Pannexin 3 is a novel target for Runx2, expressed by osteoblasts and mature growth plate chondrocytes


  • Stephen R Bond,

    1. Department of Cellular and Physiological Science, Life Sciences Institute, University of British Columbia, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Alice Lau,

    1. Department of Cellular and Physiological Science, Life Sciences Institute, University of British Columbia, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Silvia Penuela,

    1. Department of Anatomy and Cell Biology, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Arthur V Sampaio,

    1. Department of Cellular and Physiological Science, Biomedical Research Center, University of British Columbia, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • T Michael Underhill,

    1. Department of Cellular and Physiological Science, Biomedical Research Center, University of British Columbia, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Dale W Laird,

    1. Department of Anatomy and Cell Biology, University of Western Ontario, London, Ontario, Canada
    Search for more papers by this author
  • Christian C Naus

    Corresponding author
    1. Department of Cellular and Physiological Science, Life Sciences Institute, University of British Columbia, Vancouver, British Columbia, Canada
    • Department of Cellular and Physiological Science, Life Sciences Institute, 2350 Health Science Mall, Vancouver, BC, V6T 1Z3, Canada.
    Search for more papers by this author


Pannexins are a class of chordate channel proteins identified by their homology to insect gap junction proteins. The pannexin family consists of three members, Panx1, Panx2, and Panx3, and the role each of these proteins plays in cellular processes is still under investigation. Previous reports of Panx3 expression indicate enrichment in skeletal tissues, so we have further investigated this distribution by surveying the developing mouse embryo with immunofluorescence. High levels of Panx3 were detected in intramembranous craniofacial flat bones, as well as long bones of the appendicular and axial skeleton. This distribution is the result of expression in both osteoblasts and hypertrophic chondrocytes. Furthermore, the Panx3 promoter contains putative binding sites for transcription factors involved in bone formation, and we show that the sequence between bases −275 and −283 is responsive to Runx2 activation. Taken together, our data suggests that Panx3 may serve an important role in bone development, and is a novel target for Runx2-dependent signaling. © 2011 American Society for Bone and Mineral Research


Pannexins (Panxs) are a class of chordate channel proteins initially identified by their homology to the insect gap junction proteins, called innexins.1 Significant sequence similarity does not exist between the innexin/pannexin family and the connexins, which are the canonical chordate gap junction proteins. The two classes do however share a high degree of predicted structural similarity,2, 3 suggesting a potential overlap in functionality. Most mammalian tissues express at least one of the three Panx subtypes,4, 5 and they have been shown to influence a number of cellular processes, including inflammation, ischemic tolerance in neurons, and cancer.6–9 In contrast to channels formed by innexins and connexins, Panx channels do not significantly contribute to direct cell-cell gap junctional communication, but instead, tightly regulate the transfer of ions and small molecules between the cytoplasm and extracellular space.10, 11 It has also been suggested that some of the processes originally attributed to connexin “hemichannel” activity may in fact be the result of Panx channels.12 Of the three isoforms (Panx1, Panx2, and Panx3), Panx1 has the widest expression profile and shows the clearest channel activity.13 ATP in particular repeatedly has been repeatedly reported to traverse the Panx1 channel under physiological conditions.11, 14, 15 Panx2 is primarily observed in neurons of the central nervous system,13, 16 and contains a large hydrophilic domain on its C-terminal tail that sets it apart from the rest of the innexin superfamily.3 Functionally, Panx1 and Panx2 have been reported to suppress glioma growth in vivo9 and to regulate the differentiation of neuronal progenitor cells,17 although its role as a channel still requires further study.18 Panx3 has received the least attention within the Panx family, most likely owing to its restricted expression profile19 and a lack of observable channel activity during the initial descriptive experiments reported by Bruzzone et al. in 2003.13 Panx3 channel activity has since been observed in response to mechanical stimulus in vitro,5 but the functional role of Panx3 in vivo is unclear, and its mode of transcriptional regulation is undescribed.

RT-PCR screens indicate that Panx3 mRNA is present in various tissues, including skin, kidney, spleen, and brain,5, 13 some of which have been verified using immunofluorescence and Western blot.5 However, an in silico analysis of gene expression databases reveals a significant enrichment of Panx3 in skeletal tissues.20 This has been verified in a number of osteoblast cell lines as well as primary calvarial osteoblasts,21–24 and visualized more directly by immunofluorescence in the mouse cochlea, where Panx3 was found exclusively in bone.25 Interestingly, Panx3 is also expressed in hypertrophic chondrocytes during the development of axial skeleton long bones.23, 24 Osteoblasts and chondrocytes are derived from a shared osteochondral progenitor cell type, which differentiates in response to local conditions. Whereas the transcriptional programs of the two cell types are initially quite distinct, there is a convergence late in the maturation process of chondrocytes destined for endochondral ossification. These chondrocytes, located in the diaphysis of maturing long bones, cease mitosis and become terminally differentiated by switching their gene expression profile to resemble that of osteoblasts.26 We therefore hypothesized that Panx3 is regulated by one or more of the osteoblastic transcription factors utilized by chondrocytes late in their life cycle.

