We previously isolated a low bone mass mouse, Gja1Jrt/ + , with a mutation in the gap junction protein, alpha 1 gene (Gja1), encoding for a dominant negative G60S Connexin 43 (Cx43) mutant protein. Similar to other Cx43 mutant mouse models described, including a global Cx43 deletion, four skeletal cell conditional-deletion mutants, and a Cx43 missense mutant (G138R/ +), a reduction in Cx43 gap junction formation and/or function resulted in mice with early onset osteopenia. In contrast to other Cx43 mutants, however, we found that Gja1Jrt/+ mice have both higher bone marrow stromal osteoprogenitor numbers and increased appendicular skeleton osteoblast activity, leading to cell autonomous upregulation of both matrix bone sialoprotein (BSP) and membrane-bound receptor activator of nuclear factor-κB ligand (mbRANKL). In younger Gja1Jrt/+ mice, these contributed to increased osteoclast number and activity resulting in early onset osteopenia. In older animals, however, this effect was abrogated by increased osteoprotegerin (OPG) levels and serum alkaline phosphatase (ALP) so that differences in mutant and wild-type (WT) bone parameters and mechanical properties lessened or disappeared with age. Our study is the first to describe a Cx43 mutation in which osteopenia is caused by increased rather than decreased osteoblast function and where activation of osteoclasts occurs not only through increased mbRANKL but an increase in a matrix protein that affects bone resorption, which together abrogate age-related bone loss in older animals. © 2013 American Society for Bone and Mineral Research.
Connexins comprise the gap junction–forming and hemichannel-forming membrane proteins in bone and other tissues. Gap junctions allow direct cell-cell coupling and communication, whereas hemichannels, the unopposed halves of gap junctions, allow for release of extracellular signaling molecules. Connexin 43 (Cx43) is the major connexin expressed in osteoblasts and osteocytes and is also expressed in osteoclasts and bone marrow stromal cells,[1-3] and as such has been implicated in mediating cell-cell coupling and intercellular communication/signaling in the tightly regulated process of bone metabolism.
Cx43 global and conditional knockout (KO) mutant mice with decreased gap junction function are osteopenic and/or exhibit alterations in the structure/geometry of long bones.[4-9] The low bone mass phenotype was attributed in earlier studies to fewer and dysfunctional osteoblasts, or increased bone resorption by osteoclasts, or both. For example, Cx43 global and collagen 1a1 (Co1a1) promoter-Cre osteoblast-deleted Cx43 mice have fewer and dysfunctional osteoblasts producing less bone matrix and with reduced mineralization in vivo and in vitro.[5, 6] Mice with Cx43 ablation in the osteochondrogenic lineage via use of the Dermo1/Twist2 (DM1) promoter results in both dysfunctional osteoblasts and increased osteoclastogenesis and bone resorption. Mice with Cx43 deleted at later stages of the osteoblast lineage have also been generated. One osteocalcin (Ocn) promoter-Cre osteoblast-deleted Cx43 mouse line with Cx43 ablation in osteoblasts and osteocytes was reported to have no detectable change in bone mineral density (BMD) but an abrogated response to the antiapoptotic effect of bisphosphonates on osteocytes and osteoblasts, whereas another OcnCre Cx43 mouse line was reported to exhibit increased osteoclastogenesis and bone resorption due to an increase in the receptor activator of nuclear factor-κB ligand (RANKL)/osteoprotegerin (OPG) ratio in osteocytes, as well as an enhanced anabolic response to load. In these cases, although bone density was unaffected, the structural parameters of the long bone were still affected.[4, 11, 12] In other recent analyses, Plotkin and colleagues reported that in both their OcnCre Cx43 mouse strain and a dentin matrix protein-1 (Dmp1) promoter-Cre Cx43 mouse strain in which Cx43 is deleted from osteocytes only, no change in BMD is detectable (except for a slight decrease at 2 months in the Dmp1Cre Cx43 strain), but increased osteocyte apoptosis and altered osteoclast and osteoblast activity via reduced expression of OPG and sclerostin (Sost), respectively, promote geometric changes in long bones. In the +/G138R mouse, which carries a mutation in the cytoplasmic loop of Cx43 and expresses normal levels of Cx43 protein but has reduced gap junction function, osteopenia and a nonstatistically significant decrease in osteoblast number were reported. Taken together, the data from the KO and G138R missense mutants suggest differences in the mechanisms underlying changes in osteoblast and osteoclast activity upon disruption of Cx43 globally or at different stages of osteoblast development.
Through an N-ethyl-N-nitrosurea mutagenesis screen, we generated a mouse line, Gja1Jrt/ + , containing a glycine to serine mutation (G60S) in the first extracellular loop of Cx43 (denoted herein as Gja1Jrt for the allele and G60S for the mutation). Like other Cx43 mutants, this line had a bone phenotype of decreased bone mass and mechanical strength, and like G138R mice, exhibited the classical features of human oculodentodigital dysplasia (ODDD), a rare disease characterized by syndactyly, enamel hypoplasia, craniofacial abnormalities, abnormal eye development, and small stature. Although Gja1Jrt/+ mice express less than 50% of wild-type (WT) levels of Cx43 and have markedly reduced gap junction formation and function in osteoblasts and other Cx43-expressing cell types,[14, 16] we now report the unexpected finding that Gja1Jrt/+ mice have more active osteoblasts than their WT littermates. We also report that whereas young Gja1Jrt/+ mice are osteopenic, older mutant mice do not exhibit the old age–related bone loss seen in WT, and report the novel cellular and molecular basis of the osteopenia and age-related phenotypic anomaly in Gja1Jrt/+ mice.
