Low peak bone mass and attenuated anabolic response to parathyroid hormone in mice with an osteoblast-specific deletion of connexin43

Bone-forming cells are highly coupled by gap junctions formed primarily by connexin43 (Cx43) and, to a lesser degree, connexin45 (Cx45) proteins (Civitelli et al., 1993; Donahue et al., 1995; Steinberg et al., 1994). Several in vitro studies have demonstrated that Cx43 is involved in modulating the differentiation and function of bone-forming cells as well as osteocytes (Cheng et al., 2001; Lecanda et al., 1998; Schiller et al., 2001a; Schiller et al., 2001c); and work from our group indicates that Cx43 controls osteoblast gene transcription via modulation of specific signaling systems required for osteoblast gene expression (Stains et al., 2003; Stains and Civitelli, 2005b).

Although this work has laid the foundation for understanding the biology of gap junction proteins in bone, only recent studies in human and mouse genetics have brought to the fore the biologic role of Cx43 in the skeleton. We had reported that targeted ablation of the Cx43 gene in the mouse leads to a skeletal phenotype characterized by retarded intramembranous and endochondral ossification, craniofacial abnormalities and osteoblast dysfunction (Lecanda et al., 2000), providing in vivo evidence that this gap junction protein is required for normal bone development and osteoblastic differentiation (Lecanda et al., 1998; Stains et al., 2003). This notion is now further supported by findings of Cx43 mutations in patients with oculodentodigital dysplasia (ODDD), a rare congenital disease whose phenotypic features include craniofacial malformations and syndactyly (Paznekas et al., 2003; Richardson et al., 2004). A similar, though not identical phenotype has been recently reported in a mouse with a dominant negative Cx43 mutant allele, Gja1^Jrt (Flenniken et al., 2005). Interestingly, these animals have also generalized osteopenia, thus reinforcing the notion that functional Cx43 is important for bone mass accrual and maintenance. Such a premise could be tested in a full gene-deletion model, but unfortunately homozygous Cx43 null mice die shortly after birth because of severe cardiovascular malformations (Reaume et al., 1995), thus precluding the use of this model to study the consequences of complete lack of Cx43 in the adult skeleton.

In vitro studies have also shown that Cx43 is critical for bone cell response to a variety of stimuli and pharmacologic agents. For example, inhibition of gap junctional communication or Cx43 expression hinders osteoblast responses to fluid flow (Cherian et al., 2005; Saunders et al., 2001), or to mechanically induced calcium waves (Jørgensen et al., 2000). Further, the action of parathyroid hormone (PTH), an important regulator of bone remodeling also seems to be dependent on gap junctional communication. PTH increases gap junctional communication between osteoblasts by modulating Cx43 expression or function (Civitelli et al., 1998; Donahue et al., 1995), and interference with Cx43-mediated gap junctional communication using antisense oligonucleotides or chemical inhibitors disrupts both PTH-induced cAMP accumulation (Van der Molen et al., 1996) and osteoblast differentiation (Schiller et al., 2001b). Based on these findings, and considering the osteoblast dysfunction of Cx43 null osteoblasts (Lecanda et al., 2000), we hypothesized that lack of Cx43 would negatively affect skeletal responsiveness to anabolic stimuli, such as that produced by intermittent PTH administration, the only currently available modality for inducing new bone formation (bone anabolism) in patients with osteoporosis and fractures (Neer et al., 2001).

To determine the biologic importance of Cx43 in the adult skeleton, we generated a conditional Cx43 gene ablated mouse model based on the Cre/loxP system (Nagy, 2000). In this model, which overcomes the lethality of the germline Cx43 null mutation, Cre expression is driven by a 2.3-kb fragment of the α₁(I) collagen promoter, resulting in replacement of the entire Cx43 reading frame with the lacZ reporter cassette selectively in bone-forming cells (Castro et al., 2003). With this model, Cre is expressed just before birth and in cells that are already partially differentiated into osteoblasts, thus providing an osteoblast-specific and postnatal gene ablation model (Dacquin et al., 2002). We find that these animals are viable, but develop a low peak bone mass and remain osteopenic throughout their adult life, the result of a reduced ability of bone-forming cells to fully differentiate. They also exhibit a dramatically attenuated response to the anabolic effect of intermittent PTH administration. Thus, Cx43 is important not only for normal skeletal development, but also for peak bone mass accrual and adult bone homeostasis. Pharmacologic stimulation of gap junctional communication may enhance the effect of osteoanabolic agents, such as PTH.