Here, we report the expression pattern of Panx3 in developing bone with emphasis on the high levels observed in both osteoblasts and mature osteogenic chondrocytes. We further show that runt-related transcription factor 2 (Runx2), a key transcription factor for normal bone formation, binds to and transactivates the Panx3 promoter. These results build on previous reports of Panx3 expression in osteogenic cell types by qualifying the specific temporal and spatial distribution of this expression, and also identifies an important part of the regulatory mechanism controlling its transcription.

Materials and Methods

Animal Care

All experiments were performed in accordance with the guidelines established by the Canadian Council on Animal Care and were approved by the University of British Columbia Animal Care Committee.


A rabbit anti-Panx3 polyclonal antibody was developed and described previously.5 All other antibodies used during this study were obtained from the following sources: Goat anti-osteopontin (PA1-25152; Pierce, Rockford, IL, USA); mouse anti-GAPDH (5G4; HyTest, Turku, Finland); rabbit anti-Pthr1 (ab75150; Abcam, Cambridge, MA, USA); mouse anti-collagen type 1α1, developed by H Furthmayr27 (SP1.D8; Developmental Studies Hybridoma Bank (DSHB), University of Iowa, Iowa City, IA, USA); mouse anti-collagen type X, developed by TF Linsenmayer (X-AC9; DSHB); rabbit anti-Runx2 (M-70; Santa Cruz Biotechnology, Santa Cruz, CA, USA); rabbit anti-Cx43 (C6219; Sigma, St. Louis, MO, USA).

Cell culture

All cells were grown in a humidified 37°C incubator with 5% CO2. NIH-3T3 fibroblasts and HEK-293 cells were purchased from American Type Culture Collection (Manassas, VA, USA), and maintained in Dulbecco's modified Eagle medium + 10% fetal bovine serum (FBS). MC3T3-E1 (clone 4) pre-osteoblast cells were also purchased from American Type Culture Collection, and maintained in α-minimum essential medium (α-MEM) + 10% FBS for no more than 15 passages. To induce differentiation, the MC3T3-E1 cells were seeded at 100,000 cells/cm2 and supplemented with β-glycerol phosphate (10 mM) and ascorbate (50 µg/mL) for between 3 and 8 weeks as indicated. Primary osteoblasts were liberated from rat or mouse calvaria using sequential collagenase digestion as previously described,28 seeded at 50,000 cells/cm2 and maintained in α-MEM + 10% FBS until confluent. Differentiation was once again induced with β-glycerol phosphate (10 mM) and ascorbate (50 µg/ml). To confirm differentiation, alkaline phosphatase activity was detected with 0.1 mg/ml Naphthol and 0.6 mg/mL Fast Red Violet, and mineralization was visualized with Von Kossa staining (2.5% silver nitrate under bright light).


At least three CD-1 mouse embryos each, from 13, 14, 15, and 17 days post-conception (E13, E14, E15, and E17) were fixed whole for 24 to 48 hours in 4% paraformaldehyde (PFA) at 4°C. Metatarsals were harvested from pups 2 days postnatal (P2), and fixed/decalcified in phosphate buffered 12.5% EDTA and 2.5% PFA (pH 7.4) for 48 hours. Older animals (P7 to P12 months) were first perfused with 4% PFA, and harvested tissues were further fixed in 4% PFA overnight at 4°C. Fixed tissue was dehydrated sequentially through 15% and 30% sucrose for cryoprotection, then embedded in Tissue-Tek O.C.T. compound (Sakura Finetek, Torrance, CA, USA), and sectioned at 6 to 8 µm. Metatarsal sections were pretreated with 0.5% hyaluronidase (Sigma) in Hank's Balanced Salt Solution (with Mg2+ and Ca2+) for 30 minutes, to free collagen epitopes from extracellular proteoglycans.29 Sections were blocked with 7.5% bovine serum albumin fraction V (Invitrogen, Carlsbad, CA, USA) and 10 mM glycine, then permeabilized with 0.3% Triton X-100 before incubation overnight at 4°C in primary antibody. AlexaFluor 488 or 568 secondary antibodies (Invitrogen) were applied for 1 hour at room temperature, followed by mounting in ProLong Gold antifade reagent with DAPI (Invitrogen). Imaging was performed on an Axioplan2 fluorescence microscope fitted with an AxioCam MRm camera (Carl Zeiss, Thornwood, NY, USA). Sequential images of individual sections were aligned using the auto-merge algorithm in Adobe Photoshop CS4 (Adobe Systems, San Jose, CA, USA), and manually adjusted to correct misalignments. Visual inspection of the reconstructed sections was performed, and histological structures were identified by reference to the Atlas of Mouse Development, by MH Kaufman.30


RNA was harvested from fresh rat postnatal calvaria (P1) or from cultures derived from P1 rat calvaria. For the cultures, RNA was harvested at the confluence (D0) or after 9 days of differentiation (D9). Control cultures were also treated with 100 nM all-trans retinoic acid to inhibit differentiation (D9-RA). RNA was collected using Trizol Reagent (Invitrogen) followed by purification through RNAeasy columns (Qiagen, Valencia, CA, USA). RNA was labeled and hybridized to RAT 230 2.0 arrays (Affymetrix, Santa Clara, CA, USA) using the manufacturers recommended protocol, and gene expression profiles were analyzed using MAS 5.0 and GeneSpring GX.

