Pannexin 3 Inhibits Proliferation of Osteoprogenitor Cells by Regulating Wnt and p21 Signaling*

Background: The mechanism of the transition from osteoprogenitor cell proliferation to differentiation is unclear. Results: Panx3 inhibits osteoprogenitor proliferation by blocking canonical Wnt signaling and promoting p21 activation. Conclusion: A Panx3 hemichannel induces multiple Panx3 signaling pathways critical for the cell cycle exit. Significance: Our findings reveal that Panx3 is a new regulator to switch the stage from proliferation to differentiation in osteoprogenitor cells. Canonical Wnt signaling and BMP promote the proliferation and differentiation of osteoprogenitors, respectively. However, the regulatory mechanism involved in the transition from proliferation to differentiation is unclear. Here, we show that Panx3 (pannexin 3) plays a key role in this transition by inhibiting the proliferation and promoting the cell cycle exit. Using primary calvarial cells and explants, C3H10T1/2 cells, and C2C12 cells, we found that Panx3 expression inhibited cell growth, whereas the inhibition of endogenous Panx3 expression increased it. We also found that the Panx3 hemichannel inhibited cell growth by promoting β-catenin degradation through GSK3β activation. Additionally, the Panx3 hemichannel inhibited cyclin D1 transcription and Rb phosphorylation through reduced cAMP/PKA/CREB signaling. Furthermore, the Panx3 endoplasmic reticulum Ca2+ channel induced the transcription and phosphorylation of p21, through the calmodulin/Smad pathway, and resulted in the cell cycle exit. Our results reveal that Panx3 is a new regulator that promotes the switch from proliferation to differentiation of osteoprogenitors via multiple Panx3 signaling pathways.

Highly coordinated proliferation and differentiation programs regulate bone development and homeostasis. Canonical Wnt signaling plays an important role in osteoprogenitor proliferation and bone mass (see Refs. [15][16][17][18][19][20][21][22]. Wnt proteins are secreted signaling molecules that regulate many biological processes, such as proliferation, differentiation, maintenance, and survival, through ␤-catenin-dependent (canonical) and -independent pathways (noncanonical) (1)(2)(3). Canonical Wnt signaling involves protein stabilization and nuclear translocation of the downstream ␤-catenin protein, from its multimeric pro-tein complex consisting of glycogen synthase kinase-3␤ (GSK3␤), 3 APC, and Axin (4,5). In the absence of Wnt, ␤-catenin is phosphorylated by GSK3␤ in the complex and degraded through ubiquitination. Upon Wnt binding to a frizzled receptor, and its co-receptors LRP5/6, Axin, and GSK3␤ are recruited to the plasma membrane with the scaffold protein disheveled, which disrupts the protein complexes (6,7). This disruption of the protein complexes leads to the phosphorylation of GSK3␤ and inhibition of ␤-catenin phosphorylation, resulting in the stabilization of ␤-catenin and its translocation into the nucleus. ␤-Catenin in the nucleus binds to TCF/LEF transcription factors to activate Wnt/␤-catenin-responsive genes, such as CDK1 and cyclin D1, which are required for cell cycle progression (1,8,9). In addition to the membrane association mechanism, Wnt signaling is regulated by PKA and PI3K/ Akt, which phosphorylate and inactivate GSK3␤, resulting in the stabilization of ␤-catenin (10 -13).
Genetic studies in human patients with the osteoporosispseudoglioma syndrome (14) showed that loss and gain mutations in LRP5 or LRP6 result in low (14) and high bone mass (15,16), respectively. Studies in mice also support the crucial role of canonical Wnt signaling in bone development. LRP5 KO mice display inhibition of bone formation and osteoblast proliferation (17). Mice with the mutation of Dkk1, which prevents Wnt signaling by binding LRP5/6, have high bone density and an increased number of osteoblasts (18). Conditional ␤-catenin KO mice show low bone mass (19 -21). Although canonical Wnt signaling is required for osteoprogenitor cell proliferation, the mechanism of how Wnt signaling is regulated during osteogenesis is still not fully understood. Osterix (Osx) negatively regulates canonical Wnt signaling by promoting the expression of Dkk1 and inhibits osteoblast proliferation (22). However, Osx is expressed in differentiating osteoblasts, not in the transitional stage from osteoprogenitor proliferation to osteoblasts (23).
