Infection of RANKL-Primed RAW-D Macrophages with Porphyromonas gingivalis Promotes Osteoclastogenesis in a TNF-α-Independent Manner

Infection of macrophages with bacteria induces the production of pro-inflammatory cytokines including TNF-α. TNF-α directly stimulates osteoclast differentiation from bone marrow macrophages in vitro as well as indirectly via osteoblasts. Recently, it was reported that bacterial components such as LPS inhibited RANKL-induced osteoclastogenesis in early stages, but promoted osteoclast differentiation in late stages. However, the contribution to osteoclast differentiation of TNF-α produced by infected macrophages remains unclear. We show here that Porphyromonas gingivalis, one of the major pathogens in periodontitis, directly promotes osteoclastogenesis from RANKL-primed RAW-D (subclone of RAW264) mouse macrophages, and we show that TNF-α is not involved in the stimulatory effect on osteoclastogenesis. P. gingivalis infection of RANKL-primed RAW-D macrophages markedly stimulated osteoclastogenesis in a RANKL-independent manner. In the presence of the TLR4 inhibitor, polymyxin B, infection of RANKL-primed RAW-D cells with P. gingivalis also induced osteoclastogenesis, indicating that TLR4 is not involved. Infection of RAW-D cells with P. gingivalis stimulated the production of TNF-α, whereas the production of TNF-α by similarly infected RANKL-primed RAW-D cells was markedly down-regulated. In addition, infection of RANKL-primed macrophages with P. gingivalis induced osteoclastogenesis in the presence of neutralizing antibody against TNF-α. Inhibitors of NFATc1 and p38MAPK, but not of NF-κB signaling, significantly suppressed P. gingivalis-induced osteoclastogenesis from RANKL-primed macrophages. Moreover, re-treatment of RANKL-primed macrophages with RANKL stimulated osteoclastogenesis in the presence or absence of P. gingivalis infection, whereas re-treatment of RANKL-primed macrophages with TNF-α did not enhance osteoclastogenesis in the presence of live P. gingivalis. Thus, P. gingivalis infection of RANKL-primed macrophages promoted osteoclastogenesis in a TNF-α independent manner, and RANKL but not TNF-α was effective in inducing osteoclastogenesis from RANKL-primed RAW-D cells in the presence of P. gingivalis.


Introduction
Osteoclasts are large multinucleated cells that are derived from the common progenitor cells of the monocyte/macrophage lineage. Osteoclasts play a central role in bone resorption. Bone remodeling involves cooperation of osteoclasts and bone forming osteoblasts to maintain normal bone volume and calcium homeostasis. However, in inflammatory diseases such as periodontal disease and rheumatoid arthritis, the balance between osteoclasts and osteoblasts is disrupted, and osteoclasts accumulate causing local pathological bone loss.
Differentiation of osteoclasts is regulated by macrophagecolony-stimulating factor (M-CSF) and receptor activator of nuclear factor kB ligand (RANKL) [1]. RANKL, which belongs to the tumor necrosis factor (TNF) family, binds to the TNF family receptor, receptor activator of nuclear factor kB (RANK) expressed on osteoclast precursor cells, and triggers osteoclast differentiation. The signaling mechanisms underlying RANKLinduced osteoclastogenesis have been extensively studied [2].
Binding of RANKL to RANK leads to the activation of NF-kB and mitogen activated protein kinases (MAPKs), including p38 and c-Jun N-terminal kinase (JNK). Deletion of both of the NF-kB subunits p50 and p52 results in defective osteoclast differentiation and causes osteopetrosis in mice, indicating that the NF-kB pathway controls osteoclastogenesis [3]. c-Fos is also activated by RANK and plays an essential role in osteoclast differentiation [4]. NF-kB and c-Fos induce the expression of nuclear factor of activated T cells cytoplasmic 1 (NFATc1), which is auto-amplified during osteoclastogenesis [5]. NFATc1 regulates the expression of osteoclast-specific genes including calcitonin receptor, cathepsin K, and tartrate-resistant acid phosphatase (TRAP) [6]. RANK signaling also induces the expression of genes that inhibit osteoclast differentiation; e.g., IFN-b induced by RANKL inhibits osteoclastogenesis by suppressing c-Fos expression [7].
