Original Article mTORC2 signaling promotes skeletal growth and bone formation in mice

Jianquan Chen1 , Nilsson Holguin1 , Yu Shi1 , Matthew J. Silva1 and Fanxin Long1,2,3,4 1 Deaprtment of Orthopaedic Surgery; 2 Department of Medicine; 3 Department of Developmental Biology, Washington University School of Medicine, St. Louis, MO 63110 4 Corresponding author: [email protected]

Key words: mTORC2, rictor, cartilage, bone, mechanical loading, mouse Running title: mTORC2 stimulates bone formation



This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: [10.1002/jbmr.2348]

Additional Supporting Information may be found in the online version of this article.

Initial Date Submitted February 04, 2014; Date Revision Submitted July 28, 2014; Date Final Disposition Set July 30, 2014

Journal of Bone and Mineral Research © 2014 American Society for Bone and Mineral Research DOI 10.1002/jbmr.2348

Abstract Mammalian target of rapamycin (mTOR) is an evolutionarily conserved serine/threonine kinase controlling many physiological processes in mammals. mTOR functions in two distinct protein complexes, namely mTORC1 and mTORC2. Compared to mTORC1, the specific roles of mTORC2 are less well understood. To investigate the potential contribution of mTORC2 to skeletal development and homeostasis, we have genetically deleted rictor, an essential component of mTORC2, in the limb skeletogenic mesenchyme of the mouse embryo. Loss of rictor leads to shorter and narrower skeletal elements in both embryos and postnatal mice. In the embryo, rictor deletion reduces the width but not the length of the initial cartilage anlage. Subsequently, the embryonic skeletal elements are shortened due to a delay in chondrocyte hypertrophy, with no change in proliferation, apoptosis, cell size or matrix production. Postnatally, rictor-deficient mice exhibit impaired bone formation, resulting in thinner cortical bone, but the trabecular bone mass is relatively normal thanks to a concurrent decrease in bone resorption. Moreover, rictor-deficient bones exhibit a lesser anabolic response to mechanical loading. Thus, mTORC2 signaling is necessary for optimal skeletal growth and bone anabolism.

Introduction mTOR signaling has emerged as a central mechanism in controlling many physiological processes in mammalian cells (1). However, specific functions of the two mTOR-containing complexes, mTORC1 versus mTORC2, at the tissue and organ level in whole organisms are just beginning to be understood. In particular, the physiological role of mTORC2 has been understudied. In contrast to mTORC1, mTORC2 contains rictor, and phosphorylates a distinct set of downstream targets, including members of the AGC family of kinases such as Akt (at position Ser 473), serum- and glucocorticoidinduced protein kinase 1 (SGK1), and protein kinase C-á (PKC-á) (2-4). Early studies in cell culture showed that mTORC2 controls actin cytoskeleton through modulating PKC-á (5,6). More recently, mouse genetic studies have implicated hepatic mTORC2 in regulating whole-body metabolism (7). An increasing number of secreted factors including insulin/IGF1 and Wnt have been shown to activate mTORC2 (7,8). In addition, mechanical stress was shown to activate mTORC2 in vitro (9). Because both the aforementioned secreted factors and mechanical stress play important roles in skeletal biology, we hypothesize that mTORC2 signaling may regulate skeletal development and homeostasis (10-12). Much of the mammalian skeleton including long bones in the limb is formed via endochondral ossification that initiates in the embryo (13). The process begins with mesenchymal condensations that give rise to cartilage anlagen encased by the perichondrium. After the initial proliferation that increases the size of the cartilage template, cells at the center along the long axis of the template exit the cell cycle and undergo hypertrophy, whereas those at either end of the template remain mitotic and organize into two morphologically distinct regions containing round versus columnar chondrocytes. Concurrently, perichondrial cells flanking the hypertrophic region differentiate into osteoblasts that produce the bone collar. A majority of the hypertrophic chondrocytes eventually undergoes apoptosis, followed by blood vessel invasion and removal of the hypertrophic cartilage matrix. Meanwhile, osteoblast progenitors comigrating with the invading blood vessels differentiate into osteoblasts that deposit the trabecular bone (14). Thus, chondrocyte hypertrophy is tightly coupled with bone formation and is critical for skeletal growth during embryogenesis.

