Bone morphogenetic protein type IA receptor signaling regulates postnatal osteoblast function and bone remodeling.

Bone morphogenetic proteins (BMPs) function during various aspects of embryonic development including skeletogenesis. However, their biological functions after birth are less understood. To investigate the role of BMPs during bone remodeling, we generated a postnatal osteoblast-specific disruption of Bmpr1a that encodes the type IA receptor for BMPs in mice. Mutant mice were smaller than controls up to 6 months after birth. Irregular calcification and low bone mass were observed, but there were normal numbers of osteoblasts. The ability of the mutant osteoblasts to form mineralized nodules in culture was severely reduced. Interestingly, bone mass was increased in aged mutant mice due to reduced bone resorption evidenced by reduced bone turnover. The mutant mice lost more bone after ovariectomy likely resulting from decreased osteoblast function which could not overcome ovariectomy-induced bone resorption. In organ culture of bones from aged mice, ablation of the Bmpr1a gene by adenoviral Cre recombinase abolished the stimulatory effects of BMP4 on the expression of lysosomal enzymes essential for osteoclastic bone resorption. These results demonstrate essential and age-dependent roles for BMP signaling mediated by BMPRIA (a type IA receptor for BMP) in osteoblasts for bone remodeling.

Bone formation is a well characterized process; however, little is known about the molecular mechanisms that regulate bone remodeling, the physiological process through which bone mass is maintained constant. Remodeling consists of two distinct phases: initial bone resorption by the osteoclasts, followed by de novo bone formation by the osteoblasts (1). Differentiated osteoblasts are the only cells responsible for bone formation. Bone formation is thought to be regulated by hormones and by locally acting growth factors (2). Bone morphogenetic proteins (BMPs) 1 are secreted molecules and members of transforming growth factor-␤ superfamily (3,4). They were discovered by their ectopic bone formation activity when implanted locally in soft tissues (5). Over the past decade, the phenotypes of mice with mutations in genes coding for this group of proteins and their receptors uncovered the essential roles for BMPs in wide variety of developmental processes, including skeletal development and patterning (6 -9). However, despite its powerful ability to induce ectopic osteogenesis, the essential role of BMPs in bone formation and bone metabolism in the adult skeleton has not been established (10) because of embryonic lethality resulting from mutations of genes encoding the most potent BMPs for bone formation, BMP2 and BMP4, and their receptors (11)(12)(13).
We previously generated a null allele for Bmpr1a that encodes a type IA receptor for BMP (BMPRIA or ALK3). Mice homozygous for this null allele died by embryonic day 8.0 (E8.0) without mesoderm formation (13). Bmpr1a is expressed in most tissues throughout development and after birth (13,14). Expression of a dominant-negative form of BMPRIA in a cultured cell line or chick limb buds suggests that signaling through this receptor regulates apoptosis and adipocyte differentiation (15,16). Overexpression of a constitutive-active form of BMPRIA in chicken limb buds suggests that signaling through this receptor also can regulate chondrocyte differentiation (17). However, the essential role of BMPRIA in bone formation and bone metabolism in the adult skeleton is not known. To investigate the role of BMPRIA signaling at later stages of development, we generated a conditional null allele of Bmpr1a using the Cre/loxP system (18,19). Because Bmpr1a is expressed in osteoblasts (20), we designed a postnatal, differentiated osteoblast-specific disruption of Bmpr1a to elucidate the requirement of BMP signaling for bone formation (21).

