IL1β and TNFα promote RANKL-dependent adseverin expression and osteoclastogenesis

Tightly regulated resorption of bone is required for normal bone turnover, which is essential for skeletal homeostasis. Osteoclastogenesis (OCG) involves the differentiation of monocyte/macrophage lineage precursor cells to pre-fusion osteoclasts (OCs). Fusion of these cells leads to the formation of multinucleated, bone-resorbing OCs. In the pathological loss of bone that is manifest in inflammatory bone destructive diseases such as rheumatoid arthritis and periodontitis, excessive OCG and bone resorption are central features of disease progression (Graves et al., 2008; Hassanpour et al., 2014; Jiang et al., 2015). A singular feature of OCG is that multiple fusion events involving pre-OCs lead to the formation of large, multinucleated OCs. These fusion events are precisely regulated in time and space and involve the assembly of actin filaments (Chen et al., 2007) that are required for the formation of small, finger-like projections, which extend from the cell membrane and enable fusion to occur. The importance of pre-OC fusion in the formation of efficient, bone-resorbing OCs (Lees and Heersche, 1999, 2000; Lees et al., 2001) underlines pre-OC fusion as a central control locus in OC-mediated bone loss.

Actin is a high-abundance cytoskeletal protein expressed in many types of eukaryotic cells (Bray and Thomas, 1975; Pinder et al., 1975) and is involved in a wide variety of cellular processes, including locomotion, cytoplasmic streaming, protein transport, secretion, phagocytosis, cytokinesis, volume regulation, contraction and the determination of cell shape (Hall and Nobes, 2000; Quinn et al., 2002). In particular, actin assembly is required for the alterations of cell shape and plasma membrane reorganization that are central features of fusion processes that lead to the formation of many types of multinucleated cells. Actin filaments are concentrated as a dense meshwork in the cell cortex (Wolosewick and Porter, 1979) at fusion sites (Kim et al., 2007) and are arranged in distinct supramolecular structures that depend on cell type, substrate attachment and differentiation status. Clusters of actin-rich structures are enriched in the cell extensions of fusing pre-OCs (Takito et al., 2012). These structures contain Arp2/3 and cortactin, which are distributed within dense networks of actin filaments; β3 integrin, paxillin and vinculin localize to the periphery of these networks. Currently, there is little definitive information on the role and regulation of actin networks in the cell fusion events that contribute to OCG.

While it is not definitively known which proteins and regulatory processes regulate actin assembly at fusion zones in OCs, it is likely that one structurally related family of actin-binding proteins is directly involved. This family, which includes gelsolin, adseverin and flightless-1 homolog (FLII), is activated by Ca²⁺, regulates actin filament length by severing pre-existing filaments and capping the fast-growing actin barbed ends (Weeds et al., 1993). The differentiation of OCs is impaired in gelsolin-knockout mice (Witke et al., 1995). Recent data have also implicated adseverin in OCG and OC function (Hassanpour et al., 2014; Jiang et al., 2015). Although adseverin mRNA is expressed at very low levels in unstimulated OC precursor cells, after treatment with soluble RANKL (sRANKL), adseverin mRNA expression increases 23-fold within 4 days in RAW264.7 cell-derived OCs (Hassanpour et al., 2014). When adseverin is conditionally deleted in OCs of mice, bone structure, morphology and turnover are normal (Cao et al., 2017; Jiang et al., 2015), but when experimental periodontitis (an inflammatory disorder characterized by alveolar bone resorption) is induced, no bone destruction is seen (Jiang et al., 2015). These data suggest that the extent of inflammatory bone loss is modulated by the expression of actin assembly regulatory proteins, and their subsequent impact on OCG and OC function.

Much of what is known about adseverin was gleaned from the examination of stimulated and directed exocytosis in chromaffin cells of the adrenal medulla, in which the stability of the cortical actin cytoskeleton is regulated by adseverin (Bruun et al., 2000; Dumitrescu Pene et al., 2005; Ehre et al., 2005; Lejen et al., 2002, 2001; Marcu et al., 1998; Trifaró et al., 2000, 2008, 1992; Trifaró, 1999; Vitale et al., 1991; Zhang et al., 1996). Immunocytochemical analyses of chromaffin cells show a continuous subcortical actin ring that colocalizes with adseverin (Trifaró et al., 2000; Trifaró, 1999). After stimulation by nicotine, histamine or K⁺, subcortical actin filaments are locally and transiently disrupted at sites of adseverin colocalization and granule exocytosis (Trifaró et al., 2000; Trifaró, 1999; Vitale et al., 1991). Subcortical actin meshworks of fusing cells contain zones that are depleted of actin filaments but are enriched with transport vesicles, which implicates adseverin-induced actin severing in myoblast fusion (Duan and Gallagher, 2009).