Bone formation continues in postnatal life. This includes bone deposition at the periosteal surface resulting in widening of the long bones, but the molecular regulation for periosteal bone accrual is just beginning to be understood. A high-bone-mass allele of Lrp5 (A214V) and deletion of Sost both stimulate periosteal bone deposition in the mouse, implicating Wnt signaling in the process (15-17). Mechanical loading is also known to stimulate periosteal bone formation proportional to the magnitude of strain and at least in part through Wnt signaling (18-22). In all cases however, the intracellular molecules mediating the anabolic response are not well understood. In this study, by deleting rictor in the limb mesenchyme with Prx1-Cre, we find that mTORC2 signaling promotes skeletal growth via regulation of chondrocyte hypertrophy, and also contributes to the periosteal bone accrual in response to mechanical loading.

Materials and Methods Mice Prx1-Cre mice as previously reported were purchased from Jackson Laboratory (Bar Harbor, ME) (23). Rictorflox/flox mice are as previously described, and kindly provided by Dr. Jeffrey Arbeit at Washington University (24). Washington University Animal Studies Committee approved all mouse procedures.

Analyses of mouse embryos Alizarin Red/Alcian Blue staining of embryonic skeleton was performed following protocols described by McLeod (25). For histological analyses, embryonic limbs were fixed in 10% formalin, decalcified in EDTA (for E16.5 or older embryos), and embedded in paraffin. H&E, Von Kossa and Alcian Blue/Picrosirius Red staining were performed on paraffin sections following the standard procedures. In situ hybridization was performed with 35 S-labeled riboprobes as previously described (26-29). For cell proliferation assays, pregnant females were injected intraperitoneally with BrdU (0.1 mg/g body weight), and sacrificed 2 hours later. BrdU-positive cells were detected on paraffin sections with Zymed's BrdU staining kit (Zymed Laboratories, San Francisco, CA). The percentage of BrdU-positive cells was

quantified from at least three animals of each genotype. TUNEL assay was carried out on paraffin sections with In Situ Cell Death Detection Kit TMR Red (Roche, Indianapolis, IN) according to the manufacturer’s instructions.

Analyses of postnatal mice X-ray, µCT, and histomorphometry were performed as previously described (30,31). The thresholds for µCT quantification of trabecular andcortical bone parameters were set at 200/1000 and 250/1000, respectively. µCT analyses of cortical bone parameters were performed on 50 µCT slices (0.8 mm total) from the middle-shaft of femurs; trabecular parameters were assessed in 100 µCT slices (1.6 mm total) immediately below the distal growth plate of the femur.

Metabolic labeling of protein synthesis in primary chondrocytes Mouse primary sternal chondrocytes were isolated and cultured as described previously (32). Isolated chondrocytes were seeded in 6-well plates at 1 x 106 cells/well. After overnight culture, cells were infected with adenovirus expressing either green fluorescence protein or Cre at a MOI of 100 for 72 hours. Chondrocytes were then either trypsinized for cell counting followed by lysis with RIPA buffer, or used directly for metabolic labeling. Metabolic labeling was performed as previously reported (33). The amount of 35 S incorporated into protein was normalized to cell number.

Mouse BMSC cultures and osteoblast differentiation Isolation and culture of mouse bone marrow stromal cells (BMSCs) were described previously (34). Once BMSCs reached confluency at days 7–8, cells were reseeded at 0.6x105 cells/cm2 , and then infected with adenovirus expressing either GFP or Cre at a MOI of 100. After 72 hours of viral infection, BMSCs were cultured in osteogenic media (á-MEM containing 10% FBS, 1% penicillin/streptomycin, 50 ìg/ml L-ascorbic acid and 10 mM â-glycerophosphate) for 7 days (for alkaline phosphatase staining and qPCR analysis) or 14 days (for von Kossa staining). Alkaline phosphatase staining was performed as reported before (35). For von Kossa staining, cells were fixed

in cold methanol for 20 minutes, rinsed with ddH2 O, and then incubated with 5% silver nitrate solution under bright light for 30 minutes.