Subjects and Methods
Animals and ethics statement
Gja1Jrt/+ founders in a C57BL/6J background were backcrossed four generations to C3H/HeJ mice. Males from the fourth generation (C4) were crossed to FVB females to generate F1 mice; a second crossing to FVB produced F2 mice. This study was performed using litters from a cross between F2 males and C3H/HeJ females. All experimental procedures were performed in accordance with protocols approved by the Canadian Council on Animal Care and the University of Toronto Faculty of Medicine and Pharmacy Animal Care Committee.
Dual-energy X-ray absorptiometry (PIXImus; Lunar Corp., Madison, WI, USA) was used to measure bone mineral content (BMC), bone area, and BMD of femurs in mice.
Micro–computed tomography of femurs and vertebrae
The distal metaphysis of the left femurs were scanned with a Skyscan 1072 micro–computed tomography (µCT) instrument (Skyscan, Kontich, Belgium) at the Centre for Bone and Periodontal Research (www.bone.mcgill.ca) as described. Morphometric parameters were calculated with 3D Creator software.
Destructive three-point bending was performed as described, and the ultimate load, failure displacement, stiffness, and energy to failure were determined from the load displacement curve. These parameters were normalized to the cross-section of the femurs (measured with calipers) to calculate the ultimate stress, ultimate strain, Young's modulus, and toughness.
Mice were given two intraperitoneal injections of 30 mg/kg aqueous calcein prior to sacrifice as described. Polymethylmethacrylate (MMA)-embedded left femurs were cut in 3-µm sections.
The right femur in 4% paraformaldehyde (PFA) was embedded in a mixture of MMA and glycolmethacrylate (GMA) and 5-µm sections stained with 5% silver nitrate and 0.2% toluidine blue to visualize mineralized bone, osteoid, and osteoblasts. Visualization of osteoclasts in sections was tartrate-resistant acid phosphatase (TRAP) staining and counterstained with 0.4% methyl green (Vector Laboratories Inc.) and mounted in aqueous medium.
Whole blood was collected through the saphenous vein, and the plasma was separated from whole blood by centrifugation and stored at –80°C until biochemical analysis (Vita-Tech, Ontario, Canada).
Total RNA was isolated from bone and cell cultures using TriReagent (Sigma-Aldrich, St. Louis, MO, USA) and reverse transcribed using Superscript II (Invitrogen, Carlsbad, CA, USA) and random hexamers. cDNA was combined with 0.5 µM each of the forward and reverse primers (Supplemental Table 1) and iQ SYBR Green Supermix and run in the MyIQ Real-Time PCR system (BioRad Laboratories, Hercules, CA, USA). Raw data were analyzed with PCR Miner and normalized using the internal control transcript for ribosomal protein L32.
Isolation of bone marrow cells and colony-forming unit–osteoblast assay
Bone marrow cells were isolated from resected tibias and femurs, using a modification of a published method. Cells were plated in α modified essential medium (α-MEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics (1 IU penicillin, 1 μg/mL streptomycin, 50 μg/mL gentamicin, 250 ng/mL fungizone) (standard medium) at 1 × 106 nucleated cells/35-mm dish. After 3 days, the medium was changed to differentiation medium (standard medium with 50 µg/mL ascorbic acid and 10 mM β-glycerophosphate). At day 19, cultures were stained for alkaline phosphatase (ALP) activity and mineralization (Von Kossa), counted, then restained with methylene blue.
Protein isolation from bone and stromal culture and Western blotting
Long bones, cleaned of surrounding tissue, epiphysis, and bone marrow, were cut slightly below the growth plate to separate trabecular bone. Matrix proteins were extracted following a protocol modified from Goldberg and Sodek. Briefly, bones were transferred to 4.0 M guanidine HCl for 30 minutes, washed in PBS, and crushed into bone powder in liquid nitrogen. Bone powder or stromal cultures were washed in PBS then extracted with 0.5 M EDTA in 50 mM Tris/HCl pH 7.4 buffer for four sequential 24-hour extracts. Nonmatrix proteins (mbRANKL) were extracted in cell lysis buffer as described. Protein extracts (30 μg) underwent immunoblotting with antibodies of interest (Supplemental Table 2). Densitometry was done by chemiluminescence detected on film and quantified using Image J software; BSP levels were normalized against either ACTIN or OPN, and in either case was increased significantly.
Osteoclast differentiation assay
Spleens were crushed through a sterile 100-μm mesh in standard medium. Cells were collected by centrifugation, resuspended in PBS, treated with ammonium chloride to lyse the red blood cells, and plated in standard medium supplemented with cytokines RANKL and macrophage colony-stimulating factor (M-CSF) at 50 ng/mL at a density of 1 × 106 cells per well of a 12-well culture plate. Medium was changed after 3 days. On day 6, cells were fixed and stained for TRAP according to the manufacturer's instructions (Sigma).