While two other actin filament-severing proteins, gelsolin and cofilin, are expressed by OCs and have been studied in the formation of OC-specific podosome-type structures (Blangy et al., 2012; Chellaiah et al., 2000), the ubiquitous distribution, lack of upregulation during OCG, and absence of enrichment in nascent fusopods of these proteins, indicate that they are not high-probability candidates for mediating pre-OC fusion. In contrast, the involvement of adseverin in OC differentiation and function (Hassanpour et al., 2014; Jiang et al., 2015) suggests that OCG at sites of inflammatory bone loss may involve regulation of adseverin expression and function. We examined whether the pro-inflammatory cytokines IL1β and TNFα regulate RANKL-dependent adseverin expression, which in turn is required for OCG and cell fusion.

We found that sRANKL (at 10 ng/ml) initiated OCG and that the majority (∼85%) of sRANKL-stimulated mouse primary bone marrow monocytes (BMMs) at day 6 were TRAcP-positive mononuclear cells. A much smaller proportion of these cells were multinucleated and were TRAcP positive (∼14% with ≥2 nuclei; ∼2% with >4 nuclei) (Fig. 1A-C). When sRANKL concentration was increased in RAW264.7 cultures, more large multinucleated OCs were formed (Fig. 1D). For all subsequent experiments, sRANKL at 10 or 30 ng/ml was used as a baseline concentration to study additional effects of the pro-inflammatory cytokines IL1β and TNFα on OCG.

The expression of adseverin protein in day 6 OCs derived from mouse primary BMMs was dependent on the dose of sRANKL. When higher concentrations of sRANKL were used (10-150 ng/ml), there were progressive increases in adseverin expression (Fig. 1E). A dose-dependent effect of sRANKL on adseverin expression was also observed in day 4 OCs derived from RAW264.7 cells (Fig. 1F). When day 4 OCs generated from RAW264.7 cells were treated with 60 ng/ml sRANKL, immunostained adseverin colocalized with actin filaments. Colocalization was evident in the cortex of mononuclear cells and in actin belts (presumptive coalesced podosomes) at the periphery of multinucleated cells (Fig. 1G). Compared with diploid cells, multinucleated cells exhibited greater fluorescent intensity of adseverin staining in the region of interest (cell periphery; Fig. 1H).

By flow cytometry analysis of cells undergoing OCG, we found a broad range of DNA ploidy (measured by DNA content in DAPI-stained cells; Fig. 2A). In DAPI-labeled cultures that were immunostained for adseverin, flow cytometry showed that relative to the parental RAW264.7 cells, populations with a DNA ploidy of >4N exhibited >5-fold more adseverin-positive cells (Fig. 2B1, B2). We sorted day 4 OCs from RAW264.7 cells according to DNA ploidy into different fractions, deposited the sorted cells on slides and examined them by confocal microscopy (Fig. 2C). These observations and flow cytometry showed that adseverin expression was increased in cells with multiple nuclei, particularly those with >8N DNA ploidy (Fig. 2D). EGFP-adseverin (EGFP-Ads)-transfected cells (Fig. 2E-G) expressed >2.5-fold more TRAcP than control pEGFP-C1-transfected cells 2 days after sRANKL stimulation (Fig. 2I). The size of day 4 cells was 2.4 times larger than that of osteoclasts generated from EGFP-transfected control cells (Fig. 2J). There was also a 1.7-fold increase in the fusion index in EGFP-Ads cells compared with EGFP-transfected controls (Fig. 2K).

Immunostaining showed that adseverin staining was enriched at cell-cell fusion contacts (Fig. 3A), indicating that adseverin may contribute to OCG. TNFα stimulation induced osteoclast formation in the absence of sRANKL in RAW264.7 cells as demonstrated by an increase in both the average TRAcP-positive cell size and the percentage of fused cells. IL1β did not induce osteoclast formation in the absence of sRANKL, but both TNFα and IL1β significantly enhanced osteoclast formation in the presence of sRANKL relative to sRANKL alone (Fig. 3B-E). These findings were in accordance with results from murine primary bone marrow cells. None of the cytokine combinations induced multinucleated cell formation in the adseverin-knockdown samples, although increased TRAcP expression was observed.

We assessed whether RANKL-mediated adseverin expression was regulated transcriptionally by conducting adseverin promoter activity assays. Briefly, a 2 kb genomic DNA fragment (Fig. 4A) upstream of the mouse adseverin transcriptional start site was cloned into pGL4.10[Luc2], a promoter-less vector with firefly luciferase as a reporter. This construct containing the adseverin promoter was introduced to RAW264.7 cells via electroporation. Cells were co-transfected with a construct expressing Renilla luciferase under the control of the HSV-thymidine kinase promoter. The expression of firefly luciferase (the reporter gene) was normalized to Renilla luminescence. We found that adseverin promoter activity increased >2-fold in cells stimulated with sRANKL compared with vehicle-treated cells (Fig. 4B), indicating enhanced adseverin transcriptional activity in OC progenitors responding to RANKL.

