mTORC1 Plays an Important Role in Skeletal Development by Controlling Preosteoblast Differentiation

ABSTRACT

The mammalian target of rapamycin complex 1 (mTORC1) is activated by extracellular factors that control bone accrual. However, the direct role of this complex in osteoblast biology remains to be determined. To investigate this question, we disrupted mTORC1 function in preosteoblasts by targeted deletion of Raptor (Rptor) in Osterix-expressing cells. Deletion of Rptor resulted in reduced limb length that was associated with smaller epiphyseal growth plates in the postnatal skeleton. Rptor deletion caused a marked reduction in pre- and postnatal bone accrual, which was evident in skeletal elements derived from both intramembranous and endochondrial ossification. The decrease in bone accrual, as well as the associated increase in skeletal fragility, was due to a reduction in osteoblast function. In vitro, osteoblasts derived from knockout mice display a reduced osteogenic potential, and an assessment of bone-developmental markers in Rptor knockout osteoblasts revealed a transcriptional profile consistent with an immature osteoblast phenotype suggesting that osteoblast differentiation was stalled early in osteogenesis. Metabolic labeling and an assessment of cell size of Rptor knockout osteoblasts revealed a significant decrease in protein synthesis, a major driver of cell growth. These findings demonstrate that mTORC1 plays an important role in skeletal development by regulating mRNA translation during preosteoblast differentiation.

INTRODUCTION

Intramembranous ossification (most of the craniofacial skeleton) and endochondrial ossification (the vertebrate appendicular and axial skeleton) are the two cellular processes required to form the skeleton (reviewed in reference 1). In both processes, osteoblasts (OBs), the bone-forming cells of the skeleton, play an essential role. OBs differentiate from mesenchymal progenitors through distinct developmental stages marked by expression of key transcription factors, including runt-related transcription factor 2 (Runx2 product), osterix (Sp7 and Osx product), and β-catenin (Ctnnb1 product). Runx2 is the master regulator of OB lineage commitment and directly regulates Osx expression as evidenced by a lack of osterix transcript in Runx2 knockout animals (2). The homozygous deletion of Runx2 or Osx results in the formation of a complete cartilaginous skeleton, but no bone formation takes place in either the endochondral or the membranous skeleton due to the absence of OBs (2, 3).

Runx2, osterix, and β-catenin are regulated by extracellular factors to control embryonic and postnatal skeletal development. Bone morphogenetic proteins (BMPs), including BMP2 and BMP4, are potent bone anabolic factors that promote osteogenesis via the activation of SMAD signaling molecules, in turn regulating Runx2 and osterix expression and activity (4–6). Wnt ligands bind to a receptor complex comprising Frizzled and LRP5/6, inducing a signaling cascade that directs OB differentiation at the pre-OB stage, enhancing Runx2 expression (7, 8). Insulin and insulin-like growth factor (IGF-1) indirectly regulate Runx2 to promote postnatal bone accrual. Insulin downregulates Twist1 and Twist2 expression (9), negative regulators of Runx2 that bind and inhibit Runx2 activity (10). IGF-1 and insulin also inhibit Foxo1-mediated suppression of Runx2 activity by promoting Foxo1 phosphorylation and nuclear exclusion (11).

Bone anabolic signals require activation of cellular processes that govern growth. The phosphatidylinositol 3-kinase (PI3K)/Akt/mammalian target of rapamycin (mTOR) pathway is activated by a variety of extracellular factors and promotes cell cycle progression, cell survival and growth signals. In mice, the OB-specific deletion of Pten, a negative regulator of PI3K, results in a progressive increase in bone mineral density throughout life (12), whereas Akt1^−/−/Akt2^−/− mice have shorter bones (13) and delayed bone ossification (14, 15), demonstrating an important role for this pathway in skeletal development. Activation of this pathway in OBs through the deletion of Pten or expression of myristoylated-Akt (15) leads to the constitutive activation of mTOR, the nutrient-sensing pathway in mammalian cells that controls cell growth (reviewed in reference 16). mTOR forms two functionally distinct multiprotein complexes, termed mTORC1 and mTORC2. Each complex is defined by the unique adaptor proteins raptor (regulatory-associated protein of mTOR) and rictor (rapamycin-insensitive companion of mTOR), respectively (17, 18). mTORC2 regulates cell survival, metabolism, and cytoskeletal organization through the phosphorylation of several members of the AGC kinase subfamily, including Akt (reviewed in reference 19). mTORC1 plays a central role in the regulation of anabolic processes, including mRNA translation and ribosome biogenesis (reviewed in reference 20). mTORC1 regulates translation through the inhibition of 4E-BP1, a negative regulator of translation that interferes with translation initiation (21), and via activation of p70 S6K (S6K) and its downstream effectors, including ribosomal protein S6 (rS6) and eukaryotic elongation factor 2 kinase (22).

Extracellular factors that control skeletal development and bone accrual activate mTORC1. Most notably, insulin strongly activates the PI3K/Akt/mTORC1 pathway to promote anabolic cell processes, including translation, lipid biogenesis, and nutrient storage (23, 24). Insulin activates mTORC1 via inhibition of the tuberous sclerosis complex, which is a GTPase-activating protein for Rheb, a small GTPase. In its GTP-bound state, Rheb binds to and activates mTORC1 (25). Inhibition of mTORC1, using rapamycin, blocks insulin and IGF-1 stimulated osteogenesis in vitro (26, 27). Furthermore, deletion of the insulin receptor or insulin-like growth factor receptor 1 (Igflr) in OBs causes a postnatal low bone phenotype (9, 27), suggesting that mTORC1 plays an important role in transducing insulin and IGF-1 bone anabolic signals. Consistent with this view, inhibition of mTORC1 by administration of rapamycin to C57BL/6 mice resulted in a significant reduction in mineral apposition and a reduction in trabecular bone volume and thickness (27).

Taken together, these findings suggest that mTORC1 plays an important role in bone accrual. To specifically investigate this question, we deleted Rptor, the unique component of mTORC1, in pre-OBs and assessed the effects on skeletal development. Deletion of Rptor in pre-OBs blocked OB differentiation, leading to defective intramembranous and endochondrial ossification, low bone mass, and skeletal fragility.

RESULTS

OB-specific deletion of raptor in Rptor_ob^−/− mice.

Recent evidence suggests extracellular factors that regulate osteogenesis and bone accrual utilize the mTORC1 pathway to transduce anabolic signals important for pre- and postnatal skeletal development (9, 27, 28). To investigate this question, we generated mice in which Rptor, a unique and essential component of mTORC1, was disrupted in preosteoblastic cells. Male progeny from the mating of Osx1-GFP::cre-raptor^fl/wt and raptor^fl/fl mice with the genotypes Osx1-GFP::cre-raptor^fl/wt and Osx1-GFP::cre-raptor^fl/fl (here referred to as Rptor_ob^−/+ [het] and Rptor_ob^−/− [hom], respectively) were analyzed. In light of the skeletal phenotype of Osx1-GFP::cre mice (here referred to as Osx:cre) (29), age-matched male Osx:cre mice were used as controls.

To confirm bone-specific deletion of Rptor, genomic DNA was isolated from multiple tissues of Osx:cre and knockout animals, and PCR was used to amplify a product generated by cre-mediated recombination of the Rptor gene (30). A deletion-specific product was amplified from DNA isolated from the long bones of Rptor_ob^−/+ and Rptor_ob^−/− mice, confirming tissue-specific Rptor deletion (Fig. 1A). To confirm deletion at the transcriptional level, we bred Raptor^fl/fl mice with the R26-stop-EYFP reporter line and then mated the progeny with Osx:cre to generate eYFP-Rptor_ob^−/+ and eYFP-Rptor_ob^−/− mice in which cells that have expressed Osterix are labeled eYFP⁺ cells for their life span (31). RT-PCR was performed on RNA isolated from eYFP⁺ cells recovered (using fluorescence-activated cell sorting [FACS]) from the long bones of control (eYFP-Osx:cre), eYFP-Rptor_ob^−/+, and eYFP-Rptor_ob^−/− mice using Rptor-specific primers. A significant reduction in Rptor transcript was detected in eYFP⁺ cells isolated from het and hom mice relative to control, confirming gene inactivation in the majority of cells isolated by FACS (purity, ∼95% eYFP⁺) (Fig. 1B).