Western Blot

MC3T3-E1 cell lysates were collected with RIPA buffer (150 mM NaCl, 50 mM Tris-HCl pH 8.0, 0.5% Sarkosyl, 1% IGEPAL, 0.1% SDS). Protein concentration was determined using a BCA assay kit (Pierce), and 30 µg to 50 µg was separated on 10% Tris-glycine SDS-PAGE gels. The protein was transferred to Immuno-Blot PVDF (Bio-Rad, Hercules, CA, USA) and then blocked in 5% nonfat milk + 0.1% Tween20 (NFM-T). The membranes were probed at 4°C overnight, with primary antibodies diluted in 3% NFM-T, followed by HRPO-linked (Sigma) secondary antibodies for 2 hours at room temperature. HRPO activity was visualized by treating the membrane with SuperSignal West Pico Chemiluminescent Substrate (Pierce), and exposing/developing Bioflex Econo Film (Clonex, Markham, Ontario, Canada).

Promoter Alignment

The genomic sequences of the first Panx3 exon, as well as one kilobase 5′ from the start codon, were obtained for all mammalian species with an entry in the Entrez gene database, including mouse, rat, dog, cow, horse, macaque, human, chimp, and opossum. Multiple sequence alignment was performed with ClustalX2.0.3.31 MatInspector (Genomatix Software, Munich, Germany) was used to predict putative transcription factor binding motifs within the mouse Panx3 promoter,32 and these results were correlated against conserved regions identified by the ClustalX algorithm.


Three fragments from the Panx3 promoter were directionally cloned into the BlgII and BamHI sites of the promoterless pGL4.11[luc2P] luciferase reporter plasmid (Promega, Madison, WI, USA). These fragments were PCR amplified from mouse genomic DNA, using a reverse primer with a BlgII site designed to cut at +1 (5′-ATAGATCTGATCAGGCTCTGTGCC-3′), and forward primers with BamHI sites designed to cut at −206 (5′-GCGGATCCCACGGCCCTCCT-3′), −648(5′-CCGGATCCAGTGGTTTCCTCA-3′), and −1175(5′-ATGGATCCCACTGACTCCTTGC-3′). Two alternate constructs were prepared from the 648 bps construct using site-directed mutagenesis to incorporate nucleotide substitutions into the predicted Runx2 and Msx1/2 binding motifs, from −283 to −275 (AAACCACAA to CGTTTAGCT) and from −33 to −28 (TAATTG to GGATCC), respectively. A Runx2 expression vector, pCMV-OSF2, was kindly provided by Gerard Karsenty.33 A second Runx2 expression vector, pSG5-EGFP-Runx2Δrep, was engineered in house: EGFP was excised from pEGFP-N3 (Clontech, Madison, WI, USA) using BglII and NotI, blunt ended, and ligated into pSG5, which had been linearized with BamHI and BglII and also blunt ended. Runx2 was then removed from pCMV-OSF2 with EcoRI (truncating the protein by 13 amino acids, starting at residue 583), and ligated into the EcoRI site of pSG5-EGFP. The rat Panx3 cDNA was amplified from a cDNA library using primers with ClaI sites added to the 5′ end, then cloned into the ClaI site of pRK5 for use as a positive control of Panx3 expression in real-time PCR experiments, while pEF-GM-EGFP34 was used as a negative control. Finally, pCMV-SPORT6-Msx1 was purchased from Open Biosystems. All new constructs were verified by sequencing the relevant regions.

Transcription Reporter Assay

HEK-293 cells were co-transfected with the pGL4.11[luc2P] plasmids carrying Panx3 promoter constructs, a constitutively driven transcription factor or control (Runx2, Runx2Δrep, Msx1, or GFP), and a constitutively driven Renilla luciferase plasmid pGL4.73[hRluc/SV40] (Promega) to monitor transfection efficiency. Primary mouse osteoblasts were co-transfected with only the pGL4.11[luc2P]-promoter plasmids and control pGL4.73[hRluc/SV40] plasmid. Transfection was accomplished using Fugene6 reagent, as directed by the manufacturer (Roche, Indianapolis, IN, USA). After transfection, the cells were maintained in culture for 48 hours, and then the total cellular protein was extracted with passive lysis buffer (Promega). Dual-luciferase reporter assays were performed according to the manufacturer's protocol (Promega), and luciferase activity was quantified using a 96-well, dual-luciferase luminometer (Berthold Detection Systems, Pforzheim, Germany). Firefly luciferase activity was standardized against Renilla luciferase activity, and relative luciferase units from the Panx3 promoter constructs were normalized to the empty pGL4.11[luc2P] group (which was arbitrarily assigned a value of 100). Each sample was analyzed in triplicate and the values averaged, with error bars representing the standard error of at least three separate experiments. Normalized sample means were compared with one another using two-way ANOVA with subsequent Holm-Sidak pairwise multiple comparison. P-values ≤ 0.05 were considered significant.