Pannexins (Panxs) were recently identified as a new gap junction protein family (24). The Panx family consists of three members, Panx1, 2, and 3. Panx1 is ubiquitously expressed with particularly strong expression in the central nervous system. Panx2 is expressed in the central nervous system and is shown to modulate Panx1 channel activities (24 -27). Panx3 is the member that was most recently identified by genome bioinformatic analysis (24). Although Panx3 is expressed in certain soft tissues, such as skin and coronary arteries (28,29), we found high levels of Panx3 expression in developing hard tissues, including cartilage and bone (23,30).
Previously, we demonstrated that Panx3 is induced in the prehypertrophic zone and in the perichondrium of the growth plate. Panx3 inhibits PTH-mediated chondrocyte proliferation by its hemichannel activity and promotes the differentiation of chondrocytes (30). We also showed that Panx3 promotes osteoblast differentiation through its multiple pathways (23). In this study, we demonstrate that Panx3 inhibits osteoprogenitor proliferation by inhibiting Wnt/␤-catenin and PKA/CREB signaling and promotes the cell cycle exit by increasing p21 activity. Our results demonstrate that Panx3 is a new regulator that promotes the switch from proliferation to differentiation in osteoblasts.
Cell Culture-C2C12 cells were grown in DMEM (Invitrogen) containing 10% FBS (HyClone) at 37°C under 5% CO 2 . For the proliferation assay, the cells (2.5 ϫ 10 3 /ml) stably transfected with either pEF1/Panx3 or control pEF1 were cultured in DMEM in 96-well plates for up to 5 days. C2C12 cells stably transfected with the Panx3 shRNA vector were incubated with BMP2 (300 ng/ml) in 96-well plates and cultured for up to 5 days. The C3H10T1/2 cells were grown in DMEM/F-12 (Invitrogen) containing 10% FBS. For the proliferation assay, C3H10T1/2 cells transiently transfected with either pEF1/ Panx3 or pEF1 were cultured in normal media in 96-well plates for up to 4 days. The cells transiently transfected with the Panx3 shRNA vector were incubated with an osteogenic medium (DMEM/F-12, 10% FBS, 100 g/ml ascorbic acid, 10 mM ␤-glycerol phosphate, and 100 ng/ml BMP2). The cell prolifer-ation activity was evaluated using a cell counting kit (Dojindo). The absorbance was measured using a microplate reader. The primary calvarial cells were prepared from the calvaria of newborn mice and cultured in ␣-minimum essential medium (␣-MEM; Invitrogen) with 10% FBS, 100 units/ml of penicillin, and 100 g/ml of streptomycin as previously described (31). For the proliferation assay, primary calvarial cells transiently transfected with pEF1/Panx3 or pEF1 were cultured in ␣-MEM in 96-well plates for 2 days. For the TOPflash/FOPflash reporter assays, luciferase reporter plasmids were co-transfected with the pRLSV40 plasmid (32) as an internal control for transfection efficiency. Luciferase activities were assayed using the Dual-Luciferase Reporter TM assay system (Promega). Relative luciferase activities were expressed as ratios of luciferase activities of the experimental vectors to the internal control vector.
Ex Vivo Calvarial Organ Culture-Calvarial bones were isolated from either newborn C57BL/6 mice or heterozygous Axin2 LacZ mice obtained from Jackson Labs (33) and were cultured in DMEM containing 5% FBS, 50 g/ml ascorbic acid (Sigma), and 1 mM ␤-glycerol phosphate (Sigma) at 37°C in a humidified atmosphere of 5% CO 2 as previously described (34,35). One day after starting the culture, the calvarial bones were infected with either recombinant adenovirus AdCont or AdPanx3 (1 ϫ 10 9 pfu/ml) for 2 days. For the peptide inhibition assay, the Panx3 inhibitory peptide or scramble peptide (100 g/ml) was added to the calvarial culture, and the culture was incubated for 2 days. All experimental procedures were approved by the Animal Care and Use Committee of the National Institute of Dental and Craniofacial Research.
Immunostaining and X-Gal Staining-For X-gal staining, the calvarial samples were fixed with 4% paraformaldehyde overnight, then equilibrated in sucrose, and embedded in an O.C.T compound (Tissue-Tek) for cryosection. X-gal staining was performed as described previously. For immunostaining, after deparaffinization and rehydration, the sections were treated with heat-induced epitope retrieval in a pH 6.0 citrate buffer (Dako). The quantification of LacZ-and Ki67-positive cells was analyzed using ImageJ 1.40g. For the staining of cultured cells, the cells were blocked with Power block (Biocare Medical) and reacted for 2 h at room temperature with primary antibodies. Primary antibodies were detected by Alexa 488 (Invitrogen) or by Cy-3 (Jackson ImmunoResearch Laboratories)-conjugated secondary antibodies. Nuclear staining was performed using Hoechst dye (Sigma-Aldrich). The analysis was performed on an LSM 710 inverted confocal microscope (Carl Zeiss Micro-Imaging, Inc.), and co-localization was analyzed by MetaMorph (Molecular Devices).