Western blot and quantitative PCR (qPCR) Total protein was extracted from mouse forelimb buds or cell cultures using RIPA buffer. 30 µg protein samples were subsequently resolved by 10% SDS-polyacrylamide gel electrophoresis and subjected to standard western blot procedures. Antibodies for AKT , pAKT (S473), rictor, P-4EBP1(S65), 4EBP1, P-S6 (S240/244), S6, P-FoxO1(T24)/3a(T32), FoxO3a, P-NDRG1(T346) and â-actin were all purchased from Cell Signaling (Beverly, MA), and were used in 1:1000 dilution. Total RNA was extracted from E12.5 hindlimb bud tissues or cell cultures using RNAeasy mini kit (Qiagen). 1 µg of total RNA was reverse-transcribed to cDNA using iScript Reverse Transcription Supermix (Bio-Rad, Hercules, CA). Quantitative PCR (qPCR) was performed using SYBR green supermix (Bio-Rad, Hercules, CA). Gene expression was first normalized to â-actin, and then normalized to control samples. The primers used in this study are listed in supplemental Table 1.

In vivo tibial axial loading Prior to in vivo loading, WT (n=4) and RiCKO (n=4) were strain gauged on the antero-medial surface, 5 mm proximal to the tibia-fibular junction, in order to apply strain-matched loads. The estimated force-strain relationship for WT mice was ? = 343F+510 and for RiCKO was ? = 552F+784, where F represents the negative compressive forces and ? represents the peak compressive strain generated at the periosteal surface of the postero-lateral region of the tibia.

For in vivo loading, mice were anesthetized (2-3% isofluorane) and their right tibiae subjected to axial compression for 1200 cycles/day (4Hz triangle waveform with 0.1 sec rest-insertion after each cycle) using a materials testing system (Electropulse 1000, Instron, Norwood, MA, USA), as previously described (18,36,37). Mice (n=5/group) were loaded 5 days/week for 2 weeks (study days 1-5 and 8-12). After each loading session, mice received buprenorphine (0.1 mg/kg i.m.) and were returned to

their cages for unrestricted activity. The left tibia served as a non-loaded, contralateral control. To determine changes in bone mass and structure, right and left tibia were scanned before and after loading with VivaCT (VivaCT 40; Scanco, Bruttisellen, Switzerland) with the setting of 70 kV, 114 mA and 21.5 ìm voxel resolution. Analyses of cortical bone parameters were performed on 50 slices (total of 1.075 mm) starting from 5 mm proximal to the distal tibia-fibular junction. For dynamic histomorphometry, fluorochromes were injected intraperitoneally on days 5 (calcein green, 10 mg/kg; Sigma, Saint Louis, MO, USA) and 12 (alizarin complexone, 30 mg/kg; Sigma), and tibiae were harvested on day 17. Dynamic histomorphometric analysis was performed on 30µm sections taken at the mid-diaphysis, 5 mm proximal of the distal tibio-fibular junction, using commercial software (Osteo II, BIOQUANT, Nashville, TN, USA) to determine measures of bone formation. Those measures included mineralizing surface (MS/BS), mineral apposition rate (MAR), and bone-formation rate (BFR/BS) (38). Endocortical (E) and periosteal (P) surfaces were analyzed separately. To assess the local loading effect, relative (r) values were computed for each mouse as loaded (right) minus nonloaded (left).

Statistics All quantitative data are presented as mean ± STDEV with a minimum of three independent samples. Statistical significance is determined by two-tailed Student’s t-test. P-value less than 0.05 is considered statistically significant.

Results mTORC2 promotes embryonic skeletal growth To investigate the role of mTORC2 in skeletal development, we deleted rictor in early limb mesenchymal cells with Prx1-Cre (23). Briefly, Prx1Cre; rictorf/+ male mice were crossed with rictorf/f females to generate Prx1Cre; rictorf/f progenies (hereafter RiCKO). The deletion efficiency was evaluated in the limb buds of E12.5 embryos. Western blot showed that rictor levels and phosphorylation of AKT at S473, a direct readout of mTORC2 activity, were both markedly reduced

in the RiCKO limb buds (Fig. 1A). Similarly, qPCR confirmed that rictor mRNA level was reduced by about 80% in the mutant samples (Fig. S1A).