Real-Time qPCR

The total RNA was harvested from the cell culture using Trizol Reagent (Invitrogen) according to the manufacturer's directions, and 550 ng were reverse transcribed into cDNA using SuperScript III (Invitrogen). A total of 10 ng cDNA was diluted in 2X TaqMan Fast Universal PCR MasterMix (Applied Biosystems, Foster City, CA, USA) and TaqMan probes against Panx3, Runx2, or rRNA (Applied Biosystems). Real-time quantitative PCR was performed on the 7500 Fast Thermal Cycler (Applied Biosystems), using the comparative CT experimental settings of the 7500 V2.0.1 software. Panx3 and Runx2 levels were normalized against rRNA.

Chromatin Immunoprecipitation (ChIP)

Chromatin was isolated from MC3T3-E1 cells differentiated for four weeks, then queried with rabbit anti-Runx2 IgG, or non-specific rabbit IgG control as previously described.35 Briefly, the cells were cross-linked in 1% formaldehyde at room temperature for 15 minutes, followed by mechanical lysis in a Dounce homogenizer (critical for adequate recovery of MC3T3 cells, which are embedded in dense collagen at the time of collection), and shearing of the chromatin into fragments of 500 bps to 1000 bps by sonication. The chromatin was precleared with protein A Sepharose (Pierce) for 1 hour, followed by incubation with primary antibodies overnight. The immune complexes were subsequently precipitated with protein A Sepharose and reverse cross-linked overnight. The DNA was phenol/chloroform extracted, and then analyzed by PCR. PCR primers were designed to span the putative Runx2 binding site in the Panx3 promoter (Fwd: 5′-ATCAAATACAGGGCAGTTTCAGGG-'3, Rev: 5′-CACTGTGCCTTTATGCTGTCC-'3).


Panx3 is expressed in both endochondral and intramembranous bone during embryonic development

To visualize the global expression pattern of Panx3 during development, E13, E14, E15, and E17 mouse embryos were collected, fixed, sectioned, and surveyed by immunofluorescence using a Panx3 specific antibody.5 Between E13 and E15.5, Panx3 expression predominantly localizes to membranous bones of the face and upper thorax (Fig. 1A–E). The mandible and clavicle are two of the first primordia to mineralize during development, forming ossification centers as early as E14,30 although Panx3 expression appears as early as E13–E13.5 (Fig. 1A–B). By E14–E14.5, the mesenchymal condensations destined to become the parietal and frontal bones of the lateral calvaria and the maxilla also expressed Panx3 (Fig. 1C–E). All of these structures are derived through intramembranous ossification, whereby condensed mesenchymal cells differentiate into osteoblasts to fabricate bone directly.36 The development of the appendicular skeleton and parts of the axial skeleton involves a transition from cartilage to mineralized bone matrix. Longitudinal growth of this tissue is achieved through chondrocyte division, followed by hypertrophy of cells situated closest to the diaphysis.37 Inspection of cartilaginous centers of endochondral ossification (such as the dorsal ends of the ribs and diaphyses of major long bones) reveals Panx3 expression as early as E14–E14.5 (Fig. 1F). After E15, chondrogenic Panx3 expression is easily identified at the interface between terminally differentiated chondrocytes and the mineralized matrix during endochondral ossification (Fig. 1H, L, O–Q; Fig. 2A, C), particularly in appendicular long bones, while continuing to be observed in intramembranous bones as well (Fig. 1G, I–K, M–N; Fig. 2B). The specificity of the Panx3 antibody was confirmed by peptide competition (Supplemental Fig. S1).

Figure 1.

Expression of Panx3 in stages E13–E15.5 mouse embryos. Panx3 (green) was first observed in large membranous bones, such as the mandible (A) and clavicle (B), as early as E13–E13.5. By E14–E14.5, a large number of membranous bones were positive for Panx3, as illustrated by the frontal bone (C), mandible (D), and lateral calvaria (E). Panx3 was also found in the diaphyseal regions of long bones at E14–E14.5, as shown in the tibia (F). At E15–E15.5, Panx3 expression was robust in the osteoblasts of maturing membranous bones, as detected in the orbito-sphenoid bone (G), nasal capsule (I), exoccipital bone (J), mandible (K, M), and maxilla (N). Regions within cartilaginous bones also exhibited strong Panx3 expression, such as the neural arch of the C1 vertebra (H), articular and transverse processes of vertebra (L, O), rib tubercle (O), rib shaft (P), and radius diaphysis (Q). Nuclei are stained blue with DAPI.

Figure 2.

Expression of Panx3 in a stage E17–E17.5 mouse embryo. Panx3 (green) expression at E17–E17.5 was extensive throughout the developing skeleton, as shown in the basisphenoid bone (A), mandible (B), and vertebra (C). Nuclei are stained blue with DAPI.