RT-PCR-The total RNA was extracted using QG-810 and the QuickGene RNA cultured cell HC kit S (Fujifilm). The total RNA (1 g) was used for reverse transcription to generate cDNA, which was used as a template for the PCRs with genespecific primers (Table 1), as previously described (30). Real time PCR amplification was performed with iQ SYBR Green Supermix (Bio-Rad) and the Eco real time PCR system (Illumina). Real time PCR was performed for 40 cycles of 95°C for 15 s and 60°C for 1 min. Gene expression was normalized to the housekeeping gene Hprt.
Flow Cytometry Analysis-Stably transfected C2C12 cells (1.0 ϫ 10 4 cells) were seeded in a 60-mm dish and cultured with or without BMP2 (300 ng/ml) for 2 days. For the Panx3 hemichannel blocking experiment, the cells were cultured with the Panx3 antibody (1.5 g/ml) for 2 days without BMP2. For the primary calvarial cells, 8.0 ϫ 10 4 transiently transfected cells were cultured for 2 days with or without the Panx3 antibody (1.5 g/ml). The cells were then collected by centrifugation at 120 ϫ g for 5 min. DNA content was analyzed by propidium iodide staining (EMD Biosciences) with CellQuest software on FACSCalibur Station (Becton Dickinson).
Measurement of Intracellular cAMP-The cells were seeded at 1.0 ϫ 10 4 cells/well in a 96-well plate and cultured for 1 day with either DMEM for the C2C12 cells or ␣-MEM for primary calvarial cells. The cells were then incubated with media containing 0.1% albumin medium for 12 h, followed by incubation in media containing 10% serum for 1 h. The level of cAMP was determined with a Bridge-It cAMP designer fluorescence assay kit (Mediomics) and measured as previously described (30).
Western Blot Analysis-The cell lysates were prepared as previously described (30). Ten g of each protein was electrophoresed in 4 -12% SDS-polyacrylamide gel (Invitrogen) and transferred onto a polyvinylidene difluoride membrane using iBlot (Invitrogen). The membranes were immunoblotted with antibodies.
Data Analysis-Each experiment was repeated several times, and the data were analyzed using Prism 5 software. Statistical differences between two groups of data were analyzed with the Student's t test. One-way analysis of variance was used for cell proliferation assays with Wnt3a and Dkk1 (see Fig. 3A). p Ͻ 0.05 was considered to be statistically significant.

Panx3 Inhibits Osteoprogenitor Cell Proliferation-Panx3
is induced during the transition from proliferation to differentiation in osteoprogenitor cells and promotes osteoblast differentiation (23,30); therefore, we hypothesized that Panx3 regulates osteoprogenitor cell proliferation. We first examined the effects of Panx3 overexpression on the proliferation of primary calvarial cells from newborn mice. The neonatal calvarium contains progenitor cells for osteoblasts and adipocytes, of which the majority are osteoprogenitors (31,36,37). Transfection of a Panx3 expression vector (pEF1/Panx3) into primary calvarial cells cultured in proliferation media reduced proliferation compared with that of a control vector (pEF1) (Fig. 1A, panel a). We also tested the effect of Panx3 overexpression in multipotent C3H10T1/2 cells, which differentiate into osteoblasts with BMP2. Panx3 overexpression reduced the proliferation of C3H10T1/2 cells cultured in proliferation media (Fig. 1A, panel  b). C2C12 cells are osteogenic and myogenic, depending on the culture conditions. C2C12 cells differentiate into osteoblasts with BMP2 in FBS-containing media, whereas they differentiate into myoblasts in horse serum containing media. As with primary calvarial cells and C3H10T1/2 cells, Panx3 overexpression reduced proliferation of the C2C12 cells cultured in FBScontaining media (Fig. 1A, panel c).