FIG 1.

OB-specific deletion of Rptor. (A) Genomic DNA was isolated from the indicated tissues of 4-week-old Osx:cre (cre), Rptor_ob^−/+ (het), and Rptor_ob^−/− (hom) mice, and the genotypes were confirmed (upper panel). An Osx:cre-mediated gene deletion product was amplified in genomic DNA isolated from the long bones of het and hom mice (lower panel). (B) Quantitative PCR was performed on RNA extracted from eYFP⁺ cells isolated from eYFP-Osx:cre mice (control), eYFP-Rptor_ob^−/+ mice, and eYFP-Rptor_ob^−/− mice using Rptor exon 6-specific primers. Amplified product (left panel) and relative transcript levels normalized to β-actin (right panel) are shown. fl, floxed; wt, wild type; del, deletion. *, P < 0.05; **, P < 0.01.

Prenatal skeletal phenotype in Rptor_ob^−/− mice.

Rptor_ob^−/+ and Rptor_ob^−/− mice were born at the expected Mendelian frequencies. No gross phenotypic differences or difference in weight or length were evident in Rptor_ob^−/− animals compared to either wild-type (WT) littermates (data not shown) or age-matched Osx:cre mice at birth (Fig. 2A). To investigate the effect of Rptor deletion on prenatal skeletal development, we stained the skeletons of newborn mice to assess the level of mineralized bone (alizarin red) and cartilage (alcian blue). Vertebral spinous and transverse processes and bones of the pelvic cage (illium and ischium) were less mineralized in Rptor_ob^−/− mice than in Osx:cre mice (Fig. 2B). The skulls of newborn Rptor_ob^−/− mice were narrower and less mineralized (Fig. 2B). Furthermore, the level of alizarin red labeling of the jawbones of Rptor_ob^−/− mice was reduced relative to Osx:cre mice (Fig. 2B). The long bones of the forelimbs and hind limbs (the humerus and the femur, respectively) were significantly shorter in Rptor_ob^−/− mice (mean decrease of 8.5% ± 1% for both bones) (Fig. 2C and D), and the percentage of bone that was mineralized was also significantly less in new born Rptor_ob^−/− mice relative to Osx:cre mice (17.4% ± 2.2% for both bones) (Fig. 2C and D). In summary, the deletion of Rptor in osterix-expressing cells (from embryonic day 13.5 [E13.5]) (8) caused a significant reduction in the lengths of long bones and in the level of mineralization of the long bones in the skeletons of newborn mice, demonstrating an important role for mTORC1 in prenatal skeletal development.

FIG 2.

Prenatal skeletal phenotype in Rptor_ob^−/− mice. (A) Images of newborn (P0) Osx:cre (cre) and Rptor_ob^−/− (hom) mice. Scale bar, 10 mm. (B) Newborn mice were stained with alizarin red and alcian blue to visualize mineralized bone (red) and cartilage (blue). (Upper panel) Images of the pelvic cage. Scale bar, 5 mm. (Middle panels) Superior view of the skull. (Lower panels) Lateral view of the skull. Scale bar, 1 mm. (C) Forelimbs of Osx:cre and Rptor_ob^−/− mice. The length of the humerus and the percentage of the humerus that is mineralized were measured. Scale bar, 5 mm. (D) Hind limbs of Osx:cre and Rptor_ob^−/− mice. The length of the femur and the percentage of the femur that is mineralized were measured. Scale bar, 5 mm. Data are presented as means ± the SD (n = 6 per group). **, P < 0.01; ***, P < 0.001.

Reduced weight and bone length in Rptor_ob^−/− mice.

To assess the effects of OB-specific Rptor deletion on postnatal skeletal development, we analyzed animals at 4 and 12 weeks of age since these time points represent the early and later phases of long-bone growth in mice (32). At 4 weeks of age, Rptor_ob^−/− mice were smaller (Fig. 3A) and weighed significantly less (mean decrease, 22.9% ± 7.8%) than Osx:cre mice (Fig. 3B). The reduced weight of Rptor_ob^−/− mice was still evident at 12 weeks of age relative to Osx:cre and Rptor_ob^−/+ mice (15.2% ± 3.5% and 15.4% ± 3.5% reductions, respectively).

FIG 3.

Phenotypic assessment of Rptor_ob^−/− mice. (A) Images of 4- and 12-week-old mice. Scale bar, 1 cm. (B) Weights of mice. (C) Tibial lengths (distance from the proximal tibial head to the fibular notch). (D) Spinal lengths (the distance from C1 to L5 vertebrae). (E) Humeral lengths (distance from the greater tubercle to the epicondyle). Measurements are presented as means ± the SD (n ≥ 6 per group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To evaluate the macroscopic effects of OB-specific Rptor deletion on skeletal development, we measured the lengths of the spines, tibiae, and humeri at 4 and 12 weeks of age. At 4 weeks, the spines of Rptor_ob^−/− mice were significantly shorter relative to het and Osx:cre mice (10.9% ± 1.8% and 16.8% ± 2%, respectively) (Fig. 3C). Similarly, the lengths of the tibiae were significantly shorter in Raptor_ob^−/− mice (Fig. 3D), as were the lengths of the humeri (Fig. 3E). At 12 weeks, the spine length and humeral length remained reduced in Rptor_ob^−/− mice relative to Osx:cre mice (Fig. 3C and E), but tibial length had normalized to control lengths (Fig. 3D).

Low endochondral bone mass in Rptor_ob^−/− mice.

To assess the effects of OB-specific Rptor deletion on the microarchitecture of the skeleton, cortical and trabecular bone parameters were assessed using micro-computed tomography (μCT) (Fig. 4A). For the assessment of cortical bone, the tibial diaphysis was analyzed. At 4 and 12 weeks of age, cortical thickness (Ct.Th) was significantly reduced in Raptor_ob^−/− mice relative to both Raptor_ob^−/+ mice and Osx:cre mice (Fig. 4A and B). Consistent with a reduction in Ct.Th, an increase in intrameduallary diameter was observed in het and hom mice relative to Osx:cre mice at 12 weeks of age (Fig. 4C).

FIG 4.

Low postnatal bone masses in Rptor_ob^−/− mice. (A) Cross-sectional μCT images of cortical and trabecular bone (tibia) of Osx:cre, Rptor_ob^+/−, and Rptor_ob^−/− mice. (B and C) Cortical thickness (Ct.Th) (B) and intramedullary diameter (C) of the cortical bone. (D to G) Quantitative assessment of the trabecular bone of the tibia. BV, bone volume; TV, total volume; Tb.Th, trabecular thickness; Tb.N, trabecular number; Tb.Sp, trabecular seperation. (H and I) Quantitative assessment of structural indices of the trabecular bone architecture of the tibia. Tb.Pf, trabecular pattern factor; SMI, structural model index. (J and K) Femurs from control and knockout animals were subjected to 3-point load testing. (L) μCT images of the skulls of 4-week-old Osx:cre and Rptor_ob^−/− mice. (Left panels) Superior view of skulls; (right panels) lateral view of skulls. Measurements are presented as means ± the SD (n ≥ 6 per group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To analyze changes in trabecular bone parameters, a region in the secondary spongiosa proximal to the tibial growth plate was analyzed. At 4 weeks of age, bone volume normalized to the total tissue volume (BV/TV) was profoundly reduced in Rptor_ob^−/− mice (mean decrease, 29% ± 3.5%). This reduction was increased at 12 weeks, at which point the bone volume was significantly reduced in both Rptor_ob^−/+ mice (31% ± 5.2%) and Rptor_ob^−/− mice (45% ± 7%) relative to Osx:cre mice (Fig. 4D). The trabecular number (Tb.N) and thickness (Tb.Th) were reduced in Rptor_ob^−/− mice relative to the Rptor_ob^−/+ and Osx:cre mice at 4 weeks (29.4% ± 4.2% and 15.7% ± 1.2%, respectively). Notably, at 12 weeks, Tb.N was significantly lower for both Rptor_ob^−/+ mice (24.8% ± 5.9%) and Rptor_ob^−/− mice (35.5% ± 5.1%) relative to Osx:cre mice (Fig. 4E and F). Consistent with a reduction in Tb.N and Tb.Th, trabecular separation (Tb.Sp) was increased in both Rptor_ob^−/+ mice (mean increase, 44.3% ± 7.5%) and Rptor_ob^−/− mice (45.7% ± 9.4%) relative to Osx:cre mice at 4 weeks of age (Fig. 4G).

