Interleukin-32 Promotes Osteoclast Differentiation but Not Osteoclast Activation

Background Interleukin-32 (IL-32) is a newly described cytokine produced after stimulation by IL-2 or IL-18 and IFN-γ. IL-32 has the typical properties of a pro-inflammatory mediator and although its role in rheumatoid arthritis has been recently reported its effect on the osteoclastogenesis process remains unclear. Methodology/Principal Findings In the present study, we have shown that IL-32 was a potent modulator of osteoclastogenesis in vitro, whereby it promoted the differentiation of osteoclast precursors into TRAcP+ VNR+ multinucleated cells expressing specific osteoclast markers (up-regulation of NFATc1, OSCAR, Cathepsin K), but it was incapable of inducing the maturation of these multinucleated cells into bone-resorbing cells. The lack of bone resorption in IL-32-treated cultures could in part be explain by the lack of F-actin ring formation by the multinucleated cells generated. Moreover, when IL-32 was added to PBMC cultures maintained with soluble RANKL, although the number of newly generated osteoclast was increased, a significant decrease of the percentage of lacunar resorption was evident suggesting a possible inhibitory effect of this cytokine on osteoclast activation. To determine the mechanism by which IL-32 induces such response, we sought to determine the intracellular pathways activated and the release of soluble mediators in response to IL-32. Our results indicated that compared to RANKL, IL-32 induced a massive activation of ERK1/2 and Akt. Moreover, IL-32 was also capable of stimulating the release of IL-4 and IFN-γ, two known inhibitors of osteoclast formation and activation. Conclusions/Significance This is the first in vitro report on the complex role of IL-32 on osteoclast precursors. Further clarification on the exact role of IL-32 in vivo is required prior to the development of any potential therapeutic approach.


Introduction
Interleukin-32 (IL-32) is a newly described cytokine produced mainly by T-, natural killer, epithelial cells and monocytes after stimulation by Interleukin-2, Interleukin-18 or IFN-c [1,2]. It was formerly known as natural killer cell transcript 4 [3]. Although IL-32 can bind proteinase 3, a neutrophil-derived serine protease, its receptor is unknown [4,5]. IL-32 has the typical properties of a pro-inflammatory mediator by stimulating TNF-a, IL-1b and IL-8 production and by activating the NF-kB and p38 mitogenactivated protein (MAP) kinase pathways [1]. IL-32 seems to be involved in a variety of diseases such as inflammatory bowel disease, myelodysplastic syndrome and chronic myelomonocytic leukaemia, HIV infection [6][7][8]. Moreover, it has been reported recently that in a cohort of patient suffering with rheumatoid arthritis, IL-32 was significantly increased in the synovial tissue and that its levels were strongly correlated with the severity of the disease [9,10]. Moreover, IL-32 has been demonstrated to induce joint inflammation with concomitant mild cartilage damage when injected intra-articularly in murine knee joints [9]. Despite these observations, the impact of this cytokine on osteoclastogenesis remains unclear.
Osteoclasts are multinucleated cells originating from the hematopoietic lineage (CFU-GM) that specifically function in lacunar bone resorption [11,12]. Osteoclast differentiation from circulating hematopoietic precursors requires the presence of macrophage colony-stimulating factor (M-CSF) and the receptor activator for nuclear factor kB ligand -RANKL [13][14][15][16]. RANKL is a member of the tumor necrosis factor (TNF) superfamily that is expressed on osteoblasts and T-cells and interacts with its receptor, RANK, expressed on osteoclast precursors [15,[17][18][19]. It has been reported that the interaction of RANK with RANKL results in the activation of distinct intracellular pathways: NFATc1, NF-kB, Akt, and MAP kinase pathways. Osteoprotegerin (OPG) acts as a decoy receptor for RANKL and blocks RANKL-mediated osteoclast differentiation and stimulation of osteoclast resorbing activity [20,21]. Although RANKL is a crucial factor for osteoclastogenesis, RANKLindependent mechanisms have been evidenced with several proinflammatory cytokines such as TNF-a, LIGHT and IL-8 [22][23][24][25].
