Loss of Gsα in the Postnatal Skeleton Leads to Low Bone Mass and a Blunted Response to Anabolic Parathyroid Hormone Therapy*

Parathyroid hormone (PTH) is an important regulator of osteoblast function and is the only anabolic therapy currently approved for treatment of osteoporosis. The PTH receptor (PTH1R) is a G protein-coupled receptor that signals via multiple G proteins including Gsα. Mice expressing a constitutively active mutant PTH1R exhibited a dramatic increase in trabecular bone that was dependent upon expression of Gsα in the osteoblast lineage. Postnatal removal of Gsα in the osteoblast lineage (P-GsαOsxKO mice) yielded markedly reduced trabecular and cortical bone mass. Treatment with anabolic PTH(1–34) (80 μg/kg/day) for 4 weeks failed to increase trabecular bone volume or cortical thickness in male and female P-GsαOsxKO mice. Surprisingly, in both male and female mice, PTH administration significantly increased osteoblast numbers and bone formation rate in both control and P-GsαOsxKO mice. In mice that express a mutated PTH1R that activates adenylyl cyclase and protein kinase A (PKA) via Gsα but not phospholipase C via Gq/11 (D/D mice), PTH significantly enhanced bone formation, indicating that phospholipase C activation is not required for increased bone turnover in response to PTH. Therefore, although the anabolic effect of intermittent PTH treatment on trabecular bone volume is blunted by deletion of Gsα in osteoblasts, PTH can stimulate osteoblast differentiation and bone formation. Together these findings suggest that alternative signaling pathways beyond Gsα and Gq/11 act downstream of PTH on osteoblast differentiation.


Parathyroid hormone (PTH) is an important regulator of osteoblast function and is the only anabolic therapy currently approved for treatment of osteoporosis. The PTH receptor (PTH1R) is a G protein-coupled receptor that signals via multiple G proteins including G s ␣. Mice expressing a constitutively active mutant PTH1R exhibited a dramatic increase in trabecular bone that was dependent upon expression of G s ␣ in the osteoblast lineage. Postnatal removal of G s ␣ in the osteoblast lineage (P-G s ␣ OsxKO mice) yielded markedly reduced trabecular and cortical bone mass. Treatment with anabolic PTH(1-34)
(80 g/kg/day) for 4 weeks failed to increase trabecular bone volume or cortical thickness in male and female P-G s ␣ OsxKO mice. Surprisingly, in both male and female mice, PTH administration significantly increased osteoblast numbers and bone formation rate in both control and P-G s ␣ OsxKO mice. In mice that express a mutated PTH1R that activates adenylyl cyclase and protein kinase A (PKA) via G s ␣ but not phospholipase C via G q/11 (D/D mice), PTH significantly enhanced bone formation, indicating that phospholipase C activation is not required for increased bone turnover in response to PTH. Therefore, although the anabolic effect of intermittent PTH treatment on trabecular bone volume is blunted by deletion of G s ␣ in osteoblasts, PTH can stimulate osteoblast differentiation and bone formation. Together these findings suggest that alternative signaling pathways beyond G s ␣ and G q/11 act downstream of PTH on osteoblast differentiation.
Osteoporosis is one of the most common degenerative diseases of aging with an estimated 50% of postmenopausal women and 25% of older men at risk of sustaining a fragility fracture and economic costs exceeding $19 billion annually in the United States (1). The low bone mass underlying osteoporosis results from an imbalance between bone formation and bone resorption that leads to bone loss that becomes more pronounced with age (2). The most commonly prescribed therapies for osteoporosis target the inhibition of bone resorption but as such are not curative.
Anabolic therapy to increase osteoblast numbers and function and thereby stimulate bone formation is appealing as a potential route to curing osteoporosis. Administered once daily, recombinant parathyroid hormone (PTH(1-34); 2 teriparatide) reduces fracture risk and significantly increases bone mineral density and bone mass (3). Teriparatide is currently the only Food and Drug Administration-approved anabolic therapy for osteoporosis. However, the mechanisms by which intermittent PTH stimulates bone anabolism have not been fully clarified.
