Costimulation of Murine Osteoblasts with Interferon-γ and Tumor Necrosis Factor-α Induces Apoptosis through Downregulation of Bcl-2 and Release of Cytochrome c from Mitochondria

During chronic inflammation from diseases, such as periodontal disease, the proinflammatory cytokines interferon-gamma (IFNγ) and tumor necrosis factor-α (TNFα) alter bone remodeling. To elucidate the underlying molecular mechanisms, we investigated the effect of IFNγ and TNFα on the proliferation and survival of clonal MC3T3-E1 mouse osteoblasts. We found that although IFNγ or TNFα alone affected cell growth and survival only marginally, costimulation with both synergistically inhibited cell growth and reduced cell viability. The diminished cell viability was due to apoptosis, as indicated by increased TUNEL staining and elevated caspase 3, 8, and 9 activities. Western blot also showed that costimulation with IFNγ and TNFα elicited cytochrome c release and downregulated B cell lymphoma 2 (Bcl-2) expression without affecting Bcl-2-associated X (Bax) protein expression. Furthermore, stable Bcl-2 overexpression significantly alleviated cell death following costimulation. Collectively, these results suggested that IFNγ and TNFα elicited osteoblast apoptosis via cytochrome c release from damaged mitochondria, caspase activation, and Bcl-2 downregulation.

TNFα is primarily produced by macrophages and T cells but exerts diverse effects on many cell types [15,16]. For example, TNFα inhibits bone formation via multiple mechanisms, including inhibition of osteoblast differentiation and mineralization and suppression of type I collagen synthesis and alkaline phosphatase activity [14,[17][18][19]. TNFα also differentiates precursor cells into osteoclasts and promotes inflammatory bone resorption [20]. On the other hand, IFNγ is produced by T-helper 1 (Th1) cells to promote cell-mediated immunity [21,22]. Additionally, IFNγ suppresses alkaline phosphatase activity in osteoblasts, downregulates bone gla protein [23,24], and promotes differentiation of mesenchymal stem cells into osteoblasts [25]. Unlike TNFα and interleukin-1, IFNγ directly suppresses osteoclast differentiation by interfering with receptor activator of nuclear factor-κB ligand (RANKL) signaling [26,27] but indirectly stimulates osteoclast formation and promotes bone resorption through production of RANKL and TNFα by stimulating T cell activation [28].
2.3. Cell Proliferation. Cells were seeded on 3 cm culture plates (Becton Dickinson) at 5 × 10 4 cells per plate and cultured for 24 h at 37°C and 5% CO 2 in α-MEM with 10% FBS, followed by treatment with IFNγ and TNFα for various durations. The cells were then detached from cell culture plates using 0.1% actinase E and counted on a hemocytometer (Erma, Tokyo, Japan).

Cell Viability.
Cell viability was measured using a cell counting kit (Dojindo, Kumamoto, Japan) based on the formation of water-soluble formazan from the tetrazolium salt WST-8 (2-[2-methoxy-4-nitrophenyl]-3-[4-nitrophenyl]-5-[2,4-disulfophenyl]-2H-tetrazolium, monosodium salt) [39]. Briefly, cells were seeded at 1.5 × 10 3 cells per well in a 96-well microplate (Becton Dickinson), cultured for 24 h at 37°C and 5% CO 2 in α-MEM with 1% FBS, and treated with IFNγ and TNFα for various durations. The cells were then stained with 10 μl WST-8 reagent for 2 h at 37°C and 5% CO 2 and assayed at 450 nm on a microplate reader (Multiskan Bichromatic; Labsystems Diagnostics, Helsinki, Finland). Cell viability was normalized to that of unstimulated cells, which was set at 100%. Cytotoxicity was then calculated as the ratio of cell viability of unstimulated cells to that of cytokine-treated cells. Cell viability and cytotoxicity were also measured in the same manner in MC3T3-E1 cells precultured for 5 days to form a confluent monolayer.

