Activation of P2X7 receptors causes isoform-specific translocation of protein kinase C in osteoclasts

Osteoclasts are multinucleated cells of the monocyte-macrophage lineage that are responsible for the extracellular resorption of mineralized tissues, including bone, teeth and mineralized cartilage (Novack and Teitelbaum, 2008). Bone is continually remodeled by the coupled processes of resorption and formation, carried out by osteoclasts and osteoblasts, respectively, and maintenance of bone mass is strictly dependent on their balance. Hence, defects in either process, or in their coupling, leads to skeletal disorders, such as osteoporosis. In both normal and pathological conditions, bone resorption can be regulated by changes in the recruitment and differentiation of osteoclast precursors, and in the resorptive activity and life span of osteoclasts (Manolagas, 2000; Novack and Teitelbaum, 2008).

Nucleotides are important signaling molecules released in response to mechanical perturbation of many tissues and act through P2 nucleotide receptors (Burnstock, 2007). P2Y receptors signal through heterotrimeric G proteins, whereas P2X receptors are ligand-gated ion channels that are non-selective for cations and, in many cases, mediate Ca²⁺ influx (Khakh and North, 2006; North, 2002). ATP is a potent stimulator of osteoclast formation and resorptive activity (Morrison et al., 1998), although the subtypes of receptors mediating these effects are not resolved. Osteoclasts express multiple subtypes of P2Y and P2X receptors (Korcok et al., 2007), including P2X7 (Naemsch et al., 2001; Grol et al., 2009). The P2X7 receptor has been shown to play an important role in bone remodeling in vivo, based on the skeletal phenotype of P2rx7^–/– mice (Ke et al., 2003). Compared with wild type, femurs from P2rx7^–/– mice show similar length but excessive trabecular bone resorption and decreased bone density, accompanied by increased osteoclast number. The excessive bone resorption is in keeping with the known role for the P2X7 receptor in mediating cell death in some systems (Khakh and North, 2006). However, P2X7 receptors activate the pro-survival transcription factor NF-κB in osteoclasts (Korcok et al., 2004), suggesting that P2X7 has diverse roles controlling osteoclast formation and activity.

Protein kinase C (PKC) proteins make up a family of cytoplasmic serine/threonine protein kinases that are activated by diverse stimuli and phosphorylate many targets, including ion transporters, enzymes and signaling proteins. The PKC isozymes in this family can be classified into three groups: classical, novel and atypical (Dempsey et al., 2000; Newton, 2003; Reither et al., 2006). The activity of conventional protein kinase isozymes (cPKCs; α, βI, βII and γ) is stimulated by diacylglycerol, Ca²⁺ and phosphatidylserine; that of novel isozymes (nPKCs; δ, ϵ, η and θ) by diacylglycerol and phosphatidylserine; whereas atypical PKC isoforms (aPKCs; ζ and ι) are activated by phosphatidylserine, but are independent of Ca²⁺ and phorbol esters. Several PKC isozymes have been identified in osteoclasts, including α, δ and ϵ (Teti et al., 1995), with more recent evidence also revealing the expression of β and ζ in these cells (Williams et al., 2000; Rucci et al., 2005).

The activity of PKC is regulated by phosphorylation, cofactors and interaction with anchoring proteins. Catalytically competent PKC is maintained in an inactive state in the cytosol, by binding of the pseudosubstrate domain to the substrate-binding domain. In the presence of diacylglycerol or phorbol esters and Ca²⁺ (for conventional isoforms), PKC is recruited to lipid membranes, leading to dissociation of the pseudosubstrate domain, which in turn permits substrate binding and phosphorylation (Newton, 2003). PKC regulates numerous proteins in the plasma membrane, including cell-surface receptors and ion channels, in turn influencing downstream signaling activities. PKC activity can be monitored using phosphorylation assays; however, PKC activation might be transient or be localized to specific compartments within a cell. Live-cell imaging of the translocation of PKC linked to a fluorescent protein overcomes these limitations and enables investigation of spatiotemporal aspects of PKC signaling (Sakai et al., 1997; Oancea and Meyer, 1998).