Intermittent PTH treatment enhances bone formation and increases osteoblast numbers by a variety of mechanisms including stimulation of osteoblast proliferation and differentiation, inhibition of osteoblast apoptosis, and activation of quiescent lining cells (4 -11). PTH also suppresses the expression of sclerostin, an inhibitor of canonical Wnt signaling produced by osteocytes and encoded by the Sost gene (12,13). Canonical Wnt signaling plays a crucial role in regulating osteoblast differentiation and bone formation (14 -16), and patients lacking sclerostin have high bone mass (17,18). Highlighting the clinical relevance of this pathway, neutralizing antibodies targeted against sclerostin are now in clinical trials for the treatment of osteoporosis (17)(18)(19) and are being tested in preclinical models * This work was supported by National Institutes of Health Grants AR054741 of other conditions of bone fragility such as osteogenesis imperfecta (20,21).
The PTH receptor PTH1R is a G protein-coupled receptor. Stimulation of PTH1R by PTH activates a variety of G proteins including the stimulatory G protein G s . G s activates adenylyl cyclase, thereby increasing cyclic AMP (cAMP) levels and activating the protein kinase A (PKA) gene transcription pathway (22). Several lines of evidence suggest that G s is a major mediator of the anabolic actions of PTH. PTH induces the expression of several osteoblast-specific target genes including osteocalcin (Bglap) (23), Mmp13 (24), and Tnfsf11 (25) in a PKA-dependent manner. Targeting of a constitutively active mutant form of PTH1R (caPTH1R), identified in patients with Jansen metaphyseal chondrodysplasia, to osteoblasts in mice results in profound increases in trabecular bone mass (26). In vitro, this mutant version of PTH1R predominantly activates G s -dependent signaling pathways (27). Furthermore, constitutive activation of G s -dependent signaling by an engineered G scoupled receptor also significantly increases trabecular bone mass (28). In addition to G s , PTH1R couples to several other G proteins, activating phospholipase C␤ via G q /G 11 to stimulate protein kinase C and G 12 /G 13 to stimulate phospholipase D (29,30). We have previously reported that phospholipase C (PLC) signaling via PTH1R is essential for stromal cell response to PTH in growing mice (31). Activation of PTH1R also recruits ␤-arrestins 1 and 2, leading to receptor internalization and activation of ERK1/2 (32).
In the osteoblast lineage, the ␣ subunit of the heterotrimeric G s protein G s ␣ is required for normal skeletal development during embryogenesis (33)(34)(35). Deletion of G s ␣ early in the osteoblast lineage using Cre recombinase under control of the osterix (Osx) promoter (16) markedly impairs bone formation such that more immature woven bone is observed (35). There are at least two distinct roles for G s ␣ in the osteoblast lineage during skeletal development. G s ␣ is required for commitment of mesenchymal progenitors to the osteoblast lineage rather than the adipocyte lineage, likely mediated at least in part by alterations in Wnt signaling (33,35). In contrast, in cells committed to the osteoblast lineage, G s ␣ restrains osteoblast differentiation (35).
