Genetic determination of the cellular basis of the sympathetic regulation of bone mass accrual

To identify the cell type in which the sympathetic nervous system acts to regulate bone mass accrual, we relied on cell-specific gene inactivation in the mouse. To determine whether the sympathetic tone signals in osteoblasts to regulate bone mass accrual, we generated a mutant mouse strain expressing a floxed allele of Adrβ2, the adrenergic receptor whose deletion in all cells affects bone mass accrual (Elefteriou et al., 2005). These mutant mice were crossed with α1(I) Collagen (2.3-kb Col1a1 promoter)–Cre transgenic mice to delete the gene in osteoblasts only (Adrβ2_osb^−/− mice; Dacquin et al., 2002). Before analyzing the mutant mice, we verified that we had achieved a high percentage of recombination at the Adrβ2 locus in osteoblasts but in no other cell types (Fig. 1, A and B).

As it is the case in mice lacking Adrβ2 in all cells (Elefteriou et al., 2005), Adrβ2_osb^−/− mice had a normal bone mass until 12 wk of age (Fig. 1 L). However, at 24 wk of age, Adrβ2_osb^−/− mice presented a distinct high bone mass phenotype affecting both vertebrates and long bones (Fig. 1, C and D), whose severity was similar to the one observed in mice lacking Adrβ2 in all cells. Histomorphometric analysis of vertebrae showed that this high bone mass phenotype was caused by a concomitant increase in the number of osteoblasts and bone formation rate (BFR) and a decrease in the osteoclast number (Fig. 1 C). As expected, the decrease in osteoclast number did result in a decrease in bone resorption activity, as determined by serum levels of carboxy-terminal cross-linking telopeptide of type I collagen (CTx), which was decreased in Adrβ2^−/− mice (Fig. 1 H). These results established that the osteoblast is a major target of the sympathetic tone regulation of bone mass accrual.

Our previously published work (Fu et al., 2005) has established that the sympathetic tone in osteoblasts regulates phosphorylation of CREB on serine 133 and of ATF4 on serine 254. Moreover, these two events should be altered by leptin signaling in the brain, whose function is to decrease bone formation by increasing sympathetic tone (Takeda et al., 2002). In full agreement with this model, leptin intracerebroventricular (ICV) infusion decreased CREB phosphorylation in control but not in Adrβ2_osb^−/− bones (Fig. 1 E). These observations indicated that the sympathetic signaling in osteoblasts is downstream of leptin signaling in the brain.

At the molecular level, we observed that expression of CyclinD1, CyclinD2, CyclinE1, cMyc, and Per1, all genes implicated in the sympathetic regulation of bone mass accrual (Fu et al., 2005), was significantly increased in Adrβ2_osb^−/− bones (Fig. 1 F). Similarly, the expression of Cry1 and Cry2 was increased (Fig. S2 A). In contrast, the ratio of Rankl to Opg expression was decreased (Fig. 1 G). Based on these results, we anticipated that leptin ICV infusion, a procedure which decreases bone mass by decreasing bone formation parameters and increasing bone resorption parameters in WT mice, should fail to do so in Adrβ2_osb^−/− mice. That is exactly what was observed in 12-wk-old Adrβ2_osb^−/− mice (Fig. 1, I–M). In summary, the analysis of mice lacking Adrβ2 only in osteoblasts established that the sympathetic tone lowers bone mass accrual by signaling in osteoblasts through the Adrβ2. The clarity of the effect (and lack) of leptin ICV infusion in WT (and Adrβ2_osb^−/− mice) led us to use this assay as a tool to verify in the rest of the study the cell-specific nature of other aspects of the leptin/sympathetic regulation of bone mass accrual.

Previous experiments have suggested that the sympathetic tone acts, in osteoblasts, through CREB to regulate bone formation and that leptin signaling in the brain affects this process (Fu et al., 2005). To determine whether it is indeed the case, we generated, using the aforementioned strategy, mice lacking CREB in osteoblasts only (Creb_osb^−/− mice) and, as we did in all experiments, verified that we had achieved an efficient and cell-specific gene deletion before using these mutant mice (Fig. 2, A and B).