Specific osteoblast Cx43 gene deletion was demonstrated in ColCre;Cx43^–/fl mice by different approaches. PCR of genomic DNA extracted from bone revealed the expected 670-kb band corresponding to the Cx43 deleted allele, a band that was absent from extracts of tail soft tissue (Fig. 1A). Accordingly, Cx43 immunoreactive bands were barely detectable in Western blots of bone tissue extracts from conditionally deleted mice, contrasting with strong bands in wild-type equivalent littermates and fainter bands in heterozygous equivalent mice (Fig. 1B), the latter reflecting both the loss of one Cx43 allele and haploinsufficiency of the `floxed' allele (Theis et al., 2001). Substantial amounts of mRNA transcripts for the Cre transgene were detected only in bone extracts from ColCre;Cx43^–/fl mice but not in either Cx43^–/fl or Cx43^+/fl extracts (Fig. 1C). Whole-mount preparations of newborn animals revealed strong X-gal stain in areas corresponding to mineralized skeleton of ColCre;Cx43^–/fl mice, whereas very faint stain was observed in Cx43^–/fl mice, which may reflect endogenous β-galactosidase expression (Kim et al., 2004), as it is observed in animals lacking the lacZ reporter (Fig. 1D). In conditional Cx43-deleted mice, X-gal blue staining was intense in areas of more advanced ossification, such as the diaphyses of long bones, vertebral bodies, ribs, distal mandible and facial bones, whereas staining was not observed in the epiphyses of long bones, corresponding to cartilaginous growth plates, nor in the skin or internal organs (Fig. 1D). Confirming osteoblast-specific Cx43 gene deletion, X-gal stain was selectively observed in cells lining the bone surfaces of tibial cortical endosteum and trabecular surfaces of both tibia and vertebra. As expected, most osteocytes were also X-gal positive, whereas no stain was observed in bone marrow cells (Fig. 1E).

All genotypes were obtained at the expected Mendelian frequency and were viable. Whole-mount alizarin red/alcian blue staining of newborn mice did not reveal any major skeletal abnormalities in Cx43 conditional knockout mice compared with their control littermates (Fig. 2A), consistent with post-developmental deletion of Cx43. However, body weight at 1 month was significantly lower in ColCre;Cx43^–/fl mice relative to the other genotypes in both males and females, a difference that persisted until at least 6 months of age (Fig. 2B). Importantly, conditional Cx43-deficient mice exhibited significantly lower whole-body bone density by dual-energy X-ray absorptiometry (DEXA) compared with Cx43^+/fl or Cx43^–/fl littermates by two-way ANOVA (Fig. 2C). This relative osteopenia was significant as early as 1 month of age and persisted with age at least up to 6 months (P<0.05 and P<0.01, respectively). Bone mineral content (BMC) very closely resembled the bone density data, with approximately 5% lower bone mass in ColCre;Cx43^–/fl mice relative to the wild-type and heterozygous equivalent mice.

Histomorphometric analysis evidenced a markedly more reduced trabecular bone mass in ColCre;Cx43^–/fl mice, with approximately 40% reduction in bone volume/total volume and more than 50% reduction in osteoblast number relative to wild-type littermates (Fig. 3A-C). Trabecular thickness in the conditional Cx43 ablated mice was likewise reduced by ∼30%, without differences in trabecular number (Fig. 3D,E). Mineral apposition rate was reduced by ∼17%, although not statistically significantly, relative to wild-type and heterozygous equivalent littermates (Fig. 3F). By contrast, there were no statistical differences in osteoclast number among the different genotypes (Fig. 3G). The apparent discrepancy in the degree of osteopenia between DEXA and histomorphometric measurement is not uncommon (Castro et al., 2004), and reflects both a lower sensitivity of DEXA and different skeletal sites measured.

To gain further insights into the pathobiologic mechanism of this osteopenic phenotype, we studied calvaria cells isolated from genetically modified animals. Demonstrating osteoblast-specific and differentiation-dependent Cx43 gene replacement, X-gal staining was negative in ColCre;Cx43^–/fl calvaria cells upon reaching confluence, but it became progressively stronger 1 week post-confluence onward (Fig. 4A). Progressively increased X-gal stain during osteoblast differentiation is entirely consistent with the expression pattern of the promoter used to drive Cre (Dacquin et al., 2002). Accordingly, barely detectable Cx43 immunoreactive bands were detected in lysates of conditionally deleted cells, with reduced abundance of Cx43 in heterozygous equivalent cells (Fig. 4B). Likewise, Cx43 mRNA abundance, assessed by real-time PCR, was reduced by ∼90% and ∼50% in conditionally Cx43-deleted and heterozygous equivalent calvaria cells after 3 weeks in culture (Fig. 4C).

Development of alkaline phosphatase activity, a marker of osteogenic differentiation, was significantly reduced in calvarial cells derived from ColCre;Cx43^–/fl mice 2 weeks post-confluence, when it usually reaches a peak, as it occurred in the other two genotypes (Fig. 4D). Furthermore, after 2 weeks in culture the abundance of mRNA transcripts for other osteoblast-specific genes, namely osteocalcin, α₁(I) collagen, osteopontin and Cbfa1/Runx2 was reduced by more than 50%, measured by real-time PCR, relative to wild-type equivalent cells (Fig. 4E). By contrast, neither Cx45 nor N-cadherin mRNA were significantly different among the three genotypes (Fig. 4E). Importantly, ColCre;Cx43^–/fl calvaria cells were not able to produce mineralized matrix until 3 weeks in culture, whereas Cx43^+/fl and Cx43^–/fl cells were able to start mineralization after 2 weeks (Fig. 4F,G). These in vitro data strengthen the notion that Cx43 expression is necessary for full elaboration of the osteoblast phenotype.