Little is known about whether the P2X7 receptor signals through PKC. In rat parotid-acinar-salivary-gland cells, P2X7 agonists induce the tyrosine phosphorylation of PKCδ (Bradford and Soltoff, 2002). Moreover, in rat submandibular-gland ductal cells, stimulation of P2X7 receptors leads to the activation of phospholipase D (PLD). The PKC inhibitors calphostin C and chelerythrine partially block P2X7-induced activation of PLD, suggesting the involvement of PKC in this process (Pochet et al., 2003). However, it was shown subsequently that chelerythrine is an antagonist of P2X7 receptors (Shemon et al., 2004), calling into question whether P2X7 actually activates PKC in submandibular-gland ductal cells. Thus, the ability of P2X7 receptors to couple to PKC signaling warrants further study.

The P2X7 receptor has been suggested to play a role in cell fusion (Lemaire et al., 2006) and to regulate cell survival (Adinolfi et al., 2005). In osteoclasts, P2X7-receptor activation leads to an activity-dependent elevation of cytosolic free Ca²⁺ concentration ([Ca²⁺]_i) (Naemsch et al., 2001). Because PKC is a crucial transducer of Ca²⁺ signaling in other cell types (Reither et al., 2006), we investigated whether P2X7 receptors couple to activation of PKC in osteoclasts. Using RAW 264.7 cells differentiated into multinucleated osteoclast-like cells and osteoclasts derived from murine bone-marrow cells, we monitored EGFP-tagged PKC isozymes using live-cell confocal imaging. We established that P2X7 receptors induce Ca²⁺-dependent translocation of PKCα and βI, but not δ, to the basolateral membrane domain of osteoclasts. This response is unique to P2X7 receptors because BzATP failed to induce translocation of PKC in osteoclasts derived from P2rx7^–/– mice.

Freshly isolated osteoclasts from rat and mouse express functional P2X7 receptors as assessed by Ca²⁺ fluorescence and patch-clamp electrophysiological recording (Naemsch et al., 2001; Ke et al., 2003). To confirm functional P2X7 receptors on RAW-267.4-derived osteoclast-like cells, we used the Ca²⁺-sensitive dye Fura-2 to monitor changes of [Ca²⁺]_i. Stimulation of osteoclast-like cells with the relatively potent P2X7-receptor agonist BzATP (150 μM) elicited elevation of Ca²⁺ concentration (not shown).

We next used real-time imaging of live cells by confocal microscopy to evaluate the localization of EGFP-tagged PKC isozymes. We expressed PKCα, PKCβI and PKCδ, as well as EGFP alone, which served as a control. PKCα-EGFP was uniformly located in the cytoplasm in unstimulated osteoclast-like cells, although notably not in the nuclei (Fig. 1A, upper panel). Because of the three-dimensional (3D) profile of osteoclasts, it was necessary to examine the localization of PKCα-EGFP by recording a z-stack of each cell. When displayed in the x-z plane, PKCα-EGFP was evenly distributed within the cytoplasm (Fig. 1A, lower panel). After treatment with BzATP (150 μM), we observed prompt translocation of PKCα-EGFP to the plasma membrane, with fluorescence now apparent as an annular ring at the periphery when assessed midway in the cell (Fig. 1B, upper panel). Inspection of the z-stack revealed that PKCα-EGFP was strictly localized to the upper membrane of the cell, which is referred to as the basolateral membrane (Fig. 1B, lower panel).

To quantify this change in cellular localization, we measured EGFP fluorescence intensity profiles across cells, avoiding the nuclei (Fig. 1A,B, diagonal white lines in upper panels). The fluorescence intensity (FI) showed uniform distribution across the cytosol of control cells, with no noticeable enrichment at the cell periphery (Fig. 1C). By contrast, the profile in BzATP-treated cells showed marked increases in fluorescence intensity near the plasma membrane (Fig. 1D). To quantify this change, we determined the average pixel intensity in the perimembrane regions (F_m) to the average pixel intensity in the cytosol (F_c). Values of the ratio F_m:F_c exceeding 1 were taken to indicate membrane localization of EGFP, and are plotted in subsequent figures. We observed this translocation of PKCα-EGFP in response to BzATP in 88% of osteoclast-like cells studied (Table 1).