Constitutive activation of G s -dependent signaling in osteoblasts throughout embryogenesis results in a dramatic increase in trabecular bone volume (28). However, when activation of G s is delayed until birth, there is a much milder increase in bone (36), and if delayed until 4 weeks of age, there is no discernible skeletal phenotype (28). Therefore G s ␣-dependent signaling may have different functions during embryogenesis compared with postnatal skeletal homeostasis. Mice with G s ␣ deleted throughout embryonic development in Osx-expressing osteoprogenitors die before weaning (35). By administering doxycycline to suppress the expression of Cre recombinase, we delayed ablation of G s ␣ in the osteoblast lineage until birth (postnatal or P-G s ␣ OsxKO mice), allowing us to examine the role of G s ␣ in osteoblasts in the postnatal skeleton. Here we report that G s ␣ is required for the high trabecular bone mass observed with constitutive activation of PTH1R. P-G s ␣ OsxKO mice have severe osteoporosis, and when treated with anabolic PTH, there was no increase in trabecular bone volume or cortical thickness. However, osteoblast numbers and bone formation rate still increased along the trabecular surface of P-G s ␣ OsxKO mice. We also examined the effects of anabolic PTH on PLC-defective mutant (D/D) mice that express only a mutated PTH1R that activates adenylyl cyclase normally but cannot activate PLC and found that PTH significantly increased almost all parameters of bone formation in tibiae of both wild type (WT) and D/D mice. Together these data demonstrate the involvement of other downstream mediators beyond activation of PKA and protein kinase C (PKC) in the anabolic action of PTH.

Experimental Procedures
Mice-Generation of G s ␣ OsxKO mice has been described previously (34,35). Postnatal deletion of G s ␣ was achieved by administering 10 g/ml doxycycline in drinking water from conception until birth. Rosa26 (R26R, stock 3309) (37) and mTmG reporter mice (38) were obtained from The Jackson Laboratory (Bar Harbor, ME). Because G s ␣ OsxKO mice are of mixed genetic background, littermate controls (G s ␣(fl/fl) except where otherwise specified) were used for all experiments described. Generation of D/D mice has been described previously (39). Genotyping was performed on tail genomic DNA using protocols published previously (34,35,39 Histology-Mouse limbs were fixed in 10% buffered formalin, paraffin-embedded, and sectioned. Immunohistochemical analysis was performed on deparaffinized sections using biotinylated mouse anti-SOST antibody (R&D Systems). For X-Gal staining of ␤-galactosidase activity, limbs were fixed in 0.2% glutaraldehyde and 1.5% formaldehyde followed by overnight staining at 37°C in X-Gal solution containing 1 mg/ml X-Gal (Takeda, Osaka, Japan), 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl 2 , 0.01% sodium deoxycholate, and 0.02% Nonidet P-40 (10).
Skeletal Preparations-Skeletons were fixed in 95% ethanol and then stained overnight in 0.015% Alcian blue in acetic acid/ ethanol to highlight cartilage. Soft tissues were cleared in 1% KOH and then stained overnight in 0.01% alizarin red to detect mineralized bone (40).
Bone Mineral Density-Mice were anesthetized with tribromoethanol (600 mg/kg). Bone mineral density of the whole skeleton (excluding calvariae) or individual bones was measured by dual energy x-ray absorptiometry on a Lunar Piximus (GE Medical Systems, Milwaukee, WI).
Micro-computed Tomography (CT) Analysis-Assessment of bone morphology and microarchitecture was performed using a desktop high resolution CT (CT40, Scanco Medical, Brüttisellen, Switzerland) as described previously (41). Briefly, the distal femoral metaphysis (P-G s ␣ OsxKO mice) or L5 vertebrae (D/D mice) was scanned for trabecular bone assessment, and mid-diaphyses were scanned for cortical bone assessment using x-ray energy of 70 KeV, an integration time of 200 ms, and a 12-m isotropic voxel size. Trabecular bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), and trabecular number (Tb.N) were calculated using a threshold computed for control samples. For cortical bone analysis, cortical thickness (Co.Th) was assessed using a predefined fixed threshold.
Histomorphometry-Double calcein labeling was performed by injecting mice with 20 mg/kg calcein 3 and 10 days before sacrifice. Bones were fixed in 10% buffered formalin and embedded in methylmethacrylate as described previously (42). 5-m sections were stained with toluidine blue or xylenol orange. Histomorphometric analysis of undecalcified trabecular bone was performed using the Osteomeasure system (OsteoMetrics, Decatur, GA) (42).