We next tested the ability of conditional Cx43-deficient mice to respond to the anabolic stimulus provided by intermittent PTH injections. In a first study, we tested 4 doses of PTH in 5- to 6-month-old mice treated 5 days a week for 4 weeks. Because of the lower bone mass in the conditionally deleted mice relative to the other genotypes (Figs 2, 3), in these studies we monitored whole-BMC rather than bone density, to assess the absolute amount of bone gained in each group. In the wild-type equivalent Cx43^+/fl group, PTH treatment induced rapid and dose-related increments in whole-body BMC, with significant increases over baseline at 4 weeks with all doses, except the lowest one. Maximal increases were 13.1% and 13.4% in Cx43^+/fl and Cx43^–/fl mice, respectively, with significant bone gain as early as after 2 weeks of treatment (Fig. 5A,B). However, in the conditionally Cx43-deleted ColCre;Cx43^–/fl mice, only two doses of PTH resulted in statistically significant increments in bone mass, and the maximal effect obtained (9.8%) was ∼30% lower than that observed in the other two genotypes (Fig. 5C). However, taking into account the increase occurring in untreated wild-type animals (4.8%), the difference in response amplitude would be more than 40%.

Rather surprisingly, even in vehicle-treated groups we detected a basal increase in BMC, presumably reflecting continuous bone growth in these 5-6-month-old animals. Because this may confound the extent of bone gain obtained with PTH, we repeated a similar study in older mice (7.4- to 9.6-months-old), whose bone mass should be stable. In this case, we used 40 μg/kg PTH, a dose that induced maximal effects in all genotypes in the younger animals. Again, 4-week treatment with 40 μg/kg PTH induced significant changes of whole-body bone mass in Cx43^+/fl (12.5±4.7% from baseline; n=15) and Cx43^–/fl (9.3±4.6%; n=11) mice, whereas the anabolic effect of PTH was reduced by 47% in the ColCre;Cx43^–/fl group (6.7±5.3%; n=10). The changes in bone mass induced by PTH in the conditionally deleted animals were just slightly higher but not statistically different than the changes observed in a group of wild-type equivalent mice treated with vehicle (3.4±3.9%; n=9) (Fig. 6A). Region-specific analysis on BMC changes by DEXA also revealed that PTH significantly increased bone mass (>12%) at the lumbar spine only in the wild-type equivalent group, whereas no changes occurred at this site in the conditionally Cx43-deleted mice (Fig. 6B). Conversely, the anabolic effect of PTH on femur BMC was not affected by genotype, exhibiting an anabolic response of almost equal magnitude for each group (Fig. 6C).

Bone histomorphometric analysis was fully consistent with the DEXA results. After a 4-week treatment with 40 μg/kg PTH, bone volume (BV)/total volume (TV) was increased almost threefold in Cx43^+/fl mice compared with mice of the same genotype treated with vehicle. A significant increase of lesser magnitude was also observed in the heterozygous equivalent group, Cx43^–/fl, whereas BV/TV was not different in the ColCre;Cx43^–/fl group relative to the vehicle-treated group (Fig. 7A-D). Osteoblast number was increased in all genotypes with no statistical differences among groups, even though this parameter was ∼30% lower in conditionally Cx43-deleted mice relative to Cx43^+/fl littermates (Fig. 7E). Conversely, other static histomorphometric parameters of bone formation, trabecular number and thickness were significantly increased in Cx43^+/fl (∼20 and ∼40%, respectively) but not in ColCre;Cx43^–/fl or Cx43^–/fl mice (Fig. 7F,G). Cortical thickness was also highest in Cx43^+/fl mice, but the changes were not statistically significant (Fig. 7H). By contrast, osteoclast perimeter was higher in the Cx43^–/fl and ColCre;Cx43^–/fl groups, but even in this case the differences were not statistically significant (Fig. 7I).

Dynamic histomorphometric parameters of bone formation were assessed at two skeletal sites, to further investigate differences in PTH responses at the spine and femur. Abundant double-calcein labels were observed in wild-type mice after PTH treatment at both the spine and at the endosteal surface of the tibia. Double labelling was also present in the heterozygous equivalent mice, whereas in the majority of conditional knockout mice only single labels were detected (Fig. 8A,C). Consequently, mineral apposition rate (calculated in the trabecular and endosteal surfaces) was significantly lower in ColCre;Cx43^–/fl mice than in the Cx43^+/fl group in both sites (Fig. 8B,D). However, periosteal mineral apposition rate in the tibia was not significantly different among groups, even though the average was lower in ColCre;Cx43^–/fl mice (0.423±0.246 μm/day) relative to Cx43^+/fl (0.614±0.629 μm/day) and Cx43^–/fl mice (0.710±0.497 μm/day), a result very consistent with the cortical thickness data. Finally, 5-bromo-2′-deoxy-uridine (BrdU)-positive cells were observed on the bone surface in all genotypes after PTH treatment (Fig. 8E), and the osteoblast mitotic index was not different among groups (Fig. 8F).