We next analyzed the inhibitory activity of Panx3 for proliferation in neonatal mouse calvarial organ culture using a recombinant adenovirus system (Fig. 1B). The control adenovirus infection (AdCont) showed staining of endogenous Panx3-expressing cells and Ki67-positive proliferating cells. The merged images revealed that these positive staining cells did not overlap (Fig. 1B, panels a-d). With a Panx3 adenovirus (AdPanx3) infection, the number of Panx3-expressing cells increased, whereas the number of Ki67-positive proliferating cells decreased (Fig. 1B, panels e-h). The Panx3-positive cells in either AdCont-infected (Fig. 1B, panels a-d) or AdPanx3infected ( Fig. 1B, panels e-h) calvaria did not overlap Ki67positive cells, suggesting that the proliferation of Panx3-expressing cells is inhibited. The quantification also showed reduced numbers of Ki67-positive cells with the AdPanx3 infection ( Fig. 1B, panel i). We next examined the effect of the suppression of endogenous Panx3 expression by Panx3 shRNA (shPanx3) on the proliferation of C3H10T1/2 and C2C12 cells cultured in the presence of BMP2, which induces endogenous Panx3 expression. The transfection of C3H10T1/2 and C2C12 cells with shPanx3 increased the proliferation of these cells compared with that in the control shRNA transfection cells (Fig. 1C, panels a and b). These results indicate that Panx3 inhibits osteoprogenitor proliferation.
Panx3 Promotes Cell Cycle Arrest at G 0 /G 1 Phase-We examined the inhibitory activity of Panx3 in proliferation using the FACS analysis (Fig. 2). Calvarial cells overexpressing Panx3 accumulated in the G 0 /G 1 phase and were reduced in both the S and G 2 /M phases ( Fig. 2A). We also found a similar cell cycle arrest at the G 0 /G 1 phase in C2C12 cells overexpressing Panx3 (Fig. 2B, panel a). Furthermore, shPanx3 transfected cells were reduced in the G 0 /G 1 phases and increased in the G 2 /M phases (Fig. 2B, panel b). These results suggest that Panx3 expression in both primary calvarial cells and C2C12 cells inhibits proliferation by arresting the cell cycle at the G 0 /G 1 phase.
Panx3 Inhibits Wnt/␤-Catenin Signaling-Because canonical Wnt signaling promotes the proliferation of osteoprogenitor cells (17, 22, 38 -41), Panx3 may block the Wnt/␤-catenin pathway. To explore this possibility, we examined the effect of Wnt signaling on the proliferation of Panx3-overexpressing C2C12 cells (Fig. 3A). C2C12 cells produce canonical and noncanonical Wnts (data not shown) (42,43). Without exogenous Wnt3a, the proliferation of Panx3-overexpressing cells was reduced when compared with that of the control cells similar to that shown in Fig. 1A. The addition of Wnt3a increased proliferation of the control cells but did not promote proliferation of the Panx3-overexpressing cells, which suggests that Panx3 inhibited Wnt signaling. Dkk-1, an antagonist of Wnt3a/LRP5 receptor interactions, inhibited proliferation even in the absence of exogenous Wnt3a. This occurred because the cells produce endogenous Wnts (43). Dkk-1 blocked the effect of Wnts in a dose-dependent manner. However, even at the highest dose (100 ng/ml) of Dkk-1, the inhibition level of the control cell proliferation by Dkk-1 was lower than that of the Panx3overexpressing cells with or without Dkk-1 (last and first pairs of bars in Fig. 3A). This suggests that Panx3 inhibits cell proliferation through not only Wnt signaling, but also other pathways. ␤-Catenin is involved in downstream canonical Wnt signaling (1,44,45). We therefore examined the expression and localization of the ␤-catenin protein in Panx3-overexpressing C2C12 cells and in control cells (Fig. 3B). In the control cells, ␤-catenin was localized in the plasma membrane and in the nucleus, whereas in Panx3-overexpressing cells, the protein level and nuclear localization of ␤-catenin were reduced (Fig.  3B, panels a-f). The addition of Wnt3a to the control cells increased the ␤-catenin localization in the nucleus. In contrast, in Panx3-overexpressing cells, the nuclear localization of ␤-catenin was reduced (Fig. 3B, panels g-l). Quantitative analysis confirmed the reduced nuclear localization level of ␤-catenin in Panx3-overexpressing cells, compared with that in the control cells (Fig. 3B, panel m). Western blot analysis revealed that Panx3 reduced the protein level of ␤-catenin, especially in both the cytoplasm and nucleus (Fig. 3C). These results suggested that Panx3 inhibited canonical Wnt signaling by reducing ␤-catenin activity. Moreover, we analyzed ␤-catenin activity using a TOPflash reporter vector (Fig. 3D) and found that TOPflash reporter activity was reduced in the Panx3-overexpressing cells compared with that in the control cells (Fig. 3D, panel a). The reporter activity was strongly activated in the control cells by the addition of Wnt3a, whereas it was inhibited in the Panx3-overexpressing cells (Fig. 3D, panel b). We also examined ␤-catenin activity in an ex vivo culture of calvarial bone from heterozygous Axin2 LacZ mice containing an Axin2-lacZ knock-in allele, which is a target gene of ␤-catenin (33). Infection with AdPanx3 reduced the number of LacZ-positive cells compared with that of infection with AdCont (Fig. 3E,  panel a and b). The quantification showed a lower percentage of LacZ-positive cells in the AdPanx3-infected cells (Fig. 3E,  panel c). These results indicate that Panx3 inhibits Wnt signaling by reducing ␤-catenin activity.