We have shown in previous studies that prolonged applications of BzATP or a high concentration of ATP (1 mM) induced an initial transient increase followed by sustained elevation of [Ca²⁺]_i (Naemsch et al., 2001). Therefore, we characterized the kinetics of PKCα translocation in response to BzATP stimulation. Within 60 seconds of stimulation with BzATP, PKCα translocated from the cytosol to the plasma membrane, with recovery evident 8-10 minutes after stimulation even in the continued presence of BzATP (Fig. 2; supplementary material Movie 1). Membrane localization, measured as F_m:F_c, showed rapid onset of the response, as well as recovery (Fig. 2B). Note that zero time corresponds to the image of the cell under control conditions. Subsequent agonist application occurred at the time indicated by the vertical broken line in Fig. 2B, and is given in the figure legend. Near-complete restoration of cytosolic localization was evident at longer times. The average duration at half-maximal amplitude of membrane localization of PKCα in the continued presence of BzATP was 258±26 seconds (n=16 cells). Note that analysis using the ratio F_m:F_c compensated for the photobleaching of EGFP that was encountered with long-term imaging of live cells (Fig. 2A, compare panels at 0 and 750 seconds).

Although BzATP is a more potent agonist than ATP at P2X7 receptors, it is known to activate a number of P2 receptors in addition to P2X7 (North, 2002). Therefore, responses to BzATP alone do not establish the involvement of P2X7 receptors. To identify the receptor mediating the effects of BzATP on PKC translocation in osteoclasts, we first tested the effects of agonists at other P2 receptors. Osteoclast-like cells were stimulated with low concentrations of ATP (10 μM), which activates P2X4 and P2Y2 receptors, or UTP (150 μM), a P2Y2-receptor agonist, and responses were compared with those elicited by high concentrations of ATP (3 mM) or BzATP (150 μM), both of which activate P2X7 receptors. Membrane localization was assessed by live-cell imaging and quantified as described above. PKCα translocation was not stimulated by the addition of control vehicle or P2Y agonists. By contrast, translocation was observed in response to a high concentration of ATP or BzATP (Fig. 3). Moreover, translocation of PKCα-EGFP was inhibited by Brilliant Blue G (Fig. 3), a selective P2X7 antagonist (Jiang et al., 2000).

To directly examine the role of P2X7 receptors in mediating BzATP-induced translocation of PKC, osteoclasts were generated from the bone marrow of P2rx7^–/– and wild-type mice. PKCα-EGFP was expressed by adenoviral transfection, and showed typical cytoplasmic localization in untreated osteoclasts (Fig. 4A, left). Stimulation of wild-type osteoclasts with BzATP caused PKCα translocation to the membrane, with essentially the same time course and characteristics as shown above for RAW-267.4-derived osteoclast-like cells (Fig. 4A, right). Reconstructed 3D images of fixed cells at high resolution revealed selective localization of PKCα-EGFP to basolateral membranes of wild-type murine osteoclasts that were stimulated with BzATP (supplementary material Fig. S1 and Movie 2). By contrast, osteoclasts from P2rx7^–/– mice showed cytoplasmic localization of PKCα under control conditions and BzATP caused no change (Fig. 4B, left; Table 1). As a positive control, we confirmed that treatment of P2rx7^–/– osteoclasts with the PKC activator phorbol myristate acetate (PMA) induced membrane translocation (Fig. 4B, right; Table 1). Thus, pharmacological and genetic evidence establishes that nucleotide-induced PKC translocation in osteoclasts arises specifically from the activation of P2X7 receptors.

A series of control experiments was also carried out to further characterize the translocation of PKCα in osteoclast-like cells. Vehicle treatment alone had no effect on PKCα-EGFP localization (Fig. 5Ai, with representative images shown on the left and membrane localization quantified on the right; representative of 11 cells). Control expression of EGFP alone showed localization throughout the cytosol and in the nuclei (Fig. 5Aii), and no redistribution occurred upon stimulation with BzATP (representative of nine cells). Treatment of cells with PMA resulted in prompt and sustained translocation of PKCα-EGFP (Fig. 5Aiii). Inspection of the z-stack revealed that PKCα-EGFP was localized to the basolateral membrane following treatment with PMA (Fig. 5B, see lower profiles, representative of 36 cells), resembling the cellular localization induced by BzATP. The osteoclast illustrated in Fig. 5B exhibits a central vacuole, which, similar to nuclei, is devoid of PKCα-EGFP. Moreover, PKCα does not translocate to the vacuolar membrane, further emphasizing the specificity of translocation reported here.