The present study demonstrates that selective deletion of Cx43 in osteoblasts leads to a marked decrease in peak bone mass and osteopenia; it also severely attenuates the bone anabolic response to intermittent administration of PTH. These abnormalities are caused by a functional defect in bone-forming cells, which fail to increase their activity in response to the hormonal stimulus. Thus, functional Cx43 is required for normal bone mass acquisition and maintenance and it is involved in the mechanism of action of PTH-induced anabolism.

An important role for Cx43 in bone homeostasis and for the function of bone-forming cells was postulated by several in vitro studies (Civitelli et al., 1993; Donahue et al., 2000; Schiller et al., 2001a), and it was established by analysis of mice with a germline null mutation of the Cx43 gene, which exhibit delayed ossification of both endochondral and intramembranous skeleton and defective osteoblast differentiation (Lecanda et al., 2000). Craniofacial malformations are not present in ColCre;Cx43^–/fl mice, most likely because in these animals Cx43 is deleted at around birth (Dacquin et al., 2002), and thus embryonic development would be expected to be normal. However, the osteoblast defect is reproduced in ColCre;Cx43^–/fl mice, a defect that leads to significant osteopenia throughout life. Interestingly, generalized osteopenia is also present in Gja1^Jrt/+ mice, which carry a point mutation of the Cx43 gene (Flenniken et al., 2005), and whose phenotype resembles that of human ODDD, a rare autosomal dominant condition characterized by craniofacial (ocular, nasal and dental) malformations, limb dysmorphisms, spastic paraplegia and neurodegeneration (Loddenkemper et al., 2002; Schrander-Stumpel et al., 1993). Human ODDD has been linked to mutations of the Cx43 gene (Kjaer et al., 2004; Paznekas et al., 2003; Richardson et al., 2004), however both Cx43 null and Gja1^Jrt/+ mice exhibit impaired skull ossification (Flenniken et al., 2005; Lecanda et al., 2000), whereas osteosclerotic changes are described in patients with ODDD (Paznekas et al., 2003; Schrander-Stumpel et al., 1993). Such a discrepancy may be related to species differences, or to mechanisms by which different ODDD mutations affect connexin function. Nevertheless, there is now evidence from different mouse genetic models consistently demonstrating that interference with Cx43 in the postnatal skeleton leads to a low bone mass phenotype.

As noted, the cellular bases of the phenotype observed in conditionally Cx43-deleted mice suggest a defect in osteoblast differentiation and function, previously observed in the germline Cx43 null mutants (Lecanda et al., 2000), and very likely present also in the Gja1^Jrt/+ mouse (Flenniken et al., 2005). Accordingly, ColCre;Cx43^–/fl mice have a low osteoblast number, modestly decreased mineral apposition rate, delayed in vitro osteoblast differentiation, and profound deficit in osteoblast-specific gene expression. These results confirm that Cx43 is required for full osteoblast differentiation and functional activity, although in vivo interference with gap junctional communication between osteoblasts and other cells on the bone microenvironment may also contribute to the phenotype. Because the 2.3-kb fragment of the α₁(I) collagen promoter we used to delete Cx43 is expressed in committed osteoblasts (Dacquin et al., 2002), it is likely that the decreased osteoblast number in bone of conditionally deleted mice reflects a delayed differentiation rather than a decreased recruitment of new osteoblasts, a conclusion also supported by similar proliferation rates of bone cells in wild-type and deleted mice. Of course, this conclusion does not exclude other functions of Cx43 at earlier stages of osteoblast differentiation as postulated by studies in the Gja1^Jrt/+ mouse (Flenniken et al., 2005).

Although low bone mass is present in both ColCre;Cx43^–/fl and Gja1^Jrt/+ (Flenniken et al., 2005) mutants, the molecular mechanisms leading to osteopenia may be different. In the Gja1^Jrt/+ mice the mutation is germline and acts as dominant negative (Roscoe et al., 2005; Shibayama et al., 2005), whereas in our model the mutation is recessive and it occurs only in committed osteoblasts. Furthermore, osteoblasts also express Cx45 (Civitelli et al., 1993) and although this connexin forms gap junction channels of different biophysical properties than those formed by Cx43 (Steinberg et al., 1994; Veenstra et al., 1994), Cx45 might be sufficient to support some degree of gap junctional communication in the absence of Cx43. This may provide a partial compensatory mechanism for the lack of Cx43, even though Cx45 expression is not upregulated in conditionally deleted cells. By contrast, the Gja1^Jrt/+ variant may interfere with both connexins, or other interacting proteins, thus inhibiting the function of both Cx43 and Cx45 (Giepmans, 2004; Saez et al., 2003). These concepts are not at odds with the established notion that Cx45 overexpression reduces Cx43 function (Koval et al., 1995; Lecanda et al., 1998), because while in a mixed Cx43/Cx45 environment the biophysical properties of Cx45 prevail, in a Cx43 null background, as it occurs in our mouse model, the presence of Cx45 would allow a certain degree of cell-cell communication that may partially compensate for lack of Cx43.