DNA Fragmentation by Terminal Deoxynucleotidyl
Transferase dUTP Nick-End Labeling (TUNEL) Assay. The Apo Alert DNA fragmentation assay kit (BD Biosciences, San Jose, CA, USA) was used for TUNEL assays [40]. Briefly, cells were seeded in each well of a Lab-Tek chamber slide (Nalge Nunc International, Rochester, NY, USA) at 6 × 10 4 cells per well, cultured for 72 h to form a monolayer, and treated with IFNγ and TNFα for various durations. The cells were subsequently washed with PBS, fixed using 5% acetic acid in ethanol, washed again with PBS, and treated with 0.2% Triton X-100 for 5 min. After washing, the cells were equilibrated for 10 min at room temperature in 50 μl equilibration buffer supplied with the kit, followed by staining at 37°C for 1 h with 45 μl equilibration buffer, 5 μl nucleotide mix, and 1 μl TdT enzyme, which incorporates FITClabeled dUTPs onto the 3 ′ -terminal hydroxyl group of DNA fragments. Nuclei were then counterstained with propidium iodide (PI), and TUNEL-stained (apoptotic) cells were enumerated using a confocal laser microscope (Carl Zeiss, Oberkochen, Germany) at an excitation wavelength of 494 nm and an emission wavelength of 518 nm. The number of apoptotic cells was normalized to the total number of PI-stained cells.
2.6. Caspase Activity. Caspase activity was measured using an APOPCYTO caspase colorimetric assay kit (Medical & Biological Laboratories, Nagoya, Japan). Briefly, cells were seeded on 10 cm tissue culture plates at 5 × 10 5 cells per plate, cultured for 72 h at 37°C and 5% CO 2 in α-MEM with 10% FBS to form a monolayer, and exposed to IFNγ and TNFα for various durations. The cells were subsequently washed with PBS, harvested with a cell scraper (Costar, Corning, NY, USA), transferred to a 1.5 ml centrifuge tube on ice, and centrifuged for 3 min at 1700g and 4°C. The resulting pellet was resuspended in 100 μl cell lysis buffer, gently mixed, placed on ice for 10 min, and centrifuged for 5 min at 9000g and 4°C to obtain cell extracts. The extracts were assayed for total protein by the Bradford method [41] using a commercially available reagent (Bio-Rad, Hercules, CA, USA). After diluting extracts to a uniform concentration of protein, caspases were assayed for 1 h at 37°C using DEVD-pNA, IETD-pNA, and LEHD-pNA, which are cleaved by caspase 3, caspase 8, and caspase 9, respectively. The amount of liberated p-nitroaniline, a colorimetric marker, was measured by absorbance at 405 nm on a microplate reader (Multiskan Bichromatic; Labsystems Diagnostics). Caspase activity was quantified based on a p-nitroaniline standard curve. The enzymatic activity in unstimulated cells was set to 1, to which the enzymatic activity in cytokine-stimulated cells was normalized.

Mitochondrial Fractionation and Cytochrome c Release.
Cytochrome c release was evaluated according to published methods, with some modifications [42,43]. The cells were cultured as described for the caspase assay, washed three times using ice-cold PBS, washed once with PBS containing 0.02% EDTA, and disaggregated into single cells using 0.1% actinase E. The cells were then pelleted by centrifugation for 3 min at 1700g and 4°C; gently suspended in permeabilization buffer (220 mM mannitol, 68 mM sucrose, 80 mM KCl, 10 mM HEPES (pH 7.4), 2.5 mM EGTA, 1 mM EDTA, 1 mM DTT, 200 μg/ml digitonin, and 1% protease inhibitor cocktail); placed on ice for 5 min; and centrifuged for 5 min at 750g and 4°C. The resulting supernatant was used as the cytoplasmic fraction, while the pellet was resuspended in 50 μl cell lysis buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM EGTA, 2 mM EDTA, 0.2% Triton X-100, 0.3% NP-40, and 1 mM DTT) and centrifuged for 10 min at 8500g and 4°C to pellet nuclei. The resulting supernatant was used as the mitochondrial fraction. Proteins in both fractions were quantified by the Bradford method [41] and adjusted to a uniform protein concentration using sodium dodecyl sulfate (SDS) sample buffer (250 mM Tris-HCl (pH 6.8), 8% SDS, 20% 2-mercaptoethanol, 40% glycerol, and 0.04% bromophenol blue).