Panx3 is expressed by pre-hypertrophic chondrocytes, hypertrophic chondrocytes, and mature osteoblasts

To better characterize the distribution of Panx3 within developing long bones, further immunofluorescent analyses were performed on metatarsals from P2 mice using antibodies against known markers of chondrocyte differentiation (Fig. 3). Collagen type-10 (Col10α1) is a classical marker of terminally differentiated, hypertrophic chondrocytes in the growth plate.38, 39 Panx3 is observed before Col10α1 induction in the developing metatarsal cartilage, indicating that Panx3 expression precedes terminal chondrocyte differentiation (Fig. 3A). This localization pattern supports recently published in situ data, where a similar relationship between Panx3 and Col10α1 transcripts was reported in the growth plate.23 Osteoblasts and hypertrophic chondrocytes both express and secrete the glycosylated phosphoprotein osteopontin (OPN) during long bone formation.40, 41 Induction of OPN expression in the cartilage anlagen closely mimics that of Col10α1, and is also preceded by Panx3 expression. Within the perichondrium, the inner chondrogenic layer of cells at the periphery of the developing bone collar also express Panx3. OPN expression in the perichondrium overlaps that of Panx3 to a limited extent, but Panx3 appears to be reduced at the onset of OPN expression in the inner chondrogenic layer. Interestingly, Panx3 is absent from the outer fibrous layer of the perichondrium, despite an increase in OPN (Fig. 3B). As expected, the metatarsal model also illustrates Panx3 expression by osteoblasts within the mineralizing diaphysis. Collagen type-1 (Col1α1) is secreted by endochondral and perichondrial osteoblasts,42 and Panx3 expressing cells within the mineralizing diaphysis are surrounded by layers of Col1α1 containing extracellular matrix (Fig. 3C).

Figure 3.

Localization of Panx3 and markers of long bone development in a representative long bone. Endochondral ossification in the metatarsal of neonatal mouse pups is representative of long bone development. Panx3 expression precedes induction of chondrocyte hypertrophy, as delineated by Col10α1 expression (A). Osteoblasts, hypertrophic chondrocytes, and perichondrial fibroblasts all express OPN (B). Co-localization of OPN with Panx3 appears to occur in the proximal layer of chondrocytes in the bone collar, while the outer fibrous layers were negative for Panx3. Col1α1 is present in the ECM layer surrounding Panx3 positive osteoblasts that have infiltrated the diaphysis (C, arrow).

Whereas the spatial relationship between chondrocytes in the growth plate (i.e., increasing maturation from the epiphysis to the diaphysis) simplifies the process of determining when in the chondrogenic life cycle Panx3 is induced, an analogous structure does not exist for osteoblast maturation. To better characterize the temporal profile of Panx3 expression during osteoblast differentiation, MC3T3-E1 pre-osteoblast cells were cultured in differentiation-inducing conditions over an 8-week period. A robust increase in Panx3 was observed over the first 30 days of differentiation, followed by a plateau as the cultures were maintained for up to 60 days (Fig. 4A). This pattern is inversely correlated with PTH/PTH-related protein receptor (Pthr1) expression, and positively correlated with upregulation of Col1α1 (Fig. 4A). Pthr1 is involved in regulating cell growth and differentiation, with expression peaking early in MC3T3 differentiation before osteoid secretion,43, 44 whereas Col1α1 expression is consistent with the expression profile we observed in the metatarsals. This increase in Panx3 protein in MC3T3 cells is mirrored at the mRNA level, as determined by real-time PCR (data not shown). This experiment was performed twice with similar results, and differentiation was confirmed by mineralization of the cultures (Fig. 4B). Figure 4C shows that the expression pattern of Panx3 in primary rat calvarial cultures, as determined by mRNA microarray, follows the same trend as was observed in MC3T3 cells, and is very similar to established osteogenic markers (liver/bone/kidney alkaline phosphatase [Alpl], integrin-binding sialoprotein [Ibsp], and osterix [Sp7]). There is abundant expression in freshly cultured calvaria (F-RC), followed by a reduction in transcript levels as mitotically active progenitor cells quickly dominate the sub-confluent primary cultures established from these isolates (D0). Panx3 expression, along with that of Alpl, Ibsp, and Sp7, is elevated once the cultures become confluent and differentiation proceeds (Day 9, D9). Retinoic acid (RA) signaling plays two distinct roles in osteoblast development, first acting to inhibit differentiation of immature progenitor cells, and later enhancing mineralization during osteogenesis.45 Consistent with this, treatment of primary rat calvarial cultures with 100 nM RA for 9 days (D9-RA) leads to markedly decreased expression of Panx3, along with the other osteogenic markers (Fig. 4C). Treating MC3T3 cells with retinoic acid produced a similar result (data not shown). These experiments indicate that Panx3 is expressed by mature osteoblasts, and not by their earlier progenitor cells.

Figure 4.