The consequences of osteoblast-specific ablation of Cx43 are more severe under the stimulatory action of intermittent administration of PTH, reflected by the dramatic attenuation of the anabolic effect of PTH in ColCre;Cx43^–/fl mice. Interestingly, although responses of lesser magnitude were also observed at intermediate doses of PTH in Cx43^–/fl mice, in which the abundance of Cx43 in osteoblasts is reduced, the highest dose of PTH used (80 μg/kg) elicited a response similar to wild-type mice. By contrast, effects on bone mass that were maximal in wild-type and heterozygous equivalent animals were never achieved in the conditionally deleted mice, and no further gains were obtained with doses above 20 μg/kg. Thus, the gains in bone mass that can be induced by PTH are minimal, though not totally absent, when osteoblasts are deprived of Cx43 in vivo, a conclusion consistent with the notion that the anabolic response to PTH requires functional Cx43. Instead, reduced Cx43 abundance in Cx43^–/fl mice may be sufficient to support some osteoblast functions but not others. In particular, PTH upregulation of Cx43 expression (Civitelli et al., 1998) is likely to be attenuated when Cx43 is decreased, and this may contribute to attenuation of PTH anabolic effect we have seen in Cx43^–/fl mice. Similar observations have been made in the study of Cx43 function in astroglia, where the Cx43^fl allele shows haploinsufficiency for some phenotypical parameters but not for others (Theis et al., 2003). We also observed skeletal site-specific differences in Cx43 sensitivity to PTH anabolic effect in the conditionally deleted animals by regional DEXA analysis. It is possible that lack of Cx43 attenuates PTH response on trabecular bone to a greater extent than it does on cortical bone, thus potentially explaining the normal response in femur observed by DEXA in conditionally Cx43-deleted mice. Envelope-or site-specific effects of PTH have been reported, with more pronounced bone mass increments observed in the trabecular than in the cortical component (Calvi et al., 2001; Gunness-Hey and Hock, 1984; Iida-klein et al., 2002).

The attenuated osteoanabolic response to PTH is the consequence of a failure of Cx43-deficient bone-forming cells to produce new bone under the hormonal stimulus, as demonstrated by ∼70% lower mineral apposition rate in ColCre;Cx43^–/fl than in wild-type mice after a 4-week treatment with PTH, despite a significant increase in osteoblast number. Considering that osteoblast number is decreased in untreated ColCre;Cx43^–/fl mice, the results seem to indicate that the hormone is still able to stimulate osteoblast recruitment to the bone surface in conditional Cx43-deficient mice, although these cells are obviously impaired in their ability to synthesize new bone in response to PTH. Because Cx43 deletion occurs in cells that are already fully committed to the osteogenic lineage, it is likely that some of PTH effects, for example recruitment or proliferation of osteoprogenitors, occur at a stage when Cx43 deletion has not yet taken place, or are Cx43 independent. Although earlier studies indicated that intermittent PTH administration activates existing bone lining cells without affecting cell proliferation (Dobnig and Turner, 1995), an increase in bone marrow osteoprogenitor cells has been reported in response to PTH in rats (Kostenuik et al., 1999) and mice (Tanaka et al., 2004), and osteoblast number is consistently increased in mice (Iida-klein et al., 2002; Knopp et al., 2005). Furthermore, because osteoblast number was increased and osteoblast proliferation was not altered in the conditionally deleted animals, it is unlikely that our gene manipulation may have affected the anti-apoptotic action of PTH to a substantial degree (Jilka et al., 1999).

Although the molecular aspects of the interaction between Cx43- and PTH-induced bone anabolism remain to be elucidated, we had previously observed that PTH upregulates Cx43 expression and function in osteoblasts (Civitelli et al., 1998), and more recently, we demonstrated that interference with Cx43 alters transcriptional regulation of specific gene promoter elements, via MAP kinase- and protein kinase C-dependent pathways (Stains et al., 2003; Stains and Civitelli, 2005b). Because PTH signal transduction involves both of these pathways, it is possible that Cx43 is required to appropriately integrate PTH-activated signals and/or to equalize hormonal responses throughout the osteoblast network (Stains and Civitelli, 2005a). Consistent with this hypothesis, in preliminary results we find that interference with Cx43 function reduces the capacity of osteoblastic cells to increase osteocalcin gene transcription under stimulation by PTH (De Marzo et al., 2005). It is worth mentioning that the distribution of PTH receptors is not uniform throughout the bone tissue, and even within cell lines, certain signal responses are not homogeneous (Civitelli et al., 1992). The present results have interesting ramifications for development of therapeutic strategies for bone anabolism. The nature of the defect in response to intermittent PTH in our animal model makes it likely that similar attenuations of bone mass responses may occur for other anabolic agents or stimuli, i.e. mechanical load, because activation of bone-forming cell function is the ultimate requirement for manufacturing new bone. It is also reasonable to believe that the osteoanabolic response to PTH could be enhanced by increasing gap junctional communication using pharmacologic agents, thus allowing lesser doses or less frequent parenteral administration of PTH. Furthermore, the requirement of osteoblast/osteocyte Cx43 for the anti-apoptotic action of bisphosphonates (Plotkin et al., 2005), which are also widely used in the therapy of osteoporosis, can now be tested in vivo.