An assessment of trabecular structural indices revealed an increase in the trabecular bone pattern factor (Tb.Pf), a measure of connectivity that increases as the level of connections between trabecular structures decreases, in Rptor_ob^−/− mice relative to Osx:cre mice at 12 weeks of age (Fig. 4H). This increase in Tb.Pf was associated with an increase in the structural model index (SMI), a measure of the shape of the trabeculae that increases as the trabeculae transition from a plate-like to rod-like architecture, for Rptor_ob^−/− mice relative to Osx:cre mice at 12 weeks of age (Fig. 4I).

We next evaluated the biomechanical properties of OB-specific Rptor-deficient bone by subjecting isolated bones to load testing. The femurs from Rptor_ob^−/− mice displayed a reduced bending stiffness (flexural rigidity; mean decrease, 71.9% ± 0.2% and 74.8% ± 10.1%, respectively) relative to Rptor_ob^−/+ and Osx:cre controls (Fig. 4J). The femurs from Rptor_ob^−/− mice were also found to have a lower strength (ultimate load; mean decrease, 56.3% ± 10%) relative to Osx:cre controls (Fig. 4K).

We next assessed the effects of Rptor gene deletion on intramembranous ossification, the developmental process that accounts for the majority of bones of the skull. Three-dimensional reconstructions of μCT scans of 4-week-old mice revealed that the skulls of Rptor_ob^−/− mice appeared to be smaller and malformed compared to Osx:cre mice (Fig. 4L). Furthermore, regions of the skulls of Rptor_ob^−/− mice appeared to be incompletely mineralized, especially in the frontal region adjacent to the coronal suture line and also in the nasal region. The lambdoidal and sagittal sutures in the skulls of Rptor_ob^−/− mice were also incompletely mineralized, suggesting that the deletion of Rptor in calvarial preosteoblasts disrupts intramembranous ossification. In summary, the deletion of Rptor in preosteoblasts causes a significant reduction in both cortical and trabecular bone parameters that culminates in more brittle bones, highlighting an important role for mTORC1 in postnatal skeletal development.

Increased intramedullary adipose in Rptor_ob^−/− mice.

In addition to the low bone mass, high levels of intramedullary adipocytes were observed in the tibiae of Rptor_ob^−/− mice compared to Osx:cre mice (Fig. 5A). At 4 and 12 weeks of age, the numbers of adipocytes per intramedullary area (N.Adc/M.Ar) were significantly higher in Rptor_ob^−/− mice relative to Osx:cre and Rptor_ob^−/+ mice (Fig. 5B). An assessment of the area occupied by adipocytes per number of adipocytes (AdC.Ar/N.AdC) suggested that there was no difference in adipocyte size in knockout mice relative to Osx:cre mice (Fig. 5C).

FIG 5.

Elevated levels of intramedullary adipose in Rptor_ob^−/− mice. (A) Toluidine blue-stained tibial sections of Osx:cre (cre) and Rptor_ob^−/− (hom) mice. Adipocytes appear as cleared circles. Scale bar, 200 μm. (B and C) The numbers of adipocytes per area of the intramedullary space (N.Adc/M.Ar) (B) and the areas occupied by adipocyes per number of adipocytes (AdC.Ar/N.AdC) (C) were quantified using histomorphometry. (D) Two-color confocal microscopy was used to detect overlapping eYFP and perilipin staining in frozen tibial sections isolated from 4-week-old eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice. Nuclei are stained with DAPI. Scale bar, 20 μm. (E) Compact bone cells were isolated from the long bones of Osx:cre (cre), Rptor_ob^−/+ (het), and Rptor_ob^−/− (hom) mice, and the numbers of CFU-F per 10⁵ cells plated were enumerated. Measurements are presented as means ± the SD. **, P < 0.01; ***, P < 0.001.

The increase in intramedullary adiposity in Rptor_ob^−/− mice suggests that the deletion of Rptor in preosteoblasts affected the lineage commitment of the osteoadipogenic precursor pool toward the adipogenic lineage or that adipocytes arose from the trans- or dedifferentiation of Rptor knockout osteoblastic cells. To investigate the latter, we examined the origin of the adipocytes using mice bred with the eYFP reporter. Tibial sections from eYFP-Osx:cre control and eYFP-Rptor_ob^−/− mice were stained for eYFP and perilipin, an adipocyte integral membrane protein, and overlapping signals assessed using two-color confocal microscopy. Only a small proportion of adipocytes in the bone marrow of eYFP-Rptor_ob^−/− mice were found to be eYFP⁺ perilipin⁺, suggesting that most of the adipocytes did not arise from the trans- or dedifferentiation of Rptor-null cells (Fig. 5D).

We next assessed whether the deletion of Rptor in preosteoblastic cells affected the mesenchymal stem cell (MSC) pool by measuring the number of CFU/fibroblast (CFU-F) in cell populations isolated from the long bones of Rptor_ob^−/− and control mice. Enumeration of morphologically discrete fibroblastic colonies revealed a significant increase in CFU-F in Rptor_ob^−/− mice relative to Osx-cre and het mice (fold increase, 9.6 ± 4.8 and 5.2 ± 0.9, respectively) (Fig. 5E). These data suggest that deletion of Rptor in preosteoblasts stimulates progenitor cell proliferation as a mechanism to compensate for impaired bone formation in Rptor_ob^−/− mice.

Narrow and disorganized growth plate in postnatal Rptor_ob^−/− mice.

Longitudinal bone growth is primarily driven by the coordinated proliferation and differentiation of chondrocytes. At the growth plate, chondrogenic cells are arranged into vertical columns that can be stratified into horizontal layers (zones) corresponding to different phases of chondrocyte differentiation (1, 33). Osterix expression has previously been observed in the prehypertrophic chondrocytes of the developing skeleton (2, 34) and the postnatal skeleton (35). Consistent with these findings, chondrocytes within the growth plates of eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice were found to be eYFP⁺ (Fig. 6A), suggesting that the deletion of Rptor in chondrogenic cells could contribute to the reduction in limb length observed in Rptor_ob^−/− mice.

FIG 6.

Narrow and disorganized growth plate in Rptor_ob^−/− mice. (A) eYFP-labeled cells within the hypertrophic chondrocyte zones of epiphyseal growth plates of 4-week-old eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice. Nuclei were stained with DAPI. Scale bar, 50 μm. (B) Safranin O staining of acidic proteoglycan in the epiphyseal growth plate of the tibia. Scale bar, 200 μm. (C) Growth plate widths based on Safranin O staining. (D to F) Width of chondrocyte zones within the growth plate based on chondrocyte morphology. (G) Immunohistochemical staining of PCNA in the growth plates of Osx:cre and Rptor_ob^−/− mice. Scale bar, 50 μm. (H) Quantitative PCR was performed on RNA extracted from eYFP⁺ cells isolated from eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice using primers specific to the indicated genes. The relative transcript levels were normalized to β-actin. PZ, proliferating zone; HZ, hypertrophic zone; PS, primary spongiosa. Measurements are presented as means ± the SD (n ≥ 6 per group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To investigate this question, we analyzed the morphology of the tibial growth plate using sections in which acidic proteoglycan, present in cartilage tissues, was stained red (Fig. 6B). In Rptor_ob^−/− mice, chondrocytes within the proliferating zone were disorganized, with fewer cells arranged into vertical columns relative to Osx:cre mice (Fig. 6B). Furthermore, at 4 weeks of age, the width of the growth plates in Rptor_ob^−/+ and Rptor_ob^−/− mice was significantly reduced relative to Osx:cre mice (Fig. 6C). We next assessed the width of the chondrocyte zones within the growth plates of knockout and Osx:cre mice. At 4 and 12 weeks of age there was no difference in the width of the zone of resting chondrocytes (Fig. 6D), precursor cells which, when activated, proliferate and synthesize a cartilagenous matrix. In contrast, at 4 weeks of age the width of the proliferating zone (where chondrocytes undergo rapid division by mitosis) in Rptor_ob^−/+ and Rptor_ob^−/− mice and the width of the hypertrophic zone (terminally differentiated chondrocytes) were significantly reduced relative to Osx:cre mice (Fig. 6E and F). No differences in the width of the growth plate or the width of chondrocyte zones were evident in 12-week-old control and knockout mice.