Panx3 Promotes ␤-Catenin Degradation by GSK3␤ Activation-To identify the mechanism of inhibition of ␤-catenin activity by Panx3, we first analyzed the mRNA and protein levels of ␤-catenin as well as its activation level in Panx3-overexpressing C2C12 cells. We found that ␤-catenin mRNA levels did not change significantly between the control cells and Panx3-overexpressing cells (Fig. 4A). We also examined the effect of the inhibition of endogenous Panx3 expression on ␤-catenin mRNA expression by shPanx3 in C2C12 cells that had been treated with BMP2 for a short time to induce endogenous Panx3 expression. The shPanx3 did not change the ␤-catenin mRNA expression level when compared with that of the control shRNA (Fig. 4A).
Western blot analysis showed that, in contrast to the mRNA levels, ␤-catenin protein levels were reduced in Panx3-overex-pressing cells and were higher in shPanx3 transfected cells (Fig.  4B, panel a). Furthermore, the phosphorylation of ␤-catenin was promoted in Panx3-overexpressing cells, whereas its phosphorylation levels were reduced by shPanx3 (Fig. 4B, panel a). We found that the phosphorylation of GSK3␤ was inhibited in Panx3-overexpressing cells, whereas shPanx3 produced the opposite effect (Fig. 4B, panel a). Additionally, we observed similar results in primary calvarial cells (Fig. 4B, panel b). These results suggest that that Panx3 induces ␤-catenin degradation via increasing GSK3␤ activity.
Panx3 Hemichannels Inhibit Osteoprogenitor Proliferation-Hemichannels link the cytoplasm with the extracellular space and regulate cellular signaling through the release of small molecules such as ATP, and Panx3 functions as a hemichannel (23,30). Therefore, the Panx3 hemichannel may be involved in the inhibition of osteoprogenitor proliferation. To address this possibility, we first examined whether I-peptide, which is an inhibitor specific to Panx3 hemichannel function (23,30), abrogates the Panx3-mediated inhibition of cell proliferation in an ex vivo calvarial culture. The addition of I-peptide increased the number of Ki67-positive proliferating cells (Fig. 5A, panels e-h). We found that many Ki67-positive cells were also positive for Panx3. This is because the I-peptide inhibited the Panx3 hemichannel function but not the Panx3 expression and increased the proliferation of Panx3-positive cells. A control scramble peptide (S-peptide) containing the same amino acid contents of the I-peptide did not affect the proliferation state (Fig. 5A, panels a-d). Under this condition, the Panx3-and Ki67-positive cells did not overlap. The quantification confirmed the increased number of Ki67-positive cells as a result of the I-peptide addition (Fig. 5A, panel i). These results suggest that the Panx3 hemichannel plays an important role in the inhibition of osteoprogenitor proliferation.
We confirmed this Panx3 hemichannel function using the Panx3 antibody, which reacts with the extracellular domain of Panx3 and inhibits the Panx3 hemichannel (23,30). We showed that the addition of the Panx3 antibody to the culture abrogated the inhibition of Panx3-overexpressing C2C12 cell proliferation (Fig. 5B). Because intracellular ATP plays an important role in cell proliferation by regulating the intracellular cAMP levels (46, 47) (see Fig. 8), we examined the intracellular cAMP levels in C2C12 cells. Panx3 overexpression reduced the intracellular cAMP levels caused by the release of intracellular ATP via the Panx3 hemichannel (Fig. 5C, panel a), which is consistent with our previous findings (23,30). We showed that the anti-Panx3 antibody abrogated the reduced cAMP levels by inhibiting the Panx3 hemichannel function (Fig. 5, C, panel a,  and D, panel a), whereas shPanx3 increased the cAMP level in C2C12 cells and in primary calvarial cells (Fig. 5, C, panel b, and  D, panel b). These cAMP levels correlate with the proliferation states of pEF1/Panx3 and Panx3 shRNA cells (Fig. 1), and our results suggest that Panx3 reduces cell proliferation in part through the ATP/cAMP pathway by the Panx3 hemichannel.