Quantitative Real Time PCR-Total RNA was prepared from flushed long bones and primary osteoblasts using the RNeasy kit (Qiagen), and cDNA was synthesized with the SuperScript III First Strand synthesis system for real time PCR (Invitrogen). Quantitative real time PCR was performed using primers for Gnas (43), Runx2 (44), Osx (44), collagen I␣1 (ColI␣1) (40), osteopontin (Opn) (44), osteocalcin (Bglap) (45), and Sost (46) according to previously published protocols with mRNA levels normalized relative to ␤-actin expression. Total RNA samples subjected to cDNA synthesis reactions in the absence of reverse transcriptase were included as negative controls.
Serum Biochemistries-Blood was harvested by cardiac puncture at the time of euthanasia. Fasting serum levels of procollagen type I N-terminal propeptide (P1NP) and type I collagen C-terminal telopeptide (CTX; RatLaps) were measured by enzyme immunoassay (Immunodiagnostics Systems, Scottsdale, AZ) according to the manufacturer's protocol.
Statistics-Statistical analyses were performed using a twotailed Student's t test, and group differences were analyzed by two-way analysis of variance followed by post hoc Tukey's test. All values are expressed as means Ϯ S.E.

Results
To determine whether G s ␣ is required for the expansion of trabecular bone mediated by caPTH1R in vivo, we crossed G s ␣ OsxKO mice with caPTH1R transgenic mice. At 1 week of age, histological analysis of proximal tibiae revealed that trabecular bone was nearly absent in G s ␣ OsxKO mice but was greater than controls in caPTH1R mice (Fig. 1A). Skeletal preparations at 1 week of age demonstrated that healing calluses marked sites of rib fractures in G s ␣ OsxKO but not caPTH1R mice (Fig.  1B). Double mutant mice (caPTH1R;G s ␣ OsxKO mice) still exhibited rib fractures, and histological analysis revealed a paucity of trabecular bone that more closely resembled G s ␣ OsxKO mice, although there were a few remaining fragments of trabecular bone (Fig. 1, A and B).
Constitutive expression of Osx-driven Cre recombinase throughout embryonic development resulted in severe osteoporosis at birth with the majority of G s ␣ OsxKO mice dying by 3 weeks of age (34,35), precluding analysis of G s ␣ conditional knock-out mice in adulthood. However, the osterix promoterdriven GFP::Cre fusion protein is regulated by a Tet-Off tetracycline transactivator (16). Doxycycline interferes with binding of tetracycline transactivator to its target and abolishes transcription of Osx1-GFP::Cre recombinase (48,49). When doxycycline was administered in drinking water at 2 mg/ml to pregnant females from mating until 7 days after birth, caPTH1R; G s ␣ OsxKO mice resembled caPTH1R mice with significantly greater trabecular bone mass, providing evidence that doxycycline prevented the deletion of G s ␣ (Fig. 1C). However, if doxycycline was withdrawn at day 7, then by day 21, caPTH1R; G s ␣ OsxKO mice resembled G s ␣ OsxKO mice by their marked reduction in trabecular bone mass (Fig. 1D). These results demonstrate that the caPTH1R mutant receptor requires expression of G s ␣ in osteoblasts to increase trabecular bone mass and suggest that G s ␣ expression in G s ␣ OsxKO mice can be regulated by doxycycline administration.