We then examined sRANKL-induced activation of NF-κB in OC progenitors and identified the sites in the promoter region where NF-κB likely binds by computational analysis (PROMO). Putative transcription factor binding sites were identified (Fig. 4A) and luciferase assays were performed after deletion of these predicted NF-κB binding sites by point-direct mutagenesis. Sequences (490, 1118 or 1567 bp) upstream of the transcriptional start site in the luciferase reporter vector were deleted and luciferase assays were conducted to compare promoter activities between each construct and with a construct containing the full-length promoter sequence (2 kb sequence upstream of the 5′-end of the adseverin transcript). Promoter activity was halved when the first (most upstream) predicted NF-κB binding site was deleted (P<0.01; Fig. 4D) but there were no further reductions in activity when additional predicted NF-κB binding sites were progressively deleted. Adseverin promoter activity in response to sRANKL stimulation was progressively reduced with increasing lengths of deletions in the proximal 1118 bp region (Fig. 4C).

We found that in the absence of sRANKL, BMMs exhibited multinucleated cell formation after treatment with TNFα, but not with ILβ (Fig. 5A, quantified in B). Measurement of TRAcP activity also showed 3-fold increases in TRAcP activity after TNFα treatment, but not after treatment with IL1β in the absence of sRANKL (Fig. 5C). Adseverin expression was increased dose-dependently by TNFα stimulation whereas IL1β did not enhance adseverin expression (Fig. 5D). TNFα-induced adseverin expression is quantified in Fig. 5E.

As noted above, OC differentiation is initiated by trace amounts of sRANKL (10 ng/ml) in the presence of M-CSF. At this dose of sRANKL, multiple TRAcP-positive mononuclear cells and a small number of multinucleated cells were formed in BMM cultures (Fig. 1A). When we added IL1β or TNFα along with sRANKL (10 ng/ml), OC formation was augmented (Fig. 6A, quantified in B) and TRAcP activity (Fig. 6C) was increased. Without the addition of sRANKL, IL1β exerted no effect on OCG or TRAcP activity, whereas TNFα strongly enhanced both (Fig. 5).

In cultures of RAW264.7 cells (Fig. 7A1) and primary BMMs (Fig. 7A2), both IL1β and TNFα increased sRANKL-dependent (30 ng/ml) adseverin protein expression. IL1β did not increase adseverin expression in the absence of sRANKL. When replicated (four times with RAW264.7 cells), densitometry of adseverin immunoblots normalized to β-actin (Fig. 7B) showed that adseverin protein levels in sRANKL-treated cultures incubated with IL1β and TNFα were twice that of cultures treated with sRANKL alone.

We determined whether the increased adseverin protein expression after IL1β or TNFα treatment was associated with elevated adseverin promoter activity. As described above, luciferase assays were performed by co-transfection of RAW264.7 cells with the 2 kb promoter construct and cells were treated with IL1β or TNFα for 2 days after transfection. We found that TNFα, but not IL1β, could enhance adseverin promoter activity in the absence of sRANKL (Fig. 7C). In cells also treated with sRANKL, the addition of either TNFα or IL1β augmented adseverin promoter activity (Fig. 7C). RT-qPCR indicated that TNFα, but not IL1β, increased adseverin mRNA expression in the absence of sRANKL (Fig. 7D). RANKL-induced expression of adseverin at the mRNA level was strongly enhanced by both TNFα and IL1β. The cDNA used for RT-qPCR to quantify adseverin expression was also used to measure TRAcP mRNA expression. TRAcP expression was increased in cells treated with TNFα, but not with IL1β, in the absence of sRANKL. Both IL1β and TNFα enhanced sRANKL-induced TRAcP expression (Fig. 7E).

The migration of pre-fusion osteoclasts in response to cytokines was examined. These results showed that TNFα induced pre-OC migration and also enhanced cell migration in the presence of sRANKL. IL1β did not induce significant pre-OC migration nor did it enhance migration in the presence of sRANKL (Fig. 8).

Our principal findings are that the pro-inflammatory cytokines IL1β and TNFα augment RANKL-dependent adseverin expression and OCG. While previous data showed that sRANKL can enhance adseverin expression in OCs derived from RAW264.7 cells (Hassanpour et al., 2014) and primary BMMs (Jiang et al., 2015), this is the first report to demonstrate the impact of pro-inflammatory cytokines on adseverin expression and its linkage to OCG. These findings are relevant for bone homeostasis in intact animals, since the conditional knockout of adseverin in mice prevents inflammation-driven bone loss (Jiang et al., 2015).