Periodontitis is a common inflammatory disease, and bone destruction in periodontitis is usually associated with bacterial infections [8,9]. LPS, a major constituent of Gram-negative bacteria and an important TLR4 ligand, has been shown to be a potent stimulator of bone loss in vivo [10]. However, TLR ligands have been shown to have stimulatory or inhibitory functions on osteoclastogenesis in vitro. Simultaneous addition of RANKL and LPS or staphylococcal lipoteichoic acid inhibits differentiation of osteoclasts from bone marrow macrophage (BMM) [11,12]. LPS, in contrast, stimulates the late stage of osteoclastogenesis and enhances the survival and activation of osteoclasts [11,13,14]. Very recently, Zhang et al reported that P. gingivalis, which is implicated in periodontitis, differentially affects osteoclast differentiation from bone marrow macrophages depending on the stage of osteoclast differentiation [15]. In contrast, TLR ligands promote osteoclastogenesis via other cells such as osteoblasts. E. coli LPS and diacyl lipoprotein stimulate the expression of RANKL and IL-6 in osteoblasts through TLRs, and promote osteoclastogenesis in co-cultures of osteoblasts and hematopoietic cells [16,17,18]. LPS also stimulates the production of PGE2 in osteoblasts, which leads to bone resorption [19].
Down-stream signaling pathways of TLRs, other than TLR3, utilize myeloid differentiation factor 88 (Myd88). Myd88 recruits IL-1R-associated kinases leading to the activation of NF-kB and MAPK. Activated NF-kB then induces the transcription of inflammatory genes such as TNF-a and IL-6 [20,21]. P. gingivalis is a Gram-negative bacterial species, but its LPS has a unique chemical structure, and interacts with both TLR2 and TLR4. P. gingivalis LPS weakly activates TLR4 signaling, and its biological activities are primarily mediated via signaling through TLR2 [22]. On the other hand, live P. gingivalis induces cytokines and chemokines such as TNF-a, IL-6, and MCP-1, which signal through both TLR2 and TLR4 [22]. TNF-a is known as a major inducer not only of inflammation but also of bone loss. TNF-a directly acts on BMM exposed to RANKL or transforming growth factor (TGF)-b, and induces osteoclast differentiation in a RANKL independent manner in vitro [23,24]. TNF-a is a multifunctional cytokine, and also participates in restraining inflammation and tissue healing [25,26]. Thus, the role of TNF-a in osteoclastogenesis in bacterial infection is not obvious.
We have previously demonstrated that a macrophage cell line, RAW-D, a subclone of RAW264 has a high capacity to differentiate into osteoclasts [27]. In the present study, we investigated the effect of infection of RAW-D with P. gingivalis on osteoclastogenesis. Our results demonstrate that infection with P. gingivalis markedly stimulated osteoclast differentiation from RANKL-primed RAW-D cells. We found that osteoclastogenesis induced by P. gingivalis infection of RANKL-primed RAW-D cells and BMM was TNF-a independent, and we found that RANKL but not TNF-a was effective in inducing osteoclastogenesis from RANKL-primed RAW-D cells in the presence of P. gingivalis.

