TNF-α Directly Enhances Osteocyte RANKL Expression and Promotes Osteoclast Formation

Osteoimmunology peeks into the interaction of bone and the immune system, which has largely proved to be a multiplex reaction. Osteocytes have been shown to regulate bone resorption through the expression of RANKL in physiologic and pathologic conditions. TNF-α, a product of the immune system, is an important cytokine regulating bone resorption in inflammatory conditions either directly or by increasing RANKL and M-CSF expressions by osteoblasts and stromal cells. The effect of TNF-α on a wide range of cell types has been documented; however, the direct effect of TNF-α on osteocytes has not been established yet. In this study, primary osteocytes were isolated by cell sorting from neonatal calvaria of Dmp1-Topaz mice, which express the green fluorescent protein under the influence of dentin matrix protein 1 promoter. The results show that osteocytes have a significantly higher RANKL mRNA expression when cultured with TNF-α. A co-culture system of osteocytes and TNF receptors I and II deficient osteoclast precursors treated with TNF-α show a significant increase in TRAP-positive cells while cultures without TNF-α failed to show TRAP-positive cells. Additionally, in vivo experiments of TNF-α injected to mouse calvaria show an increase in TRAP-positive cell number in the suture mesenchyme and an increase in the percentage of RANKL-positive osteocytes compared to PBS-injected calvaria. Osteocytes cultured with TNF-α show up-regulation of MAPKs phosphorylation measured by western blot, and adding MAPKs inhibitors to osteocytes cultured with TNF-α significantly decreases RANKL mRNA expression compared to osteocytes cultured with TNF-α alone. We also found that TNF-α activates the NF-κB pathway in osteocytes measured as a function of p65 subunit nuclear translocation. TNF-α directly affects osteocyte RANKL expression and increases osteoclastogenesis; our results demonstrate that osteocytes guard an important role in inflammatory bone resorption mediated by TNF-α.

Osteoimmunology peeks into the interaction of bone and the immune system, which has largely proved to be a multiplex reaction. Osteocytes have been shown to regulate bone resorption through the expression of RANKL in physiologic and pathologic conditions. TNF-α, a product of the immune system, is an important cytokine regulating bone resorption in inflammatory conditions either directly or by increasing RANKL and M-CSF expressions by osteoblasts and stromal cells. The effect of TNF-α on a wide range of cell types has been documented; however, the direct effect of TNF-α on osteocytes has not been established yet. In this study, primary osteocytes were isolated by cell sorting from neonatal calvaria of Dmp1-Topaz mice, which express the green fluorescent protein under the influence of dentin matrix protein 1 promoter. The results show that osteocytes have a significantly higher RANKL mRNA expression when cultured with TNF-α. A co-culture system of osteocytes and TNF receptors I and II deficient osteoclast precursors treated with TNF-α show a significant increase in TRAP-positive cells while cultures without TNF-α failed to show TRAP-positive cells. Additionally, in vivo experiments of TNF-α injected to mouse calvaria show an increase in TRAP-positive cell number in the suture mesenchyme and an increase in the percentage of RANKL-positive osteocytes compared to PBS-injected calvaria. Osteocytes cultured with TNF-α show up-regulation of MAPKs phosphorylation measured by western blot, and adding MAPKs inhibitors to osteocytes cultured with TNF-α significantly decreases RANKL mRNA expression compared to osteocytes cultured with TNF-α alone. We also found that TNF-α activates the NF-κB pathway in osteocytes measured as a function of p65 subunit nuclear translocation. TNF-α directly affects osteocyte RANKL expression and increases osteoclastogenesis; our results demonstrate that osteocytes guard an important role in inflammatory bone resorption mediated by TNF-α.