Bone marrow stromal cells were isolated and resuspended in standard medium supplemented with 100 ng/mL of M-CSF. After 2 days, medium was changed to standard medium supplemented with 50 ng/mL M-CSF and 100 ng/mL RANKL.
Bone resorption assays
Artificial substrate assay
Osteoclasts were plated onto Corning Osteo Assay Surface plates (Corning Inc., NY). Resorption area was measured using ImageJ (version 1.44p).
Trabecular bone resorption assay
Trabecular bone was crushed into fine pieces under liquid nitrogen, using a mortar and pestle. The bone chips were washed and sonicated in ice-cold water for 4 times then transferred to ice-cold 70% ethanol and stored at –20°C until use. Prior to use, the bone powder was washed three times with sterile water and incubated overnight in α-MEM containing antibiotics solution (10× concentration). On the day of the experiment, the bone powder was washed with and then resuspended in osteoclastogenic medium and distributed into 96-well plates in excess. Bone marrow cells were plated on top of the bone chips, and cultured for osteoclast formation and resorption.
OPG and tumor necrosis factor–related activation induced cytokine (TRANCE)/RANKL in serum were assayed using a Quantikine M Murine OPG ELISA kit and a Quantikine M Murine TRANCE/RANKL ELISA kit (No.MOP00 and No.MTR00; R&D Systems, Minneapolis, MN, USA), respectively, following the manufacturer's directions. C-telopeptide fragment of collagen type I (CTX-1) levels in serum was determined from fasted mice using Serum CrossLaps ELISA (RatLaps EIA No. AC-06F1; Immunodiagnostic Systems, Fountain Hills, AZ, USA).
Results are presented as mean ± SD. Experiments were repeated at least three times. Statistical analysis was performed using GraphPad Software program InStat. Longitudinal analysis was analyzed by one-way analysis of variance (ANOVA). Unpaired t test was used for direct comparisons between mutant and WT parameters; n values presented are independent biological samples.
Gja1Jrt/+ mice have low BMD throughout life but improve with age and do not exhibit an old age–related decrease in bone mass
At birth, Gja1Jrt/+ mice were indistinguishable from their WT littermates in terms of size, but the syndactyly phenotype was evident by day 10. Consistent with our previous results from the phenotypic screen used to identify the Gja1Jrt/+ founder, Gja1Jrt/+ mice were markedly smaller than their WT littermates by 2 months of age, reflected in a lower body weight, and this persisted throughout life (data not shown). Gja1Jrt/+ mice had significantly lower BMD at all ages studied compared to age-matched WT littermates, although differences between Gja1Jrt/+ and WT mice became less pronounced with increasing age (Fig. 1A); total body BMD reached maximal value by 4 months of age and plateaued thereafter in WT; but in Gja1Jrt/+ mice, BMD continued to increase until at least 12 months of age (the oldest age quantified).
µCT analysis revealed differences in age-related changes in bone parameters between the genotypes. The percent bone volume to tissue volume (BV/TV) at 2, 4, and 8 months was significantly lower in the trabecular bone of the distal femur of Gja1Jrt/+ versus WT mice (Fig. 1B), with the most striking difference at 4 months (after which BV/TV plateaued in Gja1Jrt/+ mice) and no difference at 12 months (Fig. 1C). At 2 months, the lower BV/TV in Gja1Jrt/+ versus WT mice was due to decreased trabecular thickness (Tb.Th) alone, whereas at 4 and 8 months, both Tb.Th and trabecular number (Tb.N) were lower in the mutant bone (Fig. 1C). Consistent with these observations, the structure model index (SMI) of Gja1Jrt/+ femoral trabecular bone increased from 1.5 to 2.0 between 2 and 4 months of age, reflecting a deterioration in the 3D structure of trabeculae from a more plate-like to a more rod-like structure. Although WT mice exhibited a typical age-related decrease in trabecular BV/TV after 4 months of age, Gja1Jrt/+ mice did not.
Gja1Jrt/+ femurs were smaller than WT femurs at all ages, with significantly reduced total tissue, cortical bone and marrow areas (Fig. 2A, B; Supplemental Fig. 1). When normalized to total tissue area, femoral cortical bone area (Ct.Ar/Tt.Ar) was significantly reduced and marrow area (Ma.Ar/Tt.Ar) significantly increased in young (2–4 months) Gja1Jrt/+ versus WT mice, corresponding to reduced cortical bone thickness at 2 months in Gja1Jrt/ +. However, whereas WT cortical thickness remained constant throughout the age range studied (2–12 months), Gja1Jrt/+ cortical thickness increased over time and by 8 months had surpassed that of WT, resulting in no significant difference in Ct.Ar/Tt.Ar or Ma.Ar/Tt.Ar between genotypes in older mice (Fig. 2B). Gja1Jrt/+ femurs were also significantly shorter than those of WT littermates from 4 to 12 months of age (Table 1); however, WT femurs reached maximum length by 4 months, and Gja1Jrt/+ femurs by 8 months of age (data not shown). Mechanical testing for material and structural properties showed that Gja1Jrt/+ bones were less tough, weaker (lower ultimate stress), and less stiff (lower Young's modulus) than WT bones at 2 and 4 months of age, but more ductile (higher failure strain) than WT at 2 months; no significant differences between genotypes were seen in the older mice. However, the structural properties (ultimate load, energy to failure, and stiffness), which depend on the size and shape of the bone, were significantly decreased in Gja1Jrt/+ versus WT femoral bones at all ages tested. The average polar moment of inertia (ability to resist torsion), which usually correlates with the width of the midshaft, was significantly decreased from 4 to 12 months in Gja1Jrt/+ versus WT femurs (Table 1).