The RiCKO mice were born at the expected Mendelian ratio with no gross abnormality. However, skeletal staining of the embryos at E18.5 indicated that the limb skeletal elements of the mutant embryos were slightly shorter and thinner than their normal counterparts (Fig. 1B, S1B). Measurements of the stained bones confirmed that the total length (TL) of humerus, femur, radius or tibia was shortened by 10-15% in RiCKO embryos compared to the normal littermates (Fig. 1C, S1C). Similarly, the diaphyseal width of each bone (BW, measured at mid-point of bone shaft) was reduced by 15-25% in the mutants (Fig. 1C, S1C). Interestingly, when the cartilaginous versus bony portion (stained blue and red, respectively) of each element was measured separately, the bone shaft (BL) accounted for the entire length difference between RiCKO and control bones, whereas the cartilage length (CL) was not significantly different between the genotypes (Fig. 1C, S1C). Histology of the humerus corroborated the shortening of bone shaft in the mutant (Fig. 1D). Closer examination of the cartilage did not detect any difference in chondrocyte morphology or organization in either round or columnar zone between the genotypes (Fig. 1E). The cell density within these regions was also normal in the RiCKO embryo (Fig. 1F). Thus, mTORC2 is necessary for the optimal growth of embryonic long bones in both length and width, but does not overtly affect the morphology of growth region cartilage.

We next examined other aspects of chondrocyte biology. BrdU labeling experiments detected no defect in cell proliferation within either round or columnar chondrocytes of the proximal humerus in RiCKO embryos at E15 (Fig. 2A, B). The normal cell density within the matrix-rich cartilage as shown above suggested that mTORC2 was unlikely to play an important role in the regulation of protein synthesis. To test this notion directly, we performed metabolic labeling experiments to measure the rate of protein synthesis in primary chondrocytes with or without rictor deletion. Chondrocytes isolated from newborn rictorf/f mice were infected with adenovirus expressing either Cre or GFP as control. Western blot analyses confirmed effective Cre-mediated deletion of rictor, and the expected suppression of AKT

phosphorylation at S473 (P-AKT(S473)), a known target of mTORC2 (Fig. 2C). In contrast, phosphorylation of 4EBP1 or S6, both targets of mTORC1, was not affected (Fig. 2C). Importantly, rictor-deficient chondrocytes exhibited a normal protein synthesis rate when compared to the control (Fig. 2D). Taken together, these data indicate that mTORC2 is dispensable for chondrocyte proliferation and protein synthesis.

mTORC2 enhances chondrocyte hypertrophy The size deficit of long bones in late embryos could result from a smaller cartilage anlage at an earlier stage of embryogenesis. To explore this possibility, we examined embryos at E13.5 when the limb skeletal elements had not yet initiated bone formation, and chondrocytes were yet to undergo overt hypertrophy. At this stage, the humerus in RiCKO embryos had normal length but reduced width than the control (Fig. 3A, B). Therefore, whereas the width deficit in RiCKO bones began with the cartilage anlage, the length deficiency had a later origin.

Chondrocyte hypertrophy is a major driving force for the longitudinal growth of long bones. We therefore examined the status of hypertrophy in RiCKO versus control embryos. At E15, histology showed that the hypertrophic zone was notably shorter in the mutant than the control embryo (Fig. 3C, upper panels). In situ hybridization for Col10á1, a specific marker for early and intermediate hypertrophic chondrocytes, revealed that in the control embryo, cells within the central hypertrophic domain no longer expressed the marker, indicative of their terminal hypertrophic stage (Fig. 2C, lower panels). In contrast, all hypertrophic chondrocytes in the RiCKO embryo continued to express high levels of Col10á1 at E15. Consistent with a delay in terminal hypertrophy (eventually leading to apoptosis) in the RiCKO embryo, TUNEL assays detected apoptosis of hypertrophic chondrocytes only in the wild-type littermate, even though apoptosis occurred within the perichondrium in both genotypes (Fig. 3D). The delay in terminal hypertrophy of chondrocytes is expected to diminish the longitudinal growth of long bones in RiCKO embryos.