Western
Blot. Protein samples obtained as described were subjected to 12% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA) using a semidry transfer cell (Bio-Rad). Membranes were then blocked for 1 h at room temperature with Tris-buffered saline-Tween 20 (TBS-T; 20 mM Tris-HCl (pH 7.4), 137.5 mM NaCl, and 0.1% Tween 20) containing 5% skim milk, washed with TBS-T three times, and probed at 4°C for 12 h with rabbit antibodies to sheep cytochrome c (Sigma-Aldrich), human Bcl-2 (1 : 500 dilution; Santa Cruz Biotechnology, Dallas, TX, USA), and human Bcl-2-associated X protein (Bax; 1 : 500 dilution; Santa Cruz Biotechnology) or with goat antibodies to human β-actin (1 : 500 dilution; Santa Cruz Biotechnology). Subsequently, the membranes were washed three times with TBS-T, labeled for 1 h at room temperature with appropriate secondary antibodies conjugated to , or both (c). Cell viability was determined, and the percentage of cytotoxicity is shown above the symbols. Data represent the mean ± SEM of three independent experiments. The exponentially growing MC3T3-E1 mouse osteoblasts were also seeded in 3 cm dishes, incubated for 24 h, stimulated with or without 10 ng/ml IFNγ and/or 5 ng/ml TNFα, and analyzed at various time points (d), with time 0 indicating the addition of cytokines. Viable cells were counted using a hemocytometer. Data represent the mean ± SEM of three independent experiments. * * p < 0 01 versus untreated cultures by Student's t-test.
2.9. Establishment of Cells Stably Expressing Bcl-2. Cells stably expressing Bcl-2 were established by transforming MC3T3-E1 cells with a Bcl-2 expression vector (pCMV-Bcl-2) and a plasmid with a puromycin-resistance gene (pBabePuro). The former [44] was provided by Dr. Shie-Liang Hsieh (Institute and Department of Microbiology and Immunology, National Yang-Ming University, Taipei, Taiwan), whereas the latter was provided by Dr. Charles S. Tannenbaum (Department of Immunology, Cleveland Clinic Foundation, Cleveland, OH, USA). Briefly, the cells were seeded on 10 cm tissue culture plates at 3 × 10 5 cells per plate in α-MEM containing 10% FBS, cultured for 24 h at 37°C and 5% CO 2 , and transfected with 0.1 μg pBabePuro in PolyFect transfection reagent (Qiagen, Hilden, Germany) along with either 8 μg pCMV-Bcl-2 or 8 μg control vector (pcDNA3; Invitrogen). After 24 h, 2 μg/ml puromycin was added to the media, which were refreshed every 3 days. Antibioticresistant colonies were grown for 2 weeks, isolated using cloning rings (Iwaki, Tokyo, Japan), detached from the tissue culture plate using 0.1% actinase E, and repassaged in 24-well microplates. Twenty colonies each of the cells transfected with pCMV-Bcl-2 or pcDNA3 were harvested, and Bcl-2 protein expression was confirmed by western blot.
2.10. Statistical Analysis. Student's t-test for paired data was used to test for statistically significant differences using Prism 5 software (GraphPad Software, La Jolla, CA, USA). One-way analysis of variance (ANOVA) was used to compare multiple groups. A p < 0 05 was considered statistically significant.