|Femoral length||ns||↓ 10%***||↓ 5%**||↓ 4%*|
|Ultimate stress||↓ 51%**||↓ 42%**||ns||ns|
|Failure strain||↑ 44%**||ns||ns||ns|
|Young's modulus||↓ 65%**||↓ 47%**||ns||ns|
|Toughness||↓ 29%**||↓ 38%**||ns||ns|
|Femoral neck fracture|
|Ultimate load||↓ 28%**||↓ 32%**||↓ 32%**||↓ 21%**|
|Energy to failure||↓ 21%*||ns||ns||↓ 27%*|
|Stiffness||ns||↓ 47%**||ns||↓ 17%*|
|Failure displacement||ns||↑ 23%*||ns||ns|
|Ultimate load||↓ 37%**||↓ 42%**||↓ 27%**||↓ 20%**|
|Energy to failure||↓ 23%*||↓ 41%**||↓ 33%*||↓ 33%**|
|Stiffness||↓ 47%**||↓ 41%**||↓ 26%**||↓ 17%**|
|Failure displacement||↑ 21%**||ns||ns||ns|
|Polar moment of inertia||ns||↓ 55%***||↓ 58%***||↓ 46%***|
G60S is an activating mutation in Gja1Jrt/+ osteoblasts and results in bone matrix with abnormally high levels of BSP
Given the age-related abrogation of the osteopenia observed in the cortical and trabecular compartments of Gja1Jrt/+ mice, we next assessed osteoblast and osteocyte numbers and activity. Gja1Jrt/+ bones exhibited no evidence of changes to bone periosteal surfaces (data not shown) and there was no significant difference between genotypes in the number of osteoblasts per bone surface (data not shown), osteoblast surface per bone surface (Ob.S/BS), osteocyte number per bone area (Osy.N/BA), or number of empty lacunae in the cortical (Fig. 2D; Supplemental Fig; 1B) or trabecular (Fig. 3A) bone compartments. Mineral apposition rate (MAR) and bone formation rate (BFR) were highest in 2-month-old animals in both genotypes, but whereas no significant differences were seen between genotypes at any age tested in the trabecular compartment (Fig. 3B), both MAR and BFR were significantly lower on the endosteal surface of cortical bones of younger (2-month-old) but not older Gja1Jrt/+ versus WT mice (Fig. 2C). Notably, whereas MAR and BFR significantly declined after 2 months on the WT endosteal surface, MAR significantly declined only by 8 months of age and BFR did not decline significantly in Gja1Jrt/ +. Also, mineralizing surface per bone surface (MS/BS) was not significantly different in Gja1Jrt/+ versus WT, but by 8 months of age MS/BS was significantly higher in Gja1Jrt/+ cortical bone, reflecting a decrease with age in WT but not Gja1Jrt/+ bones.
Expression of early-, mid-, and late-osteoblast and osteocyte markers was not different in cortical bone, with the exception of lower expression of the osteocyte marker Sost in 2-month-old but not other ages of Gja1Jrt/+ versus WT mice (Supplemental Fig. 2). However, whereas Sost expression was unaffected in trabecular bone, expression of most osteoblast-associated genes, including runt-related transcription factor 2 (Runx2), osterix (Osx), Alp, Col1a1, Bsp, Ocn, and phosphate-regulating gene with homologies to endopeptidases on the X chromosome (Phex), was increased in Gja1Jrt/+ versus WT trabecular bone (4-month-old bones shown; Fig. 3D). Similarly, serum ALP, a bone formation marker, was also significantly elevated at 4, 8, and 12 months of age in Gja1Jrt/+ compared to WT mice (Fig. 3C).
G60S was also an activating mutation for stromal cell colony-forming efficiency in vitro, as evidenced by the significant increase in stromal progenitor populations isolated from the Gja1Jrt/+ versus WT bone marrow (CFU-fibroblast [CFU-F], CFU-ALP, or CFU-osteoblast [CFU-O]) (Fig. 4A). Neither stromal cell proliferation nor the sizes of individual colonies and bone nodules, ALP area/CFU-ALP and mineralized area/CFU-O, was significantly different between genotypes (data not shown).
Expression patterns of osteoblast-associated differentiation markers in cultured stromal cells harvested at day 8, 11, 14, and 19, corresponding roughly to the proliferation, differentiation, maturation-early mineralization, and late mineralization stages, were not different between genotypes during the proliferation or early differentiation stages. However, at later maturational and late mineralization stages, most osteoblast-associated markers were more highly expressed in Gja1Jrt/+ bone marrow stromal cultures (Fig. 4B). Protein extracts isolated from both trabecular bone (Fig. 4C) and from mineralized nodules of end point cultures (Fig. 4D) showed that the matrix produced by Gja1Jrt/+ osteoblasts contained strikingly elevated levels of BSP compared to WT matrix although levels of other matrix proteins, such as osteopontin (OPN) and OCN, were normal.