Activation of cPKC isozymes involves a rise in [Ca²⁺]_i, so we monitored PKCα-EGFP location concurrently with [Ca²⁺]_i by loading osteoclast-like cells with the Ca²⁺-sensitive dye Fura-red. Fura-red excitation is at 488 nm and emission at 660 nm, enabling us to simultaneously monitor PKCα-EGFP and Ca²⁺ concentration using different channels of the confocal microscope. BzATP-induced translocation of PKCα-EGFP was preceded by a rise of [Ca²⁺]_i, and the kinetics and time course of both events were closely matched (Fig. 6A). Fura-red emission intensity decreases with an increase of [Ca²⁺]; therefore, we presented the data as a ratio of initial fluorescence intensity (F₀):fluorescence intensity (F) (Fig. 6A, axis at right). We noted that some osteoclast-like cells exhibited repetitive translocation of PKCα-EGFP in the continued presence of BzATP (Fig. 6B,C, representative of 18 out of 68 cells).

P2X7 receptors are ligand-gated channels that mediate Ca²⁺ entry in osteoclasts (Naemsch et al., 2001), so we examined the dependence of PKC translocation on extracellular Ca²⁺. When osteoclast-like cells were bathed in solution containing EGTA to chelate Ca²⁺ in the medium, the addition of BzATP caused no change in localization of PKCα-EGFP (Fig. 7, representative of responses in 14 cells; Table 1). As a positive control, we then treated cells with PMA and did observe prompt translocation of PKCα to the membrane (Fig. 7A, panel at right, with time course in Fig. 7B).

To evaluate isozyme specificity, we examined responses of another conventional isozyme, PKCβI, and the Ca²⁺-independent isozyme PKCδ in osteoclast-like cells. PKCβI-EGFP was initially distributed uniformly in the cytosol, although not in the nuclei, with essentially the same spatial localization reported above for PKCα in unstimulated cells. Addition of BzATP caused prompt translocation of PKCβI-EGFP to the cell periphery, with recovery evident at later times (Fig. 8A, summarized in Table 2). Following stimulation, PKCβI-EGFP was restricted to the basolateral membrane of the osteoclast-like cells, as is shown in the z-stack images in the lower panels of Fig. 8A. This selective translocation to one membrane compartment further confirmed the similarity of this response to that of PKCα. However, we did note differences in the kinetics of the translocation, with PKCβI showing more-rapid recovery than PKCα, as illustrated in the analysis of F_m:F_c (Fig. 8B, left panel). To quantify the time course of membrane localization, we determined the duration at half-maximal amplitude (T₅₀). The duration of membrane localization of PKCβI was significantly briefer than that of PKCα (Fig. 8B, right), reflecting an isozyme-specific pattern.

In view of our findings that Ca²⁺ plays an essential role in the translocation of PKCα in osteoclast-like cells, we examined the spatial distribution of a Ca²⁺-independent, novel isozyme, PKCδ. Under resting conditions, PKCδ-EGFP was largely localized in the cytosol of osteoclast-like cells (Fig. 9A), similar to other isozymes. However, a low level of PKCδ-EGFP was consistently evident in the nuclei, in contrast to PKCα and PKCβI. Stimulation of osteoclast-like cells with BzATP caused essentially no translocation of PKCδ-EGFP to the membrane (Fig. 9A, Table 2). As a positive control, we then treated cells with PMA, which caused the prompt and reversible translocation of PKCδ-EGFP to the basolateral membrane in every cell tested.

Both native osteoclasts and RAW 264.7 cells possess multiple subtypes of P2 nucleotide receptors (Lin and Chen, 1998; Korcok et al., 2007). However, several pieces of evidence establish that the translocation of cPKC demonstrated here in osteoclasts is induced selectively by activation of P2X7, and not other P2 nucleotide receptors. ATP stimulated translocation at 3 mM, a concentration sufficient to activate P2X7, but not at 10 μM, a concentration that activates several other P2X and P2Y receptors (Burnstock, 2007). Similarly, UTP (150 μM), which activates several P2Y receptors but not P2X receptors, was ineffective. However, translocation was activated by BzATP (150 μM), which is a relatively potent, but not specific, P2X7-receptor agonist. In addition, responses were inhibited by the P2X7 blocker Brilliant Blue G. Comparison of the responses in osteoclasts derived from wild-type and P2rx7^–/– mice unambiguously revealed the involvement of P2X7 receptors in causing PKC translocation. Such specificity might be due to the unique ability of P2X7 to couple to activation of PLD and initiate lipid signaling (Panupinthu et al., 2008). By using a genetic approach, we have been able to show unequivocally that P2X7 receptors couple to PKC activation.