Infection of RANKL-primed RAW-D Cells with P. gingivalis Induces Osteoclastogenesis
We first examined whether P. gingivalis infection induced osteoclastogenesis in a mouse macrophage cell line, RAW-D. Although RAW-D has a high potential to differentiate into osteoclasts, P. gingivalis infection alone did not induce osteoclastogenesis in RAW-D cells (data not shown). Because recent studies have shown that LPS stimulates osteoclast differentiation from RANKL-pretreated osteoclast precursors [14], we stimulated RAW-D cells with RANKL for 22 h, then removed the RANKL, and infected the cells with P. gingivalis. Cells were cultured for two more days, and the effect of P. gingivalis infection on osteoclast differentiation was analyzed. After the initial 22 h of culture in the presence of RANKL, i.e., after RANKL-priming, a few mononu-clear cells positive for the osteoclast-specific enzyme TRAP were present, but no TRAP-positive multinucleated cells (MNCs) were observed, and no TRAP-positive MNCs appeared during further culture for 48 h in the absence of RANKL and P. gingivalis (Fig. 1A, left). In contrast, infection of RANKL-primed RAW-D cells with P. gingivalis induced osteoclastogenesis in an infectious dosedependent manner (Figs. 1A right, and 1B). We analyzed mRNA expression levels of several osteoclast-specific genes in unprimed or RANKL-primed RAW-D cells that were infected with P. gingivalis or were uninfected. P. gingivalis infection of RANKL-primed RAW-D cells significantly increased the expression of osteoclastspecific genes such as cathepsin K (ctsk) (Fig. 1C) and calcitonin receptors (calcr) (data not shown) in comparison with uninfected RANKL-primed RAW-D cells. Pretreatment with TNF-a instead of RANKL did not induce osteoclast differentiation (Fig. 1D). Osteoprotegrin (OPG) did not inhibit osteoclastogenesis induced by infection with P. gingivalis (Fig. 1E). Thus, RANKL-pretreatment was necessary, but concurrent presence of RANKL was not required for osteoclastogenesis in RANKL-primed macrophages induced by infection with P. gingivalis. These data indicate that infection of RANKL-primed macrophages with P. gingivalis induced osteoclast differentiation from osteoclast precursor cells.
TLR4 is not Involved in Osteoclastogenesis Induced by the Infection of RANKL-primed RAW-D Cells with P. gingivalis P. gingivalis is known to stimulate the production of TNF-a and IL-6 through TLR2 and TLR4 signals [22]. Therefore, we analyzed TLRs involved in the stimulation of osteoclastogenesis induced by P. gingivalis infection. Treatment with E. coli LPS, a TLR4 ligand, and the synthetic lipoprotein Pam3CSK4, a TLR2 ligand, stimulated osteoclastogenesis in RANKL-primed RAW-D cells ( Fig. 2A). Similarly, P. gingivalis LPS induced osteoclastogenesis in RANKL-primed RAW-D cells (Fig. 2B). We found that the concentration of P. gingivalis LPS required to stimulate osteoclastogenesis was higher than the concentration of E. coli LPS required for similar stimulation. P. gingivalis treated at 65uC for 15 min stimulated osteoclastogenesis at levels similar to live P. gingivalis, but treatment of P. gingivalis at 90uC for 5 min reduced the induction of osteoclastogenesis from RANKL-primed RAW-D cells (Fig. 2C), suggesting that some protein components of live P. gingivalis may be involved. Polymyxin B (1 mg/ml), which is a specific inhibitor of TLR4, inhibited osteoclastogenesis in RANKL-primed RAW-D cells induced by E. coli LPS, but not in cells induced by Pam3CSK4. However, the same concentration of polymyxin B (1 mg/ml) did not inhibit the induction of osteoclastogenesis in RANKL-primed RAW-D cells induced by live P. gingivalis or P. gingivalis LPS (Fig. 2D). Although a higher concentration of polymyxin B (5 mg/ml) partially inhibited the induction of osteoclastogenesis in RANKL-primed RAW-D cells by P. gingivalis LPS, these results show that the major effect of P. gingivalis on osteoclastogenesis in RANKL-primed RAW-D cells does not involve TLR 4 signaling.
TNF-a Induces Osteoclastogenesis in RANKL-primed Macrophages, but Infection with P. gingivalis Induces Osteoclastogenesis in RANKL-primed Macrophages in the Absence of TNF-a TNF-a has been shown to induce osteoclast differentiation in mouse BMM in vitro [24]. To analyze the contribution of TNF-a to osteoclastogenesis in RANKL-primed RAW-D cells infected with P. gingivalis, we analyzed TNF-a mRNA expression. Infection of unprimed or RANKL-primed RAW-D cells with P. gingivalis stimulated the expression of TNF-a. However, level of TNF-a induced in RANKL-primed RAW-D cells was only 60-65% of the level induced in unprimed RAW-D cells (Fig. 3A). In addition, the production of TNF-a protein from RANKL-primed RAW-D cells induced by P. gingivalis was markedly decreased (Fig. 3B). Because addition of TNF-a stimulated osteoclastogenesis in RANKLprimed RAW-D cells (Fig. 3C), we analyzed the capacity of P. gingivalis infection to induce osteoclastogenesis in RANKL-primed RAW-D in the presence of neutralizing antibody against mouse TNF-a. TNF-a neutralizing antibody completely blocked osteoclast differentiation in uninfected RANKL-primed RAW-D cells in the presence of mouse TNF-a (20 ng/ml) (Fig. 3C, left). In contrast, TNF-a neutralizing antibody did not block osteoclast differentiation in P. gingivalis-infected RANKL-primed RAW-D cells (Fig. 3C, right). These data indicate that TNF-a is not required for osteoclastogenesis in RANKL-primed RAW-D induced by infection with P. gingivalis.