In summary, we have demonstrated that selective Cx43 gene deletion in osteoblasts results in adult osteopenia, delayed osteoblast differentiation, and greatly attenuated osteoanabolic response to PTH, the consequence of a failure of Cx43-deficient bone-forming cells to mount a full response to the hormone. Cx43-mediated gap junctional communication represents a potential target for modulation of bone anabolic stimuli.

Development of the mouse model used in these studies has already been reported in some detail (Castro et al., 2003). Briefly, a mouse strain harboring a mutant `floxed' Cx43 allele (Cx43^fl) (Theis et al., 2001) was mated to mice expressing Cre under control of a 2.3 kb α₁(I) collagen promoter fragment (abbreviated as ColCre) (Dacquin et al., 2002), so that Cre-mediated recombination replaces the entire Cx43 reading frame with the lacZ reporter cassette. Homozygous Cx43^fl/fl mice were generated first and crossed with ColCre mice also carrying a Cx43 null allele (ColCre;Cx43^+/–). This strategy avoids potential effects of activation of Cre in the parental germ line. These crosses generate, in approximately equal numbers, the Cx43 conditionally deleted mice, ColCre;Cx43^–/fl, as well as three additional genotypes, Cx43^+/fl (wild-type equivalent), Cx43^–/fl (heterozygous equivalent), and ColCre;Cx43^+/fl (conditional heterozygous). All the mouse lines used in this project were developed in a mixed C57BL/6-C129/J background and littermate were used as controls. Mice were fed regular chow ad libitum and housed in a room maintained at constant temperature (25°C) on a 12 hours of light and 12 hours of dark schedule.

Genotyping was performed by PCR on genomic DNA extracted from mouse tails, after digestion with proteinase K, as described (Lecanda et al., 2000). The Cx43 null allele was detected using primers Cx43-5′: 5′ GGT CAA CGT GGA GAT GCA CCT GAA GCA GAT 3′; Cx43-3′: 5′ AAT CGA TTG GCA GCT TGA TGT TCA AGC C 3′ and Neo^r-5′: 5′GGA TCG GCC ATT GAA CAA GAT GGA TTG CAC 3′. Primers Cx43-5′ and Cx43-3′ amplify a 900-bp product within the Cx43 coding region. Primer Neo^r-5′ hybridizes to the neomycin resistance cassette present only in the null allele, and when used with primer Cx43-3′, it amplifies a 1.4-kb band, spanning the Neo cassette and part of the adjacent Cx43 gene (Houghton et al., 1999). PCR was performed in a final volume of 25 μL reaction; 2 mM MgCl₂, 1× PCR buffer, 0.08 mM of each dATP, dCTP, dGTP, dTTP, 1 μM primers, 2.5 U Taq DNA polymerase, 1-5 μg genomic DNA. The DNA was denatured at 94°C for 3 minutes and amplified for 35 cycles (94°C for 30 seconds, 70°C for 45 seconds and 72°C for 120 seconds) followed by a final extension at 70°C for 20 minutes.

Primers UMP (5′ TCA TGC CCG GCA CAA GTG AGA C 3′) and UMPR (5′ TCA CCC CAA GCT GAC TCA ACC G 3′) were used for the simultaneous detection of the `floxed' (Cx43^fl) and wild-type (Cx43⁺) alleles, as described (Theis et al., 2001). These primers generate a 1 kb amplicon corresponding to the Cx43^fl allele, and a 900 bp band, corresponding to the wild-type allele. In some experiments, the deleted Cx43 allele was directly identified in whole bone extracts, after homogenization and phenol/chloroform extraction. This was accomplished using primers Cx43delforw (5′ GGC ATA CAG ACC CTT GGA CTC C 3′) and Cx43delrev (5′ TGC GGG CCT CTT CGC TAT TAC G 3′), which encompass the junction between the Cx43 gene intron and the β-galactosidase coding region, thus generating a 670 bp amplicon, corresponding to the Cx43-deleted allele (Theis et al., 2001). The ColCre transgene was detected by using the primers Cre 1123-1104: 5′-AAG TGC CTT CTC TAC ACC TG-3′, Cre 982-1002: 5′-TGC TTA TAA CAC CCT GTT ACG-3′, MS1: 5′-GCT CAG CAA GCT CAC AGC AA-3′, and LM6: 5′-GAG CTT ACA CAT TTC GTC-3′. These primers generate 141 bp Cre-specific amplicon and a 448 bp Cre-negative amplicon.