Adseverin is a member of the gelsolin-like family of actin-binding proteins, which sever actin filaments in a Ca²⁺-dependent manner while remaining bound to the barbed end of newly formed filaments. These properties of adseverin underpin its central role in chondrocyte differentiation, the fusion of chromaffin granules and cortical actin rearrangement. As shown here and previously (Hassanpour et al., 2014; Jiang et al., 2015), OC progenitor BMMs do not express adseverin, whereas RAW264.7 cells express trace amounts. However, during the formation of multinucleated OCs, cellular adseverin expression increases and adseverin localizes to fusion zones of adjacent cells. Although the critical molecules responsible for regulation of OC fusion are not yet well defined, dynamic remodeling of subcortical actin filaments is critical for fusion processes (Jiang et al., 2015; Takito et al., 2012). Adseverin is likely to be important in these processes since OC formation and function is impaired in adseverin-null OC progenitor cells (Hassanpour et al., 2014; Jiang et al., 2015; Yang et al., 2008). Our current study shows that adseverin overexpression in osteoclast precursor cells resulted in the formation of large osteoclasts and numerous cells that fused in response to sRANKL, highlighting the key role of adseverin in OCG. Previous studies have shown that osteoclast size is related to resorptive potential, meaning that increased cell size results in augmented resorption in vitro (Lees and Heersche, 1999). Future in vivo studies are needed to determine whether modulation of adseverin expression can be utilized therapeutically to aid in treating diseases that involve aberrant bone remodeling such as periodontitis and Paget's disease.

OCs originate from common myeloid progenitor cells that also give rise to monocytes/macrophages and dendritic cells. Under the influence of local growth and differentiation factors and cytokine signals, OC progenitors have the potential to differentiate into OCs. In concert with M-CSF, RANKL enables OCG in vitro from primary OC progenitors (Quinn et al., 1998). While OCG in vitro can be induced by trace amounts (∼1 ng/ml) of sRANKL (Lin et al., 2007), commonly used concentrations of sRANKL are 10-200 ng/ml (Lee et al., 2015; Uchiyama and Yamaguchi, 2004). Here, we show that TNFα (but not IL1β) is a potent inducer of OCG and adseverin in the absence of sRANKL, which is consistent with earlier data that TNFα, but not IL1β, can promote the differentiation of OC progenitors into mature, TRAcP-positive multinucleated cells in the absence of RANKL (Azuma et al., 2000; Fuller et al., 2006; Kim et al., 2005; O'Brien et al., 2016). In addition, promoter analyses showed that RANKL-enhanced adseverin expression involved the transcriptional regulation of the adseverin promoter. The impact of the NF-κB binding site on the transcriptional activation of adseverin further supports the importance of NF-κB signaling in RANKL-induced OC formation (Wei et al., 2002). Collectively, these findings indicate that adseverin is an important determinant of OC formation during inflammation.

IL1β and TNFα are critically important pro-inflammatory mediators that are implicated in the enhanced expression of RANKL (Lacey et al., 1998; Yasuda et al., 1998) by osteoblasts, fibroblasts and activated T-cells (Gravallese et al., 2000; Kong et al., 1999; Kotake et al., 2001; Wei et al., 2005). In the presence of sRANKL, IL1β and TNFα can both independently co-stimulate OCG (Algate et al., 2016; Ochi et al., 2007). TNFα, but not IL1β, can induce OCG in a RANKL-independent manner (Azuma et al., 2000; Fuller et al., 2006; Kim et al., 2005; O'Brien et al., 2016) by activating OC progenitors via TNF receptor type I (TNF-R1; p55) and type II (TNF-R2; p75) (Abu-Amer et al., 2000; Azuma et al., 2000). We examined dosage effects of IL1β and TNFα on adseverin protein expression in sRANKL-treated cells and found that TNFα and IL1β, at a concentration as low as 5 ng/ml, can increase adseverin protein expression. The maximum effect of IL1β and TNFα on the promotion of sRANKL (10 ng/ml)-induced adseverin protein expression was at 15 and 30 ng/ml, respectively. In a separate experiment, TNFα at 50 ng/ml induced even more robust adseverin protein expression. The concentrations of cytokines used in these experiments were within the pathophysiological ranges that have been measured in inflamed sites in intact animals (Matsuda et al., 2016).

Adseverin knockdown prevented the formation of multinucleated cells that would otherwise be induced by TNFα or RANKL, suggesting that adseverin expression is a key step in triggering osteoclast differentiation and fusion, and may be an important regulator of the early steps in OCG. This conjecture is in agreement with previous data in which inhibition of OCG was observed after knockdown of adseverin in RAW264.7 cells (Hassanpour et al., 2014). Adseverin expression may also play a key role in inflammation-mediated bone loss. In mice in which adseverin expression was ablated and periodontitis was experimentally induced, alveolar bone loss was significantly reduced (Jiang et al., 2015). However, other results have shown no difference in bone mineral content and bone mineral density between healthy wild-type and adseverin-null mice, suggesting that adseverin knockdown does not alter bone metabolism in health (Cao et al., 2017). While further study is evidently needed to clarify the role of adseverin in health and disease, cytokine-enhanced adseverin expression may play a key role in inflammatory-mediated bone loss that is different to its role in healthy individuals. Exposure of osteoclast progenitors to increased levels of pro-inflammatory cytokines (e.g. TNFα, IL1β) as seen in periodontitis and other inflammatory diseases, may enhance adseverin expression and osteoclastic activity, thereby contributing to excessive bone resorption.