CREB has recently emerged as a more important than anticipated positive transcriptional regulator of bone formation (Yadav et al., 2008, 2009; Datta and Abou-Samra, 2009; Oury et al., 2010). In particular, it has been shown to be a transcriptional effector of the anabolic action of parathyroid hormone (Datta and Abou-Samra, 2009), while its function is inhibited by gut-derived serotonin, which is a negative regulator of bone formation (Yadav et al., 2008). Given the existence of these two important regulations, it is no surprise that at 12 wk of age, Creb_osb^−/− mice display low bone mass because of an isolated decrease in bone formation parameters (Fig. 2 C). Nevertheless, to determine whether CREB is a transcriptional mediator of the sympathetic tone-dependent regulation of bone formation, we studied the expression of genes regulated by the sympathetic tone in Creb_osb^−/− bones. As was the case in Adrβ2_osb^−/− bones, expression of genes implicated in the sympathetic regulation of bone formation, CyclinD1, CyclinD2, CyclinE1, Per1, and cMyc, was affected in Creb_osb^−/− bones (Fig. 2 E). Consistent with the decreased expression of cell cycle regulator genes, osteoblast proliferation, as measured by BrdU incorporation, was markedly reduced in Creb_osb^−/− bones (Fig. S2 B). In contrast, the ratio of Rankl to Opg expression and bone resorption measured by serum CTx levels were not affected (Fig. 2, F and G). Thus, CREB, through its expression in osteoblasts, regulates osteoblast proliferation.

Next, we asked whether leptin signaling ultimately requires Creb expression in osteoblasts to inhibit bone mass accrual. As shown in Fig. 2 (H, I, and L), long-term leptin ICV infusion failed to decrease bone formation parameters, although it could increase osteoclast number in Creb_osb^−/− mice (Fig. 2, J and L). As a result, leptin ICV infusion decreased, albeit mildly, bone volume in Creb_osb^−/− mice (Fig. 2, K and L).

The fact that leptin ICV infusion could decrease bone volume in Creb_osb^−/− mice can be explained only if one hypothesizes that the sympathetic tone can increase bone resorption independently from CREB, possibly through ATF4. If it is the case, one would expect that by deleting Creb in osteoblasts, one would blunt the inhibition of bone formation induced by the sympathetic tone, without affecting bone resorption.

If the aforementioned explanation of the bone phenotype of Creb_osb^−/− mice and of the influence that leptin exerts on it is correct, it could be verified through the use of mice lacking Atf4 in osteoblasts only (Fig. 3). As is the case in mice lacking Atf4 in all cells, Atf4_osb^−/− mice have a low bone mass phenotype caused by a decrease in bone formation parameters (Fig. 3 C; Yang et al., 2004; Elefteriou et al., 2006). The decrease in bone formation parameters could be explained, at least in part, by an alteration in osteoblast differentiation because expression of two ATF4 target genes, Bglap1 and Col1a1, was significantly reduced in Atf4_osb^−/− bone (Fig. S2 C). On the contrary, unlike what is the case in Adrβ2_osb^−/− and Creb_osb^−/− mice, expression of cell cycle regulator genes CyclinD1, CyclinD2, CyclinE1, Per1, and cMyc was not affected (Fig. 3 E). These results are consistent with the hypothesis that the expression of these genes in osteoblasts is under the control of CREB, not ATF4 (Fig. 2; Fu et al., 2005). In contrast, the ratio of Rankl to Opg expression was lower in Atf4_osb^−/− mice (Fig. 3 F). To formally verify our hypothesis, we then infused leptin ICV in control and Atf4_osb^−/− mice. Leptin ICV infusion increased the osteoclast number in control but not in Atf4_osb^−/− mice (Fig. 3, J and L). However, this procedure decreased bone formation in both control and mutant mice. This was expected because CREB signaling is intact in Atf4_osb^−/− bones. That leptin ICV infusion could decrease bone formation parameters (Fig. 3, H, I, and L) explains also why it could decrease bone mass on Atf4_osb^−/− mice (Fig. 3, K and L).