The aim of the present study was to investigate the role of IL-32 on the differentiation and maturation of human osteoclast precursors. Our findings suggested that although IL-32 induced fusion of osteoclast precursors into multinucleated TRAcP+, VNR+ cells expressing OSCAR and Cathepsin K and induced the activation of NFATc1, NF-kB, ERK1/2 and JNK, it was incapable of modulating Akt activation in a level similar to RANKL. Morphologically multinucleated cells generated in response to IL-32 did not form F-actin on dentine slices and ultimately were unable to resorb bone in vitro.

Results
IL-32 induces the differentiation of adherent PBMCs into multinucleated TRAcP+, VNR+ cells in the absence of sRANKL After 14 days in culture, the presence of newly-formed multinucleated cells were detected in response to IL-32. To determine the optimal concentration of IL-32 required to differentiate PBMCs into multinucleated cells in vitro, a dose response experiment was performed. IL-32 was capable of inducing the formation of multinucleated TRAcP and VNR positive cells as shown in Figure 1A. The number of TRAcP positive multinucleated cells was increased by a significant 4.5-fold at the lowest concentration studied (25 ng/ml) compared to M-CSF alone (361612 vs. 7965; p,0.0001) as seen on Figure 1B.
The dose dependent effect of IL-32 was statistically significant and clearly evident as illustrated in Figure 1B. A similar pattern for the number of VNR positive multinucleated cells in response to IL-32 was noted (data not shown).
The size of newly-formed multinucleated TRAcP+ VNR+ cells in response to IL-32 was also significantly greater than in M-CSFtreated culture (Fig. 1C). Interestingly, at doses higher than 50 ng/ ml, IL-32 did not markedly influence the size of the newly-formed multinucleated TRAcP+ VNR+ cells. The number of nuclei per multinucleated cells was significantly increased in IL-32-treated cultures compared to M-CSF alone with a 2-fold increase for the lowest concentration of 25 ng/ml (9.4462.74 vs. 4.6661.36; p,0.0001). The dose dependent effect of IL-32 was clearly evident as illustrated in Figure 1D. Based on the data obtained in Figures 1B-1D, the optimal dose for any subsequent experiments was chosen to be 100 ng/ml.

IL-32 is capable of inducing the formation of multinucleated cells exhibiting specific osteoclast markers
After 3 days of treatment with M-CSF/IL-32 or M-CSF/ RANKL (as a positive control), the expression of specific osteoclast markers such as NFATc1, TRAF6, OSCAR and Cathepsin K, were up-regulated as compared to M-CSF treatment alone ( Fig. 2A). The number of newly-generated multinucleated cells formed in the presence of IL-32 was statistically lower than those formed in response to sRANKL (645637 vs. 555618; p = 0.032) (Fig. 2B). In contrast, the size and the number of nuclei of the newly-formed multinucleated cells in response to IL-32 were significantly increased by 1.4-fold and 2.3-fold respectively compared to those generated in the presence of sRANKL (14666.4 mm vs. 10263.4 mm; p,0.0001 and 18.862.5 vs. 8.261.3; p,0.0001) ( Fig. 2C and 2D).
When a combination of IL-32 and sRANKL (both at 100 ng/ ml) were added to the osteoclast precursors, the number of newlyformed osteoclasts was increased by 2.2-fold compared to sRANKL-treated cultures (1417651 vs. 645637; p,0.0001) (Fig. 2B). Paradoxically, the size of the newly-formed osteoclasts in response to IL-32 and sRANKL was significantly reduced compared to IL-32-treated cultures (9663 mm vs. 14666.4 mm; p,0.0001). However, compared to sRANKL-treated cultures, the size of TRAcP positive multinucleated osteoclasts were not significantly different as those formed in the presence of IL-32 and sRANKL (9663 mm vs. 10263.4 mm; p = 0.310) (Fig. 2C). The number of nuclei per newly-formed osteoclast was significantly reduced in PBMC cultures treated with combined sRANKL/IL-32 as compared to IL-32 alone (1161.1 vs. 18.862.5; p,0.0001) (Fig. 2D).