There are few previous reports linking the activation of P2X7 receptors to PKC signaling in any cell type. In general, involvement of PKC in P2X7 signaling has been inferred from the effects of PKC inhibitors on downstream responses, such as the activation of PLD. For example, both staurosporine and prolonged treatment with phorbol ester inhibit BzATP-stimulated PLD activity in astrocytic cells (Sun et al., 1999). Similarly, in rat submandibular-gland ductal cells, P2X7 receptors were found to be coupled to the activation of a PLD, an effect that was blocked in part by the PKC inhibitors calphostin C or chelerythrine (Pochet et al., 2003).

In osteoclasts, P2X7 did not couple to the translocation of PKCδ. This contrasts with parotid acinar cells, in which P2X7 activation leads to tyrosine phosphorylation of PKCδ, although translocation or activity was not directly assessed in this study (Bradford and Soltoff, 2002). Notably, tyrosine phosphorylation of PKCδ in acinar cells is Ca²⁺-independent, clearly distinguishing it from the Ca²⁺-dependent activation of conventional PKC isoforms shown here in osteoclasts. The different effects of P2X7 activation on PKCδ in acinar cells and osteoclasts emphasize that signaling is dependent on cell type and context.

We found that P2X7 receptors couple selectively to the Ca²⁺-dependent isoforms PKCα and PKCβI, but not the Ca²⁺-independent isoform PKCδ. This is consistent with our findings that the kinetics of translocation of PKCα closely follow the P2X7-induced elevation of [Ca²⁺]_i and that translocation is dependent on the presence of extracellular Ca²⁺. It is likely that P2X7 receptors activate cPKCs through the elevation of [Ca²⁺]_i and by the production of lipid cofactors. In this regard, it is established that P2X7-receptor activation stimulates PLD activity and the production of phosphatidic acid in several cell types (Humphreys and Dubyak, 1996; Garcia-Marcos et al., 2006; Panupinthu et al., 2007). Phosphatidic acid can be hydrolyzed by phosphatidic-acid phosphatase to yield diacylglycerol, which, together with Ca²⁺, activates cPKCs (Exton, 1997). Thus, PKC can be activated downstream of PLD, but also contributes to the activation of PLD itself, creating a feedback loop that sustains diacylglycerol synthesis and PKC activation.

Our previous studies have revealed that nucleotides can induce oscillations of [Ca²⁺]_i in osteoclasts and that these are dependent on P2Y-receptor-induced release of Ca²⁺ from intracellular stores (Weidema et al., 2001). These Ca²⁺ oscillations display more-rapid kinetics than the repeated translocation of PKC that is induced by BzATP, suggesting a different underlying mechanism. There are several possible explanations for repetitive translocation of PKC. First, we have established that the initial response is dependent on influx of extracellular Ca²⁺ through the P2X7 channel. The subsequent translocation might involve intracellular Ca²⁺ release. This could arise from autocrine messengers produced by osteoclasts in response to the activation of P2X7, such as lysophosphatidic acid and eicosanoids, which we have shown are produced in osteoblasts in response to P2X7-receptor activation (Panupinthu et al., 2008). These autocrine messengers could then bind to Ca²⁺-mobilizing receptors on osteoclasts. A second possibility is that both the initial and subsequent translocations of PKC are due to Ca²⁺ entry through P2X7. There is evidence that P2X7-receptor activation in an astrocyte cell line can be negatively regulated by PKC (Hung et al., 2005). Therefore, it is conceivable that, in osteoclasts, a negative-feedback loop gives rise to repetitive pulses of Ca²⁺ influx and PKC activation.

In the present study, we found that stimulation of osteoclasts with the P2X7-receptor agonist BzATP induces translocation of PKCα and PKCβI from the cytoplasm selectively to the basolateral membrane. The mechanism of bone resorption involves polarization of the osteoclast, with formation of a specialized region of membrane, the ruffled border, apposed to the substrate (Vaananen et al., 2000). This corresponds to the apical membrane of a transporting epithelium. Our observation of selective localization of cPKCs might reflect differences in lipid content or anchoring proteins between the basolateral and apical membrane domains of osteoclasts. PKC substrates on the basolateral membrane of osteoclasts could include a broad range of transporters and cell-surface receptors. For example, the extracellular matrix initiates signaling by binding to integrins. PKCα in osteoclasts associates with αvβ3 integrin and activates ERK1/2, thereby regulating cell migration and bone resorption (Rucci et al., 2005). Interestingly, αvβ3 integrin has been reported to be localized on the basolateral membrane in osteoclasts (Duong et al., 2000).