Collectively, the analysis of cell-specific gene inactivation mouse models for two distinct transcription factors presented in Figs. 2 and 3 suggested a model whereby the sympathetic tone uses CREB and ATF4 in osteoblasts to influence bone formation and bone resorption, respectively (Fig. S1).

We have shown earlier that downstream of the sympathetic regulation of bone formation, cMyc must be a target of CREB (Fu et al., 2005). To determine whether cMyc carries out this function through its expression in osteoblasts, we used the same strategy used for the study of Adrβ2 and Creb and Atf4 functions in osteoblasts (Figs. 1–3) and again verified before using them that we had achieved an efficient and osteoblast-specific deletion in cMyc_osb^−/− mice (Fig. 4, A and B).

We should emphasize here that unlike other mutant mice, cMyc_osb^−/− mice were maintained on a mixed genetic background including C57/B6 and 129Sv/EV. This difference in genetic background explains the rather high bone volume observed in control cMyc-flox/flox and mutant cMyc_osb^−/− littermates compared with other mutant mice that were all on a pure C57/B6 background (Fig. 4 C). Notwithstanding this feature, and when compared with cMyc-flox/flox mice, cMyc_osb^−/− mice demonstrated a low bone mass phenotype at 12 wk of age that was secondary to a decrease in bone formation parameters (Fig. 4 C). The osteoclast number and activity were not significantly affected in cMyc_osb^−/− mice (Fig. 4, C and G). Consistent with the notion that cMyc may be a target gene of the sympathetic tone and CREB in osteoblasts, expression of the cyclin genes CyclinD1, CyclinD2, and CyclinE1 was also significantly decreased in cMyc_osb^−/− bones (Fig. 4 E). In contrast, the expression ratio of Rankl to Opg was not affected (Fig. 4 F). The expression of Opg was not affected in cMyc_osb^−/− bones (Fig. S3, A–D).

To determine formally that leptin inhibition of bone mass accrual requires the expression of cMyc in osteoblasts, we performed long-term leptin ICV infusion in control and cMyc_osb^−/− mice. As shown in Fig. 4 (H, I, and L), leptin ICV infusion decreased bone formation parameters in control but not in cMyc_osb^−/− mice. In contrast, and in full agreement with the ATF4 mediation of this aspect of the regulation of bone mass accrual, leptin ICV infusion increased bone resorption parameters equally well in control and in cMyc_osb^−/− mice (Fig. 4, J and L). As a result, leptin decreased bone mass in both control and in cMyc_osb^−/− mice, although the effect was more pronounced in control mice (Fig. 4, K and L).

In the last part of these experiments, we sought to establish that the leptin regulation of bone mass accrual requires the various transcription factors studied in this paper to interact in the osteoblasts. This is a particularly important aspect of our work because CREB, for instance, can be activated by a protein kinase A–dependent but also by other, unrelated pathways.

Consistent with the notion that cMyc is a downstream target gene of CREB, cMyc_osb^−/− mice showed a similar phenotype as Creb_osb^−/− mice as well as a similar response to long-term leptin ICV infusion. To verify in vivo that these genes are in the same pathway downstream of leptin signaling in osteoblasts, we generated Creb_osb^+/−;cMyc_osb^+/− compound heterozygous mice. That Creb_osb^+/−;cMyc_osb^+/− mice had decreased bone formation parameters, while bone resorption parameters were not affected (Fig. 5, A–E), supports the notion that both Creb and cMyc regulate osteoblast proliferation in osteoblasts but not osteoclast differentiation.

Next, we asked whether the high bone mass phenotype of the Adrβ2_osb^−/− mice could be corrected by removing one copy of cMyc in osteoblasts. As shown in Fig. 5 (F–J) and Fig. S4, removing one allele of cMyc in osteoblasts normalized bone formation parameters in Adrβ2_osb^−/− mice. In contrast, this manipulation did not affect the bone resorption parameters in Adrβ2_osb^−/− mice, further indicating that the sympathetic tone uses different transcriptional mediators to regulate bone formation and bone resorption. Thus, the analysis of these compound mutant mouse strains verified the existence of an interaction between Adrβ2, Creb, and cMyc in the osteoblast to allow the sympathetic nervous system to prevent bone mass accrual.