Signalling pathways involved in response to IL-32 and compared with RANKL In order to better characterize the multinucleated cells formed in response to IL-32, we investigated, using Western blot analysis, the activation of downstream signalling pathways. After 15 minutes of treatment with M-CSF or RANKL or IL-32, the activation of several intracellular pathways known to be activated by RANKL, were determined. In response to IL-32 or RANKL, PBMCs underwent a rapid activation of JNK and NF-kB as compared to M-CSF-treated cultures (Fig. 3). The phosphorylation of ERK1/2 was markedly increased by 3.6-fold in response to IL-32 compared with RANKL. Although M-CSF alone induced a strong and rapid activation of Akt, the addition of RANKL was incapable of activating Akt in a similar level, indicating a possible down-regulation of Akt activation by RANKL when associated with M-CSF (Fig. 3). Interestingly, Akt activation in IL-32 treatment increased considerably compared to RANKL exposure.

IL-32 alone cannot induce the maturation of the newlyformed osteoclast into bone -resorbing osteoclasts
In order to determine whether IL-32 alone or in combination with sRANKL have a direct effect on the maturation of osteoclasts, the expression of F-actin ring by the newly-formed multinucleated cells and the extent of lacunar resorption was examined on dentine slices of PBMC cultures after 21 days. Multinucleated cells generated in the presence of sRANKL or in combination of sRANKL/IL-32 presented an F-actin ring at their periphery as shown in Figure 4A. However, multinucleated cells generated in response to IL-32 did not similarly express F-actin ring (Fig. 4A). Furthermore, IL-32 alone was shown to have no effect on the maturation of newly-formed osteoclast as indicated in Figure 4B and 4C. However, although the mean percentage surface area resorption in response to sRANKL alone was 43.369.8%, the addition of IL-32 in combination to sRANKL did not increase the extent of lacunar resorption and surprisingly it significantly reduced this parameter by 2.6-fold (Fig. 4C).
To ascertain the mechanism by which IL-32 induced the observed decrease in the percentage lacunar resorption in sRANKL-treated cultures, the size of the lacunar pits were determined by image analysis. In cultures treated with IL-32 and sRANKL the mean diameter of the pits was significantly reduced by 1.4-fold compared to sRANKL alone (68.361.0 mm vs. 94.761.6 mm; p,0.0001) (Fig. 4D).
Multinucleated cells formed in response to IL-32 are not solely mediated through the RANK/RANKL pathway Figure 5 demonstrates the inhibitory effects of excess OPG on sRANKL-mediated and IL-32-mediated osteoclastogenesis. OPG at 250 ng/ml was capable of significantly inhibiting the number of RANKL-inducing osteoclast formed by 6.2-fold as compared to sRANKL alone (10465 vs. 645637; p,0.0001) (Fig. 5A). The size of the osteoclast generated in the presence of sRANKL and OPG was significantly reduced as compared to sRANKL alone (7362.2 mm vs. 10263.4 mm; p,0.0001) ( Fig. 5B) but the number of nuclei per osteoclast was significantly increased (Fig. 5C). Although the concentration of OPG employed was unable to completely inhibit the number of osteoclasts formation in RANKL-treated cultures, this dose was suitable in completely abolishing the activity of the newly-formed osteoclasts in the resorption assay (data not shown).
OPG treatment resulted in a moderate inhibition (1.3-fold) of the newly-formed multinucleated cells in response to IL-32 as compared to cultures where no exogenous OPG was added (423622 vs. 555618; p = 0.003, Fig. 4A). Moreover, OPG was capable of reducing the size and the number of nuclei per newlyformed multinucleated cells generated in response to IL-32 (10463.9 mm vs. 14666.4 mm; p,0.0001 and 11.662 vs. 18.862.5; p,0.0001, Fig. 5B). As the extent of lacunar resorption did not mirror the number of multinucleated cells formed in cultures treated with IL-32 (Fig. 4B), the addition of OPG to these cultures did not affect the surface area resorption by IL-32 as there was no evidence of pit formation in response to this cytokine (data not shown).