Gja1Jrt/+ osteoclast number and activity are increased in vivo, but not in vitro
No differences were found in the expression of osteoclast differentiation and fusion markers, including nuclear factor of activated T cells, cytoplasmic, calcineurin-dependent 1 (Nfatc1), calcitonin receptor (CalR), tartrate-resistant acid phosphatase (Trap), and osteoclast-associated receptor (Oscar) as assessed by quantitative RT-PCR (QPCR) of RNA isolated from bones of the two genotypes (Supplemental Fig. 3C). However, several observations suggested that mutant osteoclasts were more active in vivo than their WT counterparts in younger mice. First, osteoclast surface per bone surface (Oc.S/BS) was significantly increased at 2 months in both cortical and trabecular compartments in Gja1Jrt/+ versus WT femurs (Fig. 5A), increases that were no longer detectable in 4-month-old or older mice; indeed, in cortical bone, Oc.S/BS was significantly decreased with age in Gja1Jrt/+ versus WT. Second, expression of Cathepsin K, an osteoclast activity marker, was increased at 2 (p = 0.061) and 4 (p = 0.008) months of age in Gja1Jrt/+ versus WT bone (trabecular bone shown; Fig. 5B). Finally, although there was no significant difference between genotypes at any of the ages examined, the serum concentration of a resorption marker, the CTX-1, remained level in Gja1Jrt/+ mice, whereas in WT serum it declined significantly with increasing age (Fig. 5C).
To determine whether these differences were cell-autonomous, osteoclast cultures derived from both spleen and bone marrow of mice at 2 (Fig. 5D), 4, 8, and 12 (data not shown) months of age were cultured in the presence of RANKL and M-CSF and evaluated for osteoclast differentiation and activity. No significant differences were found in vitro in osteoclast number, size (number of nuclei), or resorption activity between Gja1Jrt/+ and WT cells irrespective of mouse age.
Age-related changes in the RANKL/OPG axis exacerbate or abrogate respectively a BSP-induced increase in osteoclastogenesis and osteoclast bone resorption in younger versus older Gja1Jrt/+ mice
Taken together, the data suggest that G60S is an osteoblast autonomous and osteoclast nonautonomous Cx43 activating mutation in Gja1Jrt/+ bone, leading us to investigate the basis of the activation of Gja1Jrt/+ osteoclasts. We showed previously that in Bsp–/– mice, bone resorption is diminished, resulting in mice with increased trabecular bone volume. We therefore asked whether the abnormally high levels of BSP in Gja1Jrt/+ bone contributed to increased bone resorption in Gja1Jrt/+ mice by plating WT osteoclasts onto bone fragments generated from WT, Gja1Jrt/ + , Bsp + /+ (WT littermates of Bsp–/– mice), and Bsp–/– mice. WT osteoclasts exhibited significantly higher resorption activity, assessed via CTX-1 concentration in cell culture medium, when plated onto Gja1Jrt/+ mouse bone fragments than on WT. Conversely, osteoclasts had lower resorption activity when plated onto Bsp–/– bone fragments than when plated onto Bsp + /+ strain-matched bone fragments (Fig. 6A). This finding was further supported by QPCR expression analyses which showed that WT osteoclasts plated onto Gja1Jrt/+ bone fragments had a significantly increased expression of Cathepsin K over those plated onto WT bone fragments (Fig. 6B).
The finding that the osteopenic phenotype is present early and at first worsens precipitously at 4 months but then becomes less pronounced in older mutant mice relative to WT, despite the fact that BSP overproduction is sustained throughout the lifespan of Gja1Jrt/+ mice, prompted us to examine the OPG/RANKL signaling system at different ages. Although there was no difference between genotypes in the Rankl/Opg gene expression in either cortical or trabecular bone (Supplemental Fig. 3A, B, respectively), in trabecular bone, mbRANKL significantly increased from 2 to 4 months of age in Gja1Jrt/+ (2.4-fold change; p = 0.02), consistent with the dramatic decline in bone volume from 2 to 4 months, whereas there was no age-related change in mbRANKL in WT samples (Fig. 6C). In older mice (8 months old), relative serum concentrations of OPG were higher in the mutant compared with that in the WT mice, an observation not found in younger mice (2 months old), whereas serum concentrations of RANKL were similar between the genotypes at both ages (Fig. 6D). The increased OPG, along with unchanged RANKL, in older Gja1Jrt/+ versus WT mice may thus contribute to the relative protection of the older Gja1Jrt/+ mice against a further age-related decrease in BMD.
Gja1Jrt/+ mice, which carry a G60S missense mutation in Cx43, express less than 50% of WT levels of Cx43 and have markedly reduced gap junction formation and function in osteoblasts and other Cx43-expressing cell types.[14, 16] Like other Cx43 mutants with loss or reduction in Cx43 gap junction formation and or function, Gja1Jrt/+ mice exhibit early-onset osteopenia and changes in the structural and biomechanical properties of bone, and like G138R Cx43 missense mutation knockin (+/G138R) mice, exhibit the classical features of human ODDD. We report here a longitudinal study of Gja1Jrt/+ mutant mice, which recapitulate some phenotypic traits of other Cx43 loss-of-gap junction function models, but also exhibit novel and age-related bone phenotypes, and we show that the mechanism underlying the osteopenia in these mice results from activation of osteoblast activity, which also protects mice from further old age-related bone loss.