A total of 157 mice were used for basal phenotypic characterization. For the PTH studies, 90 (45 males and 45 females) 5- to 6-month-old animals of three genotypes, Cx43^+/fl, Cx43^–/fl and ColCre;Cx43^–/fl, were subcutaneously injected (5 days a week for 4 weeks) with either vehicle (0.9% saline containing 0.1% BSA and 0.001N HCl; n=12) or human recombinant PTH (1-34) (Teriparatide®, Eli-Lilly, Indianapolis, IN) at doses of 10 μg/kg (n=19), 20 μg/kg (n=18), 40 μg/kg (n=18), and 80 (n=23) μg/kg of body weight. Mice were weighed on the second week of treatment, and the amount of PTH injected was adjusted for any change in weight. A second group of older mice (7.4- to 9.6-month-old; n=45) was also treated with 40 μg/kg PTH with the same modalities as just detailed.

After sacrifice, newborn mice were skinned, eviscerated and maintained for 24 hours in ethanol 100%. After fixation in acetone for 24 hours, the carcasses were stained in a solution containing alizarin red 0.1%, alcian blue 0.3%, acetic acid and 70% ethanol (1:1:1:17). They were then transferred to a 1% KOH solution in 20% glycerol until they were cleared and then stored in glycerol for analysis of cartilage and bone, as described (Lecanda et al., 2000; McLeod, 1980). For whole-mount X-gal staining, carcasses of newborn mice were fixed for 2 hours in 2% formaldehyde, 0.02% paraformaldehyde, 5 mM EGTA, 0.1 mM MgCl₂ and 0.1 M NaPO₄, pH 7.3, then washed in a solution containing NP-40 and stained for 3 hours in X-gal substrate (5-bromo-4-chloro-3-indolyl-D-galactopyranoside) 1 mg/ml, as described (Frendo et al., 1998). They were then transferred to a 1% KOH solution in 20% glycerol until they were cleared and then stored in glycerol. For X-gal staining of bone sections, tibiae or lumbar spine were fixed in 2% paraformaldehyde and 0.02% glutaraldehyde for 1 hour, and then decalcified in 4% EDTA for 17 days. Decalcified bones were incubated in 0.1% (v/v) X-gal substrate (see above) for 12 hours, post-fixed in 4% paraformaldehyde and embedded for paraffin sectioning (Hens et al., 2005). Sections were counterstained with eosin.

Total body BMC and BMD were monitored by DEXA using a PIXImus scanner (GE/Lunar, Madison, WI), under anesthesia with 100 mg/kg ketamine and 10 mg/kg xylazine i.p., as described (Castro et al., 2004). Heads were excluded from the analysis by masking. Region-specific BMD was also measured at the spine and femur, by identifying regions of interest corresponding to the L1-L6 area or the entire femur, respectively. In the latter case, animals were positioned with the femur at a 45° angle with the tibia, and values for both femurs averaged. Calibration was performed daily with a standard phantom as suggested by the manufacturer. The precision of whole-body BMD, assessed by the root mean square method is 1.34% (coefficient of variation) (Castro et al., 2004).

Mice were labeled twice by injection of calcein (15 mg/kg i.p., Sigma-Aldrich) on days 7 and 2 before euthanasia, and bone samples were prepared according to previously described methods, with some modifications (Castro et al., 2004). Briefly, dissected tibiae or lumbar spine were fixed in 70% ethanol and either decalcified in 14% EDTA for 14 days and embedded in paraffin, or left undecalcified and embedded in methyl methacrylate. Plastic sections were stained using the Masson trichrome technique, and tartrate-resistant acid phosphatase activity stain was used for paraffin sections, which were counterstained with methyl green and thionin for identification of osteoclasts and osteoblasts (Liu and Kalu, 1990). Eight μm sections were left unstained for dynamic bone histomorphometry. Quantitative histomorphometry was performed in an area 175-875 μm distal to the growth plate using the OsteoMeasure software program (Osteometrix, Atlanta, GA) in an epifluorescence microscopic system, as detailed elsewhere (Castro et al., 2004). The following parameters of bone remodeling were estimated (Parfitt et al., 1987): trabecular bone volume as a percentage of total tissue volume, trabecular thickness (in μm), trabecular number (per μm), trabecular separation (in μm), osteoblast perimeter per bone perimeter (in percent) or osteoblast number per trabecular area (in number/mm²), osteoclast perimeter per bone perimeter (in percent), and mineral apposition rate (in μm/day), calculated as the mean distance between two fluorescent labels divided by the number of days between the labels.