Costimulation with IFNγ and TNFα Suppresses
Proliferation of MC3T3-E1 Mouse Osteoblasts. IFNγ (Figure 1(a)) and TNFα (Figure 1(b)) alone marginally decreased cell proliferation at 0.1 ng/ml, with viability decreasing by only~20%, even at 10 ng/ml. However, stimulation with varying concentrations of IFNγ in the presence of 10 ng/ml TNFα significantly decreased cell viability to~20% at both 1 ng/ml and 10 ng/mL IFNγ (Figure 1(c)). Similar results were obtained at 5 ng/ml TNFα (data not shown). These results indicated that stimulation with either IFNγ or TNFα did not strongly affect MC3T3-E1 cell proliferation, but costimulation with both led to synergistic suppression of cell growth.
Time course analysis following exposure to IFNγ and TNFα (Figure 1(d)) indicated slight suppression of the growth of MC3T3-E1 cells stimulated with either IFNγ or TNFα alone, although cells continued to proliferate, even after 72 h. By contrast, costimulation with both cytokines led to marked suppression of cell growth by 72 h.

Effects of IFNγ and TNFα Costimulation on Confluent
MC3T3-E1 Cells. With fresh media provided every 3 days to prevent nutrient depletion, MC3T3-E1 cells were grown for 5 days to form a confluent monolayer, which exhibited osteoblastic properties [37,45,46]. As measured by WST-8 assay (Figure 2), stimulation of these monolayers with either IFNγ or TNFα alone did not strongly affect cell viability, whereas costimulation with both led to a time-dependent loss of cell viability to between~20% and~25% within 72 h. These results indicated that the costimulation with IFNγ and TNFα was cytotoxic to MC3T3-E1 monolayers.

Costimulation with IFNγ and TNFα Causes DNA
Fragmentation in MC3T3-E1 Cells. To investigate the mechanisms associated with cytotoxicity, confluent MC3T3-E1 monolayers were stimulated with IFNγ and TNFα for 72 h, fixed, and TUNEL stained to detect DNA fragmentation, a hallmark of apoptosis, along with PI staining of nuclei and imaging with a confocal laser microscope (Figure 3(a)). The costimulation with IFNγ and TNFα increased the proportion of TUNEL-stained cells to 80% at 72 h (Figure 3(b)). Additionally, nuclei were more condensed in cells treated with both cytokines than in unstimulated cells (Figure 3(a), middle panel). These results suggested that IFNγ-and TNFα-induced cytotoxicity was a consequence of apoptosis.

Costimulation with IFNγ and TNFα Increases Caspase
Activity in MC3T3-E1 Cells. Apoptosis is induced following activation of cysteine-aspartic proteases (caspases), which cleave various intracellular molecules, including caspaseactivated DNase, which fragments chromosomal DNA [47,48]. Therefore, we investigated whether caspases mediated the apoptotic effects observed following IFNγ and TNFα costimulation. Caspase 3, 8, and 9 activities were measured over time in confluent MC3T3-E1 cells against peptide substrates labeled with a chromogen (Figures 4(a)-4(c)).  We observed no caspase activation within the first 12 h after costimulation with IFNγ and TNFα; however, caspases 3, 8, and 9 were activated after 24 h, with activity significantly increasing after 36 h (p < 0 01). These results implied that the costimulation of MC3T3-E1 cells with IFNγ and TNFα induced apoptosis via caspase activity.