G60S is unique in being an osteoblast-autonomous activating mutation
Several Cx43 mutant mouse models have been described previously, including a global Cx43 deletion, conditional-deletion of Cx43 in bipotent osteochondroprogenitors (DM1Cre), osteoblasts (Col1a1Cre), mature osteoblasts-osteocytes (OcnCre),[4, 8] and osteocytes (Dmp1Cre),[8, 9] and a mutant Cx43 knockin (+/G138R). In all these cases except the Dmp1Cre-osteocyte-specific Cx43 deletion,[8, 9] a reduction in Cx43 gap junction formation and/or function resulted in mice that displayed varying degrees of osteopenia, as seen also in the Gja1Jrt/+ model. The early osteopenic phenotype has usually been attributed to a reduction in osteoblast number and/or function and/or changes in RANKL/OPG signaling that increase osteoclast formation and activity; differences in the different models have been attributed to different consequences of loss of Cx43 function in less or more mature progenitor or osteoblast-osteocyte populations. We report that the G60S mutation does not abrogate but instead activates osteoblast function in appendicular bones and in stromal populations as manifested by increases in MAR-BFR-MS/BS, and increased expression of many osteoblast-associated genes and increased production of BSP, mbRANKL, OPG, and ALP proteins. It should also be noted that while a previous study suggested that terminal differentiation is diminished in neonatal G60S calvarial osteoblast cells in culture (based on lower Bsp and Ocn expression versus WT cells), our studies showed that Gja1Jrt/+ calvarial cells are indistinguishable from WT cells in vitro with regard to ALP production, mineralization, and expression of osteoblast-associated markers tested including Bsp and Ocn (data not shown). The reasons for the discrepancies are unclear but may include differences in cell isolation, culturing conditions, and/or mouse strain variations resulting from independent breeding of successive generations.
In addition to increased osteoblast activity, mesenchymal progenitor and osteoprogenitor numbers were also increased in stromal cells isolated from Gja1Jrt/+ compared to WT mice, thus suggesting a role for Cx43 in stromal cell commitment, maintenance of precursor populations, and/or controlling the overall subpopulation makeup of the stroma. It was recently reported that mesenchymal and osteoblastic progenitors were increased in the bone marrow of Col1a1Cre;Cx43fl/fl mice relative to WT littermates; the increases were attributed to downregulated expression of Sost, a factor that prevents mesenchymal stem cell (MSC) proliferation. Similarly, osteoblast activity was reported to be increased in Dmp1Cre-Cx43 ablated mice also as a consequence of downregulation of SOST due to increased apoptosis of osteocytes. However, we found no evidence for increased osteocyte apoptosis, altered osteocyte number, or altered number of empty lacunae at any age in cortical or trabecular bone compartments, or for altered Sost expression in cortical and trabecular bone in Gja1Jrt/+ versus WT, with the exception of a decrease in Gja1Jrt/+ cortical bone at 2 months; similarly, Sost expression was increased at later differentiation stages in Gja1Jrt/+ stromal cell cultures, consistent with the increased osteoblastogenesis observed. Additionally, in Gja1Jrt/+ cells, proliferation rates and self-renewal capacity in vitro were unchanged (data not shown). It remains to be determined what other factors affect the MSC microenvironment, but we previously reported biphasic expression of BSP during mesenchymal cell differentiation, with early upregulated or “primed” expression of BSP in very primitive osteoprogenitors, suggesting that BSP overexpression in Gja1Jrt/+ mice may be a factor contributing to MSC commitment. Further studies are ongoing to dissect how the G60S mutation elicits this effect specifically on stromal progenitor populations, but altered response to mechanical loading and hormonal signals may also play roles (see the final section of the Discussion).
In contrast to osteoblasts, the G60S mutation had no detectable cell autonomous effect on osteoclasts. This is consistent with previous studies on Cx43-deficient mice in which changes in the RANKL/OPG ratio (increased osteocyte-derived Rankl/Opg mRNA ratio in OcnCre;Cx43fl/fl mice, decreased Opg mRNA levels in DM1Cre;Cx43–/fl osteoblasts, and decreased OPG expression by osteocytes in DMP1Cre;Cx43fl/fl mice) have been reported to underlie the increased osteoclastogenesis and bone resorption and contribute to the osteopenia and/or altered bone structure in these mutants. We found no evidence to suggest that the Rankl/Opg changes were specific to changes in expression by osteocytes, as evidenced by no significant difference in Rankl/Opg in mutant versus WT cortical bone. However, we did find an osteoblast-dependent upregulation of osteoclast activity in Gja1Jrt/+ mice resulting from changes in the RANKL/OPG signaling axis, with increased mbRANKL protein but normal levels of serum OPG in young Gja1Jrt/+ mice, concomitant with increased osteoclast number and activity, and phenotypic traits that changed with aging (see the final section of the Discussion). It is also important to consider that the Gja1Jrt/+ osteoclasts may be affected by changes in RANKL/OPG expression by other G60S Cx43-expressing cells that are not in the osteoblast lineage, such as stromal cells, fibroblasts, or activated T-cells.