We next analyzed mouse BMMs to determine whether TNF-a is involved in osteoclast differentiation in RANKL-primed macrophages in primary culture. We found that, as was the case in RANKL-primed RAW-D cells, P. gingivalis infection of RANKL-primed BMM stimulated osteoclastogenesis (Fig. 4A). Neutralizing antibody against anti-mouse TNF-a blocked osteoclast differentiation induced by TNF-a but did not inhibit

Expression of NFATc1 but not c-fos is up-regulated During Osteoclastogenesis in RANKL-primed RAW-D Cells Induced by Infection with P. gingivalis
To assess the contribution of osteoclast signaling molecules in osteoclastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis, we analyzed mRNA expression levels of NFATc1, c-Fos, and IFNb in unprimed RAW-D cells treated or not treated with RANKL or P. gingivalis, and in RANKL-primed RAW-D cells retreated or not retreated with RANKL or P. gingivalis. Initial RANKL treatment of RAW-D cells stimulated the expression of both NFATc1 and c-fos. However, expression of both NFATc1 and c-fos declined if the cells were further cultured in the absence of RANKL or P. gingivalis for 24 h (Fig. 5A and B, compare bars 3 and 4). Retreatment of RANKL-primed macrophages with RANKL increased expression of both NFATc1 and c-fos compared to unretreated cells. In contrast, infection (i.e., retreatment) of RANKL-primed cells with live P. gingivalis increased expression of NFATc1 but not of c-fos ( Fig. 5A and  B). INFb is known to inhibit osteoclastogenesis by down-regulating c-Fos [7]. RANKL stimulated the expression of IFNb, whereas P. gingivalis infection of RAW-D cells reduced the expression of IFNb. Retreatment of RANKL-treated RAW-D cells with P. gingivalis also decreased the expression of IFNb mRNA (Fig. 5C). Together, these data suggest that expression of NFATc1 is important in osteoclastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis, and the down-regulation of IFNb expression induced by P. gingivalis may facilitate induction of osteoclast differentiation. NFATc1 and p38 MAPK Signaling but not JNK or NF-kB Signaling are Required for Osteoclastogenesis in RANKLprimed RAW-D Cells Induced by Infection with P. gingivalis To elucidate the mechanism of osteoclast induction, we analyzed the effect of inhibitors of NFATc1, MAPK, and NF-kB pathways on osteoclastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis. All inhibitors suppressed osteoclastogenesis in RANKL-primed RAW-D cells induced by retreatment with RANKL (Fig. 6A). Inhibitors of NFATc1 (FK506) and p38 MAPK (SB203580), similarly inhibited osteo- clastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis (Fig. 6A). In contrast, inhibitor of NF-kB (Bay11.7082) did not significantly inhibit osteoclastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis. The concentrations of these inhibitors used in the differentiation assays did not inhibit the proliferation of RAW-D cells (Fig. 6B). These results indicate that osteoclastogenesis in RANKL-primed RAW-D cells induced by infection with P. gingivalis is dependent on the expression of NFATc1 and p38 MAPK but is not dependent on expression of NF-kB, whereas osteoclastogenesis induced in these cells by RANKL is dependent on NF-kB as well as on NFATc1 and p38 MAPK.

