fusion to form mature osteoclasts has been shown to involve supportive factors expressed by stromal/osteoblastic lineage cells, namely M-CSF and RANKL.(32,33) To characterize the ability of OABs to promote osteoclastogenesis, OP cells were co-cultured for 72 h in vitro with or without OABs derived from primary osteocytes or MLO-Y4 osteocyte cell line cultures, in the absence of exogenous M-CSF and RANKL.
In comparison with OP cultures that were fixed 24 h after isolation from the bone marrow (T0 control), OP cell number decreased by 35% in nontreated control cultures over the 72-h experimental period (Fig. 3A). In contrast, after 72 h, cell number was maintained at T0 levels in cultures treated with both RANKL and M-CSF (Fig. 3A), highlighting the known dependency of osteoclastogenesis on these factors and the diminishing nature of the untreated cultures.
Interestingly, addition of either primary or MLO-Y4 cell line–derived OABs induced a dose-dependent increase in both the OP cell number (Figs. 3A and 3C) and the proportion of TRACP+ cells (Figs. 3B and 3D) after 72 h in culture relative to the untreated 72-h controls. Both primary (Figs. 3A and 3B) and cell line–derived OABs (Figs. 3C and 3D) at the lowest test density of 50,000 OAB/ml maintained the OP cell population at T0 control levels and increased the proportion of TRACP+ cells above those in T0 cultures, indicating a cell supportive and differentiation activity of OABs. The highest primary OAB density used (200,000 OAB/ml) dramatically increased both OP cell number and TRACP+ cells 2.6- and 5-fold, respectively, compared with T0 control and 4.3- and >40- fold compared with the precursor-depleted 72-h controls, indicating the presence of a strong osteoclastogenic activity (Figs. 3A, 3B, and 3G). In a similar way, we observed that the highest MLO-Y4 cell line–derived OAB density of 200,000 OAB/ml also dramatically increased both OP cell number and TRACP+ cells 2- and 8-fold, respectively, compared with T0 control and 3- and >40-fold compared with the precursor depleted 72-h controls, indicating the presence of a strong osteoclastogenic activity (Figs. 3C, 3D, and 3F). Moreover co-culture of OP with primary or cell line–derived OABs at the highest density enhanced the formation of TRACP+ cells by 2.5- and 2-fold, respectively, compared with OPs treated with M-CSF and RANKL (Figs. 3B, 3D, and 3H).
Having determined the effects of OABs on osteoclast formation from OPs, we went on to test the ability of the cells generated in this way to undertake authentic bone resorption, as would be expected from mature bone-resorbing osteoclasts. When co-cultured with either primary or MLO-Y4 cell line–derived OABs on dentin slices for 72 h, OP cells underwent both increased osteoclastic differentiation and increased resorptive function (as evidenced by measures of total resorption area) compared with that of untreated controls (Figs. 4A–4D).
To study the relevance of the mode of osteocyte death in the response, we co-cultured OPs with necrotic osteocyte cells. There was no evidence of an osteoclastogenic response (Fig. 5) or a resorptive response (Fig. 4A) when OPs were co-cultured with necrotic osteocytes or soluble products released in these cultures (Figs. 4A, 4G, and 5), indicating that the osteoclastogenic activity was associated specifically with the apoptotic osteocyte rather than just the dying cell.
In vitro osteoclastogenic activity is not associated with apoptotic osteoblasts
To determine the importance of soluble factors associated with the dying osteocytes and/or OABs in the osteoclastogenic response, OPs were cultured in the presence of media conditioned by primary or cell line–derived apoptotic osteocytes (aOCM). After 72 h of culture, aOCM did not induce an increase in either the proportion of TRACP+ cells (Fig. 5) or their resorptive activity compared with the T0 control (Fig. 4A), pointing to the existence of an OAB-associated signal, either membrane bound or encapsulated within the vesicle itself.