Gja1Jrt/+ is the first Cx43 mutant mouse in which unusually high levels of matrix BSP have been reported and linked to increased osteoclast activity
Although gene expression of most osteoblast-associated markers was upregulated, only BSP content and not that of other osteoid proteins (eg, OPN, OCN) was increased in the trabecular bone matrix and mineralized nodules of stromal cultures of Gja1Jrt/+ compared to WT mice. Whether this is due to preferential degradation of matrix proteins other than BSP, more robust sequestration of BSP into the bone matrix, or other possibilities, is currently not known. Additionally, this does not preclude the possibility that the content of other untested matrix proteins may be altered, contributing to the abnormal composition and enhanced resorption rate of the Gja1Jrt/+ bone matrix. In any case, the in vitro resorption assay we developed indicated that the abnormal composition of the Gja1Jrt/+ bone matrix was a significant factor in Gja1Jrt/+ osteopenia, with high BSP promoting high resorption. RANKL and human recombinant BSP were shown to act synergistically to induce osteoclastogenesis and bone resorption in vitro. We have also reported that Bsp–/– mice, in contrast to the Gja1Jrt/+ mice, had higher trabecular bone density and lower bone turnover. Similarly, many histological features of the Gja1Jrt/+ skeleton correlate with observations in BSP-overexpressing cytomegalovirus (CMV)-BSP transgenic mice, which display a decrease in trabecular bone due to an increased Oc.S/BS. Thus, from both our in vivo analyses of young Gja1Jrt/+ bones and in vitro resorption assays testing the Gja1Jrt/+ bone matrix, we conclude that the decrease in Gja1Jrt/+ bone volume results from increased osteoclastogenesis and bone resorption at least in part in response to increased matrix BSP.
No other studies have yet reported a bone matrix containing abnormal levels of bone proteins in Cx43 mutant mice[4-6, 13]; however, we predict that the disordered collagen bundles recently reported in cortical bone matrix of DM1Cre;Cx43–/fl mice and the decreased mineralization of the femoral diaphysis in DMP1Cre;Cx43fl/fl mice may reflect abnormal content of the noncollagenous proteins, in particular BSP. This may also be a factor in the recently reported OCNCre;Cx43fl/– mice, in which reduced quality of the bone matrix and its decreased material properties were linked to improper maturation of collagen cross-links (decreased fraction of nonreducible to reducible collagen cross-links) in the cortical bone matrix. We also suggest that matrix anomalies, in particular changes in BSP, may contribute to the alteration in the osteogenic bone marrow niche recently reported in Col1a1Cre;Cx43fl/fl mice. Taken together, the data indicate a need for additional analysis of the matrix in various Cx43 mutant mouse lines and the role of matrix anomalies, ie, niche anomalies, in both altered osteoclast and altered osteoblast activities.
Gja1Jrt/+ mice are protected from old age–related diminution in BMD and exhibit age-related improvement in femoral bone structural and material parameters
Although many phenotypic traits are seen in the Gja1Jrt/+ mouse model that parallel those reported in various other Cx43 mutant mouse models, some traits are unique, including certain phenotypic changes with aging and the underlying mechanisms. For example, lower BMD versus WT controls has been reported in almost all Cx43 mutant models with loss of gap junction function, and low BMD, where observed, persists up to at least 12 months of age,[4, 6, 7, 13] as it does in Gja1Jrt/+ mice. Similar to what has been described for DM1Cre;Cx43–/fl and DM1Cre;Gja1+/fl(G138R mice, the greatest difference in Gja1Jrt/+ versus WT BMD was seen at younger ages. Notably, however, no significant difference in trabecular parameters (BV/TV, Tb.N, and Tb.Th) was seen in 12-month-old Gja1Jrt/+ versus WT femurs, and Gja1Jrt/+ femoral cortical thickness surpassed that of age-matched WT bones by 8 months of age. Corresponding age-related changes in the structural and material properties of Gja1Jrt/+ long bones were also seen. Thus, similar to observations made in DM1Cre, Col1a1Cre, OcnCre, and Dmp1Cre Cx43-conditionally deleted mice,[4, 7-9, 11, 12, 34] Gja1Jrt/+ femurs exhibited reduced structural and material properties versus WT at younger ages. Gja1Jrt/+ bone material quality improved in older mice, although the structural properties remained lower versus WT, most likely because the Gja1Jrt/+ femoral total tissue cross-sectional area remained smaller than WT throughout life. This is in contrast to what is seen in OcnCre;Cx43fl/– and DMP1Cre;Cx43fl/fl mice, in which structural parameters were unaffected even though material properties were reduced, presumably because the benefit of the increased femoral cross-section offset the lower bone quality in these mutants.