A key cytokine that regulates osteoclast formation and activity is receptor activator of NF-κB ligand (RANKL), which binds to its receptor (RANK) on osteoclasts. Inhibitors of PKC block osteoclastogenesis by suppressing RANKL-induced activation of NF-κB (Wang et al., 2003). This observation is consistent with PKC interacting with components of the RANK signaling cascade, which is initiated by RANKL binding to RANK on the basolateral membrane of osteoclasts. In addition, P2X7 receptors are reported to be regulated by PKC (Hung et al., 2005), raising the possibility that P2X7 receptors in the basolateral membrane are themselves substrates for PKC. Given that the intracellular N- and C-termini of the P2X7 receptor possess potential phosphorylation sites for PKC (Boue-Grabot et al., 2000; Smart et al., 2003), PKC might feed back to regulate P2X7 activity. Lastly, PKC isoforms play important roles in modulating additional signaling pathways, such as activation of phospholipase A₂ (Lin and Chen, 1998). Thus, specific localization to the basolateral membrane might be crucial for positioning PKCs near their appropriate substrates in osteoclasts.

PKC has been implicated the regulation of several aspects of osteoclast function, although the details remain controversial. For example, the PKC activator PMA increases the number of multinucleated osteoclasts in an in vitro culture system when given during the last 12 hours of culture, indicating a role for PKC in the control of the fusion of osteoclast precursors (Fan et al., 1996). In keeping with these findings, expression of PKCβI and II increases during osteoclast formation in mouse bone-marrow cell cultures, and knockdown or inhibition of PKCβII suppresses differentiation, fusion and bone resorption (Lee et al., 2003). By contrast, PMA is reported to inhibit early stages of osteoclastogenesis, an effect thought to be due to the suppression of RANKL-induced activation of NF-κB (Wang et al., 2003).

Similarly, the role of PKC in the regulation of osteoclastic resorption is unclear. Some studies indicate that PKC stimulates resorptive activity (Teti et al., 1992) and that PKCα functions downstream of αvβ3 integrins in activating cell migration and osteoclastic resorption (Rucci et al., 2005), whereas others provide evidence for inhibitory effects of PKC on resorptive activity (Murrills et al., 1992; Moonga et al., 1996; Moonga and Dempster, 1998). It has also been recently shown that PKC is involved in the regulation of osteoclast survival (Pereverzev et al., 2008). Thus, the actions of PKC in controlling osteoclast activity are complex. It is possible that osteoclast function is regulated in a temporally and spatially specific manner by particular PKC isoforms, accounting for the diversity observed in previous studies.

The murine leukemic monocyte macrophage cell line RAW 264.7 was obtained from the American Type Culture Collection (Manassas, VA) and maintained in Dulbecco's modified Eagle Medium (DMEM, Sigma, D-7777) buffered with HCO₃^– containing 10% heat-inactivated fetal bovine serum (FBS, Invitrogen-BRL, Gaithersburg, MD) and 1% antibiotic solution (penicillin 10,000 units/ml; streptomycin 10,000 μg/ml; amphotericin B 25 μg/ml, Invitrogen) at 37°C and 5% CO₂. RAW 264.7 cells were cultured at a density of 1.3×10⁴ cells/cm² and treated with recombinant human RANKL (huRANKL-LZ, 100 ng/ml, kindly provided by Amgen, Thousand Oaks, CA) for 4 days to give rise to multinucleated osteoclast-like cells.

The P2rx7^–/– mouse, generated as described (Solle et al., 2001), was obtained from Pfizer Global Research and Development (Groton, CT). Both P2rx7^+/+ (wild-type) and P2rx7^–/– (knockout) mice were maintained by crossbreeding of P2rx7^+/– mice that have mixed genetic background (129/Ola × C57Bl/6 × DBA/2). Genotypes were confirmed using PCR. These studies were approved by the Council on Animal Care at the University of Western Ontario. Bone-marrow cells from 6- to 10-week-old male wild-type and P2rx7^–/– mice were used to prepare osteoclasts. Cells isolated from femurs and tibias were suspended in α-minimum essential medium (α-MEM), buffered with HCO₃^– (Invitrogen) containing 10% serum, 1% antibiotics and cultured in T75 tissue-culture flasks (15 × 10⁶ cells per flask) with human recombinant M-CSF (PeproTech, 25 ng/ml). After 24 hours, non-adherent cells were removed and resuspended in α-MEM with M-CSF (50 ng/ml) and huRANKL-LZ (100 ng/ml), and plated at 10 × 10⁴ cells/cm² on plastic or glass-bottom dishes (MatTek Corp., Ashland, MA). These cells were cultured for another 3 days, the media then changed and, after 24 hours of additional culture, preparations were processed for tartrate-resistant acid phosphatase (TRAP) staining or used for adenoviral transfections, as described below.