Osteoclastic cell-cell fusion requires that pre-OCs meet and establish cell-cell contact. Increased cell fusion mediated by adseverin in vitro may be the result of increased cell migration, which in turn enhances the probability of pre-OC and OC encounters. We found that IL1β cannot directly stimulate migration of osteoclast precursors, in contrast to TNFα, which promoted chemotaxis. Despite these findings, IL1β enhanced RANKL-dependent osteoclast formation and adseverin expression, suggesting that IL-1 plays an important role in post-migratory events in OCG. The ability of IL1β to enhance RANKL-dependent osteoclast formation may be due to augmented cell fusion when pre-OCs are already in close proximity to one another. Notably, IL1β and TNFα synergistically stimulate fibroblasts to produce monocyte chemoattractant protein-1 [MCP-1), a protein that is chemotactic for OC precursors (Ozaki et al., 1996)]. Furthermore, MCP-1 and complement intermediate product 5α, which are both abundant in inflamed periodontal tissues, exhibit a concentration-dependent effect on OC chemotaxis (Niwa et al., 2013). The chemotactic response of OCs to various cytokines differs depending on the stage of osteoclast differentiation (Niwa et al., 2013). Future studies should ascertain whether IL1β can induce pre-OC migration through indirect mechanisms possibly involving cross-talk between multiple cell types, and whether the effect is time- and concentration-dependent.

In cultured cells undergoing OCG in vitro, a pool of OC progenitors responds asynchronously to RANKL stimulation; as a result, a wide range of large and small OCs with varying levels of cell differentiation co-exist (Trebec et al., 2007). Separation of large OCs from small OCs in mixed cell populations is technically challenging, in part because there are no distinguishing cell membrane markers for OC subpopulations of different sizes. Accordingly, flow or magnetic cell isolation of OC subpopulations based on membrane marker expression is not feasible. While low-speed centrifugation of cells over fetal bovine serum gradients has been used to isolate cultured OCs (Trebec et al., 2007), the purity of the isolated cell fractions is low. In contrast, we used a separation method based on DNA content. Our flow cytometric method provides an alternative way to isolate OC subpopulations based on ploidy status, and demonstrated higher adseverin content in cells with multiple nuclei. This approach could be useful for studying OC differentiation and maturation.

We found that RANKL-induced upregulation of adseverin protein expression in OCs is mediated in part at the transcriptional level, since adseverin promoter activity was markedly induced by sRANKL stimulation. In addition, although not statistically significant, there was a trend of enhanced adseverin promoter activity after TNFα stimulation in the absence of sRANKL. Furthermore, and as expected, there was no increase of adseverin promoter activity by IL1β in the absence of sRANKL, which is consistent with the failure of IL1β alone to induce adseverin expression and OC formation.

Bone destruction in inflammatory disorders has been observed for decades (Spector et al., 1993) and studied intensively (Hardy and Cooper, 2009). Several mechanisms have been proposed to explain the relationship between chronic inflammation and loss of bone homeostasis. Our findings showing that adseverin expression is driven by inflammatory mediators, and that adseverin regulates cell fusion in OCG associated with bone loss in periodontitis, suggest suggestING that adseverin may be an important potential therapeutic target for treating inflammatory-related bone disorders.

All animal experiments were approved by the University of Toronto Animal Care Committee. OCG was studied in mouse primary bone marrow monocytes (BMMs) harvested from 2-month-old wild-type C57Bl/6 male mice or RAW264.7 cells (ATCC TIB-71, passage 3-11) as described (Hassanpour et al., 2014; Jiang et al., 2015). Cells from the mouse monocyte/macrophage cell line RAW264.7 can differentiate into OCs (Hsu et al., 1999) and are widely used to study OCG. Primary BMMs were differentiated into OCs by treatment with recombinant murine M-CSF (PeproTech, cat. no. 315-02) and murine sRANKL (aa157-316) expressed and purified from BL21/DE3 E. coli harboring pGEX-4T1-sRANKL (SalI/NotI) for 6 days. RAW264.7 cells, free of contamination from other cell lines and bacteria, were cultured in the presence of sRANKL for the indicated incubation period and conditions, but without exposure to exogenous M-CSF as they synthesize and release M-CSF endogenously. The methods used to generate adseverin stable knockdown and wild-type control RAW264.7 cell lines were described previously (Hassanpour et al., 2014). Total BMMs were seeded in triplicate in 8-well chamber slides (Falcon, 1×10⁶ cells/well, 700 µl medium), 96-well plates (0.5×10⁶ cells/well, 200 µl medium) or 6-well plates (2-5×10⁶ cells/well, 4 ml medium). Cells were exposed to M-CSF (20 ng/ml) and/or sRANKL (60 ng/ml), with/or without pro-inflammatory cytokines TNFα (PeproTech cat. no. AF-315-01A) or IL-1β (R&D Systems, cat. no. 201-LB). Cell culture medium and cytokine(s) at the same concentration were replaced every other day. Cell differentiation in the chamber slides was assessed by positive tartrate-resistant phosphatase (TRAcP) expression. Cells cultured in 96-well plates were lysed for TRAcP solution assays, and cells cultured in 6-well plates were lysed for immunoblot analysis. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) and minimum essential medium-α (α-MEM) with 10% fetal bovine serum (FBS) and antibiotics (164 IU/ml penicillin G, 50 µg/ml gentamicin and 0.25 µg/ml fungizone).