We next assessed the proliferative state of chondrocytes within the growth plate of 4-week-old Rptor_ob^−/− mice as a possible mechanism to account for the marked reduction in the size of the proliferating and hypertrophic zones. Using immunohistochemistry (IHC), proliferating cell nuclear antigen (PCNA) expression, a nuclear marker of proliferating cells (36), was clearly evident in chondrocytes within the growth plate of Osx:cre mice but was otherwise absent in the growth plate in chondrocytes from Rptor_ob^−/− mice (Fig. 6G).

We next assessed whether Rptor deletion in osterix-expressing chondrocytes affects the stages of chondrocyte differentiation. Chondrocyte maturation stimulates mineralizeation of the cartilage and ingress of the vasculature, allowing OB recruitment and differentiation (37). To achieve this, we assessed the transcript levels of immature and mature chondrocyte markers in eYFP⁺ cells recovered from the long bones of eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice. There was no difference in the expression of early chondrocyte markers Sox9 and Col2a1 (isoform 2) in Rptor_ob^−/− mice relative to Osx:cre mice (Fig. 6H). However, the transcript levels of Col10a1 and Mmp13, markers of mature and terminally differentiated chondrocytes, were significantly reduced in Rptor_ob^−/− mice relative to Osx:cre mice (Fig. 6H). In summary, these findings suggest that mTORC1 plays an important role in long bone growth by regulating chondrocyte proliferation and differentiation.

Reduced OB function in Rptor_ob^−/− mice.

Bone mass is maintained through the coordinated action of bone-resorbing osteoclasts (OCs) and bone-forming OBs. To determine whether the low bone mass phenotype of Rptor_ob^−/− mice was due to a reduction in OB numbers and/or an increase is OC numbers, OB and OC cells were enumerated in toluidene blue-stained tibial sections. No significant differences in the numbers of OBs or OCs per bone perimeter (N.Ob/B.Pm and N.Oc/B.Pm, respectively) were found in knockout animals relative to Osx:cre mice at 4 and 12 weeks of age (Fig. 7A and B). Similarly, there were no differences in the bone surfaces occupied by either OBs or OCs (Fig. 7C and D). No change in OB and OC numbers or in the surface areas occupied by OBs suggests that the phenotype primarily results from defective OB function. In support of this hypothesis, the mineral apposition rate (MAR) was significantly reduced in Rptor_ob^−/− animals relative to Osx:cre and het mice (Fig. 7E and F). The low MAR in Rptor_ob^−/− mice was associated with a significant reduction in the bone formation rate (BFR) relative to both Osx:cre and Rptor_ob^−/+ mice (Fig. 7G). Furthermore, the serum osteocalcin levels, a marker of OB function, were significantly reduced in Rptor_ob^−/− mice at 4 weeks of age relative to age-matched Osx:cre mice (Fig. 7H). These data suggest that the deletion of Rptor in preosteoblastic cells has a significant effect on OB function that could be associated with either a functional defect in mineralizeation and/or a defect in osteogenesis.

FIG 7.

Reduced OB function in Rptor_ob^−/− mice. (A and B) OBs lining the bone perimeter (N.OB/B.Pm) (A) and OB perimeter per bone perimeter (OB.Pm/B.Pm) (B) were measured in tibial sections using histomorphometry. (C and D) OCs lining the per bone perimeter (N.Oc/B.Pm) (C) and the OC perimeter per bone perimeter (Oc.Pm/B.Pm) (D) were measured in tibial sections using histomorphometry. (E) Dually labeled osteoid after consecutive intraperitoneal injections (3 days apart) of calcein. Arrowheads indicate the interlabel distance. (F) The mineral apposition rate (MAR) was calculated by dividing the interlabel distance by the time interval. (G) The bone formation rate (BFR) was calculated as an expression of the amount of newly formed mineralized bone. (H) Serum osteocalcin levels at 4 weeks of age. Measurements are presented as means ± the SD (n ≥ 7 per group). *, P < 0.05; **, P < 0.01; ***, P < 0.001.

OBs from Rptor_ob^−/− mice have an immature transcription profile and reduced osteogenic potential.

To determine whether the reduction in OB function in Rptor_ob^−/− mice was due to a block in OB differentiation, we assessed the expression of OB differentiation markers in tibial sections from knockout and control mice. Osterix expression was clearly evident in cells of the primary spongiosa immediately adjacent to the zone of hypertrophic chondrocytes in the growth plate of Osx:cre controls but was otherwise absent in sections derived from Rptor_ob^−/− mice (Fig. 8A). Similarly, osteocalcin expression, a marker of mature OBs, was evident in tibial sections from Osx:cre controls but was absent from Rptor_ob^−/− mice (Fig. 8A).

FIG 8.

Stalled osteogenic differentiation in Rptor_ob^−/− OBs. (A) Hematoxylin and eosin staining (H&E) and IHC staining of osterix and osteocalcin in the primary spongiosa of tibial sections from 4-week-old Osx:cre (cre) and Rptor_ob^−/− (hom) mice. Scale bar, 50 μm. (B) Quantitative PCR was performed on RNA extracted from eYFP⁺ cells isolated from eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice. The expression levels of the indicated genes relative to β-actin are shown. (C) Calvarial cells isolated from 4-week-old Osx:cre and Rptor_ob^−/− mice were cultured under osteogenic conditions, and the levels of acid-soluble mineral were quantified and expressed relative to the total DNA. Mineralized matrix was stained using alizarin red. HZ, hypertrophic zone; PS, primary spongiosa. Measurements are presented as means ± the SD from three independent experiments.*, P < 0.05; **, P < 0.01; ***, P < 0.001.

To gain further insight into the osteoblastic phenotype of the Rptor knockout cells, we analyzed the transcript level of genes involved at different stages of osteogenesis in eYFP⁺ cells isolated from the long bones of eYFP-Osx:cre and eYFP-Rptor_ob^−/− mice using reverse transcription-PCR. The expression of Runx2 and Sp7 (osterix), transcription factors essential for the initial stages of osteogenesis, and Alpl, a marker of early osteogenic differentiation, was significantly elevated in eYFP⁺ cells isolated from Rptor_ob^−/− mice relative to eYFP-Osx:cre mice (Fig. 8B). In contrast, the expression of late OB markers, including integrin-binding saloprotein (Ibsp), osteopontin (Spp1), osteonectin (Sparc), and osteocalcin (Bglap), was significantly downregulated in Rptor_ob^−/− mice relative to Osx:cre mice (Fig. 8B). These data suggest that the deletion of Rptor in preosteoblasts, using the Osx-cre driver, stalls OB differentiation, resulting in osteogenic cells with an immature transcript profile.

We next investigated how the immature OB phenotype of Rptor knockout cells affects OB function. Calvarial cells, isolated from knockout and control mice, were cultured under osteogenic conditions and the ability to form a mineralized matrix was assessed. Deletion of Rptor caused a significant reduction in mineralized matrix production as evidenced by a decrease in the level of acid-solubilized calcium and calcium-dependent alizarin red staining relative to Osx:cre mice (Fig. 8C). These data suggest that mTORC1 function in preosteoblastic cells is essential for osteogenic differentiation.

Mechansisms leading to low bone mass in Rptor_ob^−/− mice.