To our knowledge, Gja1Jrt/+ is the only mouse model in which loss of gap junction function leads to increased cortical thickness with age, and age-related elimination or abrogation of the reduced cortical thickness, increased marrow space and reduced material properties compared to that observed in younger Gja1Jrt/+ mice versus WT. In most Cx43 conditional-deletion models, changes in the cortical parameters and bone geometry reflect increased endocortical resorption and increased or unaffected periosteal bone formation.[4, 7-9, 11, 12, 34] In contrast, in Gja1Jrt/+ mice, cortices in younger (2 months old) mice are thinner due to decreased bone formation and increased bone resorption on the endosteal surface, with no detectable change in periosteal parameters. The correction of the cortical bone and marrow area proportions and thickening of the cortical bone in older Gja1Jrt/+ mice results from a significant decrease in endosteal bone resorption hand-in-hand with maintenance of endosteal bone formation and mineralization parameters higher than those seen in WT at the same ages. Our data suggest an age-related switch in the Rankl-Opg signaling mechanism, with increased mbRANKL and increased resorption in younger Gja1Jrt/+ mice, followed by increased serum OPG in older mice, reducing bone resorption, allowing cortical and trabecular bone thickness to increase over time. As already mentioned, the changes in Gja1Jrt/+ mice do not appear to reflect changes in osteocyte-specific changes in either Rankl-Opg signaling or Sost expression.
In addition to age-related effects on osteoclasts and resorption, expression of osteoblast-associated genes is higher in Gja1Jrt/+ versus WT mice at all ages tested and up to at least 1 year; among upregulated genes was ALP, which would be expected to contribute to increased total mineral deposition, and increases in BMD with age. An expanded osteoprogenitor population, as seen in Gja1Jrt/+ versus WT stromal cell cultures, presumably contributes to maintenance if not expansion of the bone itself in older Gja1Jrt/+ mice; an increased capacity to generate active osteoblasts could contribute to the lack of further decreases in BMD, Tb.N, and BV/TV and increased cortical thickness seen in aging Gja1Jrt/+ mice. The age-related changes in the Gja1Jrt/+ bone phenotype may arise as a consequence of age-related changes in Cx43 gap junction formation and/or function, altered responsiveness to mechanical load or hormonal and molecular signals, skeletal site-specific differences in sensitivity to disruption of Cx43, and/or other factors. Recently, for example, Cx43 deficiency has been shown to result in an increased responsiveness to mechanical load,[4, 12] with several studies demonstrating that cells from different skeletal locations are differentially sensitive to loss of Cx43, particularly within the endocortical and periosteal surfaces of the cortical bone.[4, 7, 8] Specifically, in response to mechanical loading, DM1Cre;Cx43fl/fl mice displayed an increased periosteal, but decreased endocortical (BFR) response, Col1a1Cre;Cx43–/fl mice a decreased endocortical (BFR) response, and OcnCre;Cx43–/fl mice both enhance periosteal and endocortical (BFR) response versus WT.[4, 12] After loading, DM1Cre;Cx43fl/fl also experienced a significant increase in trabecular BV/TV, an effect not seen in WT mice. Interestingly, Lloyd and colleagues very recently showed that after unloading via hind limb suspension, OcnCre;Cx43–/fl mice experienced an attenuated response (less of a decline in trabecular bone parameters and no suppression of periosteal and endosteal bone formation) versus WT. Other recent studies in rat stromal and osteoblastic cells have indicated that Cx43 gap junction formation and the capacity for gap junction intercellular communication in response to a hormonal signal (PTH) is significantly decreased as a function of age. Whether and how the G60S mutation in Cx43 alters responses to hormonal or mechanical stimuli remains to be determined, but the differences observed in the Gja1Jrt/+ versus other loss-of-function bone phenotypes support the view that the mechanisms are multifactorial and, reflect a complex summation of positive and negative effects across the diversity of bone cell populations and maturational stages affected.[4-8, 10-13, 27, 34]
In summary, we report that the G60S mutation results in a cell autonomous activation of the osteoblast population and further link the resulting production of abnormally high matrix levels of BSP protein to increased osteoclastogenesis and bone resorption (Fig. 7). Our results also show that the G60S mutation, unlike Cx43 knockout mutations, exerts a significant age-related enhancement of trabecular and cortical bone volume and quality.
All authors state that they have no conflicts of interest.
This work was supported by a Canadian Institutes of Health Research (CIHR) operating grant (FRN 69198 to JEA) and the Quebec Transgenic Research Network (to JEH), as well as scholarship support from CIHR-Skeletal Regenerative Medicine Team (to TZ), Queen Elizabeth II-GSST (to TZ); Department of Medical Biophysics (University of Toronto) (to TZ); Canadian Arthritis Network (to RAZ), and CIHR/Osteoporosis Society (to RAZ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We thank members of the Centre for Modeling Human Disease for their support and discussions (Celeste Owen, Nishma Kassam, Ann Flenniken, Zorana Berberovic, Bill Stanford, and Nicole Anderson); Usha Bhargava, Ruolin Guo, Nick Basso, Feryal Sarraf, and Lee Wei for expert technical assistance; and Cindy Lee for contributions made during a summer studentship and research elective.
Authors' roles: Study design and conduct: TZ, FC, LM, RAZ, MG, JEH, and JEA. Data collection: TZ, FC, LM, and RAZ. Data analysis: TZ, FC, LM, RAZ, and JEH. Data interpretation: TZ, FC, LM, RAZ, MG, JEH, and JEA. Drafting manuscript: TZ, FC, and JEA. Revising manuscript content and approving final version of manuscript: TZ, FC, LM, RAZ, MG, JEH, and JEA. TZ, FC, LM, RAZ, MG, JEH, and JEA take responsibility for the integrity of the data analysis.