INTRODUCTION
The bone is a dynamic tissue, it is continuously broken down and built through the process of bone remodeling in which bone cell populations serve to achieve a balance between episodes of resorption and deposition (1). Osteoclasts, which descend from the hematopoietic stem cell lineage are the sole cells responsible for resorbing bone (2). In bone osteolytic diseases such as rheumatoid arthritis and periodontitis (3), the balance is lost and equilibrium shifts in favor of bone resorption. Molecular signals act together with cellular components to regulate bone resorption, macrophage colony-stimulating factor (M-CSF) is the first in line to induce osteoclastogenesis by binding to its c-fms receptor to promote differentiation and maturation of osteoclast precursors (4). Receptor activator of nuclear factor κB Ligand (RANKL) is a member of the tumor necrosis factor superfamily and secreted by osteoblasts, bone marrow stromal cells (5), and lymphocytes (6). RANKL is obligatory for bone resorption (2). RANKL interacts with its receptor RANK on the surface of osteoclast precursors and drives their differentiation to bone-resorbing osteoclasts, and its effect is diminished by osteoprotegerin (OPG), a soluble decoy receptor secreted by osteoblasts and stromal cells (7). Tumor necrosis factor-α (TNF-α) is one of the most versatile cytokines, it is a product of the immune system secreted by macrophages, and its role in bone inflammatory diseases is well documented as pro-resorptive aiding in disease progression (8). TNF-α produces an array of cellular responses by binding to two receptors TNF receptor I (TNFR I) and TNF receptor II (TNFR II) (9). TNF-α stimulates osteoclastogenesis by increasing the production of M-CSF and RANKL in marrow stromal cells (10,11) and osteoblasts (12) but other researchers reported a mechanism independent of RANKL in which TNF-α directly stimulates osteoclast formation in M-CSF-dependent bone marrow macrophage cultures treated with TNF-α (13), others reported that this happens when cells are primed with RANKL first (14).
Osteocytes are terminally differentiated osteoblasts which became embedded in their secreted matrix (15,16) and account for 90% of the bone cell population, they reside within lacunae and communicate with each other and other cell types through processes extending to the bone surface (17). Osteocytes were shown to function in regulating mineral metabolism, remodeling the perilacunar matrix, and as a mechanosensory cell (18). Recently two groups demonstrated that osteocyte RANKL is the most important in physiologically supported osteoclastogenesis in the developing skeleton (19,20). Osteocyte RANKL was shown to be important postnatally in osteocyte-specific RANKL deficient mice as these mice aged; an osteopetrotic phenotype became increasingly evident. Also, specific deletion of RANKL in osteocytes resulted in mice with a severe osteopetrotic phenotype and protection against loss of bone that accompanies unloading of mechanical force (20). Others reported that deletion of osteocyte RANKL confers protection against infection-induced periodontal bone loss (21) and an increase in cancellous bone mass in osteogenesis imperfecta mice (22). It is clear that osteocyte RANKL influences bone resorption both in health and disease.
An increasing body of research has linked TNF-α to osteocyte RANKL in inflammation-induced bone loss. A study where TNFα antagonist was used resulted in a marked decrease in the number of RANKL-positive osteocytes and osteoclast formation in diabetic rats with periodontitis (23). Blockage of TNF-α in a co-culture of an osteocyte rich fraction and bone marrow cells treated with Pasteurella multocida toxin resulted in a decreased number of RANK positive cells, a marker of osteoclasts (24). TNF-α and osteocyte RANKL are linked to inflammationinduced bone loss, but whether TNF-α has a direct effect on osteocytes is not clear. In this study, we provide evidence that TNF-α can directly affect osteocyte RANKL expression by activation of downstream MAPKs phosphorylation and induces osteocyte osteoclastogenic ability both in vivo and in vitro.

Mice and Reagents
Eight-week-old C57BL6/J wild-type (WT) were purchased from CLEA Japan Inc. (Tokyo, Japan). B6.129S-Tnfrsf1a tm1Imx Tnfrsf1b tm1Imx /J (TNF receptor I, II deficient) mice and C57BL/6-Tg(Dmp1-Topaz)1Ikal/J mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). All animal procedures were performed in accordance with Tohoku University regulations. Recombinant murine TNF-α was prepared in our laboratory as described previously (10). Recombinant mouse M-CSF was purified from an M-CSF expressing cell line (25).