MATERIALS AND METHODS
Mice-The generation of Bmpr1a conditional null mice was reported elsewhere (19). Briefly, one loxP site was placed in intron 1, and two others were placed in intron 2 flanking a PGK-neo cassette (fn allele). After germ line transmission, mice heterozygous for the fn allele were mated with CMV-Cre transgenic mice to remove the neo cassette by Cre-dependent recombination in vivo (fx allele). Both the fn allele and the fx allele behaved as wild type, indicating that the presence of the PGK-neo cassette or the loxP sites in the Bmpr1a locus did not reduce Bmpr1a activity (19). For the studies reported here we used both the fn and fx alleles. Mice heterozygous for Bmpr1a null allele (ϩ/Ϫ) were bred to Og2-Cre transgenic mice (21) to generate Bmpr1a(ϩ/Ϫ),Og2-Cre(ϩ) mice. These mice were bred to Bmpr1a fn/fn or fx/fx mice. Theoretically, 25% of the progeny should be transheterozygous for fn (or fx) and null alleles and also carry the Og2-Cre transgene to execute osteoblastspecific disruption of Bmpr1a. All experimental procedures were performed according to ethical guidelines approved by local authorities.
Osteoblast Culture-10 6 bone marrow cells collected from femurs were plated in mineralization medium (␣-minimal essential medium, 0.1 mg/ml ascorbic acid, 5 mM ␤-glycerophosphate) in 35-mm culture dishes and cultured in 5% CO 2 at 37°C (22). After 4 weeks, the cultures were scored for osteoblastic colonies by staining for alkaline phosphatase to assess osteoblast differentiation and von Kossa stain for mineral deposition (22).
Bone Histomorphometry-Animals were injected with calcein (5 mg/kg of body weight) twice at 4-day intervals and sacrificed 2 days later. Histomorphometric analyses were carried out according to standard protocols (23).
Measurement of Bone Mineral Density-Whole right femora were dissected and completely cleaned of all tissue for peripheral duel photon x-ray absorptiometry (pDXA) determination of bone mineral content and bone mineral density (BMD) of the whole femur, central femur (central 50%), and distal 25% of the femur. These data were obtained using a Norland TM Saber pDXA scanner set at a resolution of 0.1 ϫ 0.1 mm and a scan speed of 2 mm/s. Mouse Calvarial and Tibia Organ Culture-The calvaria and tibia from aged mice (12 months old on average) that were homozygous for fx were excised and cut in half along the sagittal suture (calvaria) or perpendicular to the long axis (tibia). Each portion was placed in a 24-well tissue culture dish containing 0.5 ml of BGJb medium (Invitrogen) supplemented with 5% fetal bovine serum and 150 g/ml vitamin C (Sigma) (43). Cre-dependent recombination was induced by infection with recombinant adenovirus that expresses Cre recombinase (a gift from Dr. Robert Sobol), with 10 8 plaque-forming units/ml in culture for 7 days. Bone samples were subsequently cultured for 11 days with or without recombinant BMP4 (100 ng/ml, R&D Systems). Tibia were flushed to remove bone marrow before RNA extraction.
RNA Isolation-Frozen calvaria and tibia from organ cultures were pooled (2 samples/pool, n ϭ 3/group) together, crushed, and homogenized in TRIzol reagent (Invitrogen) using a Polytron PT 10 -35. Total RNA was extracted from bone samples using a Purescript RNA Isolation Kit (Gentra Systems) followed by DNase treatment on RNeasy Micro-columns (Qiagen) according to manufacturer's instructions.

Disruption of the Type IA Receptor for BMPs in a Differentiated
Osteoblast-specific manner-Osteocalcin2 (Og2) is specifically expressed in differentiated osteoblasts only after birth (24,25). To direct Cre recombinase in postnatal differentiated osteoblasts, we used transgenic mice that carry a 1.3-kb mouse Og2 upstream region ligated to the Cre recombinase gene (21). Osteoblast-specific Cre activity was examined by crosses with CAT-lacZ mice that report Cre activity by ␤-galactosidase expression (26). In Og2-Cre, CAT-lacZ double heterozygotes, all bones stained positively for ␤-galactosidase activity (Fig. 1A), resembling the ␤-galactosidase pattern of Og2-lacZ transgenic mice (27). ␤-Galactosidase activity was observed only in the osteoblasts on trabecular bone near the growth plates and on cortical bone surfaces (Fig. 1, B, C, E, and F). No ␤-galactosidase activity was detected in the soft tissues, bone marrow, or chondrocytes, indicating that expression of Cre in Og2-Cre mice was highly specific for osteoblasts.