In this study, we used genetic and pharmacological means to formally identify the osteoblast as the major cell type in which the sympathetic tone, under the control of leptin signaling in the brain, acts to inhibit bone mass accrual. Results of this investigation are represented schematically in Fig. S1. We cannot exclude, based on our results, the possibility that osteocytes may also be a target of the sympathetic tone. We should emphasize here that our study is restricted to the means whereby leptin signaling in the brain, as represented by leptin ICV infusion, regulates bone mass. The reason for this focus on this aspect of leptin regulation of bone mass accrual is that a neuron-specific inactivation of the leptin receptor in the mouse recapitulates the bone phenotype of mice lacking leptin signaling, whereas its osteoblast-specific inactivation fails to do so (Shi et al., 2008).

Once it was realized that there was a central control of bone mass (Ducy et al., 2000), the next question was to decipher how the brain signals to the bone. It became rapidly apparent that the sympathetic nervous system is a critical mediator of this homeostatic function exerted by the brain. This signaling inhibits bone mass accrual by, at the same time, preventing bone formation and favoring bone resorption. The evidence supporting this view was genetic and relied on the study of a classical gene inactivation mouse model. Indeed, mice lacking the β2 adrenergic receptor in all cells had a high bone mass secondary to an increase in osteoblast proliferation and bone formation and a subsequent decrease in bone resorption (Elefteriou et al., 2005). It was also supported by a clinical aspect such as the low bone mass of patients affected with reflex sympathetic dystrophy (Schwartzman, 2000).

Several transcription factors have been implicated in mediating this function of the sympathetic tone, and it has long been assumed, although never tested, that this regulation takes place in the osteoblasts (Elefteriou et al., 2005, 2006; Fu et al., 2005). However, because many of these molecular players such as Adrβ2, CREB, and cMyc are by no means osteoblast-specific molecules, there was an urgent need to demonstrate that indeed it is in osteoblasts that this pathway operates. What the experiments presented in this study establish is that it is by acting in the osteoblasts that the sympathetic tone regulates bone mass accrual. It is also through its expression in osteoblasts that CREB regulates osteoblast proliferation downstream of the sympathetic tone. Likewise, it is through its expression in osteoblasts that ATF4 favors Rankl expression and osteoclast differentiation downstream of the sympathetic tone. It is noteworthy that these two transcription factors have distinct functions. Indeed, CREB regulates osteoblast proliferation but does not affect osteoclast differentiation, whereas ATF4 affects osteoclast differentiation but is not involved in osteoblast proliferation.

In considering the overall sympathetic mode of regulating bone mass accrual, it is quite remarkable that the sympathetic tone recruits two different transcription factors in the same cell to regulate each arm of bone remodeling. Remarkably, this specificity is conserved when one looks at the main regulator of this function of the sympathetic nervous system, leptin. Indeed, long-term ICV infusion of this hormone in mutant mice lacking, in osteoblasts only, either Creb or Atf4 verified that leptin and the sympathetic tone recruit each of these two transcription factors for different purposes.

Looking more globally at these regulations of bone mass by leptin, it underscores the importance of the various roles played by CREB at several steps in this pathway. Indeed, CREB is mediating serotonin regulation of the sympathetic tone in the ventromedial hypothalamic nuclei, a regulation inhibited by leptin signaling in the brain. Thus, CREB acts upstream of the sympathetic tone (Oury et al., 2010). The present study shows that it also acts downstream of it and in osteoblasts.

This work, by showing that the bulk of the sympathetic regulation of bone mass accrual occurs in osteoblasts, raises the question of its pharmacological relevance. The work of Bonnet et al. (2008) showing that a low dose of β blockers acting through Adrβ2 could prevent gonadectomy-induced bone loss without affecting other function regulated by the sympathetic nervous system is in full agreement with the fact that mice lacking only one allele of Adrβ2 display high bone mass. As such, it is certainly important if this pathway could be exploited further for the purpose of treating low bone mass disease.