To investigate the mechanism leading to the stall in osteogenesis, we isolated calvarial cells from newborn raptor^fl/fl mice and used a tamoxifen-inducible Cre lentivirus to generate Rptor knockout OBs (Rap^−/−). The deletion of Rptor caused a reduction in basal, insulin-stimulated and IGF-1-stimulated phosphorylation of the key mTORC1 substrates p70 S6K and 4E-BP1 and ribosomal protein S6, a p70 S6K substrate, findings consistent with reduced mTORC1 activity (Fig. 9A). When cultured under osteoinductive conditions, the loss of mTORC1 function led to a significant decrease in osteogenic potential, as evidenced by the reduced expression of osteoblast-related genes in Rap^−/− osteoblasts relative to control (flox) cells (Fig. 9B). Despite this reduced osteogenic potential, no differences in Runx2 RNA or protein levels were detected in Rap^−/− OBs relative to control cells (Fig. 9B and C). In contrast, the levels of osterix (Sp7), a Runx2-regulated transcription factor essential for osteogenesis, were significantly reduced in Rap^−/− osteoblasts (Fig. 9B to D). These findings suggested that Runx2 transcriptional activity is reduced in Rap^−/− cells, and this may account for the reduction in osteogenic potential of these cells.

FIG 9.

Mechanisms leading to low bone mass in Rptor_ob^−/− mice. Calvarial cells isolated from Rptor^fl/fl mice (P4) were infected with a lentivirus harboring a tamoxifen-inducible cre recombinase. Infected cells were treated with tamoxifen (Rap^−/−) to stimulate Rptor gene deletion. Control cells (flox) were treated with vehicle. (A) The effect of Rptor gene deletion on mTORC1 substrate activation at steady-state and following stimulation with either insulin or IGF-1 was assessed by Western blotting. (B) Assessment of osteogenic gene expression in vehicle (flox)- or tamoxifen (Rap^−/−)-treated cells after 21 days of osteogenic induction. (C) Western blots showing Runx2 and osterix protein levels in vehicle (flox)- or tamoxifen (Rap^−/−)-treated cells after 7, 14, and 21 days of osteogenic induction. (D) Densitometry showing osterix protein levels normalized to β-actin in vehicle (flox)- or tamoxifen (Rap^−/−)-treated cells after 7, 14, and 21 days of osteogenic induction. (E) Western blots showing the activation status of STAT1 and STAT3 in vehicle (flox)- or tamoxifen (Rap^−/−)-treated cells. (F) Western blots showing the subcellular localization of the indicated proteins in vehicle (flox)- or tamoxifen (Rap^−/−)-treated cells. (G) Metabolic labeling in vehicle (flox), tamoxifen (Rap^−/−)- or vehicle plus AZD8055 (flox+AZD)-treated cells. (H) Cell size analysis in vehicle (flox)-, tamoxifen (Rap^−/−)-, or vehicle plus AZD8055 (flox+AZD)-treated cells. Measurements are presented as means ± the SD from three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To investigate how mTORC1 regulates Runx2 activity, we looked at known mTORC1 substrates and their relationship with Runx2 in osteoblasts. From this investigation, we identified STAT1 and STAT3 as potential candidates. STAT1 is a known substrate of mTORC1 (38, 39) and unphosphorylated STAT1 has been shown to attenuate Runx2 activity by retaining Runx2 in the cytoplasm (40). STAT3 is essential for normal skeletal development (41, 42), and STAT3 activation (serine 727 phosphorylation) is dependent on mTORC1 activity (43). Based on these findings, we investigated the phosphorylation status of STAT1 and STAT3 in control and Rptor-null osteoblasts, and this revealed no significant change in the phosphorylation status of Ser727 of either protein in control versus Rap^−/− osteoblasts (Fig. 9E). Furthermore, an assessment of the subcellular localization of Runx2, STAT1, and STAT3 failed to reveal any significant changes in Rap^−/− OBs relative to control cells (Fig. 9F).

In light of the significant effect of Rptor gene deletion on the phosphorylation status of S6K and 4E-BP1 (Fig. 9A), we looked at the protein synthetic capacity of these cells as a possible mechanism to explain the reduction in bone mass in Rptor_ob^−/− mice. Protein synthesis was significantly reduced in Rap^−/− OBs relative to control OBs (Fig. 9G). Furthermore, this reduction was similar to the decrease observed in cells treated with AZD8055, a potent inhibitor of both mTORC1 and mTORC2 (44), suggesting that the mTOR signaling plays a substantial role in the control of ribosome biogenesis and protein synthesis in OBs. We next examined the effect of Rptor gene deletion on cell size since mTORC1 and its substrate S6K are key determinants of this process (45, 46). Deletion of Rptor or treatment of control cells with AZD8055 significantly reduced the size of calvarial OBs (Fig. 9H). Taken together, these data suggest that deletion of Rptor in preosteoblasts reduces the protein synthetic capacity and sizes of these cells, which is likely to inhibit their ability to differentiate into mature OBs.

DISCUSSION

In this study, we investigated the role of mTORC1 in pre- and postnatal skeletal development by deleting Rptor, an essential component of mTORC1, in preosteoblasts. The deletion of Rptor using Osx:cre had a profound effect on postnatal bone accrual. By 4 weeks of age, cortical bone thickness (Ct.Th) was reduced by 32.3%, and the trabecular bone volume (TB/BV) was reduced by 29%, decreasing to 45% by 12 weeks. Notably, this reduction occurred relative to Osx:cre mice, which already display reduced cortical thickness and trabecular bone volume compared to wild-type mice (29). These data demonstrate that mTORC1 plays an important role in postnatal bone accrual.

The low-bone-mass phenotype of Rptor_ob^−/− mice was associated with a significant decrease in the mineral apposition rate and the bone formation rate. This was not due to a reduction in OB numbers but to a decrease in OB function. In vitro, calvarial OBs isolated from Rptor_ob^−/− mice and Rptor knockout OBs, generated using lentiviral Cre, were shown to have a reduced mineralizing capacity and reduced osteogenic potential. To investigate the mechanism, we bred our mice lines with an eYFP reporter strain that allowed us to purify the osteoblastic cells in which Rptor was deleted. Using this approach, we found that OB differentiation is stalled at an early phase in the osteogenic program, as evidenced by a decrease in the expression of late OB markers and an increase in the expression of Runx2, Sp7, and ALPL. Runx2 controls OB differentiation by regulating the expression of key osteoblastic genes, including Osx (2) and the genes for integrin-binding saloprotein (Ibsp), osteopontin (Spp1), osteonectin (Sparc), and osteocalcin (Bglap) (47, 48). Mechanistically, mTORC1 controls mRNA translation and ribosome biogenesis (21), suggesting that the suppression of translation in Rptor_ob^−/− OBs reduced the ability of preosteoblasts to proceed with the osteogenic program once Osx expression was initiated by Runx2. Consistent with this view, we showed that Rap^−/− OBs have a reduced osteogenic potential that was associated with a significant reduction in the synthetic properties of these cells, as measured by the level of protein synthesis and assessment of cell size (Fig. 9). These findings are iconsistent with an important role for mTORC1 function in the control of mRNA translation during the early phase of osteogenic differentiation.

Suppression of insulin and IGF-1 signaling is likely to play an important role in the skeletal phenotype of Rptor_ob^−/− mice. In vitro, both insulin and IGF-1 regulate primary mTORC1 substrates in OBs, and this was significantly reduced upon knockout of Rptor (Fig. 9). Mice in which the insulin receptor (Ob-ΔIR) or insulin-like growth factor receptor (Igflr^−/−) is deleted in OBs have a low-bone-mass phenotype (9, 27). In Ob-ΔIR mice, the failure to accumulate trabecular bone is due, in part, to a significant reduction in OB numbers, whereas the low bone mass in Igflr^−/− mice appears to be due to a failure of Igflr-null OBs to fully differentiate into mature osteocalcin-positive cells. IGF-1 activates the PI3K/Akt/mTOR pathway, and IGF-1-induced osteogenic differentiation of Sca-1⁺ MSC progenitor cells is blocked by rapamycin (27), suggesting that the low-bone-mass phenotype in Rptor_ob^−/− mice could be due, in part, to impaired IGF-1 signaling. mTORC1 has also been shown to play a role in the anabolic effects of Wnt7b using mice in which the Wnt7b cDNA, flanked by a floxed transcriptional stop, was knocked-in to the Rosa26 locus (R26-Wnt7b) (28). Osx:cre-mediated activation of Wnt7b expression, initiated postnatally, promoted bone formation by stimulating OB number and function. In this postnatal context, deletion of Rptor in compound mice overexpressing Wnt-7b (Osx:Cre; R26-Wnt7b; Raptor^fl/fl) reduced Wnt-7b-induced bone mass by inhibiting OB function (as evidenced by decreased mineral apposition rate and bone formation rate) but not OB numbers (28). These results are consistent with our findings that demonstrate an important role for mTORC1 in the control of OB function.