OPG inhibits osteoclast formation and resorption in sRANKL/IL-32 combined treatment
In cultures treated with IL-32 and sRANKL, the number of multinucleated osteoclasts formed was markedly reduced in response to excess concentration of OPG as compared to untreated cultures (Fig. 6A). The size and the number of nuclei of the newly-formed osteoclasts were, however, unaffected in cultures treated with sRANKL, IL-32 and OPG as compared to conditions in the absence of OPG ( Fig. 6B and 6C). OPG treatment of sRANKL/IL-32 PBMC cultures resulted in a complete inhibition of lacunar resorption as shown in Figure 6D.  these cytokines were not detectable in the culture supernatant. After 48 hrs, a similar pattern of cytokine release was evidenced with the detection of high levels of TNF-a, IL-6, LIGHT, MIP-1a, VEGF, IFN-c and IL-4. Here again in the unstimulated culture none of these cytokines were evidenced. Soluble RANKL was undetectable in the supernatant of IL-32-treated or unstimulated cultures after 24 hrs and 48 hrs.

Discussion
Osteoclasts are multinucleated cells responsible for bone resorption and are derived from hematopoietic precursor cells that circulate in the blood [26][27][28]. It is currently thought that two critical factors supplied by osteoblasts, M-CSF and RANKL, are essential for the differentiation and maturation of osteoclast precursors [11,29]. Although M-CSF defective mice (op/op) show an osteopetrotic phenotype, it can spontaneously reverse with age, suggesting that alternative osteoclastic pathways do exist [30][31][32]. Hepatocyte growth factor (HGF), vascular endothelial growth factor (VEGF) and Flt3 ligand have all been shown to support osteoclast formation in the absence of M-CSF [33][34][35]. Furthermore mice deficient in either RANKL or its receptor RANK also show an osteopetrotic phenotype that is caused by the complete lack of osteoclast in their bones [36,37]. Although no osteoclasts can be identified in the bone of RANKL or RANK deficient mice, this may not be simply because of the complete failure of osteoclastogenesis. Indeed, RANKL is a survival factor for differentiated osteoclasts [38] and it is plausible that impaired osteoclast differentiation superimposed on a shortened lifespan might also explain the observed phenotype in RANKL or RANK deficient mice [39]. As such, alternative RANKL-independent pathways (e.g. LIGHT, TGFb and TNF-a) have been reported to support osteoclastogenesis in the presence of excess osteoprotegerin, an inhibitor of RANKL-RANK interactions [23,24,40]. The emergence of the osteoimmunology field has demonstrated that activated T cells directly modulate osteoclastogenesis and bone resorption [41,42], and that T cell products, such as IL-17, TWEAK, GM-CSF and IFN-c, can regulate osteoclast formation [43][44][45]. The present study sought to determine the role of IL-32, a newly described cytokine presenting characteristic of proinflammatory cytokine and involved in a variety of inflammatory disorders [1,[6][7][8][9][10] in osteoclast differentiation and activation. We have demonstrated that multinucleated cells formed in response to IL-32 expressed several specific markers of osteoclast such as activation of the NF-kB and MAP kinase pathways, expression of TRAcP and VNR, up-regulation of NFATc1, OSCAR and Cathepsin K which were also observed in RANKL-treated cultures. However, IL-32-treated multinucleated cells were unable to induce bone resorption in vitro. One explanation for the lack of bone resorption could be attributed to the failure of the multinucleated cells generated in response to IL-32 to express Factin ring which ultimately is required for anchorage of the cells prior to bone resorption.