OABs induce osteoclast formation in the presence of OPG in vitro and in vivo
The preceding data show the ability of OABs to specifically modulate osteoclast formation and function, whereas the OAB dose response studies suggest that osteoclast formation was dependent on increasing concentrations of OAB-associated molecules. Expression of RANKL and its soluble inhibitor OPG by stromal/osteoblastic cells in the bone microenvironment has been shown to play a key role in osteoclastogenesis.(33,34) RANKL was found to be present on the surface of both healthy MLO-Y4 and, to a moderate extent (based on immunocytochemistry), ABs derived from MLO-Y4 (Fig. 6). Interestingly, as also shown here, healthy cultures of the MLO-Y4 osteocyte cell line have been shown to be capable of promoting osteoclast formation when in contact with OPs through a RANKL-dependent mechanism,(20) whereas soluble factors released from healthy osteocyte cultures have not been associated with osteoclastogenic activity (Figs. 4A and 5).(20)
To test the hypothesis that apoptotic osteocytes can support osteoclastogenesis in a similar RANKL-dependent manner, we exposed co-cultures of OPs and OABs (100,000 OAB/ml) to a range of concentrations of the RANKL inhibitory decoy receptor OPG (50–400 ng/ml; Fig. 7). We observed that, whereas osteoclastogenesis induced by healthy osteocytes was inhibited by OPG at concentrations of 50 ng/ml and above (Fig. 7A), osteoclast formation engendered by either primary (Fig. 7B) or cell line–derived (Fig. 7C) OABs (100,000 OAB/ml) was not reduced by OPG treatment across this concentration range, indicating a potential RANKL-independent mechanism of action. To substantiate these data, the OPG concentration range was extended to 50–1000 ng/ml in MLO-Y4 cell line OAB experiments (Fig. 7C), showing no inhibition of the osteoclastogenic response in the presence of these high (>400 ng/ml) OPG concentrations. In addition, a wider range of OAB concentrations down to 6250 OAB/ml were incubated with a fixed known anti-osteoclastogenic concentration of OPG (100 ng/ml).(20) OAB concentrations as low as 25,000 OAB/ml significantly stimulated osteoclastogenesis (p = 0.002), and a clear dose dependency was noted (Fig. 7D). OPG at a concentration of 100 ng/ml was incapable of blockade of the OAB engendered response at any of the stimulatory OAB concentrations used (Fig. 7D), indicating an inability of OPG to block the OAB-induced osteoclastogenic response. Furthermore, we observed that in vivo injection of murine calvaria with OABs in the presence of OPG at 100 ng/ml did not prevent the resorption induced by OABs alone on the previously bone-forming surface of parietal bones (Fig. 7E).
OABs induce osteoclast formation in a TNF-α–dependent manner
In an attempt to identify alternative osteoclastogenic molecular pathways that might be used by OABs, we studied the role of the multifactorial cytokine TNF-α. TNF-α has been shown to either cooperate with RANKL(35,36) or to work independently of RANKL to drive osteoclast formation.(37,38) The co-addition of a TNF-α–specific blocking antibody did not abrogate the osteoclastogenic effects of healthy osteocytes (Fig. 8A). In contrast, osteoclastogenesis driven by either primary (Fig. 7B) or cell line (Fig. 7C) –derived OABs was significantly reduced by all concentrations of anti-TNF-α in a dose-dependent manner. Co-administration of anti-TNFα and OABs in vivo significantly reduced the resorption on the bone-forming surface (Fig. 8D). Cytokine analysis of both aOCM and lysed OAB medium indicated that the source of TNF-α production was not the apoptotic osteocytes or the OAB, pointing to the likely production of TNF-α by OP target cells on interaction with OABs (Fig. 8D). Because TNF-α is implicated in the regulation of osteoclast formation and resorptive activity,(39,40) our findings suggest that OABs are stimulating local release of TNF-α by OP cells, thereby targeting bone resorption to specific anatomical sites in bone.
Previous studies in vivo have shown that the targeted destruction of damaged bone was preceded by osteocyte apoptosis in the region of damage, suggesting that signals released as a result of osteocyte apoptosis might initiate the process of targeted bone resorption.(3) Here we have, for the first time, shown a stimulatory effect on osteoclastic bone resorption after the localized application of primary or MLO-Y4 osteocyte–derived ABs on murine parietal bones in vivo and in further studies using osteoclast precursor cultures in vitro.
The application of purified ABs derived from both primary murine osteocytes and the MLO-Y4 murine osteocyte cell line had a dramatic stimulatory effect on the number and the differentiation state of osteoclastic precursor cells isolated from murine bone marrow. The magnitude of both the proliferation and the differentiation response in the osteoclastic precursors was shown to be dependent on the numbers (density) of ABs applied over a range of 25,000–200,000 OAB/ml. This density range is equivalent to between 0.6 and 5 apoptotic bodies per osteoclast precursor cell, and interestingly, is similar to the density of ABs needed to engender a robust response in professional phagocytes.(41,42) A stimulation of bone resorption was also initiated 5 days after application of OABs to murine calvariae in vivo. Osteoclastic cells and classical Howship's lacunae (resorption pits) were evident in treated bones and points to a genuine osteoclast-mediated bone resorptive response.