RNA Preparation and Real-Time RT-PCR Analysis
Osteocytes were cultured in α-MEM with or without TNF-α 100 ng/ml for 3 days. To test the effect of MAPKs inhibition, Osteocytes were pre-incubated with 10 µM of SB 203580 (InSolution TM SB 203580 -EMD Millipore, MA, USA) a p38 MAPK inhibitor, 10 µM of U0126 (InSolution TM U0126-EMD Millipore, MA, USA) a MEK1/2 inhibitor targeting ERK1/2 MAPK, and 10 µM of JNK Inhibitor II (InSolution TM JNK Inhibitor II-EMD Millipore, MA, USA) an inhibitor of JNK I, II, and III. Total RNA was obtained from lysed cells by using RNeasy Minikit (QIAGEN, Hilden, Germany) according to manufacturer's instructions. Total RNA was used to synthesize cDNA using the SuperScript R IV First-Strand Synthesis System (Invitrogen TM , CA, USA) according to the manufacturer's instructions. mRNA expression values were measured using the Thermal Cycler Dice Real-Time System (Takara, Shiga, Japan). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a reference gene. The following primers were used: . The purity of osteocytes obtained through cell sorting was checked against the following primers:

Preparation of Osteoclast Precursors and Co-culture System Set-Up
The bone marrow of long bones of WT and TNFR I, II deficient mice was flushed and cultured in α-MEM containing 10% FBS, 100 IU/ml penicillin G, 100 µg/ml streptomycin, and M-CSF 100 ng/ml for 4 days. Attached cells were harvested by trypsinization and used as osteoclast precursors. Osteocytes 2.5 × 10 4 /well were cultured overnight in α-MEM and harvested WT or TNFR I, II deficient osteoclast precursors 5 × 10 4 /well were added on top of osteocytes (2:1), and the co-culture was maintained either without M-CSF, with M-CSF alone (100 ng/ml), TNF-α alone (100 ng/ml), M-CSF+TNF-α (100 ng/ml), or M-CSF+RANKL 100 ng/ml in a 96 well plate. To test the effect of OPG on osteoclastogenesis, osteocytes and TNFR I, II deficient osteoclast precursors were co-cultured with M-CSF+TNF-α (100 ng/ml) or M-CSF+TNF-α+OPG (100 ng/ml), OPG was purchased from R&D Systems (MN, USA). To test the significance of cell-to-cell contact, an osteocyte-conditioned medium was prepared as follows: osteocytes were cultured with M-CSF+TNF-α (100 ng/ml) for 1 day, the medium was then aspirated and added to TNFR I, II deficient osteoclast precursor, as a positive control TNFR I, II deficient osteoclast precursor were cultured with M-CSF+RANKL (100 ng/ml). The co-cultures were maintained for 4 days after which they were terminated by fixation with 4% paraformaldehyde and stained with tartrate-resistant acid phosphatase (TRAP) staining consisting of acetate buffer (pH 5.0), naphthol AS-MX phosphate, fast red violet LB salt, and 50 mM sodium tartrate. Cells were considered osteoclasts if they were TRAP-positive and had 2 or more nuclei.