Mouse adseverin complementary DNA (cDNA; a gift from Johan Robbens) (Robbens et al., 1998) was amplified by polymerase chain reaction (PCR), using a primer pair: 5′-GGTTCCGCGTGGATCCATGGCGCAGGAGCTGCAGCACCCC-3′ (forward, F) and 5′-GATGCGGCCGCTCGAGTTTACCACCTGCTGGAGTCCCAGC-3′ (reverse, R). The amplified PCR product was ligated to the BamHI/XhoI site of pGEX-4T-2 using In-fusion HD enzyme premix (Clontech). The insert was confirmed by sequencing (ACGT Corp., Toronto, ON). The construct was transferred into BL21/DE3 E. coli for protein expression and purification. Cells were lysed and glutathione S-transferase (GST)-tagged protein was prepared. GST-free product was prepared by cleavage of the purified, GST-fused adseverin with thrombin and then by chromatography on a glutathione-Sepharose 4B column. This purified protein was used to prepare rabbit polyclonal antibodies, which were collected with protein A beads.

The sequence of mouse adseverin promoter was obtained from Ensembl (GenBank acc. no. U04354). Genomic DNA extracted from RAW264.7 cells was used as a template to amplify this region (−1 to −2000 bp upstream of transcription start site) using DNA polymerase (Phusion high-fidelity) and primer pair: 5′-GCGCCTCGAGAGGTCTATGACTATGGAGTATTGATATTCT-3′ (F) and 5′-GCGCAAGCTTGACCGCGAACCTGCTCTCCCGGCTTTAAAG-3′ (R). A terminal adenosine overhang was added to the resulting PCR-amplified 2 kb fragment. The fragment was inserted into pGEM T-easy vector and further cloned into the XhoI/HindIII site of pGL4.10[Luc2] (Promega). The resulting construct is described as pGL4.10[luc]-Ads-Promo-2 kb. The validity of the promoter was confirmed by sequencing (ACGT Corp., Toronto, ON).

Nuclear factor-kappa B (NF-κB)-binding sites in the promoter region of mouse adseverin genomic DNA were predicted by an online tool (PROMO: http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3). A modified site-directed mutagenesis method (Liu and Naismith, 2008) was used to generate constructs containing either a truncated promoter region or deletion of the predicted NF-κB binding sites. The truncated constructs were designated as −496, −1118 or −1567 based on the number of base pairs deleted upstream from the transcription start site. In order to locate the potential transcription factor binding sites (TFBS), putative binding sites were removed from pGL4.10[luc]-Ads-Promo-2 kb using the following respective primer pairs: 5′-CCTTTGTACAAAGCTTGGCAATCCGGTACTGTTGGTAAAGCCAC-3′ (F) and 5′-AAGCTTTGTACAAAGGTACGTCAGGTTTTCTAGAAAAGATAAAA-3′ (R); 5′-TGCTAGGGAGAAGCTTGGCAATCCGGTACTGTTGGTAAAGCCAC-3′ (F) and 5′-AAGCTTCTCCCTAGCAAGCCATTAATAGTAAAGGCACACTTAGA-3′ (R); 5′-GTCTAACAAAAAGCTTGGCAATCCGGTACTGTTGGTAAAGCCAC-3′ (F) and 5′-AAGCTTTTTGTTAGACAGGTTCGCATGTAGCACATGTAGCCCAC-3′ (R).

The predicted first three NF-κB binding sites (site 1-GGCTTCCC; 2-TCAGGTTCCC; 3-GGGAAGGAAC) upstream of adseverin transcription start site in pGL4.10[luc]-Ads-Promo-2 kb were progressively deleted using primer pairs 5′-CGTGGCTGTTTTCCAGCCCCGCCTCCGCTTCTGGCGAGCCGCTT-3′ (F) and 5′-CTGGAAAACAGCCACGCCCGAACCCGGGCGGGAACCTGAATCCC-3′ (R); 5′-AAACTGGGATGCCCGGGTTCGGGCGTGGCTGTTTTCCAGCCCCG-3′ (F) and 5′-CCGGGCATCCCAGTTTGGACCAAGTTAACCAAAACTCTAGTAAA-3′ (R); 5′-AACACCAGCATGGGGTGACAGGGACCAGTGTGGACCCGCCCGAG-3′ (F) and 5′-ACCCCATGCTGGTGTTCACCCTCTGATGCCAGGGTGAGGAGAAA-3′ (R). The resulting constructs were used to locate NF-κB binding sites activated by RANKL.