Panx3 expression is induced during osteoblast differentiation. (A) Panx3 levels progressively increase over several weeks of MC3T3-E1 differentiation, reaching a maximum at 1 month. There was a positive correlation between the expression of Panx3 and the mature osteoblast marker Col1α1, and an inverse correlation between Panx3 and parathyroid hormone receptor (Pthr1) expression. (B) Differentiation was confirmed by measuring alkaline phosphatase activity (gray), and mineralization (von Kossa; black). (C) cDNA microarray data shows increased expression of Panx3 in freshly isolated calvarial osteoblasts (F-RC) compared to primary cultures derived from these cells (D0). This is the same pattern observed for classical markers of osteogenesis (Ibsp, Sp7, and Alpl). When the osteoblast cultures are induced to differentiate for 9 days (D9), Panx3, Ibsp, Sp7, and AlpI are all upregulated. Treatment with retinoic acid (D9 RA) inhibits the transition of osteoblast progenitor cells into mature osteoblasts, preventing the upregulation of the aforementioned genes.

The Panx3 promoter is responsive to Runx2

Multiple sequence alignment of the first 1000 bps immediately 5′ from the Panx3 start codon reveals conserved binding sites for transcription factors associated with bone development, such as Barx2,46 the vitamin D receptor/retinoid-X receptor heterodimer (VDR/RXR),47 Msx1/2,48 and Runx249 (Supplemental Fig. 2). Three progressively longer stretches of the Panx3 promoter (206 bps, 648 bps, and 1175 bps), were cloned into the pGL4.11[luc2P] luciferase reporter plasmid (Fig. 5A). These plasmids were co-expressed in HEK-293 cells with either full-length Runx2, or with a Runx2-ΔRep construct truncated by 13 amino acids on the C-terminal end to delete a repressive domain.50 Compared with empty pGL4.11[luc2P] and the 206 bps truncation construct, there was a 2- to 3-fold increase in luciferase expression generated by the 648 bps and 1175 bps constructs when co-transfected with either of the Runx2 plasmids (Fig 5B). The longer truncations both contain a highly conserved Runx2 binding motif (AACCACA51) 275 bps from the transcriptional start site. To test the significance of the putative Runx2 binding site more directly, its sequence was disrupted in the 648 bps construct. This reduced the Runx2 induced luciferase expression to the same levels recorded from the empty pGL4.11[luc2P] control vector. A binding motif for the homeobox proteins Msx1/2 ([C/G]TAATTG52) is also highly conserved in the Panx3 promoter. Msx1 is important for normal skeletal patterning and growth, especially in relation to osteoblast activity,53 and can function as both a transcriptional repressor54 or an activator.55 Luciferase expression appeared elevated on co-expression of the promoter constructs with Msx1. However, this level was unchanged when the Msx1/2-binding sequence was disrupted, implying that Msx1 on its own does not significantly enhance expression by binding to this particular location. As expected, Runx2 continued to induce luciferase expression despite disruption of the Msx1/2-binding motif. To ensure that the increased luciferase induction was not an artifact of Runx2 over-expression, primary calvarial osteoblasts were also transfected with the promoter truncation constructs after 6 days of differentiation (Fig. 5C). Even without exogenous Runx2, luciferase expression was enhanced 18- and 20-fold over the promoterless pGL4.11[luc] control by the 648 bps and 1175 bps constructs. To show a direct interaction between Runx2 and the Panx3 promoter in differentiated MC3T3 cells, a ChIP assay was used. After pull down and reverse crosslinking, the DNA was analyzed by PCR using primers spanning the Runx2-binding site in the Panx3 promoter. The promoter fragment was highly enriched by incubation with the Runx2 antibody compared with IgG control (Fig. 5D). To test whether Runx2 is sufficient to induce Panx3 expression, the NIH-3T3 cell line (which does not express appreciable levels of Panx3) was transfected with the Runx2 and Runx2-ΔRep plasmids. Real-time qPCR was used to monitor Runx2 and Panx3 expression, and whereas high levels of Panx3 mRNA were observed in cells transfected with pRK5-Panx3, there was no measurable increase in Panx3 in the Runx2 transfected cells when compared with controls transfected with GFP (Fig. 5E).

Figure 5.