Costimulation with IFNγ and TNFα Increases
Cytochrome c Release and Suppression of Bcl-2 Expression in MC3T3-E1 Cells. Various apoptotic signals are eventually integrated in mitochondria, which then become permeable and release cytochrome c [49][50][51][52][53][54][55][56]. Therefore, cytosolic and mitochondrial fractions from confluent MC3T-E1 monolayers costimulated with IFNγ and TNFα were separated by SDS-PAGE, and the levels of cytochrome c were analyzed by western blot ( Figure 5). Cytochrome c significantly accumulated in the cytoplasm beginning at 12 h and up to 48 h after costimulation, while cytochrome c levels diminished in the mitochondrial fraction. These results suggested that costimulation with IFNγ and TNFα increased mitochondrial membrane permeability and induced cytochrome c release. Additionally, we examined the expression of Bcl-2 and Bax, both of which regulate mitochondrial membrane permeabilization [56][57][58][59]. Bcl-2 expression was stable up to 12 h after costimulation with IFNγ and TNFα but diminished starting at 24 h after stimulation. By contrast, Bax was stable throughout the experiment. These results implied that costimulation with IFNγ and TNFα shifted the equilibrium between Bcl-2 and Bax in mitochondria, resulting in permeabilization of the mitochondrial membrane.