Panx3 Hemichannels Reduce PKA/CREB Signaling and ␤-Catenin Activity-Intracellular cAMP activates downstream PKA/CREB signaling, which induces the expression of genes involved in the progression of cell proliferation (48). To further delineate the Panx3 hemichannel pathway, which inhibits cell proliferation, we analyzed the downstream molecules of cAMP signaling in either pEF1/Panx3 or shPanx3 transfected C2C12 cells (Fig. 6A) and in primary calvarial cells (Fig. 6B). Panx3 overexpression reduced the phosphorylation of PKA and CREB (Fig. 6, A, panel a, and B, panel a, left panels; A, panel b, and B,  panel b, left panels), whereas the addition of the Panx3 antibody to the Panx3-overexpressing C2C12 cells blocked the reduction in the PKA/CREB phosphorylation levels (Fig. 6A, panel a, middle panel, and panel b, left panels). The inhibition of endogenous Panx3 expression by Panx3 shRNA in C2C12 and primary calvarial cells increased the levels of both PKA and CREB phosphorylation, compared with that of the control shRNA transfected cells (Fig. 6, A, panel a, and B, panel a, right panels; A,   panel b, and B, panel b, right panels). These results suggest that Panx3 inhibits proliferation of both C2C12 and primary calvarial cells through the PKA/CREB signaling pathway.
Because GSK3␤ kinase activity is inhibited through the phosphorylation of GSK3␤ by cAMP-dependent PKA (11,12), Panx3-mediated ␤-catenin degradation may be regulated through decreased cAMP/PKA signaling by the Panx3 hemichannel, which promotes GSK3␤ kinase activity resulting in an increase in ␤-catenin phosphorylation for the degradation. We examined ␤-catenin activity in an ex vivo calvarial culture using Axin LacZ mice with I-peptide to inhibit the Panx3 hemichannel. We found that an addition of I-peptide increased the number of the LacZ-positive cells, indicating increased ␤-catenin activity in the ex vivo calvarial culture (Fig. 6C, panels a and b). The quantification confirmed the increased number of LacZ-positive cells by the addition of I-peptide (Fig. 6C, panel c). This result indicates that the Panx3 hemichannel inhibits ␤-catenin signaling through the cAMP/PKA pathway.
Panx3 Inhibits the Activity of Cell Cycle Regulators-We further analyzed the activity of cell cycle regulators downstream of the PKA/CREB pathway (Fig. 7). PKA/CREB signaling induces the expression of E2F1, cyclin D1, and CDK4, which promote cell cycle progression (49). The cyclin D1-CDK4 complex induces the phosphorylation of Rb, which dissociates Rb from the Rb-E2F1 complex. Free E2F1 induces gene expression for cell cycle progression (50). Wnt/␤-catenin signaling activates genes for cell cycle molecules, such as c-Myc and cyclin D1, and promotes cell growth (8,22). In both the Panx3-overexpressing primary calvarial cells and C2C12 cells, cyclin D1 protein levels and Rb phosphorylation levels were reduced, whereas Panx3 shRNA cells showed the opposite effects (Fig. 7A, panels a and  b). Panx3 overexpression also reduced E2F1 mRNA levels (Fig.  7A, panel c). These results suggest that Panx3 promotes cell cycle arrest at the G 0 /G 1 phase by inhibiting the activity of the cell cycle molecules involved in cell cycle progression from the G 1 to the S phase.