RANKL but not TNF-a Enhances Osteoclastogenesis in RANKL-primed RAW-D Cells in the Presence of P. gingivalis
To determine the role of RANKL and TNF-a in osteoclastogenesis induced by infection with P. gingivalis, we further analyzed the effect of RANKL and TNF-a on osteoclastogenesis in RANKL-primed RAW-D cells in the presence or absence of P. gingivalis. RANKL retreatment stimulated osteoclastogenesis in a dose-dependent manner in RANKL-primed RAW-D cells (Fig. 7A, left). RANKL retreatment coupled with live P. gingivalis infection further enhanced osteoclastogenesis (Fig. 7B, left). Similarly, TNF-a retreatment of RANKL-primed RAW-D cells strongly stimulated osteoclast formation in RANKL-primed RAW-D cells (Fig. 7A, right). However, TNF-a retreatment coupled with live P. gingivalis infection did not increase osteoclast formation (Fig. 7B, right). These data indicate that TNF-a stimulates osteoclastogenesis in osteoclast precursor cells in the absence, but not in the presence, of P. gingivalis. On the other hand, RANKL enhances osteoclastogenesis in osteoclast precursor cells both in the presence or absence of P. gingivalis.
We analyzed mRNA expression of p55 TNF receptor, p75 TNF receptor, TLR2, and TLR4 in RANKL-primed or untreated RAW-D cells and in RANKL-primed RAW-D cells retreated with P. gingivalis or RANKL by semi-quantitative RT-PCR. Expression of p55 and p75 TNF receptors was increased by P. gingivalis and was not down-regulated in RANKL-primed RAW-D cells. Expression of both TLR2 and TLR4 was also detected in RANKL-primed RAW-D cells (Fig. 7C). These data suggest that the inability of TNF-a to enhance osteoclastogenesis in RANKLprimed RAW-D in the presence of P. gingivalis may not be a consequence of the loss of expression of TNF-a receptors.

Discussion
Osteoclasts are responsible for inflammatory bone loss caused by diseases such as periodontitis and rheumatoid arthritis. Periodontitis is a common infectious inflammatory disease associated with resorption of alveolar bone. It has previously been reported that bacterial components such as E. coli LPS induce bone loss in vivo and affect osteoclastogenesis in vitro [10,16,17,18]. Recent studies have shown that E. coli LPS inhibits osteoclastogenesis from bone marrow macrophages in early stages of the process but stimulates osteoclastogenesis in later stages. [14]. Hotokezaka et al. have also shown that peptidoglycan in addition to E. coli LPS induce osteoclastogenesis and especially cell fusion in RANKLtreated RAW264.7 cells in the presence of a low concentration of U0126, a MAPK-ERK kinase (MEK)/ERK inhibitor [28]. While our experiments were ongoing, Zhang et al reported that P. gingivalis differentially modulates osteoclast differentiation from bone marrow macrophages depending on the stage of osteoclastogenesis [15]. However, it is possible that, if the adherent cells are not positively depleted, primary bone marrow macrophage preparations will be contaminated with small numbers of other cell lineages such as T cells and stromal cells that may affect osteoclast differentiation [23]. Direct effects of pathogens such as P. gingivalis on osteoclastogenesis are still not clear. In addition, although cytokines are considered to be involved in the pathology of periodontitis [29,30], the contribution of TNF-a produced by macrophages to osteoclastogenesis is obscure. In the present study, using the RAW-D macrophage cell line, we demonstrated that infection of RANKL-primed macrophages with P. gingivalis markedly promoted osteoclastogenesis in the absence of TNF-a.
Zhang et al. showed that P. gingivalis increases osteoclastogenesis two-fold compared to RANKL alone as assessed by analysis of the formation of TRAP-positive MNCs. We found a more marked effect of P. gingivalis infection on osteoclastogenesis from RANKLprimed RAW-D cells. If RANKL-primed RAW-D cells were cultured without stimulation, they did not form TRAP-positive cells, whereas, if RANKL-primed RAW-D cells were infected with P. gingivalis they formed numerous TRAP-positive MNCs. We also showed evidence that P. gingivalis infection induces expression of several osteoclast-specific genes such as cathepsin K and calcitonin receptors. As we used a macrophage cell line, RAW-D, it is apparent that the effect of P. gingivalis infection is the result of direct action on RAW-D cells. We also showed that P. gingivalis LPS similarly affects osteoclast induction in RANKL-primed RAW-D macrophages.