An important step in the process of osteoclast differentiation is the induction of NFATc1 above a critical threshold [39]. This is consistent with the findings that overexpression of NFATc1 is sufficient to induce osteoclast differentiation [46]. In the present study, IL-32 was capable of up-regulating the expression of NFATc1 compared to M-CSF-treated cultures or even M-CSF/ RANKL-treated cultures. Moreover, in response to IL-32, these cells expressed high levels of OSCAR and Cathepsin K, two markers that are specific for osteoclasts [47,48]. According to the different markers expressed by the cells in response to IL-32, it is reasonable to conclude that these multinucleated cells are likely to be ''immature osteoclasts''. Recently, Kim et al. [39] have reported that TNF-a is capable of inducing multinucleation of osteoclast precursors and expression of osteoclast phenotypic markers (TRAcP, F-actin) in RANK-deficient cells, but was unable to induce evidence of lacunar bone resorption. In the present study, although PBMCs released pro-osteoclastic mediators (TNFa, IL-6, LIGHT, MIP-1a and VEGF), the lack of lacunar resorption in IL-32-treated cultures could be explained by the fact that IL-32 stimulated the release of known osteoclastic inhibitors, i.e. IL-4 and IFN-c. Our results herein indeed indicated that this is the case as we demonstrated a significant increase in the release of IL-4 and IFN-c from IL-32-stimulated PBMCs as compared to unstimulated cells.
Tumour necrosis factor receptor-associated factor 6 (TRAF6) has been reported to be important for osteoclast activation, i.e. lacunar bone resorption [49]. The complex role of IFN-c in osteoclastogenesis has been previously addressed by Takayanagi et al. [50]. They have shown that IFN-c induces rapid degradation of TRAF6, which results in strong inhibition of the RANKL-induced activation. We therefore speculated that the inhibitory effects of IL-32 alone or in combination with soluble RANKL could partly be attributed to TRAF-6 degradation. However, surprisingly, we found that TRAF6 is not degraded but it is overexpressed in response to IL-32 treatment compared to RANKL. Recently, Gao et al. have shown that IFN-c exhibits a ''direct'' anti-resorptive effect by blunting osteoclast differentiation. However, IFN-c can act ''indirectly'' as a pro-resorptive factor by stimulating T-cells to express RANKL and TNF-a [51]. In the present study, we used PBMCs as a source of osteoclast precursors and although cells were washed thoroughly to eliminate non-adherent cells (mainly B-and T-cells), it is plausible that some T-cells could have been present in the culture and contributed to osteoclastogenesis. This hypothesis is also reinforced by the evidence that the addition of excess OPG to the IL-32-treated cultures led to a marked decrease in the number and size of newly-formed multinucleated cells. Although we were unable to detect any soluble RANKL in the supernatant of IL-32-treated cultures, it is conceivable to suggest that effects of IL-32 could have been partially mediated through a RANKL-dependent mechanism. We have previously observed that IL-32 increase the expression of membrane-bound RANKL in T-cell cultures (unpublished data) and it is likely that a few numbers of RANKL-expressing T-cells may have been present in the PBMC cultures. As IL-32 is known to be produced by PBMCs in response to IFN-c [52], the effects of IFN-c in combination with IL-32 on T-cells could have contributed to the inhibitory effects of OPG observed herein.
The downstream signalling of RANK/RANKL interactions has been extensively studied in the last decade. It has been shown that the binding of RANKL to its receptor activated NF-kB, MAP kinase and Akt pathways. However, downstream pathways involved in response to IL-32 in osteoclasts are not fully elucidated. In our study, we found that PBMCs treatments with M-CSF/IL-32 or M-CSF/RANKL dramatically increased the activation of NF-kB and JNK pathways compared to M-CSF-treated cultures. However, the activation of the Akt pathways appeared more complicated. M-CSF or M-CSF/IL-32 treatments were capable of strongly activating Akt pathways compared to M-CSF/RANKL. These results appeared controversial with the consensus that RANKL activates Akt [11,53]. However, most of the studies showing Akt activation in response to RANKL have been done in in vitro models of bone marrow co-cultures or using murine cell line, RAW264.7 cells. In these studies, treatment of serum-starved osteoclast precursors with soluble RANKL resulted in a significant activation of the Akt pathway as compared to cultures in the absence of RANKL. This is in conflict with the present findings whereby after exposure of M-CSF-treated PBMCs to sRANKL, Akt activation was down-regulated compared with M-CSF alone. Interestingly, M-CSF/IL-32 treatment of PBMCs resulted in a similar level of Akt activation compared to M-CSF treatment. It has also been extensively shown that M-CSF, via its receptor cfms, is a strong activator of PI-3 Kinase which in turn activates Akt [54,55]. In the present study, we demonstrated that M-CSF/IL-32 treatment exhibited levels of Akt activation similar to M-CSF treatment and it is questionable whether Akt activation in response to IL-32 or RANKL results in the same downstream effectors (Fig. 8).