TRAcP staining was performed as described (Lojda et al., 1979). Briefly, cells were rinsed twice with pre-warmed phosphate buffered saline (PBS), fixed with 4% paraformaldehyde (PFA) in PBS for 20 min at room temperature (RT), and incubated in TRAcP staining solution at 37°C for 15 min. Cells were counterstained with DAPI (4′,6-diamidino-2-phenylindole) and imaged with a Nikon microscope (Eclipse E1000). TRAcP solution assays were performed as described (Tintut et al., 2002). Cells cultured in 96-well, flat-bottom tissue culture plates were washed once with pre-warmed PBS and lysed at room temperature (RT) while shaking in 80 µl/well cold citrate buffer (pH 4.8) containing 80 mM sodium tartrate and 0.1% Triton X-100. After lysis, cells were incubated in the same buffer with 20 mM p-nitrophenyl phosphate (80 µl/well) for 3-5 min while shaking. The reaction was stopped by adding 40 µl of 0.5 M NaOH. Optical densities of solutions were obtained at 405 nm with a microplate reader.

RAW264.7 cells (2×10⁶) were co-transfected with adseverin promoter constructs expressing firefly luciferase (2 µg) and with a pRL-TK loading control construct expressing Renilla luciferase (20 ng or 2 µg; provided by Andras Kapus, St. Michael's Hospital, Toronto, ON), using Amaxa Nucleofector (program D-32) and Kit V (VCA-1003, Lonza). The transfected cells were incubated in 6-well plates for 2 days and treated with vehicle or with sRANKL (10 to 120 ng/ml) for 4-6 h in the presence or absence of IL1β or TNFα. In some experiments, cells were treated with these cytokines before transfection and re-stimulation. Cells were lysed and luciferase activity was measured using a dual luciferase assay kit (E1960, Promega) with a Lumat 3 luminometer (LB9508, Berthold Technologies). Luciferase activity associated with the adseverin promoter constructs was normalized to the activity of the internal transfection control (Renilla luciferase).

Cells were rinsed with pre-warmed PBS, fixed with 4% paraformaldehyde, rinsed three times with PBS, incubated in 100 mM glycine in PBS for 10 min to reduce autofluorescence, permeabilized with 0.1% Triton 100 in PBS (5 min, RT) and blocked with 1% bovine serum albumin (BSA) in 0.1% Triton X-100 in PBS for 30 min. Adseverin expression was detected by indirect immunofluorescence with rabbit anti-mouse adseverin antibody (1:1000; produced in our lab) in 1% BSA and 0.1% Triton X-100 in PBS at RT for 1 h. After extensive washing, cells were incubated for 1 h in Alexa Fluor 488-conjugated goat anti-rabbit IgG solution (Molecular Probes, 1:1000 in BSA/Triton/PBS) at RT. Actin filaments were stained with tetramethylrhodamine B isothiocyanate (TRITC)-conjugated phalloidin (Sigma, cat. no. P-1951). Cells were counterstained with DAPI.

OCs derived from RAW264.7 cells at days 4-6 were detached with EDTA (2 mM) in pre-warmed PBS and fixed with 70% ethanol at −20°C for 20 min, washed with PBS, incubated with adseverin antibody (1:1000 in 0.5% FBS in PBS), stained with goat anti-rabbit IgG-Alexa Fluor 488 (1:1000 in 0.5% FBS/PBS) and counterstained with DAPI. The sorted cells were deposited on the center of glass slides directly or on slides by Cytospin for microscopy. In some experiments, labeled cells were analyzed by flow cytometry to estimate adseverin expression.

Mouse adseverin cDNA was subcloned into pEGFP-C1 (Clontech, provided by Richard Collins and William Trimble, SickKids, Toronto, ON) XhoI/BamHI site, using the following primer pair: 5′-CCGCTCGAGAAATGGCGCAGGAGCTGCAGCACCCCGAGT-3′ (F) and 5′-CGCGGATCCCTATTACCACCTGCTGGAGTCCCAGCCCAGA-3′ (R). The sequence of the adseverin cDNA insert was confirmed by ACGT Corp. (Toronto, ON). RAW264.7 cells (1×10⁶) were transfected with either pEGFP-C1 vector (2 µg) or pEGFP-C1-Ads (2 µg) using Amaxa Nucleofector, Cell Line Nucleofector Kit V and program D-032. The expression of EGFP and EGFP-Ads in the transfected unstimulated cells was confirmed by western blot analysis and by fluorescence microscopy 2 and 3 days after transfection, respectively. The transfected cells were incubated in 800 µl of complete DMEM containing 30 ng/ml sRANKL in triplicate, using 8-well chamber slides (1×10⁴ cells/well) for 4 days to induce osteoclast formation. Cell culture medium was changed on day 2. The cells were fixed with 4% PFA and TRAcP stained on day 4. Randomly obtained images (n=30) of OCs derived from three independent pEGFP-C1-Ads or pEGFP-C1 transfections were obtained. The sizes of all TRAcP-positive cells with ≥3 nuclei in the selected images were measured using ImageJ. The cell fusion index (number of nuclei in multinucleated osteoclasts/total nuclei per microscopic field×100) in the selected fields was also determined.