Cytokine production from macrophages infected with P. gingivalis is down-regulated in both TLR2 and TLR4 knockout mice, although the reduction is greater in TLR2 knockout mice [15,22]. On the other hand, it has been shown that TLR2 is crucial for inflammatory bone loss in an experimental model of periodontitis induced by infection with P. gingivalis [31]. Zhang et al. showed the importance of TLR2 and Myd88 in the inhibition of RANKL-induced osteoclastogenesis by P. gingivalis infection in unprimed bone marrow macrophages. In this study, using polymyxin B, we showed that TLR4 was not necessary for the induction of osteoclastogenesis by P. gingivalis infection. These results suggest that TLR2 is involved in osteoclastogenesis in RANKL-primed RAW-D cells induced by P. gingivalis infection. We also found that a high amount of P. gingivalis LPS was necessary to induce osteoclastogenesis, indicating that the signal through P. gingivalis LPS is weak or other components of P. gingivalis may be involved. Heat treatment of P. gingivalis reduced the stimulatory activity suggesting that some protein components may be involved. Several virulence factors of P. gingivalis such as LPS, lipoproteins, Fim A fimbriae, hemagglutinins, and cysteine proteinases (gingipains) are considered to have pathological roles in periodontitis [32]. P. gingivalis Fim A fimbrial proteins signal through TLR2 and induce the production of inflammatory cytokines in macrophages [22,33]. However, Fim A proteins have been shown to be activated by heat treatment at 95uC for 10 min [34]. Further investigation is necessary to elucidate the contribu-tion of TLRs in osteoclast formation induced by live P. gingivalis in RANKL-primed macrophages.
Using specific inhibitors, we found that different signaling pathways are required for osteoclastogenesis in RANKL-primed RAW-D cells induced by P. gingivalis than in cells induced by RANKL. Although inhibitors of both of NFATc1 and NF-kB inhibited osteoclastogenesis induced by RANKL, a specific inhibitor of NFATc1, but not an inhibitor of NF-kB, inhibited osteoclastogenesis induced by infection with P. gingivalis in RANKL-primed RAW-D cells. Osteoclastogenesis induced by P. gingivalis in RANKL-primed RAW-D cells was dependent on NFATc1 but not NF-kB. Interestingly, Zhang et al. showed the existence of a negative feedback loop between NF-ATc1 and NF-kB [15]. In their study, NFATc1 signaling was activated, but P. gingivalis-induced NF-kB activation was down-regulated in RANKL-primed macrophages. Although the mechanism is unknown, our results support Zhang et al.'s hypothesis. It remains to be determined how NFATc1 is activated in osteoclastogenesis induced by P. gingivalis in RANKL-primed macrophages. In osteoclastogenesis, NF-kB and c-Fos are recruited to the NFATc1 promoter and contribute to its transcription in early phases of osteoclastogenesis; however, in later phases NF-kB and c-Fos are not involved as NFATc1 stimulates its own transcription through an auto-amplification loop [35]. Therefore, in RANKL-primed macrophages, it is possible that the expression level of NFATc1is sufficient to auto-amplify its own expression. However, P. gingivalis may employ additional mechanisms to maintain the localization and expression of NFATc1. It has been reported that peptidoglycan and LPS activate phospholipase C c2 (PLCc2), leading to intracellular calcium mobilization in BMM [36], which may activate NFATc1.