In this study, we found that Rptor_ob^−/− mice undergo a significant reduction in limb length and mineralizeation at birth and postnatally. Osx expression has previously been reported in the growth plate of prenatal (2, 34, 49) and postnatal mice (35) and, consistent with these findings, chondrocytes were labeled eYFP⁺ in eYFP-Rptor_ob^−/− and control eYFP-Osx:cre mice. Labeling of chondrocytes has also been observed in several lineage tracing studies using reporter mice and Osx:cre mice (8, 50–52). In the developing skeleton, Osx transcripts are found transiently in differentiating chondrocytes at E13.5 and weakly in the prehypertrophic chondrocytes of E15.5 embyros (2). At E18.5, osterix protein is evident in prehypertrophic chondrocytes and is abundant in preosteoblasts in the primary spongiosa but absent from the more mature hypertrophic cells (53). This pattern of Osx expression overlaps with mTORC1 activity in the developing skeleton. Activated prS6, a major substrate of mTORC1, is evident in the proliferating columnar chondrocytes and prehypertrophic chondrocytes of embyros from E16.5 to E18.5. The activation status of prS6 is markedly reduced in the hypertrophic zone but is reactivated in chondrocytes at the final stage of hypertrophy (53, 54). Activated prS6 is also abundantly evident in OB precursor cells within the primary spongiosa at E18.5 (53). Recent evidence suggests that correct regulation of mTORC1 activity during chondrogeneis is essential for the normal progression of prehypertrophic chondrocytes to a terminal hypertrophic fate. Deletion of Lkb1, a negative regulator of mTORC1 (55) in immature chondrocytes using Col2α1-Cre causes a rapamycin-sensitive expansion of immature chondrocytes and the formation of enchondroma-like tumors (53). Deletion of mTOR or raptor in the developing head and limb mesenchyme at E9.0, prior to skeletogenesis, using Prx-1-cre mice has also been shown to profoundly affect chondrogenesis (54). Most notably, mTORC1 is important for the transition of chondrocytes to the hypertrophic program and may be involved in the phases of hypertrophy that drive the increase in dry cell mass (54, 56). Given the importance of mTORC1 activity in chondrogenesis, deletion of Rptor in osterix-expressing prehypertrophic chondrocytes may have affected chondrocyte differentiation by prematurely terminating mTORC1 activity. We did note that Col10a1 and Mmp13 expression was reduced in eYFP⁺ chondrocytes isolated from the long bones of eYFP-Rptor_ob^−/− mice, suggesting that fewer terminally differentiated chondrocytes were present in Rptor_ob^−/− mice. It was also evident that deletion of Rptor in osterix-expressing hypertrophic chondrocytes caused an overall reduction in chondrocyte proliferation, as measured by a decrease in the expression of PCNA (Fig. 6G), suggesting that decreased chondrocyte proliferation may also account for the decrease in expression of mature chondrocyte markers.

Recently, gain-of-function mouse models have been used to investigate the role of mTORC1 in osteogenesis. Deletion of either Tsc1 in preosteoblasts using Osx:cre (ΔTsc1 mice [57]) or Tsc2 in mature osteoblasts using Ocn:cre (ΔTsc2 mice [58]), results in constitutive activation of mTORC1. In both models, hyperactivation of mTORC1 leads to an increase in postnatal bone mass that was evident in cortical and trabecular regions of the long bones and calvarial bones of the skull. The increase in bone mass is associated with an increase in OB number; however, in both models, OB function is impaired. In vitro, the addition of low-dose rapamycin (10 pM to 100 pM) to ΔTsc1 or ΔTsc2 OBs is able to rescue the impaired osteoblast function, implicating mTORC1 as a negative regulator of osteogenesis. However, in ΔTsc2 mice (58) the impaired OB function can be attributed, at least in part, to a reduction in insulin signaling as mTORC1 promotes negative-feedback inhibition of insulin signaling via mTORC1-p70 S6K-mediated phosphorylation and degradation of IRS-1 (59). Hyperactivation of this feedback loop in ΔTsc2 OBs may contribute to the reduction in OB activity, as seen in mice in which the insulin receptor (Ob-ΔIR) is deleted in OBs (9). Constitutive activation of mTORC1 in preosteoblasts (ΔTsc1 mice) causes more severe skeletal abnormalities than those observed in ΔTsc2 mice, possibly reflecting the earlier embryonic time point of Cre activation (Osx promoter E13.5 days postcoitus [dpc] versus Ocn promoter E17 dpc [2, 26]). The addition of rapamycin to ΔTsc1 mice, from E13.5 dpc, rescues the bone defects, suggesting that mTORC1 function must be switched off to allow osteogenesis to proceed correctly. In contrast to gain-of-function mouse models, the evidence from loss-of-function studies suggests that mTORC1 may play a role in promoting osteoblast differentiation. Mice with the hom deletion of mTOR in mature osteoblasts (Ocn:cre ΔmTOR) demonstrate reduced tibial trabecular bone volume and trabecular number (58). This phenotype can be attributed to impaired mTORC1 function since mice with homozygous deletion of Rictor, an essential component of mTORC2, in mature osteoblasts have no tibial bone phenotype (60). As reported here, Rptor_ob^−/− mice have significantly reduced mineral apposition and bone formation rates and no change in OB number per bone perimeter (N.OB/B.Pm, Fig. 7), suggesting impaired OB function. Analysis of the transcriptional profile of OBs isolated from Rptor_ob^−/− mice revealed an immature bone phenotype, suggesting that mTORC1 signaling promotes osteoblast differentiation from preosteoblasts, a finding supported by the recent work of others (61). Taken together, the findings from loss- and gain-of-function studies indicate that temporal regulation of mTORC1 plays an essential role in the osteogenic program.

Clinically, rapamycin (Sirolimus) and the related analogue (Everolimus) are used as immunosuppressants to prevent organ transplant rejection and to treat a variety of cancers (62). Given the role of mTORC1 in osteogenesis, it is conceivable that the use of inhibitors in the pediatric context could lead to adverse skeletal outcomes. To date, despite extended treatment periods (2 to 5 years), these concerns have not been borne out with respect to rates of linear growth in young children (age < 5 years) (63, 64). Nonetheless, continued assessment of bone mineral density is warranted in these children. Notably, in adults, adjuvant treatment of estrogen receptor-positive (ER⁺) breast cancer with Everolimus appears to have bone-protective effects (65), which have been attributed to its potent antiosteoclastic properties rather than to a bone anabolic effect.

In summary, our studies have revealed a novel role for mTORC1 in osteogenesis. Notably, mTORC1 is required for the progression of the osteogenic program from the preosteoblast stage to the mature osteoblast stage. Mechanistically, mTORC1-mediated control of mRNA translation is likely to be a key determinant governing this progression. Finally, our studies further highlight the importance of pathways that coordinate cellular growth and energy metabolism in the accrual of bone mass.

MATERIALS AND METHODS

Transgenic mice.

All mice were bred and housed at the SA Pathology Animal Care Facility, and studies were performed with Institutional Ethics approval (SA Pathology Animal Ethics Committee approval Project 63/10). Raptor^fl/fl mice (30), Osx1-GFP::cre mice (8), and R26-stop-EYFP mice (31) were used in this study. The following forward and reverse primer pairs were used for genotyping: Raptor^fl/fl, 5′-ATGGTAGCAGGCACACTCTTCATG-3′ and 5′-GCTAAACATTCAGTCCCTAATC-3′; Osx1-GFP::cre, 5′-CGCGGTCTGGCAGTAAAAACTATC-3′ and 5′-CCCACCGTCAGTACGTGAGATATC-3′; and R26-stop-EYFP, 5′-AAAGTCGCTCTGAGTTGTTAT-3′ and 5′-GCGAAGAGTTTGTCCTCAACC-3′. Detection of Rptor gene deletion was performed using 5′-ATGGTAGCAGGCACACTCTTCATG-3′ and 5′-CTCAGAGAACTGCAGTGCTGAAGG-3′, as previously described (30). Genomic DNA was isolated from tissues using a DNeasy blood and tissue kit (Qiagen, Melbourne, Australia).