We have also evidence of a marked increase in the phosphorylation of ERK1/2 in M-CSF/IL-32 compared to M-CSF/RANKL treated cultures. It is well known that activated ERK1/2 translocates to the nucleus and activates its target to promote the expression of specific genes (reviewed in [56]). It is plausible to suggest that the downstream targets of activated ERK1/2 in M-CSF/IL-32 are different to those known for M-CSF/RANKL (Fig. 8). This is also reinforced by the morphological differences (e.g. differences in cell size and number of nuclei, lack of F-actin ring) observed in the multinucleated cells generated in presence of IL-32 or RANKL.
Our current findings suggest that IL-32 was capable of inducing osteoclast differentiation in a manner partially independent of the RANK/RANKL pathway. However, although IL-32 could increase the release of pro-inflammatory mediators known to positively influence osteoclastogenesis, it was unable to induce the activation of these newly-formed multinucleated cells into boneresorbing osteoclasts and had a direct inhibitory effect on osteoclast activation in vitro. It is worth noting that IL-32 has direct effects on other cell types such as epithelial cells, T-cells, natural killer cells and monocytes. Although the present study only addressed its direct role on osteoclast precursors, IL-32 could also indirectly modulate osteoclastogenesis in vivo. The complex role of IL-32 in the patho-physiology of bone disorder therefore requires further clarification prior to the development of any potential therapeutic approach.

Materials and Methods
Recombinant IL-32a, M-CSF and OPG were purchased from R&D systems Europe (Abingdon, UK) and soluble RANKL was purchased from PeproTech (London, UK). Recombinant cytokines were aliquoted and stored at 280uC on day of purchase. All chemicals for immuno-histochemical staining were purchased from Sigma-Aldrich (Poole, UK).

Preparation and culture of human peripheral blood mononuclear cells (PBMCs)
The study was approved by the Oxfordshire Research Ethics Committee (CO1-071). PBMCs were isolated from 6 normal healthy volunteers as previously described [57]. Briefly, donors were free of prescribed and over the counter prescription. Antecubital venous blood was supplied by National Blood Service (UK). Blood was diluted 1:1 in a-minimal essential medium (MEM) (Invitrogen, Paisley, UK), layered over Histopaque (Sigma-Aldrich Chemicals, Poole, UK), and centrifuged (6936 g) for 20 min. The interface layer was resuspended in MEM then centrifuged (6006 g) for a further 10 min after which the resultant

Assessment of osteoclast formation
To assess the extent of osteoclast formation and activation, isolated human PBMCs from whole blood were cultured on glass coverslips and dentine slices as described previously [57]. Briefly, 5610 5 PBMCs were added to 4 mm diameter dentine slices and 6 mm diameter glass coverslips in MEM containing 100 UI/ml penicillin, 100 mg/ml streptomycin and 10% FCS. After 2 h incubation, cultures were vigorously rinsed in medium to remove non-adherent cells, and then maintained in 1 ml MEM/FCS with 25 ng/ml recombinant human M-CSF with or without 100 ng/ml recombinant human sRANKL for up to 21 days. To assess whether IL-32 could influence osteoclastogenesis, PBMCs were cultured in the following conditions: Characterization of newly-formed osteoclast on glass coverslips and dentine slices Tartrate resistant acid phosphatase (TRAcP). After 14 days in culture, the expression of TRAcP (one of the known osteoclastic markers [59]) on the adherent cells on the coverslips were examined histochemically, as described previously [59]. Briefly, coverslips were removed from the culture wells and rinsed promptly in PBS buffer, fixed with formalin (10% in PBS buffer) for 10 minutes and rinsed in distilled water. TRAcP was histochemically revealed by a simultaneous coupling reaction using Naphtol AS-BI-phosphate as substrate and Fast violet B as the diazonium salt. The coverslips were then incubated for 90 minutes at 37uC in the dark, rinsed three times in distilled water and the residual activity was inhibited by 4% NaF for 30 minutes. Coverslips were then rinsed in distilled water, counterstained with DAPI for 20 minutes and allowed to dry before mounting using an aqueous medium. TRAcP positive cells, with more than three nuclei, were considered as osteoclasts. The number of newly generated osteoclasts and the number of nuclei per osteoclast (after counterstaining with DAPI) were assessed using light microscopic examination. The size of the osteoclasts was determined by image analysis using the ImageJ Freeware (NIH, Bethesda, MD).