Panx3 ER Ca 2ϩ Channel Promotes the Smad1/5 and p21 Pathway for Cell Cycle Exit-The Panx3 ER Ca 2ϩ channel is activated by PI3K/Akt signaling and increases intracellular Ca 2ϩ levels, ([Ca 2ϩ ] i ), following the activation of calmodulin (CaM) signaling (23). In these processes, the external ATP released from the cells through the Panx3 hemichannel binds to purinergic P2 receptors, which activate PI3K/Akt signaling. It is known that the CaM pathway activates Smad1/5 signaling, which regulates p21, a cell cycle inhibitor (51,52). Therefore, we examined whether Panx3 promotes p21 activation via Smad signaling in primary calvarial cells and in C2C12 cells. The Panx3 overexpression resulted in the activation of Smad1/5 and p21 (Fig. 7B, panels a-c). The activation of these proteins was blocked by PPADS, an inhibitor of the P2 receptors, and by the Panx3 antibody, which inhibited the Panx3 hemichannel (Fig.  7B, panels b and c). Panx3 overexpression also promoted the expression of p21 mRNA (Fig. 7B, panel d) and protein levels (Fig. 7B, panels a-c). Because the Panx3 ER Ca 2ϩ channel promotes [Ca 2ϩ ] i by Akt signaling, we examined the effects of Akt activation on the phosphorylation levels of the Smad1/5 and p21 pathway by transfecting the Akt constitutive active (Akt-CA) and dominant negative (Akt-DN) vectors to Panx3-overexpressing C2C12 cells. The Akt-CA promoted increased phosphorylation of the Smad1/5 and p21 over that of the mock vector transfection. In contrast, Akt-DN inhibited the activation of the Smad1/5 and p21 by Panx3 overexpression (Fig. 7C). These results suggest that Panx3 promotes the cell cycle exit by the activation of Smad1/5 and p21 through the Panx3 function as an ER Ca 2ϩ channel.

DISCUSSION
In this study, we demonstrate that Panx3 plays an important role in the transition from proliferation to differentiation in osteoprogenitor cells. We found that Panx3 inhibits osteoprogenitor proliferation and promotes the cell cycle exit. Fig. 8 presents a schematic diagram of the mechanism of Panx3 actions that negatively regulate the proliferation state of osteoprogenitor cells. At the proliferation stage, canonical Wnt signaling stabilizes and translocates ␤-catenin to the nucleus, where it activates genes for cell cycle progression (Fig. 8A). When Panx3 is induced by BMP at the early transition stage, Panx3 functions as a hemichannel to release intracellular ATP into the extracellular space. This Panx3 hemichannel activity induces all subsequent Panx3 signaling pathways involved in the inhibition of proliferation. It reduces cAMP/PKA signaling, resulting in ␤-catenin degradation via GSK3␤ activation, which leads to a reduction in the ␤-catenin/TCF/Lef-dependent transcription of the genes necessary for proliferation. The Panx3 hemichannel also reduces activity of CREB, a DNA-binding protein factor downstream of cAMP/PKA, resulting in the reduced transcription of the CREB target genes required for cell cycle progression. In addition, the binding of ATP released from the Panx3 hemichannel to ATP receptors activates the Panx3 ER Ca 2ϩ channel/CaM/CaM kinase/Smad1/5 pathway via PI3K/Akt signaling, following an increase in the transcription and activation of p21. All of these signaling pathways are abrogated by the inhibition of the Panx3 hemichannel. Thus, the Panx3 hemichannel is the most critical first step to switch proliferation to the cell cycle exit.
Wnt signaling plays an important role in many aspects of tissue development, including osteogenesis (45). In bone formation, Wnt signaling promotes osteoprogenitor cell proliferation and mineralization while inhibiting osteoclastogenesis (40,53). Wnt3a enhances the cell proliferation of the osteoblastic cell lines C2C12, MC3T3-E1, and C3H10T1/2 (39,54,55). Wnt1, 2, and 3a also induce alkaline phosphatase activity in osteogenic cells (55). Thus, Wnt signaling regulates osteoprogenitor cell proliferation and differentiation. However, the mechanisms of Wnt signaling regulation in the cell proliferation stage and in the late differentiation stage remain unclear. BMPs and Wnt signaling interactions regulate cell proliferation and differentiation during development (56). BMPs antagonize the Wnt-mediated proliferation of osteoprogenitor cells and promote osteoblast differentiation by inducing Runx2 (57,58). Subsequently, Runx2 induces Osx that inhibits Wnt signaling by inducing Dkk1 (22). In this paper, our data show a different pathway for Wnt signaling inhibition by Panx3. We previously demonstrated that BMP2 induces Panx3 during osteoblast differentiation (23). The suppression and overexpression of Panx3 inhibits and promotes the expression of Osx and other osteoblast markers, respectively (23). These results indicate that Panx3 is an upstream molecule of Osx. Thus, Panx3 inhibits Wnt signaling first, followed by Osx-mediated inhibition during osteoblast differentiation to ensure the differentiation process. [Ca 2ϩ ] i plays a critical role in cell proliferation and differentiation (59). The ER serves as the major Ca 2ϩ storage space in the cell. Panx3 functions as an ER Ca 2ϩ channel, which promotes the osteoprogenitor cell cycle exit (Fig. 7). CaM is the major transducer of Ca 2ϩ signaling in many cell types. In osteoblasts, upon binding to Ca 2ϩ , CaM interacts with and activates protein factors such as calmodulin kinase II and calcineurin and induces differentiation (60 -62). We found that Panx3 stimulated the phosphorylation of Smad1/5 and increased both the protein levels and phosphorylation levels of p21 (Fig. 7B). Our results indicate that the Panx3 ER Ca 2ϩ channel regulates the CaM/Smads/p21 signaling pathway (Fig. 8). The anti-Panx3 antibody and PPADS inhibited the Panx3-promoted Smad1/5 activation and p21 expression, suggesting that the Panx3 hemichannels and P2 receptors are involved in these processes. The increase in p21 contributes to the Panx3-promoted cell cycle exit (Fig. 2). These results suggest that Panx3-promoted Ca 2ϩ signaling activates Smad/p21 signaling, which promotes the cell cycle exit.