Skeletal staining.

Alizarin red and alcian blue staining of new born mice was performed as previously described (66). The percentage of the bone (humerus and tibia) that was mineralized was calculated by dividing the length of the mineralized (alizarin red) region of the bone by the total length of the bone using ImageJ (67). Images were aquired using bright field optics.

High-resolution μCT scanning.

X-ray microtomography was performed using a SkyScan 1076 (Bruker, Kontich, Belgium). Calvariae and tibiae were scanned at 74 kV/100 mA using a 0.5-mm aluminum filter, a 0.8 rotation step, and two-frame averaging with an isometric resolution of 8.65 μm/pixel. Three-dimensional reconstruction of the scan data (smoothing, 1; beam hardening, 30%) was performed using NRecon (SkyScan). Analysis of the bone microarchitecture was performed using CTAn (SkyScan). Cortical bone measurements were calculated as the average of measurements in the lateral, medial, caudal, and cranial orientation from three slices of cortical bone. Slices were selected in relation to the primary spongiosa. For analysis of the proximal tibia, a trabecular region of interest was manually defined to exclude the cortex. Transverse micro-computed tomography (μCT) slices of the tibia were acquired using a 6.5-μm slice increment. A total of 150 slices (0.97 mm) were analyzed for each tibia, commencing 20 slices distal to the primary spongiosa. Images were reconstructed and thresholded using a specimen-specific threshold. For trabecular bone regions, the percent bone volume (BV/TV), trabecular thickness (Tb.Th.), trabecular number (Tb.N), trabecular separation (Tb.Sp), structure model index (SMI), and trabecular pattern factor (Tb.Pf) were calculated using CTAn. To ensure that the analysis of trabecular bone was conducted in corresponding regions of the secondary spongiosa proximal to the growth plate, the distance from the primary spongiosa and the area analyzed was normalized to the length of the tibia. Calvarial images were acquired using Avizo Fire (FEI Visualization Sciences Group, Burlington, MA). Tibial length measurements were performed using CTAn software (SkyScan) and taken from the proximal tibial head to the fibrual notch. Humeral lengths were measured using CTAn, with measurements taken from the greater tubercle to the epicondyle. For spine length, the distance from cervical 1 (C1) to lumbar 5 (L5) was measured in whole-body X-ray images (LX-60; Faxitron, Lincolnshire, IL) and quantitated using ImageJ.

Histology.

Methyl methacrylate processing and embedding of nondecalcified bone and staining of sections were conducted as previously described (68, 69). Briefly, sections (5 μm) were deplasticized with acetone for 15 min and then rinsed in water. Toluidine blue sections were stained in 2% toluidine blue solution for 3 min and washed twice with sodium phosphate-citrate buffer. Safranin O sections were stained in 0.2% Fast Green (Sigma) for 5 min and then stained in 0.1% Safranin O stain for 5 min. Quantitation of images was conducted using OsteoMeasure XP (OsteoMetrics, Inc.). Histomorphometric analysis of OB numbers (N.Ob/B.Pm), OB-occupied bone surface (Ob.S/BS), osteoclast numbers (N.Oc/B.Pm), osteoclast-occupied bone surface (Oc.S/BS), the number of adipocytes per intramedullary area (total area − bone area) (N.AdC/M.Ar), and the percentage adipose area per intramedullary area (AdC.Ar/M.Ar) was conducted on toluidine blue-stained sections. The region of interest for the proximal mouse tibiae was defined as a region of secondary spongiosa distal from the primary spongiosa to a distance of 1.8 mm.

Double fluorochrome labeling and mineral apposition rate.

To measure bone formation in vivo, mice were subcutaneously injected with calcein at 20 μg per g of body weight and then injected again after a 3-day interval. At 24 h after the second injection, the tibias were collected, fixed in 10% formalin, and then embedded in methyl methacrylate as previously described (68). Dynamic measurements of bone formation were conducted on calcein-labeled sections of the tibiae in the secondary spongiosa using the Osteomeasure XP system (Osteometrics, Inc., Atlanta, GA), and the distance between the two calcein-labeled mineralization fronts was quantitated. The MAR was calculated by dividing the interlabel distance by the time interval.

Immunohistochemistry.

Bones were fixed in formalin for 1 week, demineralized in EDTA (14%, wt/vol), and embedded in paraffin. Sections (5 μm) were subjected to antigen retrieval (10 mM citrate buffer [pH 6.0], 95°C) for 10 min and then cooled to room temperature. Endogenous peroxidase was quenched with 0.5% H₂O₂ (vol/vol) in methanol for 30 min, and sections were blocked in 3% normal horse serum in phosphate-buffered saline (pH 7.4; PBS) and then incubated with the primary antibody at 4°C overnight. Sections were washed in PBS and then incubated with species-specific biotinylated secondary antibodies (anti-goat [BA-5000] and anti-rabbit [BA-1000] antibodies; Vector Laboratories, Burlingame, CA) for 1 h at room temperature, followed by 30 min of incubation with streptavidin-HRP (Vector Laboratories). Antibody binding was visualized using diaminobenzidine (DAB) according to the manufacturer's instructions (Vector Laboratories). The antibodies used for IHC were as follows: osterix (ab22552; Abcam, Cambridge, United Kingdom), eGFP (A600-101-215; Rockland, Limerick, PA), eYFP (A600-401-215; Rockland), osteocalcin (sc-18319; Santa Cruz Biotechnology, Santa Cruz, CA), and PCNA (sc-7907; Santa Cruz Biotechnology).

Confocal microscopy.

For immunofluorescence staining, paraffin-embedded sections were antigen retrieved as described above and blocked in 1% bovine serum albumin, 2.5% goat serum, and 2.5% horse serum in PBS (blocking buffer), followed by incubation with primary antibodies overnight at 4°C. Sections were washed with ice-cold PBS and then incubated with fluorescently tagged species-specific secondary antibodies (anti-chicken [Alexa Fluor 488 conjugated, ab150173; Abcam] and anti-rabbit [Alexa Fluor 647 conjugated, ab150067; Abcam] antibodies) at room temperature for 1 h. The sections were washed and then briefly fixed for 10 min (0.1% formalin in PBS), and the nuclei were labeled with DAPI (4′,6′-diamidino-2-phenylindole; 300 nM in PBS) for 5 min. Sections were mounted with ProLong Gold antifade reagent (Invitrogen).

For frozen sections, the bones were fixed overnight in formalin at 4°C and then demineralized in EDTA (14% [wt/vol] in PBS) and 10% sucrose at 4°C until no calcified tissue was detectable by X-ray examination. Tissues were frozen in optimum-cutting-temperature resin (OCT; Sakura, Alphen aan den Rijn, Netherlands) using liquid nitrogen chilled isopentane then sectioned using a cryostat (CM 3050s; Leica, Sydney, Australia). Sections (8 μm) were air dried and then stained and mounted as described above. Fluorescent labeling was detected using a LSM400 confocal microscope (Zeiss, Oberkochen, Germany). Antibodies to eYFP (ab13970; Abcam) and perilipin (catalog no. 9349; Cell Signaling, Danvers, MA) were used for immunofluorescence analyses.

Biomechanical testing.

Three-point bending analyses to failure were conducted using femurs isolated from 4-week-old Osx:cre, Rptor_ob^−/+, and Rptor_ob^−/− mice as previously described (70).

Isolation of eYFP⁺ cells and real-time PCR.

Compact bone cells (CB-MSCs) were obtained from the long bones of 4-week-old eYFP-Osx:cre, eYFP-Raptor_ob^−/+, and eYFP-Raptor_ob^−/− mice in accordance to published methods (71). In brief, the tibias and femurs from 8 to 12 mice were excised, cleaned thoroughly, and then crushed in ice-cold PBS supplemented with 2% fetal calf serum (FCS) and 2 mM EDTA, washed several times, chopped into small fragments, and digested into type 1 collagenase (Worthington Biochemical Corporation, Lakewood, NJ) at 3 mg/ml at 37°C for 45 min with shaking. The resulting cell suspension was filtered through a 70-μm-pore size strainer, and then eYFP⁺ cells were isolated by FACS using a MoFlo Astrios (Beckman Coulter, Brea, CA). Total RNA was isolated from eYFP⁺ cells by using a RNAqueous micro kit (Life Technologies, Melbourne, Australia), and cDNA was prepared from 70 ng of total RNA using Sensiscript reverse transcriptase according to the manufacturer's instructions (Qiagen). Real-time PCRs were performed using RT² SYBR green ROX reagent (Qiagen) in a Rotor-Gene 3000 thermal cycler (Corbett Research/Qiagen). Forward and reverse primer pairs, designed to amplify across intron-exon boundries, are listed in Table 1. Changes in gene expression were calculated relative to β-actin.