Cells were characterized by staining for TRAP activity, which was carried out using an Acid Phosphatase Kit (Sigma-Aldrich) according to the manufacturer's instructions. To assess resorptive activity, cells were cultured on Osteologic discs (BD Biosciences, USA) or dentine discs in the continued presence of RANKL or M-CSF and RANKL for 6 days. To evaluate resorption pits on dentine, slices were sonicated in water to remove osteoclasts, sputter coated (Denton Desk II AuPD sputtering system) with gold palladium and imaged using a LEO 1530 field emission scanning electron microscope.

Multinucleated cells were generated from either murine bone marrow or RAW 264.7 and stained for TRAP activity (supplementary material Fig. S2A,D). When these cells were seeded on Ca²⁺-phosphate-coated discs or dentine, resorptive activity was evident as clear areas on Ca²⁺ phosphate (supplementary material Fig. S2B,E) and resorption pits on dentine (supplementary material Fig. S2C,F), confirming the identity of these cells as functional osteoclasts.

The PKCβI- and PKCδ-EGFP fusion-protein expression vectors were gifts of Stephen S. G. Ferguson (Robarts Research Institute, London, Canada) and the pPKCα-EGFP vector was obtained from Clontech. RAW 264.7 cultures were seeded at a density of 0.5 × 10⁴/cm². After 48 hours, cells were transfected using FuGENE-HD or FuGENE 6 (Roche Diagnostics, Laval, Quebec, Canada) according to manufacturer's protocols with various EGFP-tagged PKC isoforms or control vector (pEGFP-N1, Clontech) in the presence of RANKL. Transfected osteoclast-like cells were studied by confocal microscopy 4 days post-seeding, while bathed in HEPES-buffered, HCO₃^– free DMEM (Sigma D-2902) supplemented with 3.5 g/l glucose, 10% serum and 1% antibiotics.

Adenoviruses expressing PKCα-EGFP fusion or EGFP proteins were created using `ViraPower Adenoviral Expression System' according to the manufacturer's protocols (Invitrogen). pAd/CMV/V5-DEST Gateway and pENTR1A vectors were used. An EcoRI-NotI fragment containing the EGFP coding sequence and an XhoI-NotI fragment containing PKCα-EGFP were isolated from the pEGFP-N1 and pPKCα-EGFP vectors (Clontech), respectively, and ligated into the pENTR1A vector. Crude viral lysates were applied to bone-marrow cultures on day 5 of differentiation for 20 minutes at 37°C, followed by change of medium. Cultures were then maintained with RANKL (100 ng/ml) and M-CSF (50 ng/ml) for an additional 2 days before studying the cells. Preliminary experiments were performed to optimize transfection conditions for each virus preparation.

Cells were examined using a Zeiss LSM 510 META confocal microscope using W Plan-Apochromat objective lens (40× magnification, numerical aperture 1.2) with excitation using the 488 line of an Ar⁺ ion laser. EGFP emission was detected within the 500- to 530-nm bandwidth, and Fura-red in the 630- to 700-nm bandwidth. For live-cell imaging, image stacks were recorded every 5 seconds for a period of 10-15 minutes after the addition of agonists. Imaging experiments were performed at 25-28°C using LSM image browser (Carl Zeiss Vision Imaging Systems). In some experiments, cells were loaded with Fura-red-AM dye (Invitrogen). Measurement of [Ca²⁺]_i using Fura-2 was carried out as described (Weidema et al., 2001).

Data are representative traces or means ± s.e.m. The n value is the number of osteoclasts or the number of samples tested, as indicated. Differences between two groups were assessed using t-tests. Differences among three or more groups were evaluated by one-way analysis of variance (ANOVA) followed by a Bonferroni multiple comparisons test. Differences were accepted as statistically significant at P<0.05.