An important observation in the current report is that TNF-a is not required for osteoclastogenesis in RANKL-primed RAW-D RAW-D cells were stimulated with or without RANKL (50 ng/ml) or P. gingivalis for 22 h. RANKL-primed RAW-D cells were then retreated with or without RANKL (50 ng/ml) or P. gingivalis for 24 h. Total RNA was prepared, cDNA was synthesized, and real-time PCR analysis was performed using NFATc1 (A), c-fos (B), or IFNb (C) Taqman probes. Statistical significance was determined with Student's t test. **P,0.01, *P,0.05 compared with unprimed control or RANKL-primed control. doi:10.1371/journal.pone.0038500.g005 cells induced by P. gingivalis infection. Several reports have shown that RANKL treatment of macrophages reduces the production of inflammatory cytokines induced by LPS [15,37]. RANKL pretreatment also protected mice from LPS-induced death from endotoxic shock caused by excess amounts of pro-inflammatory cytokines [38]. In the current study, we have shown that P. gingivalis infection stimulates osteoclastogenesis in the presence of neutralizing antibody against TNF-a. In addition, we showed that osteoclastogenesis induced by P. gingivalis infection was not significantly inhibited by a specific inhibitor of NF-kB, adding further support to the conclusion that TNF-a is not involved. In contrast, Ukai et al. recently reported that culture supernatant from P. gingivalis-stimulated macrophages induced osteoclastogenesis from BMM, and TNF-a was involved [39]. Although our data conflict with their results, this discrepancy can be attributed to differences in the culture conditions or presence of live P. gingivalis in our experiments. Our finding that TNF-a induces osteoclastogenesis in RANKL-primed RAW-D cells in the absence of stimulation with TLR ligands is similar to the results of Ukai et al.
More interestingly, TNF-a did not enhance osteoclastogenesis in RANKL-primed RAW-D induced in the presence of P. gingivalis. At present, it is unknown why RANKL-primed macrophages are not responsive to TNF-a in the presence of P. gingivalis. This is likely to be the result of differences in the endogenous signaling response of RANKL-primed macrophages to TNF-a in the presence or absence of P. gingivalis. It has been reported that some negative regulators of NF-kB activation, such as A20, are induced by LPS [40]. Pretreatment of macrophages with TNF-a suppressed cytokine induction by inhibiting LPSinduced NF-kB signaling; this effect was mediated by A20 and glycogen synthase kinase 3-a (GSK3) [25]. Our results suggest that there may be similar regulatory mechanisms operating between signaling through TNF receptor and TLRs, which may be an important regulatory mechanism for protection from excessive inflammation.
In contrast to TNF-a, retreatment with RANKL promoted osteoclast differentiation from RANKL-primed macrophages in the presence of P. gingivalis. It has recently been shown that RANKL antagonists and OPG inhibit bone loss in experimental periodontitis [41,42], indicating the importance of RANKL and OPG balance in bone loss in periodontitis. In addition, a previous study showed that a higher percentage of T and B cells expressed RANKL in bone resorption lesions of diseased gingival tissues [43]. Consistent with previous results, our data indicate that RANKL plays an important role in bone destruction in the presence of pathogen.
Our current work suggests a possible role of TNF-a and RANKL in osteoclastogenesis induced by infection with P. gingivalis (Fig. 8). Infection of osteoclast precursor cells with P. gingivalis markedly stimulates osteoclastogenesis in an NFATc1 dependent but NF-kB independent manner. TNF-a has the ability to promote osteoclastogenesis in the absence of direct stimulation with pathogen, but does not promote osteoclastogenesis in osteoclast precursor cells in the presence of P. gingivalis. In contrast, RANKL effectively stimulated osteoclastogenesis from osteoclast precursor cells in the presence or absence of P. gingivalis. Our results identify differential effects of RANKL and TNF-a on osteoclastogenesis in RANKL-primed macrophages in the presence of pathogen, which will be useful in devising strategies to regulate bone loss in infection-induced inflammatory diseases. AT activation inhibitor, FK506, was obtained from LC Laboratories (Woburn MA). P38 MAPK inhibitor, SB203580, was purchased from Calbiochem (San Diego, CA). IkB-a inhibitor, BAY11-8082, was obtained from InvivoGen (San Diego, CA). Neutralizing antibody against anti-mouse TNF-a (MP6-XT3) and isotype control antibody, rat IgG1 (eBRG), were from eBioscience (San Diego, CA). Male C57BL/6 mice (aged 5-10 weeks) were purchased from Kyudo Co. (Saga, Japan).

Ethics Statement
All experiments were reviewed and approved by the Laboratory Animal Care and Use Committee of Saga University, permit number (20-026-4).