The apoptotic osteocyte-derived particles used were shown to resemble classical ABs at both the morphologic and the biochemical level, having been purified on the basis of apoptosis-specific surface markers. We went on to study the requirement for the apoptotic rather than the necrotic mode of death in this response. Challenge of osteoclast precursors with necrotic osteocytes did not engender an osteoclastogenic response, pointing to the likely importance of the apoptotic process or the apoptotic particle in the response. The existence of phagocytic macrophage responses that are specific to apoptotic rather than necrotic cells is well established(43,44) and includes an increase in the release of anti-inflammatory cytokines TGF-β and IL-10 and a decrease in the release of the pro-inflammatory cytokine TNF-α.(41,45) The importance of such a specific response in bone cells has not previously been reported.
The OAB-mediated response in osteoclast precursors was shown to be independent of the source of murine osteocytes used. ABs derived from both the primary and the SV40 transformed cell line produced an identical response both in vivo and in vitro, providing reassurance as to the authenticity of this activity in genetically “normal” cells.
Fundamental to the potential physiological meaning of these data is the finding that ABs derived from the related but phenotypically distinct osteoblast cells did not engender a precursor-supportive or osteoclastogenic response in the osteoclast precursor cell cultures. In line with these findings, we showed that, in contrast to OABs, osteoblast-derived ABs did not promote osteoclastic resorption of bone in our in vivo model. Whereas the phenotypic specificity of this response has not been the primary target of our experiments, it does lead us to a number of important observations. First of all, the death of an osteocyte might have a very specific physiological meaning in bone. Furthermore, the osteoclastogenic response to OABs in vivo is unlikely to be caused by inflammation engendered by the allogeneic nature of the injected material, because the primary and transformed osteocyte cells engendered the same response, whereas the primary osteoblasts from the same source as the primary osteocytes engendered none.
The equivalence in protein content of the ABs from all cell sources used indicates that some phenotype-related molecular specificity was responsible for the observed responses. The authors are unaware of any previous direct demonstration of the phenotypic specificity in the response of a phagocyte to an AB. However the knowledge that ABs are capable of carrying on their surface, molecules that relate to their cell of origin, would tend to support this possibility.(46) Indeed it opens up an intriguing area of future research endeavor. The applicability of a signaling system that can distinguish between the death of two closely related cell types in the highly heterogeneous cell environment in bone is clear. It would seem logical and/or biologically efficient for there to exist a differential response to the production of ABs by cell phenotypes in a tissue where apoptosis is a common phenomenon.
The data collected here indicate that the pro-osteoclasto- genic activity associated with osteocyte apoptosis is not present in the soluble fraction. It would seem that the activity is associated with the AB itself. ABs are associated with a range of molecules carried on their surface, some of which are known to facilitate specific responses in phagocytes.(41,45,47,48) Whereas a wide range of molecules exist within the membrane-bound AB, the range and physiological significance of these remains to be determined. In addition, it has been recognized that the molecules carried by ABs are capable of modifying immune responses and potentially contributing to various autoimmune diseases.(49,50) In the studies described here, we did not distinguish between a requirement for a physical interaction of the AB with the phagocyte or a full phagocytosis of the particle for initiation of the osteoclastogenic response, although both phenomena were seen in our cells.
In a similar way, factors present in the soluble fraction derived from both necrotic and healthy osteocytes were incapable of promoting osteoclastogenesis.
On the other hand, the physical contact of osteoclast progenitors with healthy osteocytes was shown to promote formation of functional osteoclasts. These data are in agreement with the work of Zhao et al.,(20) in which healthy MLOY4 osteocytes promoted osteoclastogenesis in a contact and RANKL-dependent manner. Here we also confirm the RANKL dependency of this response.