Adrβ2_osb^−/−, Creb_osb^−/−, Atf4_osb^−/−, cMyc-flox/flox, and α1(I) Collagen (2.3-kb Col1a1 promoter)–Cre mice were previously described (de Alboran et al., 2001; Dacquin et al., 2002; Hinoi et al., 2008; Couillard and Trudel, 2009; Yoshizawa et al., 2009). cMyc-flox/flox mice were provided by F.W. Alt (Harvard Medical School, Boston, MA) and M. Trudel (Université de Montréal, Montréal, Québec, Canada), and Creb-flox/flox mice were provided by G. Schütz (German Cancer Research Center, Heidelberg, Germany). Adrβ2-flox/flox, Creb-flox/flox, Atf4-flox/flox, and cMyc-flox/flox littermates were used for the control mice for Adrβ2_osb^−/−, Creb_osb^−/−, Atf4_osb^−/−, and cMyc_osb^−/− mice, respectively. For Creb_osb^+/−;cMyc_osb^+/−mice analysis, Creb-flox/+;cMyc-flox/+ mice were used as controls. Creb_osb^+/− mice, Creb-flox/+, cMyc_osb^+/− mice, and cMyc-flox/+ mice were independently analyzed. For Adrβ2_osb^−/−;cMyc_osb^+/− mice analysis, Adrβ2-flox/flox;cMyc-flox/+ mice were used as control. For Adrβ2_osb^−/− mice, Adrβ2-flox/flox mice were used as control. cMyc_osb^+/− mice and cMyc-flox/+ mice were independently analyzed. Genotyping was performed by PCR analysis of genomic DNA. All mutant mouse strains were on C57BL6/J except cMyc_osb^−/− mice, which are on C57BL6/J and 129Sv/EV. All procedures involving animals were approved by the Columbia University the Institutional Animal Care and Use Committee and conform to the relevant regulatory standards.

Bone marrow–derived osteoblasts were prepared as previously described (Takahashi et al., 1988; Ferron et al., 2010). In brief, mice were sacrificed at the age of 2–3 wk. Bone marrow from tibia was flushed out with PBS, and cells were plated at 0.4 million cells/cm² in α-MEM containing 15% FBS. 4 d after plating cells, medium was replaced by α-MEM containing 10% FBS, 5 mM β-glycerophosphate, 100 µg/ml ascorbic acid, and 10 nM dexamethasone and cultured for 7 d. Medium was changed every other day.

Bone histomorphometry was performed as previously described (Chappard et al., 1987; Parfitt et al., 1987). In brief, lumbar vertebrae were dissected, fixed for 24 h in 10% formalin, dehydrated in graded ethanol series, and embedded in methyl methacrylate resin according to standard protocols. Von Kossa/Von Gieson staining was performed using 7-µm sections for bone volume over tissue volume measurement. BFR was analyzed by the calcein double labeling method. Calcein (Sigma-Aldrich) was dissolved in calcein buffer (0.15 M NaCl and 2% NaHCO₃) and injected twice at 0.125 mg/g body weight on days 1 and 4, and then mice were killed on day 6. 5-µm sections were cleared in xylene and used for BFR measurements. For analysis of the parameters of osteoblasts and osteoclasts, 5-µm sections were stained with toluidine blue and tartrate-resistant acid phosphatase, respectively. Histomorphometric analyses were performed using the OsteoMeasure analysis system (OsteoMetrics).