Histological Examination
WT mice were subjected to subcutaneous supra-calvarial injection of either TNF-α 3.0 µg/100 µl or PBS for 5 consecutive days, on the 6th day the mice were sacrificed and the calvaria was excised. The calvaria was fixed in 4% paraformaldehyde overnight at 4 • C and demineralized in 14% EDTA for 3 days at room temperature. The calvaria was trimmed from soft tissue and cut into 3 pieces perpendicular to the sagittal suture, and dehydrated then embedded in paraffin. Paraffin-embedded calvaria specimens were cut into 5 µm-thick sections using a microtome (Leica Biosystems, Wetzlar, Germany). Histological sections were deparaffinized, rehydrated, and blocked with 3% skimmed milk for 30 min at 37 • C then stained with anti-RANKL antibody (FL-317 SCBT rabbit polyclonal IgG) diluted to 1:50 in can get immunostain solution B (Toyobo, Osaka, Japan) overnight at 4 • C. Sections were washed and incubated with Histofine R Simple Stain TM Mouse MAX PO (R) (Nichirei Bioscience, Tokyo, Japan) for 1 h at room temperature then stained with hematoxylin as a counterstain. RANKL-positive osteocytes were counted as a percentage from the total number of osteocytes per 400 × 400 µm 2 section area. For osteoclast formation examination, calvarial sections from WT mice were deparaffinized, rehydrated, and stained with TRAP staining then counter-stained with hematoxylin. TRAP-positive cells with two or more nuclei were counted and corrected to total bone surface area.

Preparation of Cells for Flow Cytometry
Bone marrow cells (BMC) from WT mice and fractions 2-5 from Dmp1-topaz mouse calvaria were suspended in α-MEM 100 IU/ml penicillin G, 100 µg/ml streptomycin. Cells were washed with 1% BSA in PBS, fixed in 4% paraformaldehyde for 3 min, and washed with 1% BSA in PBS 3 times. Cells were incubated with PE-conjugated anti-TNFR I antibody (CD120a mAb, GeneTex, GTX16198) at 1 µg/ml and PE-conjugated anti-TNFR II (CD120b mAb, BD Biosciences, 550086) at 1 µg/ml for 30 min on ice in the dark and washed 3 times with 1% BSA in PBS. Stained cells were analyzed with FACSAria TM II (BD Biosciences, NJ, USA).

Immunofluorescence
Osteoblasts from fraction 2 and osteocytes were cultured overnight in α-MEM 100 IU/ml penicillin G, 100 µg/ml streptomycin. Cells were washed with PBS, fixed with 4% formaldehyde for 15 min, washed 3 times with PBS and blocked with 3% BSA in PBS for 1 h. Cells were incubated with PEconjugated anti-TNFR I antibody (CD120a mAb, GeneTex, GTX16198), and its isotype control (IgG2a isotype control mAb, GenTex, GTX35058) at 1 µg/ml and PE-conjugated anti-TNFR II (CD120b mAb, BD Biosciences, 550086), and its isotype control (IgG1λ1 isotype control mAb, BD Biosciences, 554711) at 1 µg/ml on ice in the dark for 1 h. Cells were washed 3 times with PBS before imaging. For testing the NF-κB pathway activation, osteocytes were cultured on collagen-coated (Nitta Gelatin Inc., Osaka, Japan) 8-chamber slide (Lab-Tek, Thermo Fisher Scientific, IL, USA) with or without TNF-α 100 ng/ml for 1 h. Cells were washed with PBS, fixed in 4% formaldehyde for 15 min and washed, then permeabilized with 0.1% Triton X-100 in PBS for 10 min, washed and blocked with 3% BSA in PBS for 1 h at room temperature (R/T). Cells were then incubated with anti-p65 antibody at (1:100) in 3% BSA (C-20 SCBT rabbit polyclonal IgG) at 4 • C overnight. Cells were washed with PBS and incubated with Alexa Fluor 555 (goat anti-rabbit  IgG) (1:400) in 3% BSA (Life Technologies, NY, USA) for 1 h in the dark at R/T, then washed and incubated with DAPI for 5 min. Fluorescent cell imaging was done using a fluorescence microscope (Olympus IX71, Tokyo, Japan).

Statistical Analysis
The data were analyzed using student's t-test for two data sets and scheffe's test for multiple data sets. Data were presented as a mean ± standard deviation (SD). Results are representative of three or more independent experiments throughout the paper.