Pups transheterozygous for the conditional null and null alleles of Bmpr1a, and hemizygous for the Og2-Cre transgene, were recovered after birth. As shown in Fig. 2A, they were smaller than their control littermates at 2 months of age. Smaller body size was recognizable as early as 3 weeks of age and became more prominent at 6 or 12 weeks (Fig. 2B). Their body weight remained below normal for up to 6 months. Because the proportion of the mature osteoblasts was too small for detection of the Cre-dependent recombination by Southern blot, floxed and recombined alleles were differentiated by PCR. All of the mice that carried the Og2-Cre transgene showed at least one recombined allele that had deleted exon 2 (Fig. 2C).
Reduced Osteoblast Function in the Mutant Mice-Contact x-ray photography of 3-month-old mice showed no overt changes in bone shape compared with controls (Fig. 3A). However, irregular calcification was found, most prominently in femurs (Fig. 3,  B and C, arrow). Histological analysis showed decreased bone trabeculae (Fig. 3, D and E) with no growth plate abnormalities.
Osteopontin and osteocalcin were expressed normally in the mutant bones (data not shown). To investigate the cause of the altered bone formation, we performed a morphometric analysis of undecalcified histological sections (23), which revealed a decrease in bone volume in the mutant mice that was about half of that in their control littermates (Fig. 3F). Bone formation rate was reduced in mutants in comparison with controls (Fig. 3G). In addition, the numbers of osteoblasts and osteoclasts in tibia and spine did not differ significantly between mutants and their controls (data not shown). . The number of animals is shown above each bar. p Ͻ 0.005. C, PCR detection of Cre-dependent recombination. DNA extracted from bone was amplified with allele-specific PCR primers shown. All mice analyzed here were heterozygous for the fn allele. All of the mice that carried the Og2-Cre transgene showed the deletion of exon2. D10 and D8 are ES cell clones heterozygous for ⌬exon2-neo and ⌬exon2 alleles, respectively, and D10 is mosaic for fn and ⌬exon2-neo (19).
Consistent with the morphometric analysis, in in vitro culture of bone marrow cells from mutant and control mice, the osteoblastic colonies derived from Bmpr1a fx/Ϫ,Og2-Cre(ϩ) mice were smaller than those derived from control cells and stained poorly for mineral deposition (Fig. 4, A and B). Fewer alkaline phosphatase-positive colonies and von Kossa-positive colonies were observed in the culture derived from bone marrow cells in mutant mice (Fig. 4C). Taken together, the in vivo and in vitro analyses suggest that osteoblast-specific Bmpr1a disruption leads to less osteoblast function and loss of bone formation, resulting in decreased bone remodeling and decreased body size in mutant mice.
Reduced Osteoclast Activity in Aged Mutant Mice-Although younger mutant mice (up to 6 months) weighed less than controls (Fig. 2), the difference became smaller as they got older (data not shown). Intriguingly, BMD in 10-month-old mutant mice was significantly higher than that of controls (Fig. 5A,  WT/Sham and KO/Sham, B, and D). Because there were no significant differences in BMD observed at 3 months of age (data not shown), these findings suggested that loss of BMP signaling in osteoblasts might lead to down-regulation of osteoclast function as the mice aged. Histomorphometric analysis of femurs of 10-month-old mice confirmed the higher bone volume (BV/TV (bone volume per total volume)) in mutant mice compared with controls (Fig. 5F, WT/Sham and KO/Sham). Dynamic histomorphometric analysis with double calcein labeling documented that bone turnover (BFR/BS (bone formation rate per total bone surface)) is decreased in mutant mice (Fig. 5G, WT/Sham and KO/Sham), indicating that increased bone mass in mutant mice is due to decreased bone resorption. These results suggest that, in aged animals, loss of BMP signaling via BMPRIA in osteoblasts may lead to suppression of osteoclast function and bone resorption. To gain further insights into the role of BMPRIA and BMP signaling in adult bone metabolism, we evaluated the effects of ovariectomy (OVX) that activates bone turnover and osteoclastic bone resorption, leading to bone loss. Interestingly, the loss of BMD in the OVX group, compared with the sham operated group, was more significant in mutant mice (Fig. 5A). Histological observations agreed with the BMD findings (Fig. 5, B-E). More trabecular bone is observed in the distal femur of the mutant mice compared with the controls, and in both mutant and control mice, trabecular bone was severely reduced by OVX. Quantitative histomorphometric analysis confirmed these findings by documenting the significant reduction of bone volume (BV/TV) in OVX groups to Sham-operated groups with more significant effects of OVX observed in mutant mice (Fig. 5F). More significant bone loss induced by OVX in mutant mice likely results from the decreased osteoblast function, observed in vivo and in vitro analyses in young animals, that became evident in high bone turnover induced by OVX. Taken together, the increased bone mass and the greater degree of bone loss induced by OVX in the mutant mice suggest that the activity of osteoclasts as well as osteoblasts is reduced in the aged osteoblast-specific Bmpr1a-deficient mice, leading to the complex skeletal phenotype.