TABLE 1.

Oligonucleotides used for real-time PCR

Gene	Accession no.	Sequence (5′–3′)		Size (bp)
		Forward	Reverse
Rptor	NM_028898.2	TGGGTCTTCAACAAGAACTACACT	TCTGGGCAAGTGGATGGTTT	199
Runx2	NM_001146038.2	ATGATGACACTGCCACCTCTG	ATGAAATGCTTGGGAACTGC	120
Sp7	NM_130458.3	ATGGCGTCCTCTCTGCTTG	GTCCATTGGTGCTTGAGAAGG	211
Bglap	NM_007541.3	AAGCAGGAGGGCAATAAGGT	TCAAGCCATACTGGTCTGATAGC	142
Spp1	NM_001204201.1	AGCAAGAAACTCTTCCAAGCAA	GATTCGTCAGATTCATCCGAGT	130
Sparc	NM_009242.5	ACTACATCGGACCATGCAAATAC	CGTACAAGGTGACCAGGACAT	108
Ibsp	NM_008318.3	CAGTCCAGGGAGGCAGTG	GGAAAGTGTGGAGTTCTCTG	158
Col10a1	NM_009925.4	CATCTCCCAGCACCAGAATC	GCTAGCAAGTGGGCCCTTTA	142
Sox9	NM_011448.4	CACAAGAAAGACCACCCCGA	GGACCCTGAGATTGCCCAGA	209
Mmp13	NM_008607.2	AGCAGTTCCAAAGGCTACAACT	CCTCGGAGACTGGTAATGGC	269
Col2a1(II)	NM_001113515.2	GCGACCGGGAGCATATAACT	GCCCTAATTTTCGGGCATCC	290
Actb	NM_007393.4	TTGCTGACAGGATGCAGAAG	AAGGGTGTAAAACGCAGCTC	236
Alpl	NM_007431.3	GCCTTACCAACTCTTTTGTGC	GGCTACATTGGTGTTGAGCTT	116
Gli1	NM_010296.2	TACCATGAGCCCTTCTTTAGGA	GCATCATTGAACCCCGAGTAG	116

Primary cell cultures and lentiviral transduction.

Primary calvarial OB cultures were established from 4-week-old Osx:cre, Rptor_ob^−/+, and Rptor_ob^−/− mice and Rptor^fl/fl mice (P4) by using serial collagenase digestion as previously described (26). Cells were cultured in α-modified Eagle medium (Sigma-Aldrich, St. Louis, MO) supplemented with 20% FCS, 2 mM l-glutamine, 1 mM sodium pyruvate, 15 mM HEPES buffer, penicillin (50 U/ml), and streptomycin (50 μg/ml) as the growth medium. OB cultures generated from raptor^fl/fl mice (passages 2 and 3) were infected with a lentivirus harboring a tamoxifen-inducible self-deleting Cre recombinase (LEGO-CreERT2-iG2) in the presence of 4 μg of Polybrene/ml as previously described (72). Calvarial cells with stable lentiviral integration were purified using FACS on the basis of eGFP expression. After in vitro expansion, the cells were treated with 0.5 mM 4-hydroxytamoxifen (4-OHT; Sigma-Aldrich) or vehicle (0.05% ethanol) for 8 days. The 4-OHT-treated cultures are labeled “Rap^−/−” to signify Rptor gene deletion, and the vehicle-treated cultures are labeled “flox” to signify intact floxed Rptor allelles. No 4-OHT or vehicle was included in cultures beyond the 8-day Cre induction period.

CFU-F assays.

For CFU-F assays, compact bone MSCs were isolated from the long bones of 4-week-old mice as previously described (73) and plated at concentrations of 1 × 10⁵, 3 × 10⁵, and 6 × 10⁵ cells/well in duplicate in six-well plates. Plated cells were incubated for 9 days in a hypoxic in vitro cabinet (Coy Laboratory Products) at 37°C before staining for CFU-F colonies with 0.1% (wt/vol) toluidine blue in 1% paraformaldehyde as previously described (74). Aggregates of >50 cells were scored as CFU-F-derived colonies, and aggregates of >10 and <50 cells were scored as clusters.

Osteogenesis assays.

Primary calvarial osteoblast cultures were seeded into 24- or 6-well plates (9 × 10³/cm²) in growth medium supplemented with β-glycerol phosphate (10 mM) and l-ascorbic acid 2-phosphate (100 μg/ml) (osteogenic induction medium). The medium was changed twice weekly for 21 days, and the mineral content was quantified by measuring the concentration of Ca²⁺ in an acid-solubilized matrix using an Arsenazo kit (Thermo Fisher Scientific, Inc., Middletown, VA), normalized to the number of cells per well using a Quant-iT PicoGreen double-stranded DNA (dsDNA) assay (Life Technologies). The mineralized matrix was stained with alizarin red S (2% [wt/vol]) for 1 h, rinsed with water, and allowed to air dry. For lentivirus-infected OB cultures, RNA was harvested using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA), and real-time PCR was performed as described above.

Western blotting and subcellular fractionation.

Protein lysates were prepared from vehicle- and tamoxifen-treated lentivirus-transduced cultures, and proteins (50 μg) were resolved by SDS-PAGE, transferred to a polyvinylidene difluoride nylon membrane, and Western blotted as previously described (69). For insulin and IGF-1 pulse experiments, the cells were incubated under serum-free conditions for 3 h and then pulsed with 100 nM insulin or IGF-1 (ProSpec-Tany TechnoGene, Ltd., Ness-Ziona, Israel) at 100 μg/ml for 10 min. Nuclear and cytoplasmic extracts were prepared using NE-PER extraction reagents according to the manufacturer's instructions (Thermo Fisher Scientific). Antibodies to the following (catalog numbers are in parentheses) were used for Western blotting: Raptor (2280), P-p70 S6K T389 (9206), p70 S6K (9202), P-S6 S240/S244 (2215), S6 (2317), P-STAT1 S727 (8826), STAT1 (14994), P-STAT3 S727 (9134), STAT3 (9139), Runx2 (8486), and histone H3 (4499) (Cell Signaling Technologies, Beverly, MA); osterix (ab22552; Abcam); HSP90α/β (sc-7947; Santa Cruz Biotechnology); and β-actin (A5441; Sigma-Aldrich). Serum osteocalcin levels were measured by using a commercial enzyme-linked immunosorbent assay (Alfa Aesar, Ward Hill, MA) according to the manufacturer's instructions.

Metabolic labeling and cell size assessment.

Metabolic labeling experiments were performed as previously described (75). In brief, vehicle- and tamoxifen-treated lentivirus-infected calvarial cultures were seeded into 6-well plates (2 × 10⁵ cells/plate) and cultured for 15 h, a time frame in which there is no significant difference in the proliferative capacity of control versus Rap^−/− cells (data not shown). Growth medium was replaced with Cys-Met-free Dulbecco modified Eagle medium (Thermo Fisher) supplemented with 10% dialyzed FCS for 1 h. AZD8055 (1 μM; Santa Cruz Biotechnology) was added after 30 min. [³⁵S]methionine-cysteine (Perkin-Elmer, Waltham, MA) was then added (to a final concentration of 10 μCi/ml), and the cells were cultured for 1 h. [³⁵S]methionine-cysteine incorporation was determined using scintillation counting and normalized to the total cell number. Analysis of the cell size was performed as described previously (76).

Statistical analysis.

All data are graphed as means ± the standard deviations (SD). Statistical analyses were performed using a one-way or two-way analysis of variance with Dunnett's post hoc test or an unpaired Student t test using Prism (GraphPad Software, Inc., La Jolla, CA). In all cases, a P value <0.05 was considered statistically significant.