Runx2 transactivates the Panx3 promoter. (A) Illustration of the luciferase reporter constructs created to determine the effect of Runx2 and Msx1 on the Panx3 promoter. The putative Runx2 and Msx1/2 binding motifs are both present in the 1175 bps and 648 bps constructs, but the 206 bps construct was sufficiently truncated to eliminate the Runx2 site. The predicted Runx2 and MSX binding site sequences were each modified in the 648 bps plasmid using site-directed mutagenesis, to create ΔRunx2 and ΔMSX respectively. (B) Over-expression of Runx2 in HEK-293 cells significantly enhanced luciferase expression from the 648 bps, 1175 bps, and ΔMSX plasmids compared with the empty control, but not from the 206 bps and ΔRunx2 plasmids (two-way ANOVA with Holm-Sidak pairwise multiple comparison, *p-value < 0.001). Over-expression of Msx1 enhances luciferase expression from the 648 bps, ΔMSX, and ΔRunx2 constructs over empty control (two-way ANOVA with Holm-Sidak pairwise multiple comparison, †p-value ≤ 0.003), but there is no difference between these constructs, indicating that removal of the putative Msx1/2 binding site has no significant effect. (C) The same pattern was found when the Panx3 promoter constructs were placed in primary osteoblasts without exogenous transcription factor over-expression. The 648 bps, 1175 bps, and Δ-MSX constructs all generated significantly more luciferase expression than each of the 206 bps, Δ-Runx2, and Empty constructs, whereas no difference existed between the 206 bps, Δ-Runx2, and Empty constructs (One-way ANOVA with Holm-Sidak pairwise multiple comparison, p-value < 0.001). (D) The Panx3 promoter was shown to directly associate with Runx2 in differentiated MC3T3 cells using a ChIP assay. PCR amplification of the Runx2 binding region in the Panx3 promoter resulted in a higher-intensity band on a DNA gel when the template was chromatin immunoprecipitated with a Runx2 antibody, as opposed to IgG control. (E) Despite the pro-expression effects Runx2 imparts on the Panx3 promoter, over-expression of Runx2 is insufficient to induce an upregulation of Panx3 in fibroblast (as assessed by real-time PCR, two-way ANOVA with Holm-Sidak pairwise multiple comparison, p-value > 0.05). Baseline expression of Panx3 and Runx2 was arbitrarily set at a value of 10, using fibroblasts transfected with GFP as a normalizing control. The sensitivity of the Panx3 PCR reaction was confirmed by including a sample transfected with pRK5-Panx3. All experiments were performed at least three times.

The distribution of Runx2 is limited, but it is not exclusive to osteogenic cell types. Sebaceous glands in the skin and lactating mammary gland epithelium also express Runx2, which in turn induces the expression of molecules normally found in developing bone, such as OPN and Ihh.56, 57 Both the lactating mammary glands of postpartum females and the sebaceous glands in the tail sections of P7 animals tested positive for Panx3 expression (Fig. 6A–B). During our global immunofluorescence surveys, we also unexpectedly observed Panx3 expression in the small intestinal epithelium, as shown in a section from a P2 animal (Fig. 6C).

Figure 6.

Panx3 is present in tissues known to express members of the Runx family. Immunofluorescence revealed Panx3 expression (green, all panels) in lactating mammary tissue (A), in sebaceous glands (B), and in the small intestinal epithelium (C).


The pannexin family has received an ever increasing amount of attention since its discovery 10 years ago, as it represents a unique class of channel proteins with what appears to be a diverse role in cellular functionality. The majority of work to date has focused on Panx1,58 and only recently have more detailed studies begun to explore the properties of Panx29, 17 and Panx3.23, 24, 59, 60 A number of reports have now shown that Panx3 is expressed by, and plays a role in, the differentiation of osteogenic cell types,21–25 so the focus of this study was to specifically identify when expression occurs during development, and to understand some of the regulatory underpinnings of this expression pattern.

Using immunofluorescence we were able to assess the spatial and temporal distribution of Panx3 in embryonic mice during development. Morphological analysis reveals Panx3 expression at sites of both intramembranous ossification and endochondral ossification just before the initiation of mineralization, beginning as early as E13–E13.5. Given this expression pattern, future studies into the function of Panx3 in vivo should be directed at processes subsequent to mesenchymal condensation and skeletal patterning. In the growth plate, we confirmed that induction of Panx3 expression precedes the onset of chondrocyte mineralization by comparing the localization of Panx3 protein with other well-characterized markers of bone development, such as Col10α1 and OPN. Growth plate chondrocytes go through a number of distinct differentiation steps, as they first construct a cartilaginous anlagen of the future mature bone, followed by hypertrophy, mineralization, and eventually apoptosis.37 Hypertrophy is best characterized at the molecular level by the onset of Col10α1 expression,61 which we only observe in the large centermost cells of the diaphysis. The detection of Panx3 in the smaller pre-hypertrophic chondrocytes indicates that its expression is induced prior to mineralization. OPN has a similar distribution as Col10α1 in hypertrophic chondrocytes,62 but is also found in the chondrogenic inner layer of the perichondrium. The cells in this region are strongly influenced by soluble Indian Hedgehog (Ihh) diffusing out from prehypertrophic and early hypertrophic chondrocytes. Ihh stimulates the cells in the lower perichondrial layer to switch from a chondrogenic to an osteoblastic phenotype.63 some of which are retained in the peripheral tissue to deposit layers of mineralizing osteoid in the creation of the bone collar, whereas others migrate inward to facilitate trabecular bone formation.64 Both of these osteoblast populations appear to express Panx3, as illustrated by positive staining in a thin proximal layer of cells at the bone collar, as well as in irregularly shaped Col1α1 secreting cells inside the matrix of the diaphysis. Furthermore, cultured osteoblasts express greater levels of Panx3 as they differentiate. MC3T3-E1 differentiation involves a well-described transition from fibroblast-like pre-osteoblasts to mature, mineralizing osteoblasts over several weeks.65 Panx3 was found to increase 6- to 7-fold during MC3T3-E1 differentiation, and normal maturation of these cells was confirmed by an increase in Col1α1 expression and concomitant decrease in Pthr1. The same pattern was observed in the transcriptome of primary osteoblast cultures, where the expression of Panx3 mirrored that of classical osteogenic markers (Alpl, Ibsp, and Sp7), both during normal induction of differentiation, and when osteogenesis was inhibited with RA. From a functional perspective, it has recently been reported that Panx3 expression strongly influences osteoblast differentiation, acting as an Akt-sensitive calcium channel on the ER, as an ATP channel at the plasma membrane, and as a gap junction channel facilitating calcium waves between cells.24 Panx3 has also been shown to enhance maturation to the terminally differentiated hypertrophic state of cultured chondrocytes.23 In this case, it was suggested that Panx3 channel activity reduces the intracellular concentration of ATP, thus reducing the synthesis of cAMP on activation of Pthr, and ultimately hindering cAMP-responsive element-binding protein (CREB) driven proliferation.23 Although not addressed in the aforementioned report, it is worth stating that parathyroid hormone, cAMP, and CREB are also important mediators of osteoblast proliferation and differentiation.66 While further work is still required to determine the relative impact each of these mechanisms has in vivo, during both intramembranous and endochondral ossification, it is clear that an intimate link exists between osteogenesis and the presence of Panx3.