To further investigate the above mentioned hypothesis, we used a bone organ culture system to evaluate the impact of ablation of Bmpr1a on the genes associated with osteoblasts and osteoclasts. Calvarias from aged mice, that were homozygous for floxed Bmpr1a, were treated with adenoviral Cre recombinase and effects of BMP4 on the expression of marker genes were assessed by real-time PCR. Cre-dependent deletion of Bmpr1a was confirmed by significant reduction of the levels of wild-type transcript (Fig. 6A) and induction of the Bmpr1a transcript lacking exon 2 (Fig. 6B). In the control culture, expression levels of TRAP, Mmp9, and Ctsk, lysosomal en-zymes secreted from osteoclasts, in calvaria were all increased with BMP4 treatment. These results are consistent with the role of the BMPRIA signaling pathway in osteoclastic bone resorption postulated by the reduced bone resorption observed in mutant mice. In contrast, in calvaria following treatment with Cre recombinase that inactivates Bmpr1a, BMP4 treatment showed no effects on the levels of expression of TRAP, Mmp9, and Ctsk (Fig. 6, C-E). Basal levels of these enzymes were also reduced by inactivation of Bmpr1a. Interestingly, the expression of calcitonin receptor, a marker of osteoclast, was not induced by BMP4 nor regulated by treatment with Cre recombinase (data not shown), suggesting the specific effects of the BMP-BMPRIA signaling pathway in expression of lysosomal enzymes essential for osteoclastic bone resorption. Expression of RANKL that is expressed in osteoblasts to support osteoclast function was not changed by Cre treatment (data not shown). Similar results were observed by organ culture of tibia from the same animals (data not shown). Analysis of marker genes for osteoblasts, such as type I collagen or osteocalcin, did not show significant changes under these experimental conditions (data not shown). DISCUSSION Our results show that signaling through the BMP type IA receptor in differentiated osteoblasts plays an important role in bone formation when mice are relatively young (Ͼ6 months of age). The mutant mice showed reduced body weight, lower bone volume, and reduced bone formation rate. This signal is apparently not be essential for osteoblast differentiation or proliferation but is important for the production of bone matrix by differentiated osteoblasts. Intriguingly, the differentiated osteoblast-specific disruption of the BMP type IA receptor leads to an opposite phenotype when mice become older (at 10 months of age). Bone volume in the mutant mice was higher than that of control littermates even though bone formation rate was still lower in the mutants. The mutant mice that were ovariectomized showed the same BMD and bone volume as the controls that were ovariectomized, suggesting that BMP signaling through the type IA receptor in the differentiated osteoblasts may be important for regulating osteoclast activity. This hypothesis is supported by the induction of lysosomal enzymes essential for osteoclastic bone resorption by BMP4 in calvarial organ culture that is abolished following inactivation of Bmpr1a.