Osteocyte Isolation and Characterization
In order to extract osteocytes from their calcified bony matrix, we processed the calvaria using collagenase and EDTA, and then we run the digests through FACS to separate GFP-positive from GFP-negative populations ( Figure 1A). The isolated cells had a characteristic osteocyte-like phenotype (Figure 1B). PCR analysis of the isolated cells using Dmp1, SOST and Kera confirmed that GFP-positive cells are highly expressive of Dmp1 and SOST which are genes highly expressed in osteocytes, GFPnegative cells showed the opposite with faint expression of Dmp1 and SOST and a significantly higher expression of Kera which is differentially expressed in osteoblasts ( Figure 1C). This result confirms the successful separation of osteocytes from other cell populations.

Osteocytes Express TNF Receptor I and TNF Receptor II
Flow cytometry analysis shows that osteocytes (GFP+) stained for TNFR I and II express both receptors (Figure 2A). Unstained BMC population were used to account for background fluorescence, which shows that BMC population falls below the threshold level for both PE and GFP, while BMC stained for TNFR I and II used as positive controls confirm the expression of both receptors on their surface, and that osteocytes express both TNFR I and II on their surface. Results for immunofluorescence staining using osteocytes stained for TNFR I and II indicated that osteocytes express both receptors on their surface, unstained osteocytes and osteocytes stained with isotype controls do not show any fluorescent activity (Figure 2B). Osteoblasts stained for TNFR I and II also express both receptors while unstained osteoblasts and osteoblasts stained with isotype controls do not show fluorescent activity (Supplementary Figure 1).

TNF-α Induces Osteocyte RANKL Expression in vitro
We examined the change in RANKL, OPG, and M-CSF expression levels in osteocytes upon incubation of osteocytes with TNF-α for 3 days. PCR analysis revealed that osteocytes cultured with TNF-α expressed a significantly higher RANKL mRNA ( Figure 3A) but no difference in OPG expression ( Figure 3B). The ratio of RANKL/OPG was significantly higher in osteocytes cultured with TNF-α (Figure 3C), while M-CSF revealed no difference between osteocytes cultured with or without TNF-α ( Figure 3D). The significantly higher ratio of RANKL/OPG indicates that TNF-α plays a role in osteocyte osteoclastogenic ability through supporting osteoclast precursor survival and differentiation.

TNF-α Induced Osteocyte-Supported Osteoclast Formation in Co-culture
To test whether osteocytes cultured with TNFR I, II-deficient osteoclast precursors in the presence of TNF-α successfully supports the generation of multinuclear TRAP-positive osteoclasts; we cultured osteocytes and TNFR I, II-deficient osteoclast precursor with TNF-α or without TNF-α in the presence of M-CSF. While osteocytes were able to induce osteoclast formation in TNF-α+M-CSF treated wells, osteoclastogenesis failed without the addition of M-CSF even with the addition of TNF-α ( Figure 4A). Co-cultures of WT osteoclast precursor and osteocytes only supported osteoclast formation after adding M-CSF and either TNFα or RANKL (Supplementary Figure 2). To test whether osteoclast formation in co-cultures treated with TNF-α is the result of RANKL expressed by osteocytes, we added OPG to co-cultures, which completely inhibited the formation of osteoclasts, this indicates that osteoclast formation was a result of RANKL expressed by osteocytes treated with TNF-α ( Figure 4B). We also cultured TNFR I, II-deficient osteoclast precursor with an osteocyte conditioned medium treated with M-CSF+TNF-α, the conditioned medium failed to induce osteoclast formation, however adding RANKL to TNFR I, II osteoclast precursor induced osteoclast formation, which highlights the need for membranebound RANKL in osteocytes to induce osteoclastogenesis ( Figure 4C).

TNF-α Induces RANKL Expression in Osteocytes in vivo
TRAP staining of calvaria revealed a higher number of osteoclasts in TNF-α injected calvaria compared to PBS injected calvaria (Figures 5A,B). Immunohistochemical examination of mouse calvaria injected with TNF-α revealed a higher expression of RANKL-positive osteocytes compared to calvaria injected with PBS ( Figure 5C). The number of RANKL-positive osteocytes was significantly higher in the TNF-α injected group compared to the PBS injected group (Figure 5D).