Total bone marrow cells were incubated in complete α-MEM overnight, collected and cultured in 60-mm-diameter Petri dishes (5×10⁶ cells/dish). Cells were stimulated with M-CSF (20 ng/ml; 2 days) to promote differentiation and attached cells were treated with sRANKL (60 ng/ml), M-CSF+sRANKL+TNFα (10 ng/ml) and M-CSF plus sRANKL+IL1β (20 ng/ml) for 4 days. Medium was changed on day 2 of incubation. Total RNA (1 µg) extracted from cells (RNeasy Mini Kit; Qiagen) was reverse transcribed to cDNA for quantitative PCR as described (Wang et al., 2013). TRAcP mRNA expression was quantified with the use of a primer pair 5′-ACGGCTACTTGCGGTTTCA-3′ (F) and 5′-TCCTTGGGAGGCTGGTCTT-3′ (R) (GenBank acc. no. NM_001102405, amplicon size 138 bp) (Wang et al., 2008). A primer pair 5′-CCTATGGTGACTTTTACGTCGG-3′ (F) and 5′-CTCATCCTGGGAACACTCCTT-3′ (R) was used for adseverin amplification (GenBank acc. no. NM_001146196, amplicon size 116 bp) and primer pair 5′-CCTTCCGTGTTCCTACCCC-3′ (F) and 5′-GCCCAAGATGCCCTTCAGT-3′ (R) (Wang et al., 2008) was used to amplify the internal control gene GAPDH (GenBank acc. no. M32599, product size 131 bp).

Cells were lysed with RIPA buffer [with 1× proteinase inhibitor cocktail (Sigma) and 1 mM phenylmethylsulfonyl fluoride] for 10 min on ice, scraped, transferred to sterilized Eppendorf tubes, vortexed and the lysates were centrifuged (16,000 g at 4°C for 5 min) to remove cell debris. The total protein concentration in supernatants of cell lysates was measured with Pierce BCA protein assay kit (Thermo Scientific, cat. no. 23225). Samples were then mixed with Laemmli buffer, boiled for 10 min and loaded (10-20 µg/lane) on 10% acrylamide gels for separation. Separated proteins were transferred to nitrocellulose membranes and blocked with 5% fat-free milk powder in Tris-buffered saline with Tween-20 (TBST) buffer at RT for 1 h. Membranes were incubated sequentially with primary antibody (diluted in blocking buffer) overnight at 4°C with shaking and later with fluorescently labeled secondary antibodies (diluted in TBST; 1 h at RT while shaking). Membranes were washed for 3×10 min with TBST after antibody incubations. The primary antibodies were rabbit anti-mouse adseverin (1:20,000; produced in our lab) and anti-mouse β-actin (clone AC-15, Sigma, cat. no. A5441, 1:8000); secondary antibodies were IRDye 800CW goat anti-rabbit IgG (1:10,000) and IRDye 680RD goat anti-mouse IgG (1:10,000). Membranes were imaged with an Odyssey Fc detection system (Li-Cor Biotechnology, Lincoln, NE).

RAW264.7 cells were stimulated with 60 ng/ml sRANKL for 2 days in 10 cm² Petri dishes to induce pre-OC formation. Cells were detached by scraping, counted and re-plated onto polycarbonate membrane Transwell supports with 8 µm pores (Costar, Corning, NY; 5×10⁴ cells/well in 200 µl medium), which were pre-equilibrated in 600 µl complete DMEM overnight in 24-well plates. The cells were incubated for 2 h to allow for adhesion to the membranes. Chemoattractants were added below the membranes at the following concentrations and combinations: 0 ng/ml, 100 ng/ml sRANKL, 30 ng/ml IL1β, 60 ng/ml TNFα, 100 ng/ml sRANKL+30 ng/ml IL1β and 100 ng/ml sRANKL+60 ng/ml TNFα. After 20 h, cells were fixed and stained with Alexa Fluor 488 phalloidin (Invitrogen Molecular Probes, cat. no. A12379) and DAPI for microscopy.

Means and standard errors were calculated for all continuous variables. Bartlett's test was used to test for homogeneity of variance. Analysis of variance (ANOVA) was used to examine differences between multiple groups. Post hoc comparisons between groups were evaluated with Tukey's test. The statistical analysis was performed with either Excel Data Analysis Add-in (Microsoft Office Professional Plus 2010) or R programming language (version 3.4.2). Bar plots were coded in R using the bar plot function. All experiments were conducted with at least three biological replicates and in some instances up to six biological replicates (e.g. flow cytometry).

We thank Morris Manolson for the pGEX-4T1-sRANKL construct.