Vitronectin receptor (VNR). Adherent cells on coverslips were fixed with 4% paraformaldehyde after 14 days of culture and stained immuno-histochemically by an indirect immunoperoxidase technique. This technique utilized the monoclonal antibody 23C6 (Serotec, Oxford, UK), which is directed against the VNR, CD51, an osteoclast-associated antigen [60].
F-actin. Adherent cells on dentine slices were fixed in 4% paraformaldehyde after 21 days of culture, washed in PBS and stained histo-chemically with phalloidin-TRITC (500 ng/ml) for 45 minutes. Dentines slices were washed thoroughly in PBS prior to mounting using 70:30 (v/v) water:glycerol and examined with an Olympus BX40.
Lacunar resorption. After 21 days in culture, the dentine slices were removed from the culture wells, placed in NH 4 OH (1N) for 30 minutes and sonicated for 5 minutes to remove any adherent cells. After rinsing in distilled water, the dentine slices were stained with 0.5% (v/v) toluidine blue (pH 5.0) prior to examination by light microscopy. The surface of each dentine slice was examined for the presence of lacunar resorption and the extent of eroded surface on dentine slices was determined using image analysis on each dentine and data was expressed as the percentage of surface area resorbed.

Western blot analysis
PBMCs were cultured on 25 cm 2 flasks with M-CSF (25 ng/ml) alone or M-CSF and soluble RANKL (100 ng/ml) or M-CSF and IL-32 (100 ng/ml) for the indicated period of time. Cells were washed in cold PBS and homogenised in lysis buffer containing 50 mM Tris-Hcl pH 7, 100 mM NaCl, 50 mM NaF, 3 mM Na 3 VO 4 , protease inhibitor cocktail (Sigma-Aldrich) and 1% Nonidet P-40. Samples were spun at 13,000 rpm for 30 min at 4uC, the supernatant was collected and protein concentration was determined using a BCA protein assay kit (Pierce Biotechnology, Rockford, IL). Samples (20-50 mg per lane) were run on a 10% acrylamide and blotted onto a PVDF membrane. The membranes were washed in Tris buffered saline (TBS) and blocked with 5% bovine serum albumin. Samples were incubated overnight with one of the following specific antibodies for IkBa, Phospho-IkBa, ERK1/2, Phospho-ERK1/2, JNK, Phospho-JNK, Akt, Phospho-Akt, TRAF6 (Cell Signalling), Cathepsin K, NFATc1, OSCAR (Santa Cruz Biotechnology) and b-actin (Sigma-Aldrich). Subsequently the membranes were washed in TBS and incubated with the appropriate secondary antibodies coupled to HRP. Immunoreactive bands were visualised using an ECL kit (Amersham, UK). The degree to which the different markers were induced was determined by normalizing the specific signal to that of actin using ImageJ Freeware (NIH).