It has been shown that alteration of BMP signaling during embryogenesis affects skeletalgenesis (6 -9, 28 -31). Gene disruption of the type IB receptor for BMPs (Bmpr1b) shows defects in the proximal and middle phalanges (32,33), which resembles that in Gdf5 mutant mice, brachypodism (7). A transgenic mouse line that expresses a dominant-negative form of Bmpr1b in an osteoblast-specific manner shows more severe abnormalities in the skeletal system than Bmpr1b-deficient mice such as reduced BMD, bone volume, and bone formation rate up to 3 months after birth (34). A transgenic mouse line that expresses Noggin, one of the BMP antagonists, in a differentiated osteoblast-specific manner shows reduced bone formation up to 6 months after birth (35). Recently, linkage of osteoporosis to chromosome 20p12 where BMP2 resides was demonstrated in a Danish population (36). These results suggest that BMP signaling in the osteoblasts plays important roles in bone formation during younger stages of life, consistent to the results presented here.
It has been reported that osteoblasts play essential roles in supporting the generation and activity of osteoclasts that are critical for bone resorption (37). The stimulatory effects of BMP4 in lysosomal enzymes expressed in osteoclasts in calvarial organ culture is in agreement with previous reports docu- FIG. 5. Impact of ovariectomy on bone in Bmpr1a-deficient mice. OVX was performed on 8-month-old mice, and its impact was examined 8 weeks after the surgery. A, mutant mice in the sham group showed higher BMD than normal littermates in the sham group (**, p ϭ 0.0186). BMD was significantly reduced in OVX groups compared with respective sham groups (*, p Ͻ 0.001). B-E, histological analysis of 10-month-old femurs. B and C, Bmpr1a fx/Ϫ,Og2-Cre(Ϫ); D and E, Bmpr1a fx/Ϫ,Og2-Cre(ϩ). Ovariectomized mice (C, E) showed reduced bone (toluidine blue staining) compared with respective sham operation mice (B, D). F, BV/TV was significantly reduced in OVX groups compared with the respective sham groups (*, p Ͻ 0.001). Mutant mice in the sham group showed higher BV/TV than normal littermate in the sham group (**, p ϭ 0.0186). G, bone BFR/BS in the sham group was significantly reduced in Bmpr1a-deficient mice compared with normal littermate (***, p ϭ 0.036). Measurement of BFR/BS in the OVX groups was not relevant because of too little amounts of bone trabecular. menting the stimulatory effects of BMPs on osteoclast function in vitro (38 -42). BMPs may act directly on osteoclasts (41) or alternatively via osteoblasts (40) because both types of cells express receptors for BMPs. Although we cannot exclude the possible role for Cre-dependent inactivation of Bmpr1a in osteoclasts in the loss of BMP induction of lysozomal enzymes in organ culture, these results along with in vivo results suggest that BMP signaling in osteoblasts controls bone resorption by supporting osteoclast function in aged mice. Our results suggest that BMP functions in differentiated osteoblasts are altered in an age-dependent manner, initially for bone formation at younger ages, then more predominantly for supporting osteoclast function as animals age.
FIG. 6. Effect of ablation of Bmpr1a on expression of lysosomal enzymes essential for osteoclastic bone resorption. Calvarias were removed from aged mice that were homozygous for floxed Bmpr1a and then cultured with or without recombinant Adenovirus that expressed Cre for 7 days. Subsequently, calvarias were cultured with or without 100 ng/ml BMP4, then the RNA was extracted. Gene expression was measured by real time PCR. A, reduction of a wildtype transcript from Bmpr1a (*, p Ͻ 0.0015). B, emergence of a transcript from Cre recombined allele of Bmpr1a (*, p Ͻ 0.0001). C-E, levels of lysosomal enzymes expressed in osteoclasts. Up-regulation of TRAP, Mmp9, and Ctsk with BMP4 (*, p Ͻ 0.02) were abolished by treatment with Cre recombinase (**, p Ͻ 0.003). The levels of expression of each gene are presented as fold differences relative to the levels expression of each gene in the control group.