TNF-α Activates the NF-κB Signaling Pathway in Osteocytes
To evaluate NF-κB pathway activation, we relied on immunofluorescence imaging of osteocytes cultured with TNF-α ( Figure 7A). We assessed NF-κB p65 subunit nuclear localization and found that the number of activated osteocytes to total osteocyte number increased significantly in cells treated with TNF-α compared to control (Figure 7B).

DISCUSSION
Osteocytes have gained popularity ever since their isolation from bone became possible (28), from there osteocytes were shown to regulate mineral metabolism (29), remodel the extracellular matrix (30), act as an endocrine cell (31), a mechanosensory cell (32), and activate osteoclastogenesis through osteocyte apoptotic bodies (33). Recently, osteocyte-secreted RANKL has been considered as the most relevant to physiologic bone resorption (19,34). However, it became clear that osteocyte RANKL is expressed in pathologic conditions as well, leading to bone damage in a range of diseases (34,35), as do the findings of this study. In this study, we tested the direct effect of TNFα, an important inflammatory cytokine on osteocytes and we found that TNF-α directly increases RANKL expression as well as osteocyte-aided osteoclastogenesis both in vitro and in vivo. First, we obtained pure cultures of primary osteocytes by cell sorting. To test the purity of the isolated population, we selected Dmp1, a gene predominantly expressed by osteocytes and was reported to exhibit the highest expression level in osteocytes compared to osteoblasts, and Kera, a gene which encodes keratocan and has the highest change in negative expression as osteoblasts mature into their osteocytic phenotype, Dmp1 and Kera were reported to have the highest difference in their expression levels between osteocytes and osteoblasts amongst genes associated with extracellular matrix and secreted proteins (27).We also measured SOST expression which has been shown to be expressed in mature osteocytes as sclerostin, an antagonist for the Wnt/β-catenin pathway (18). Moreover, Dmp1 and SOST have been used as markers of osteocytes while Kera as a marker of osteoblasts (19), thus providing a rational for using the expression of Dmp1, SOST, and Kera to assess the success of osteocyte isolation.
Two cell surface receptors, p55 and p75 TNFRs mediate the action of TNF-α (36). We tested the expression of TNF receptors I and II on osteocyte surface using flow cytometry and immunofluorescent staining, the results of both of these experiments indicate the presence of both receptors on osteocyte surface, these data provided a basis for testing the direct effect of TNF-α on osteocytes bearing TNFR I and II when co-cultured with osteoclast precursors deficient in these two receptors. We set up a co-culture system of osteocytes and WT osteoclast precursors to check the effect of TNF-α or RANKL on osteoclast precursors and osteocytes when both are able to respond to TNF-α to confirm osteoclastogenesis. Then we used osteoclast precursor isolated from TNFR I, II deficient mice which do not respond to TNF-α due to the lack of TNFR I and II on their surface, while osteocytes were isolated from Dmp1-Topaz mice which respond to TNF-α, confirmed by the presence of TNFR I, II on their surface. This set-up allowed us to test the effect of TNFα on osteocytes alone. Cells cultured with M-CSF and TNF-α FIGURE 6 | Effect of TNF-α on ERK1/2, P38, JNK MAPKs, and AKT phosphorylation in osteocytes. Osteocytes were incubated with TNF-α 100 ng/ml for 0, 5, 15, 30, 60 min. 0 indicates that no TNF-α was added. Cells were lysed and analyzed by western blotting, using antibodies for (A) phospho-ERK1/2, ERK1/2, (Continued) FIGURE 6 | (B) phospho-p38, p38 (C) phospho-JNK, JNK (D) phospho-AKT, AKT. β-actin was used as a loading control and band density was measured using Image J software, n = 3. (E) RANKL mRNA relative expression as measured by qRT-PCR. Osteocytes were cultured with TNF-α (100 ng/ml) and MAPKs inhibitors, U0126, SB 203580, and JNK inhibitor II (SP600125) at 10 µM for 3 days. Data are expressed as mean ± SD. Statistical significance was determined