Hong et al. (50) reported that high doses of doxycycline can lead to prolonged suppression of Cre expression even after withdrawal of doxycycline, whereas lower doses of doxycycline allow improved titration of Cre expression. We therefore tested doxycycline at a 200-fold lower dose (10 g/ml) in drinking water starting from conception. To determine whether Cre is expressed after withdrawal of doxycycline, we crossed G s ␣ OsxKO mice to R26 reporter mice (37) in which a floxed stop cassette precedes the gene encoding ␤-galactosidase. Expression of Cre recombinase results in expression of ␤-galactosidase, which is detected by staining with X-Gal. Chen et al. (51) reported that the extraskeletal expression of Osx1-GFP::Cre is limited to the olfactory bulb, gastric, and intestinal epithelial cells; therefore we focused on X-Gal staining in bone. Following withdrawal of doxycycline at birth, we found abundant X-Gal staining of osteocytes and osteoblasts in Osx1-GFP::Cre ϩ mice, indicating Cre expression ( Fig. 2A). In controls maintained on doxycycline until 6 weeks of age, X-Gal staining in the bone was significantly but not completely reduced (Fig. 2B). Despite incomplete suppression of Cre recombinase by doxycycline, the survival of P-G s ␣ OsxKO mice, which in the absence of doxycycline do not survive to weaning, improved to 100% at 3 weeks of age (Fig. 2C).
To assess the effects of postnatal G s ␣ deletion, we administered 10 g/ml doxycycline throughout embryogenesis until birth (P0) and examined the skeleton at 6 weeks of age (Fig. 2D). Control mice did not express the Cre transgene; however, expression of the Cre transgene in wild-type mice did not significantly affect skeletal formation or bone mineral density, although Cre-expressing mice were slightly smaller (data not shown), consistent with previous reports of a transient decrease in body weight of Osx-Cre transgenic mice that resolves by 12 weeks of age (52). Histological analysis of P-G s ␣ OsxKO mice at 6 weeks of age demonstrated low trabecular bone mass (Fig. 2E), and dual energy x-ray absorptiometry revealed reduced bone mineral density in the humerus (Fig. 2F). Serum calcium and phosphate levels did not differ between P-G s ␣ OsxKO and con-trol littermates (data not shown). We have previously reported that Gnas mRNA levels are reduced by almost 90% in sorted Osx1-GFP::Cre ϩ cells of G s ␣ OsxKO mice in which G s ␣ is deleted throughout embryogenesis (34). However, the yield of Osx1-GFP::Cre ϩ cells from mineralized adult bone is extremely low. We therefore prepared RNA from bone samples flushed of marrow and found that Gnas mRNA levels were reduced by 50% (Fig. 2G). Because G s ␣ is ubiquitously expressed, the remaining Gnas mRNA detected was likely largely expressed in non-osteoblast lineage cells.
As another approach to determine the effectiveness of G s ␣ deletion, we examined expression of sclerostin. Sclerostin is an inhibitor of canonical Wnt signaling that is suppressed by PTH.
Constitutive ablation of G s ␣ in the osteoblast lineage up-regulates Sost, the gene encoding sclerostin (35). We therefore examined Sost mRNA and sclerostin protein levels in P-G s ␣ OsxKO mice. Sost mRNA increased Ͼ15-fold in bones of female P-G s ␣ OsxKO mice (Fig. 2H). Immunohistochemical analysis for sclerostin protein also revealed a marked increase in sclerostin expression in P-G s ␣ OsxKO osteocytes (Fig. 2I), suggesting deletion of G s ␣ with high efficiency from the osteoblast lineage.