Osteoblast-enriched calvaria cultures were prepared from newborn mice by sequential collagenase digestion as described (Castro et al., 2004; Lecanda et al., 2000), and grown in α-modified essential medium (αMEM; Mediatech, Herndon, VA), supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals, Norcross, GA) and 100 IU/ml penicillin and 100 μg/ml streptomycin (Sigma Chemicals, St Louis, MO). Approximately 3-5 calvariae were pooled to prepare the cell cultures used in each experiment. Cx43 gene deletion was assessed in differentiating osteoblasts by β-galactosidase activity after fixation in 2% paraformaldehyde, and incubation in a solution containing 1 mg/ml X-gal substrate (see above), as described (Castro et al., 2003). Osteogenic differentiation was assessed by monitoring alkaline phosphatase activity and in vitro mineralization by von Kossa staining in the presence of 50 μg/ml ascorbic acid and 10 mM β-glycerophosphate, using standard techniques (Castro et al., 2004; Lecanda et al., 2000; Shin et al., 2000). Enzymatic activity was normalized for total protein content (Bio-Rad protein assay kit) and expressed as nmol of p-nitrophenol produced from p-nitrophenyl phosphate per minute per mg of protein. Mineralization was quantitated by calculating the surface area covered by dark stain per well, using digital image-processing software (IPLab v.3.5; Scanalytics, Rockville, MD), as previously described (Lecanda et al., 2000).

In vivo cell proliferation was assessed by BrdU incorporation, determined by immunoassay, according to the manufacturer's instructions (5-Bromo-2′-deoxy-uridine labeling and detection Kit III, Roche Molecular Biochemicals). For in vivo labeling, 100 μg BrdU (Sigma, St Louis, MO, USA) per gram of body weight in PBS was injected i.p. 2 hours before sacrifice. Longitudinal, 5 μm sections of paraffin-embedded tibiae, prepared as described above, were rehydrated and incubated for 10 minutes with 30% H₂O₂ in absolute methanol (1:9) and processed in denaturing and blocking solutions following the manufacturer's protocol. BrdU incorporated into nuclei was detected by immunostaining (Zymed Laboratories, South San Francisco, CA). Slides were counterstained with hematoxylin and high-power field images of the cancellous bone were examined by optical microscopy. All BrdU-positive (dark-brown) nuclei in the secondary spongiosa were counted. The percentage of BrdU-positive nuclei versus total nuclei was calculated as mitotic index.

For whole cell lysates, calvaria cells were grown on 100 cm² Petri dishes and were extracted in a buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 20 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS and protease inhibitors. For whole bone extracts, one femur was homogenized in TRIzol (Gibco) and incubated for 5 minutes at room temperature. After precipitation of DNA with ethanol, proteins were extracted from the phenol-ethanol supernatant by adding 1.5 ml of isopropanol per 1 ml of TRIzol reagent. The protein pellet was washed three times in 0.3 M guanidine hydrochloride in 95% ethanol, and dissolved in 1% SDS. Proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes (Invitrogen, Carlsbad, CA). Western blots were processed at room temperature as described (Castro et al., 2003; Lecanda et al., 2000) using an anti-Cx43 antibody (Sigma, St Louis, MO) at 1:8000 dilution or anti-GAPDH antibody, and visualized by enhanced chemiluminescence (ECL) detection.

As already reported (Stains et al., 2003; Stains and Civitelli, 2005b), confluent calvaria cells or bone tissue were extracted using TRIzol (Gibco) and total RNA (1 μg) was reverse transcribed using Superscript II reverse transcriptase and oligo(dT)₁₅ primers. Real-time PCR analysis was performed using the SYBR green PCR method according to manufacturer's instruction (PE Biosystems, Foster City, CA). The primers used in this study have all been reported (Mbalaviele et al., 2005; Stains et al., 2003). GAPDH (PE Biosystems) was used as internal control. The cycle number at which the fluorescence exceeded the threshold of detection (CT) for GAPDH was subtracted from that of the target gene product for each well (ΔCT). Transcription levels relative to Cx43^+/fl controls was defined as (2^–ΔΔCT), where ΔΔCT equals the genotype ΔCT minus the ΔCT of Cx43^+/fl cells. All real-time PCR experiments were performed at least three times.

Group means were analyzed by ANOVA after establishing normal distribution of data and homogeneity of variances. Where significant overall differences were observed by one-way ANOVA, the Tukey Kramer test or other post-hoc analyses were applied for multiple group comparisons. For repeated measures (PTH studies), a two-way ANOVA was applied, keeping treatment or genotype and time as independent variables. Analyses were performed using SPSS v.12.0.0 (SPSS, Chicago, IL), with the level of significance for comparison set at P<0.05. All data are expressed as the mean ± s.d. (unless otherwise indicated).

Supported by NIH grant R01 AR041255 (R.C.) and by funds from Barnes-Jewish Hospital Foundation (R.C.). Work in Bonn was supported by grants of the German Research Association (SFB 400/E3 and Wi270/22-3.4) and the Funds of the Chemical Industry (to K.W.). D.J.C. was partially supported by a grant from the Sung-Am Cultural Foundation. C.H.M.C. was a post-doctoral Fellow of CAPES Foundation, Ministry of Education, Brazil. M.T. received a stipend of the Graduierten Kolleg: Pathogenese von Krankheiten des Nervensystems. Part of this work has been presented at the 2005 International Gap Junction Conference, Whistler, BC, Canada, 13-18 August 2005, and at the 27th annual meeting of the American Society for Bone and Mineral Research, Nashville, TN, 23-27 September 2005.