Bacterial Strain and Culture
P. gingivalis ATCC33277 (kindly provided by Dr. K. Nakayama, Nagasaki University, Japan) was cultured and maintained on enriched trypto-soy agar plates containing 5 mg/ml hemine, defibrinated sheep blood, and 1 mg/ml menadione, at 37uC in an anaerobic atmosphere. For the preparation of P. gingivalis for cell stimulation, bacteria were harvested, centrifuged, and washed in PBS. The number of bacteria (CFU/ml) was determined by measuring the optical density (OD) at 600 nm. cells were stimulated with RANKL (50 ng/ml) for 22 h and then re-stimulated with RANKL or TNF-a in the absence (A) or presence (B) of live P. gingivalis. The culture was stained for TRAP activity after 48 h of retreatment, and TRAP-positive MNCs were counted. Data are expressed as mean 6 S.D. of four independent cultures. Statistical significance was determined with Student's t test. **P,0.01, compared to cultures without RANKL or TNF-a. (C) Expression of mRNAs for p55 and p75 TNF receptors, TLR2, and TLR4 in RAW-D cells treated with or without P. gingivalis for 22 h, or RANKLprimed RAW-D cells retreated with or without RANKL or P. gingivalis for 24 h. Total RNA was isolated, and mRNA expression was assessed by semiquantitative RT-PCR using specific primers as described in Materials and Methods. doi:10.1371/journal.pone.0038500.g007 Figure 8. Possible role of TNF-a in osteoclastogenesis in the presence or absence of P. gingivalis. Macrophages respond to infection with P. gingivalis by producing TNF-a, which stimulates osteoclastogenesis in osteoclast precursor cells in the absence of P. gingivalis (A). However, osteoclast precursor cells primed with RANKL do not produce TNF-a and respond differentially to various stimuli. (B) Cells that are not stimulated do not differentiate into osteoclasts. (C) Cells that are continuously re-stimulated with RANKL differentiate into osteoclasts in an NFATc1-and NF-kBdependent manner in the presence of P. gingivalis. (D) Cells that are infected with P. gingivalis differentiate into osteoclasts in an NFATc1-dependent and NF-kB-independent manner. TNF-a does not stimulate osteoclastogenesis in osteoclast precursor cells in the presence of P. gingivalis, whereas RANKL stimulates osteoclastogenesis in the presence or absence of P. gingivalis. doi:10.1371/journal.pone.0038500.g008

Cell Culture and Osteoclastogenesis
The murine macrophage cell line RAW-D (a subclone of RAW264) [27], was cultured in a-MEM containing 10% FCS. For stimulation with TLR ligands and P. gingivalis, RAW-D cells were precultured with RANKL (50 ng/ml) for 22 h. The cells were then replated in 96 well plates at a density of 4.5610 4 cells/ml. Cells were then infected with live P. gingivalis at indicated multiplicities of infection (MOI), or incubated in the presence or absence of RANKL (5 ng/ml) for 1-2 days. For stimulation of primary macrophages, bone marrow cells were cultured in the presence of 10 ng/ml M-CSF for 3 days to generate BMMs, which were then stimulated with 50 ng/ml RANKL for 22 h, then infected with P. gingivalis and cultured in the presence of M-CSF (10 ng/ml) for 2 days [5]. At the end of the culture, cells were fixed and stained with a commercial kit for the osteoclast marker, TRAP (Sigma-Aldrich, St. Louis, MO). TRAP-positive cells with 3 or more nuclei were counted as multinucleated cells (MNCs).

Cell Viability Assay
RANKL-primed RAW-D cells were infected with P. gingivalis, and cultured in the presence of various concentrations of signaling inhibitors for 2 days. Cell viability was evaluated using Cell Counting kit-8 (CCK-8) (Dojin Laboratories, Japan) reagent. A 1/ 10 volume of reagent was added to each well, and the cells were incubated at 37uC for an additional 2 h. Absorbance at 450 nm was then measured.

ELISA Analysis
Culture supernatants were collected, and protein levels of TNFa were measured by enzyme-linked immunosorbent assay using an ELISA kit (eBioscience, San Diego, CA) according to the manufacturer's instructions.