To examine the role of G s ␣ in the anabolic response to PTH, control and P-G s ␣ OsxKO mice were injected with once daily PTH(1-34) at 80 g/kg/day beginning at 8 weeks of age for 4 weeks (Fig. 3A). In vehicle (PBS)-treated mice, histological analysis of the proximal humerus (Fig. 3B) and CT analysis of distal femur trabecular bone (Fig. 3C) revealed low trabecular bone mass in P-G s ␣ OsxKO male and female mice compared with controls. In females, PTH treatment increased BV/TV and Tb.N in WT compared with vehicle-treated mice, but this was not observed in P-G s ␣ OsxKO mice where PTH treatment instead decreased Tb.N (Fig. 3, D and E). In males, PTH did not increase BV/TV or Tb.N in either WT or P-G s ␣ OsxKO mice (Fig.  3, D and E). Trabecular thickness was significantly greater in PTH-treated WT and P-G s ␣ OsxKO mice of both genders (Fig.  3F), and trabecular separation was increased in PTH-treated male and female P-G s ␣ OsxKO mice as compared with WT controls (Fig. 3G). Cortical thickness was reduced in both male and female P-G s ␣ OsxKO mice compared with WT controls; PTH treatment had no effect on this parameter (Fig. 4, A and B). Serum levels of P1NP, a marker of bone formation, did not differ in female mice, whereas in males P1NP was lower in PBStreated P-G s ␣ OsxKO mice than in WT, but PTH did not have any significant effect (Fig. 4C). In summary, in both male and female P-G s ␣ OsxKO mice, PTH failed to increase trabecular bone volume or cortical thickness but did enhance trabecular thickness. In addition, PTH increased trabecular spacing in P-G s ␣ OsxKO but not control mice.
Histomorphometric analysis of distal femora from male mice confirmed that BV/TV and trabecular number were lower in PBS-treated P-G s ␣ OsxKO compared with WT mice and that PTH treatment had no effect on either of these parameters (Table 1). Again, PTH increased trabecular thickness in WT and P-G s ␣ OsxKO mice; no significant change in trabecular separation was observed. PTH treatment increased both osteoblast surface and osteoblast numbers in control mice and, surprisingly, in P-G s ␣ OsxKO male mice. Bone formation rate as assessed by double calcein labeling was also greater in both control and P-G s ␣ OsxKO male mice treated with PTH. The failure of intermittent PTH to increase bone volume in P-G s ␣ OsxKO mice despite increased osteoblast numbers and bone formation rate could potentially be explained by high osteoclast activity. However, PTH treatment did not significantly alter osteoclast surface in either WT or KO mice (Table 1). There were no differences in serum levels of CTX, a marker of bone resorption, in either female or male mice (Fig. 4D).
To determine whether the effects of PTH on osteoblast numbers and bone formation rate in P-G s ␣ OsxKO mice were mediated by direct action of PTH on osteoblasts, we isolated calvarial osteoblasts from mice with two floxed alleles of G s ␣. Infection of G s ␣(fl/fl) calvarial osteoblasts in vitro with adenovirus encoding Cre recombinase leads to efficient (Ͼ90%) deletion of G s ␣ (35). We treated G s ␣(fl/fl) calvarial osteoblasts infected with Cre recombinase or ␤-gal control and then cul-

. Postnatal deletion of G s ␣ leads to osteopenia and increased sclerostin expression.
A, X-Gal staining for ␤-galactosidase activity in 6-week-old G s ␣ OsxKO mice treated with doxycycline (Doxy) from plug until delivery (100ϫ magnification). B, X-Gal staining of 6-week-old P-G s ␣ OsxKO mice treated with doxycycline until 6 weeks of age (100ϫ). C, survival frequency in control (WT no doxy; white bar; n ϭ 18) and G s ␣ OsxKO (KO no doxy; hatched bar; n ϭ 8) mice without doxycycline treatment or in control (WT ϩ doxy; gray bar; n ϭ 7) or P-G s ␣ OsxKO mice (KO ϩ doxy; black bar; n ϭ 10) on 10 g/ml doxycycline from plug until birth. D, pregnant dams were administered 10 g/ml doxycycline in drinking water from plug until delivery. The resulting mice were analyzed at 6 weeks of age. E, H&E-stained sections of proximal tibia from female 6-week-old mice treated with 10 g/ml doxycycline until birth. F, bone mineral density (BMD) is significantly reduced in the humerus of P-G s ␣ OsxKO mice. gm, gram. G, Gnas mRNA levels in femurs of WT and P-G s ␣ OsxKO mice at 6 weeks. H, Sost mRNA levels in calvariae of 6-week-old mice. I, sclerostin immunostaining in cortical bone of WT and P-G s ␣ OsxKO tibiae at 6 weeks of age. n ϭ 5, WT; n ϭ 6, P-G s ␣ OsxKO . *, p Ͻ 0.05. Error bars represent S.E. JANUARY (53). In control osteoblasts, PTH increased mineralized nodule formation after 14 days, and as reported previously, deletion of G s ␣ led to markedly enhanced mineralization (35). Treatment with PTH further increased mineralized nodule formation by G s ␣deficient osteoblasts (Fig. 5, A and B). Mineralization can be increased by increased cell density. However, cell numbers were increased only transiently at day 3 in PTH-treated control osteoblasts (Fig. 5C). Examination of expression of markers of osteogenic differentiation confirmed that PTH treatment of G s ␣-deficient osteoblasts significantly increased expression of a range of osteoblast marker genes including Osx, ColI␣1, Opn, and Bglap as compared with PTH-treated control osteoblasts (Fig. 5D). This provides additional evidence that PTH can stimulate osteogenic differentiation in G s ␣-deficient osteoblasts.