Discussion
Inflammatory bone loss in periodontal disease is due not only to increased bone resorption by osteoclasts but also to suppressed bone formation by osteoblasts [5][6][7][8]. In this study, we examined the effect of the inflammatory cytokines IFNγ and TNFα on the proliferation and viability of MC3T3-E1 mouse osteoblasts and investigated the underlying molecular mechanisms. The data showed that costimulation with IFNγ and TNFα promoted apoptosis, as indicated by increased DNA fragmentation according to TUNEL staining. Nuclear condensation, another hallmark of apoptotic cells, was also observed based on PI staining. Additionally, IFNγ and TNFα costimulation activated caspases, which are effectors of apoptosis, and induced cytochrome c release from mitochondria. Furthermore, the expression of Bcl-2, a protein that regulates mitochondrial membrane permeability, diminished after the IFNγ and TNFα costimulation. Collectively, these findings strongly suggested that IFNγ and TNFα induced apoptosis in MC3T3-E1 cells through mitochondrial damage. Mitochondria are master regulators of apoptosis and contain apoptosis-inducing proteins, such as cytochrome c and Smac/Diablo, between the outer and inner membranes [49][50][51][52][53][54][55][56]. These proteins leak into the cytoplasm when an apoptotic signal permeabilizes the mitochondrial membrane. Subsequently, released cytochrome c binds apoptotic protease-activating factor-1 (Apaf-1) in the presence of ATP, thereby inducing apoptosis by activating caspases 9 and 3, initiator and apoptotic caspases, respectively. Additionally, Bcl-2 proteins control mitochondrial membrane permeability and are subdivided into two major functional groups [56][57][58][59][60][61][62]. The first group includes Bax and Bcl-2 homologous antagonist/killer (Bak1), which promote mitochondrial membrane permeabilization by inducing structural changes in voltage-dependent anion channels in the mitochondrial outer membrane. By contrast, Bcl-2 sequesters Bax to suppress membrane permeabilization. Accordingly, we investigated the kinetics of Bcl-2 and Bax in MC3T3-E1 cells costimulated with IFNγ and TNFα ( Figure 5). The data showed that Bcl-2 expression was suppressed in the mitochondrial fraction after 24 h of stimulation, although no major changes in Bax expression were observed, implying that the costimulation with IFNγ and TNFα might alter the equilibrium between Bcl-2 and Bax, liberate Bax, and promote mitochondrial membrane permeabilization. Conversely, Bcl-2 overexpression significantly alleviated cell death induced by IFNγ and TNFα costimulation ( Figure 6). These results suggested that the apoptotic effects of IFNγ and TNFα were dependent upon Bcl-2/Bax equilibrium. Interestingly, cytochrome c release was observed beginning at 12 h after costimulation, whereas Bcl-2 expression was suppressed at 24 h ( Figure 5). Caspase activation was also observed at 24 h, coinciding with decreased Bcl-2 levels. These results suggested that the initiating event after IFNγ and TNFα costimulation was cytochrome c release from damaged mitochondria, followed by caspase activation and Bcl-2 degradation [42,63]. The initial mitochondrial damage might be due to increased NO concentration, given that IFNγ and TNFα costimulation induces the expression of NO synthase-2 (Nos2) in osteoblasts [29][30][31][32][33]35]. Endogenous reactive oxygen species (ROS) might also contribute to mitochondrial damage [64], because IFNγ induces the expression of NADPH oxidase-1, which generates ROS, such as superoxide [65]. Indeed, preliminary experiments with the antioxidant N-acetyl-L-cysteine partially reduced cell death and caspase activation induced by IFNγ and TNFα costimulation (data not shown). Taken together, these observations suggest that although IFNγ or TNFα alone only modestly affected osteoblast viability, the costimulation with both might boost NO and ROS concentrations, thereby causing mitochondrial membrane damage and cytochrome c release. Alternatively, the costimulation might suppress the expression of intracellular antioxidants, such as superoxide dismutase and catalase, thereby allowing ROS accumulation and promoting mitochondrial damage. In any case, the mechanisms underlying the observed mitochondrial damage will be the subject of future research.
It is important to confirm that the costimulation with IFNγ and TNFα leads to induction of apoptosis in other osteoblastic cell lines. Although few studies have described osteoblastic apoptosis induced by TNFα and IFNγ costimulation in cell lines other than MC3T3-E1, a previous study has shown that stimulation of rat osteoblast cell line ROS 17/28 with a mixture of TNFα, IFNγ, and LPS induced apoptosis [34]. One reason why limited publications of this nature exist is because human osteoblastic cell lines such as MG-63 and Saos-2 were established from osteosarcoma, resulting in the loss of their normal physiological properties, and many tumor cells have been shown to acquire antiapoptotic pathways. In contrast, MC3T3-E1 cells were established from newborn mouse calvaria and have been shown to possess the features of normal osteoblasts. Further study using human osteoblastic cell lines with normal physiological properties will be necessary to corroborate the osteoblastic apoptosis induced by costimulation with IFNγ and TNFα.
Although IFNγ directly inhibits osteoclastogenesis by interfering with the RANKL-RANK signaling pathway and inducing apoptosis mediated by Fas/Fas ligand signaling [27,66], accumulating evidence suggests that Th1-derived IFNγ and TNFα induced by acquired immune response promote net bone loss in pathological conditions such as periodontitis [28,[67][68][69][70]. A mouse model of alveolar bone loss induced by periodontopathic bacteria Porphyromonas gingivalis showed that CD4 + T cell-derived IFNγ and IL-6 were important effectors of bone loss associated with periodontal disease [67]. Peripheral blood mononuclear cells (PBMCs) from patients with chronic periodontitis have been shown to produce large amounts of inflammatory cytokines including IFNγ and TNFα and differentiate to osteoclasts with high resorption activity by RANKL alone [68]. Another study of mouse models of bone loss has demonstrated that IFNγ promotes osteoclast formation through stimulation of antigen-dependent T cell activation and secretion of RANKL and TNFα by T cells [28]. Interestingly, the Th1-derived cytokine has also been implicated in orthodontic tooth movement, which induces local inflammation in the periodontium, by increasing the number of osteoclasts [71]. Besides osteoclastogenesis, increased apoptosis of osteoblasts induced by IFNγ and TNFα has been demonstrated in mouse models of periodontitis [69,70]. These lines of evidence along with our study indicate that IFNγ and TNFα promote net bone degradation through increased bone resorption by osteoclasts and deficient bone formation by osteoblasts in inflamed periodontal tissue.

Conclusion
Here, we showed that costimulation with the inflammatory cytokines IFNγ and TNFα induced apoptosis in MC3T3-E1 osteoblasts via cytochrome c release from mitochondria, caspase activation, and Bcl-2 suppression. These findings advance our understanding of the molecular mechanism driving inflammatory bone loss in periodontal disease.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
No potential conflict of interest was disclosed.