However, in contrast to these data, the osteoclastogenic action of osteocyte apoptotic bodies seemed to occur in a RANKL-independent manner, despite moderate evidence of the presence of RANKL on both healthy osteocytes and purified osteocyte ABs. Application of the RANKL inhibitor and decoy receptor of RANKL, OPG, at a wide range of concentrations including and far exceeding at least 10 times higher than those known to confer inhibition of RANKL activity in osteoclast precursor cultures failed to block the osteoclastogenic activity of OABs. In addition, known inhibitory concentrations of OPG could not block the osteoclastogenic effect of any of the OAB concentrations used, including the minimal number needed for any response. The inability of OPG at 100 ng/ml to block the response to OABs was also shown in our in vivo model. This in vivo model has been used previously to determine the osteoclastogenic activity of a number of compounds.(30,31) Whereas the OPG was administered at a concentration known to block osteoclastogenesis in vivo and in vitro, it is possible that extremely high doses might block the effect, although the known 1:1 relationship between OPG and RANKL blockade would suggest that extremely high RANKL concentrations would have to be present for this to be the case. The cumulative evidence derived from our in vitro and in vivo findings would suggest a RANKL-independent mechanism of action for OAB. Similar RANKL independent responses have been noted in osteoclast progenitors and have involved the multifactorial cytokine TNF-α,(37,38) which has been characterized as an important mediator of osteoclastogenesis that may be implicated in both physiological and pathologic bone destruction.(36) Here we found within 72 h of OAB application a large increase in TNF-α concentration in the media of osteoclast precursor cultures. Importantly, measurement of TNF-α levels in media containing either apoptotic bodies or osteoclast progenitors alone did not show significant levels of the cytokine. These data suggest that the source of TNF-α in the OAB-stimulated cultures is unlikely to be the OABs themselves and is probably the stimulated osteoclast precursor cells, pointing to the production of TNF-α in the progenitors on stimulation with OABs. The concentration of TNF-α in the bone microenvironment in vivo has been estimated to be ∼100–200 ng/g and increase up to 600–1500 ng/g on stimulation with pathogenic agents.(51) In the bone microenvironment, cell sources that are likely candidates for TNF-α production include the osteoblastic cells,(40) adipocytes,(52) endothelial cells,(53) T lymphocytes, and the monocytes/macrophages,(54) which have been shown to affect bone resorption in vivo, whereas studies have also indicated the capacity of TNF-α to regulate osteoclastic activity in an autocrine fashion in vitro.(55) Importantly, addition of a TNF-α–neutralizing antibody both in vitro and in vivo significantly blocked the OAB-related osteoclastogenic and the bone resorptive response. This response was shown to be dose dependent in vitro and capable of completely blocking TNF-α stimulation in the OP cultures and the formation of TRACP+ cells. In vivo, the blockade of calvarial resorption engendered by OAB was in the order of 70%. Whereas the absolute role of RANKL in the osteoclastogenic response to OABs cannot be entirely excluded, the likelihood is the majority of the response is related to TNF-α activity.
The likelihood that either live, RANKL-expressing, or dying AB-producing osteocytes can communicate such osteoclastogenic signals in vivo should be considered. In principle, cell to cell contact–dependent RANKL delivery to marrow cavity resident osteoclast progenitor cells by live osteocytes is only possible at the canalicular endings/openings on the bone surface and is only likely to influence cells that are already adhered to the bone surface. Because the majority of the bone surface is covered by a layer of mesenchymal lineage cells, it would seem unlikely but not impossible that osteoclast precursors would frequently encounter such a signal. In contrast, OABs might potentially carry signals to more remote sites and will not involve a cell to cell contact dependency to do so. Measurements based on our SEM images show that the diameter of the ABs produced in our experiments ranged from 20 nm to 1 μm. The methodologies used do not preclude the possibility that we have not measured the smallest of the ABs that might have been either lost in the preparation of material for SEM and below the resolution of the technique. In principle, significant numbers of these would be capable of traveling within the lacuna/canalicular system in bone because the canaliculi are up to 700 nm diameter in the mouse(56) and would become devoid of cell processes (that take up ∼60% of the diameter) and pericellular matrix on death of the osteocytes within them. In addition, the mechanically driven flow of fluid would act to drive particles through these narrow bore tubes where significant fluid pressures are calculated to exist.(57) Hence, we would consider it unlikely that the small ABs produced by osteocytes would all remain resident in osteocytic lacunae. On the other hand, others have noted within a small proportion of lacunae the accumulation of mineralized nodules forming a micropetrosis.(58) It is possible that the micropetrotic particles are derived from ABs or necrotic particles that have not been cleared rapidly from the canalicular/lacuna system, pointing to some level of inefficiency in the clearance system under as yet undefined circumstances.
In conclusion, the in vivo and in vitro data presented here support the hypothesis that the presence of apoptotic osteocytes might dramatically alter the bone environment in terms of localized resorptive activity. In this way, the site-specific apoptotic death of osteocytes might underlie the mechanism by which targeted remodeling is initiated in bone. Such a precise stimulation of osteoclastic bone resorption would provide a further meaning to the apoptotic process in bone, beyond its ability to remove dead/damaged cells from a tissue.
This work was sponsored by SHEFC and a College of Medicine and Veterinary Medicine studentship, University of Edinburgh. We thank Alistair Gracie, Division of Immunology, Infection and Inflammation, Royal Infirmary, Glasgow, for the cytokine analysis.
- 212003 Generating murine osteoclasts from bone marrow. Methods Mol Med 80: 129–144., , ,