Low Bone Mass in Mice Lacking G s ␣ in the Postnatal Skeleton
In addition to the G s ␣-linked adenylyl cyclase-PKA signaling pathway, PTH is known to activate the G q /G 11 -linked PLC-PKC signaling pathway. PLC-defective mutant (D/D) mice express only a mutated PTH1R that activates adenylyl cyclase normally but cannot activate PLC (39). To determine whether PLC signaling pathway via the PTH1R plays a role in bone anabolic action of PTH, we administered intermittent PTH to both PTH1R WT and D/D mice. Although a lowered bone mass was previously noted in growing tibiae of D/D mice at 6 and 10 weeks of age, at 16 weeks neither the tibia (determined by histomorphometry) nor the femur (examined by CT) of D/D mice exhibited a significantly lower bone mass at baseline (Fig.  6 and Table 2). Intermittent PTH increased trabecular bone volume in the primary spongiosa in both WT and D/D mice; this area was excluded from standard measurement of histomorphometry (Fig. 6, A and B). Histomorphometry revealed no changes in bone architecture parameters (BV/TV, Tb.Th, and Tb.N) in the areas measured (Table 2). However, parameters of bone formation including mineral apposition rate, bone formation rate, bone volume and tissue volume, osteoid surface, osteoblast surface, and osteoblast number were increased by intermittent PTH in both WT and D/D mice (Table 2). In L5 vertebrae, as measured by CT, PTH increased trabecular bone volume only in WT, but not D/D, mice (Fig. 6C). Trabecular number was not significantly altered by PTH treatment in either WT or D/D mice (Fig. 6D). PTH increased trabecular   (Fig. 6E) but had no effect on trabecular separation (Fig. 6F).
We measured serum P1NP as a bone formation marker and serum CTX as a bone resorption marker (Fig. 6, G and H). Both P1NP and CTX were significantly increased by PTH in WT and D/D mice; the level of P1NP in PTH-treated D/D mice was lower than that of PTH-treated WT mice (Fig. 6G).

Discussion
We have found that G s ␣ signaling in the osteoblast lineage is important not only for embryonic skeletal development but also for adult skeletal homeostasis. Furthermore, we have demonstrated that G s ␣ is a downstream mediator of anabolic PTH signaling in vivo by two approaches. First, we crossed mice lacking G s ␣ in osteoblasts to mice expressing caPTH1R in osteoblasts and found that the postnatal absence of G s ␣ markedly attenuates the ability of caPTH1R to increase trabecular bone. Second, we administered intermittent PTH to adult mice lacking G s ␣ in osteoblasts and found that in the absence of G s ␣ the anabolic actions of intermittent PTH on trabecular bone are blunted in vivo.