In developing bone, separate transcriptional programs drive osteochondral progenitor cells to mature into osteoblasts or chondrocytes. To become osteoblasts, transcription factors like Msx2, Dlx5/6, Runx2, and Sp7 are important.67 while Sox5, Sox6, and Sox9 will push the progenitor cells toward chondrogenesis.68, 69 Many of the molecular markers of differentiated osteoblasts are also expressed by hypertrophic chondrocytes in the epiphyseal growth plate, where a transcriptional switch takes place to initiate ossification.70 The promoter of Panx3 was scrutinized for highly conserved protein-binding motifs that could link its expression to the osteogenic transcriptional regime. Of the motifs identified as being related to the bone transcriptional program, the highly conserved Runx2-binding site at −275 bps stood out, since Runx2 is an important mediator of both osteoblast differentiation71 and the mineralization of growth plate chondrocytes.72 The Panx3 promoter was indeed found to be responsive to Runx2 induction, and subsequent mutagenesis confirmed that the putative motif at −275 bps was responsible. Taken together, these observations strongly support our hypothesis that Panx3 expression is regulated by Runx2 activity. However, Runx2 over-expression did not induce Panx3 expression in fibroblasts, suggesting that transcriptional cofactors are probably involved. Indeed, Runx2 has previously been reported to work in conjunction with auxiliary transcription factors like MSX2 (inhibitory) and DLX3/5 (excitatory).35 This dependence on other factors could also explain why the Panx3 promoter constructs were nearly 10 times more responsive in primary osteoblasts than they were in Runx2 over-expressing HEK-293 cells.

As explained in the “Results” section, sebocytes and lactating mammary gland epithelium are both non-mineralizing tissues regulated by Runx2.56, 57 We have now shown that these tissues also express Panx3, supporting the premise that Panx3 is regulated by Runx2, and further suggesting that the correlation between Panx3 expression and mineralization is not necessarily causative. At the very least, it indicates that Panx3 expression outside the context of osteogenic cell types is insufficient to induce mineralization. The presence of Panx3 in the small intestinal epithelium was not an anticipated result, because it is not a known site of Runx2 expression. It does however express a second member of the Runx family (Runx3),73 and the functional redundancy between these two transcription factors74 may partly explain the expression observed. It seems clear that Panx3 does not indiscriminately induce ossification, but it may still facilitate significant movement of ATP and calcium within and between cells in vivo, as the recent in vitro work by Iwamoto et al.23 and Ishikawa et al.24 suggests. This could have the pro-osteogenic effect of reducing intracellular cAMP, but the possibility remains of an anti-osteogenic effect via activation of P2Y2 receptors, which have been shown to inhibit maturation of osteoblasts.75 It is anticipated that these conflicting scenarios will be better assessed through the generation and characterization of a Panx3 null mouse.

In conclusion, this study describes the temporal and spatial expression profile of Panx3 in mice during development, with an emphasis on the induction pattern observed in maturing osteoblasts and hypertrophic chondrocytes. This is also the first study to describe the transcriptional regulation of the Panx3 promoter, which we have found to contain a functional enhancer element specific to the transcription factor Runx2.


All the authors state that they have no conflicts of interest.


This study was supported by grants from the Canadian Institute of Health Research (CCN, DWL). SRB was supported by a trainee studentship from the Michael Smith Foundation for Health Research. CCN and DWL hold Canada Research Chairs.

Authors' roles: SRB was the primary author of the manuscript and performed all experiments and analyses except where otherwise noted. AL acquired the images for Figure 2. SP and DWL generated the Panx3 antibody and provided critical feedback through multiple drafts of the manuscript. AVS and TMU created the pSG5-EGFP-Runx2Δrep construct and performed the microarray experiments. CCN provided important suggestions regarding design of the study and provided critical feedback through multiple drafts of the manuscript.