IP3 receptors (IP3Rs), which consist of three members (IP3R1, 2, and 3), are ubiquitous ER Ca 2ϩ channels, and they release Ca 2ϩ from the ER upon its binding to IP3 (63). Three types of IP3Rs are expressed in osteogenic C2C12 cells (data not shown). The expression levels of IP3Rs did not change during the osteoblast differentiation (data not shown). Ryanodine receptors, which are another ER Ca 2ϩ channel, are not expressed in C2C12 cells and are not induced during osteoblast differentiation (23). We previously showed that PPADS strongly blocks the Panx3 ER Ca 2ϩ channel, but not IP3Rs, in FIGURE 7. Panx3 reduces cell cycle signaling pathways and increases p21 activation. A, Western blotting with antibodies to Panx3, cyclin D1, phospho-Rb (P-Rb), Rb, and ␣-tubulin. Panels a and b, pEF1 and pEF1/Panx3 transfected cells were cultured for 2 days in ␣-MEM for primary calvarial cells (panel a) or for 1 day in DMEM for C2C12 cells (panel b). sh control and shPanx3 transfected cells were cultured with BMP2 (300 ng/ml) for 2 days for calvarial cells (panel a) or 1 day for C2C12 cells (panel b). pEF1 and pEF1/Panx3 transfected C2C12 cells were cultured in DMEM for 2 days. sh control and shPanx3 transfected cells were cultured with BMP2 (300 ng/ml) for 2 days. Panel c, E2F1 mRNA levels were then observed by quantitative PCR. B, Panels a-c, phosphorylation of Smad1/5 and p21 cultured in ␣-MEM for pEF1 and pEF1/Panx3 transfected calvarial cells (panel a) and DMEM with PPADS or anti-Panx3 antibody in pEF1 and pEF1/Panx3 transfected C2C12 cells (panels b and c). Panel c, quantification of protein levels for P-Smad1/5 (upper panel) and P-p21 (lower panel). Panel d, quantitative PCR for p21 mRNA levels on the same experimental condition as E2F1. C, Panx3-promoted Smad1/5 and p21 activation via Akt signaling pathways. Panx3 overexpressed C2C12 cells transfected with Akt CA or Akt DN or mock vector were cultured in DMEM for 1 day. **, p Ͻ 0.01. Error bars represent the means Ϯ S.D., n ϭ 3.
C2C12 cells. The reduction of Smad1/5 and p21 activity by PPADS indicates that the Panx3 ER Ca 2ϩ channel primarily regulates p21 activation to inhibit osteoprogenitor cell proliferation (Fig. 7B). Mice lacking either IP3R2 or IP3R3 are viable and show no obvious abnormalities in skeletal development. Mice lacking both IP3R2 and IP3R3 are born with a normal appearance but begin losing body weight after weaning because of defects in exocrine secretion (64). Thus, the Panx3 ER Ca 2ϩ channel likely plays a major role in osteogenic proliferation as well as in osteogenic differentiation.
In summary, we have shown that Panx3 inhibits osteoprogenitor proliferation through the multiple signaling pathways induced by the Panx3 hemichannel. Our results reveal that Panx3 promotes the switch from osteoprogenitor proliferation to osteoblast differentiation.