Trabecular bone architecture of distal tibia was assessed by using a µCT system (VivaCT 40; SCANCO Medical AG). Tibia bone specimen was stabilized with gauze in a 2-ml centrifuge tube filled with 70% ethanol and fastened in the specimen holder of the µCT scanner. 100 µCT slices, corresponding to a 1.05-mm region distal from the growth plate, were acquired at an isotropic spatial resolution of 10.5 µm. A global thresholding technique was applied to binarize grayscale µCT images in which the minimum between the bone and bone marrow peaks in the voxel gray value histogram was chosen as the threshold value. The trabecular bone compartment was segmented by a semiautomatic contouring method and subjected to a model-independent morphological analysis (Hildebrand et al., 1999) by the standard software provided by the manufacturer of the µCT scanner.

Osteoblast proliferation was assessed in vivo in newborn pups at day 5.5 by injecting thymidine analogue BrdU intraperitoneally. Animals were killed 4 h after BrdU injection. Femora were removed and fixed in 10% formalin at 4°C for 24 h. Bones were demineralized in 25% EDTA at 37°C. Demineralized bones were dehydrated and embedded in paraffin. BrdU labeling was measured in 4-µm sections in the femoral head region using a commercially available kit (Invitrogen). Sections were counterstained with hematoxylin. Proliferating osteoblasts were analyzed using a 40× objective on a microscope (DMLB; Leica) outfitted with a charge-coupled device camera (DXC-S500; Sony) and an OsteoMeasure analysis system.

Levels of serum CTx were measured using a commercial kit (Immunodiagnostic systems). Western blotting was performed as previously described (Yang et al., 2004; Yoshizawa et al., 2009). In brief, proteins from bone marrow–derived osteoblasts or bones were obtained by homogenizing cells or bones in lysis buffer (25 mM Tris-HCl, pH 7.4, 10 mM Na₃VO₄, 100 mM NaF, 10 mM Na₄P₂O₇, 10 mM EGTA, 10 mM EDTA, and 1% NP-40 containing proteinase inhibitor cocktail [Roche]), followed by centrifugation at 14 krpm for 20 min at 4°C. The supernatant was used as extracted proteins. The concentration was measured by Bradford method, and 30 µg of protein was subjected to Western blotting. Anti–phospho-CREB and anti–total CREB antibody were purchased from Cell Signaling Technology. Anti-Adrβ2, anti-cMyc, and anti-ATF4 were purchased from Santa Cruz Biotechnology, Inc.

Animals were anesthetized with avertin and placed on a stereotaxic instrument (Stoelting). The calvaria was exposed, and a 0.7-mm hole was drilled upon bregma. A 28-gauge cannula (Brain infusion kit II; Alza) was implanted into the third ventricle. The cannula was secured to the skull with cyanoacrylate and attached with Tygon tubing to an osmotic pump (Alza) placed in the dorsal subcutaneous space of the animal. The rate of delivery of leptin (Sigma-Aldrich) was 0.25 µl/h (8 ng/h) for 28 d. In all strains examined, the leptin ICV infusion started at 8 wk of age, and mice were sacrificed at 12 wk of age.

Bone marrow–flushed long bones were homogenized in TRIZOL reagent (Invitrogen), and total RNA was extracted according to the manufacturer’s instruction. Real-time PCR was performed on DNaseI-treated total RNA converted to cDNA using Taq SYBR green Supermix with ROX (Bio-Rad Laboratories) on an MX3000 instrument (Agilent Technologies); β-actin amplification was used as an internal reference for each sample except Rankl, for which Opg was used for an internal reference.

Results are given as means ± SEM. Statistical analyses were performed using unpaired, two-tailed Student’s t tests.

Fig. S1 shows a schematic representation of the peripheral mediation of the leptin-dependent sympathetic regulation of bone mass accrual. Fig. S2 shows gene expression and BrdU incorporation analysis. Fig. S3 shows Opg expression in Adrβ2_osb^−/−, Creb_osb^−/−, Atf4_osb^−/−, and cMyc_osb^−/− mice. Fig. S4 shows bone histomorphometric analysis of cMyc_osb^+/− mice.

We thank Dr. Frederick W. Alt and Dr. Marie Trudel for cMyc-flox/flox mice and Dr. Günther Schütz for Creb-flox/flox mice.

This work was supported by grant DK58883 from the National Institutes of Health (to G. Karsenty).

The authors have no competing financial interests.