Mice with postnatal ablation of G s ␣ in osteoblast progenitors (P-G s ␣ OsxKO mice) have severe osteoporosis affecting both trabecular and cortical bone. Osteoclast surface and CTX levels are not elevated, indicating that their low bone mass is not due to enhanced bone resorption. These findings are consistent with our previous studies demonstrating failure of bone formation in mice with deletion of G s ␣ in osteoprogenitors throughout embryonic development (35).
PTH increased osteoblast numbers and bone formation rate in P-G s ␣ OsxKO mice, a paradoxical finding because PTH failed to increase overall bone mass in those mice. A proportionally greater increase in resorption in response to PTH by P-G s ␣ OsxKO mice would be one explanation; however, there was no significant increase in osteoclast surface, CTX levels, or trabecular separation in P-G s ␣ OsxKO mice treated with intermittent PTH, and trabecular thickness was increased. An alternative explanation is that by 8 weeks of age when anabolic PTH was started P-G s ␣ OsxKO mice have insufficient trabecular bone surface on which new bone can be formed; studies are now underway to begin PTH treatment earlier before bone loss has progressed to such an advanced degree in P-G s ␣ OsxKO mice. Finally, in cultured osteoblasts, PTH can maximally activate PKA at concentrations significantly lower than required to increase total cAMP levels, suggesting that only a fraction of the total cAMP that can be generated by PTH stimulation is required for PKA activation (54). It is also important to note that G s ␣-mediated PTH1R signaling in other tissues such as the kidney is presumably preserved. Careful analysis of postnatal extraskeletal Osx-Cre expression by Chen et al. (51) revealed that although Osx-Cre targets cells in the olfactory bulb, gastric, and intestinal epithelia there is no expression in the kidney, liver, or other organs. Therefore indirect effects of PTH, for example on systemic mineral metabolism, may also influence skeletal homeostasis. However, we found that PTH does have direct effects in osteoblasts that are altered by the removal of G s ␣.
The mechanisms by which intermittent PTH stimulates osteoblast differentiation and function are incompletely understood. Because PTH treatment has a positive effect on osteoblast surface in G s ␣-deficient mice, it is likely that PTH also stimulates signaling pathways downstream of PTH1R independent of G s ␣. These may include pathways activated by other PTH1R-coupled G proteins including G q /G 11 and G 12 /G 13 (29,30). We found that although intermittent PTH in older mice did not significantly increase trabecular bone volume in long bones of either WT or D/D mice within the time frame of this experiment it did significantly increase almost all parameters of bone formation in both WT and D/D mice. Our data suggest that PLC signaling through the PTH1R is therefore not required for PTH-stimulated bone turnover in mature mice. However, our findings do not rule out the possibility that in the absence of G s ␣ activation of G q/11 may mediate some of the anabolic effects of PTH.
In addition to PTH1R-coupled G proteins, ␤-arrestins have also been implicated in the anabolic response to PTH, and the anabolic effect of PTH is blunted in the absence of ␤-arrestin2 (55). Initially identified based on their roles in G protein-coupled receptor desensitization, growing evidence suggests that effects mediated by stimulation of endogenous PTH secretion were not ruled out. Studies are now underway to address the role of non-G s ␣-mediated signaling pathways downstream of PTH1R.
Author Contributions-J. Y. W. conceived and coordinated the study and wrote the paper. J. Y. W., G. N., and E. S. designed, performed, and analyzed the experiments shown in Fig. 1. J. Y. W. and P. A. designed, performed, and analyzed the experiments shown in Fig. 2. P. S. and R. C. designed, performed, and analyzed the experiments shown in Figs. 3 and 4. I. J. P. and N. A. S. designed, performed, and analyzed the experiments shown in Table 1. H. S. designed, performed, and analyzed the experiments shown in Fig. 5. J. G. and H. M. K. designed, performed, and analyzed the experiments shown in Table 2. M. C. and L. S. W. generated the G s ␣-floxed mice and contributed to experimental design and data interpretation. All authors reviewed the results and approved the final version of the manuscript.