by scheffe's test (n = 3, *P < 0.05, **P < 0.01). showed an increase in TRAP-positive activity while cells cultured without TNF-α or with M-CSF alone did not show any TRAPpositive activity. Previous reports mentioned that TNF-α can directly induce osteoclastogenesis independent of RANKL (13), and to test whether TNF-α enhanced osteoclastogenesis due to an increase in RANKL expression or independent of it, we added OPG a decoy RANKL receptor which blocks RANKL/RANK interaction, and it completely inhibited the formation of osteoclasts, thus affirming that osteoclastogenesis mediated by TNF-α in osteocytes is the result of RANKL expression. One study reported that osteocyte have a 10x higher expression of RANKL mRNA compared to osteoblasts in an unstimulated state, however, in co-culture of osteocytes and osteoclast precursor, osteoclasts failed to appear unless stimulated with prostaglandin E 2 (PGE 2 ) and 1,25 dihydroxy vitamin D 3 [1,25(OH) 2 D 3 ], their conclusion was that despite the high expression of RANKL, downregulation of OPG expression is required for efficient osteoclastogenesis (19). Our result demonstrates that the increase in the ratio of RANKL/OPG in cultures stimulated with TNFα is the result of a higher RANKL expression and not the downregulation of OPG expression. Furthermore, similar to published reports, a conditioned medium of osteocytes treated with TNF-α failed to support osteoclast formation, highlighting the importance of membrane-bound RANKL (19). Osteocytes co-cultured with TNFR I, II deficient osteoclast precursor failed to form osteoclasts without M-CSF, even with the addition of TNF-α, which indicates that osteocytes are not a significant source of M-CSF. Our in vitro results also show that osteocytes cultured with TNF-α for 3 days exhibit no difference in M-CSF mRNA expression. Our interest in M-CSF expression in osteocytes stems from the findings of previous studies which noted that TNF-α increases the pool of osteoclast precursors available due to an increase in M-CSF production in stromal cells (11), and that M-CSF levels are increased in the synovial fluid around loose joint prosthesis (37) and in the serum of patients with rheumatoid arthritis (38), both of which exhibit a marked increase in TNF-α expression.
TNF-α was shown to increase the expression of RANKL in osteoblasts, stromal cells (5) and lymphocytes (6), and in this report we show that TNF-α increases osteocyte RANKL expression in vivo and in vitro by regulating MAPKs phosphorylation and possibly NF-κB activation. TNF-α is a potent activator of ERK, JNK and p38 MAPKs, and their interaction with TNF-α has functions upstream and downstream of TNF-α signaling (39). In this report, we tested the downstream inhibition of TNF-α action on ERK1/2, p38 and JNK MAPKs activation via selective U0126, SB203580 and a JNK inhibitor II (also known as SP600125) (40), respectively which attenuated RANKL mRNA expression in osteocytes treated with TNF-α.
NF-κB is involved in the transcriptional activation of inflammatory-related genes induced by TNF-α, and its activation is crucial for the protection of cells from apoptotic cell death induced by the TNFR I cytoplasmic death domain activation. In an unstimulated state, IκB proteins interact with NF-κB units to mask their nuclear translocation, in this report TNFα significantly increased the nuclear presence of the NF-κB p65 unit in osteocytes nuclei after 1 h of treatment. AKT is also a pathway when activated works as an antiapoptotic signal, TNF-α stimulates the AKT pathway in a cell-type-specific manner (41), we found no difference in AKT phosphorylation, however whether the AKT pathway has a role in TNF-α signaling and RANKL expression in osteocytes or not needs further experimentation.
The results of this study can be summarized by asserting the role of osteocytes in their potential contribution to bone destruction in inflammatory bone diseases by increasing RANKL expression. Osteocytes could be a target therapeutic factor in future studies.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Care and Use Committee of Tohoku University.