mTORC2 augments bone formation In addition to the smaller long bones, whole-mount skeletal staining also revealed a deficit in ossification of the skull in RiCKO embryos. This was most evident at E16.5 and E17.5 but remained detectable at E18.5 (Fig. 4A, S1B). Because the skull is targeted by Prx1-Cre, this finding is consistent with a direct role of rictor in bone formation. To ensure that rictor deletion disrupted normal mTORC2 signaling in osteoblasts, we cultured calvarial osteoblasts from neonatal rictorf/f mice and infected them with adenovirus expressing Cre or GFP. Western blot analyses showed that rictor deletion markedly suppressed P-Akt(S473) and P-NDRG1(T346), both readouts for mTORC2 activity (Fig. 4B). On the other hand, phosphorylation of FoxO1/3a was not notably impaired by rictor deletion. To test the function of mTORC2 directly, we performed osteoblast differentiation assays in vitro, with bone marrow stromal cells (BMSC) isolated from postnatal mice. We confirmed that after seven days of culture in osteogenic media, BMSC from wild type mice markedly upregulated osteocalcin (also known as Bglap), a marker for mature osteoblasts. Interestingly, upon differentiation BMSC also increased the expression of mTORC2 components, including rictor, Mlst8, Sin1 and mTOR (Fig. 4C). We then isolated BMSC from rictorf/f mice and infected them with adenovirus expressing Cre or GFP. After seven days of culture in osteogenic media, rictor-deficient cells exhibited a markedly lower level of alkaline phosphatase (AP) activity (Fig. 4D, upper panels). After fourteen days, the rictor-deficient cells failed to undergo mineralization that was readily detectable by von Kossa staining in the normal culture (Fig. 4D, lower panels). Furthermore, molecular analyses by qPCR in cells after seven days of differentiation confirmed a markedly lower mRNA level of Alpl, along with several other osteoblast markers including Sp7, Ibsp, and Bglap in the rictor-deficient cells, even though Runx2, Col1á1 or Twist1, which encodes an antagonist of Runx2, were relatively normal (Fig. 4E). In addition, the rictordeficient cells expressed significantly less Rankl (also known as Tnfsf11), even though other regulators of osteoclastogenesis such as Opg (also known as Tnfrsf11b) and M-CSF (also known as Csf1) were relatively normal (Fig. 4F). These results indicate that mTORC2 signaling directly promotes osteoblast differentiation, and may indirectly stimulates osteoclastogenesis.

The finding above predicts that rictor deletion may reduce bone mass in postnatal mice. To investigate this possibility, we examined RiCKO versus littermate control mice at six weeks of age. X-ray images showed that both the tibia and the femur remained shorter in RiCKO mice (Fig. 5A, B). Moreover, ìCT imaging revealed that both outer and inner diameters of the femur diaphysis were smaller in the mutant than the wild type (Fig. 5C). Quantitative ìCT analyses in the femur showed that total cross-section area (Tt. Ar), cortical bone area (Ct. Ar) and cortical thickness (Ct. Th) were all reduced in the mutant by 27%, 32% and 22%, respectively, even though the percentage of cortical bone area over total cross-section area (Ct.Ar/Tt.Ar) was not altered (Fig. 5D). In contrast to the cortical bone, the trabecular bone parameters did not change in the mutant (Fig. 5E, F). The decrease in cortical bone mass was likely due to impaired osteoblast activity, as dynamic histomorphometry demonstrated a significant decrease in mineral apposition rate (MAR), but no change in the percentage of mineralizing surface over total bone surface (MS/BS), at both the periosteal and endosteal surface in femora of RiCKO mice (Fig. 6A). As a result, the bone formation rate (BFR/BS) was significantly reduced at both surfaces in the mutant. A similar decrease in MAR and BFR/MS was observed in trabecular bone with no change in MS/BS or osteoblast number, but this was concurrent with a decrease in osteoclast number, likely explaining the relatively normal trabecular bone mass (Fig. 6B,C). Histology also revealed that the RiCKO mice possessed less bone marrow fat at six weeks of age (Fig. 6D). In keeping with fewer osteoclasts in trabecular bone, the serum level of CTX-I, a cleavage product of type I collagen reflecting total bone resorption activity, was reduced in the RiCKO mice (Fig. 6E). Consistent with an overall decrease in bone formation, the mutant mice exhibited a significant reduction in the serum level of amino-terminal propeptide of type I procollagen (PINP) (Fig. 6F). Overall, mTORC2 signaling is necessary for optimal bone accrual in postnatal mice.

mTORC2 mediates bone anabolic response to mechanical stress The suboptimal bone accrual rate in postnatal RiCKO mice implies that the mutant bones may exhibit a lesser anabolic response to mechanical stress, a primary stimulus for bone formation. To test this hypothesis, we subjected the tibia of four-month-old mice to axial loading. In vivo ìCT analyses before

loading indicated that the cortical bone size and thickness remained smaller in the RiCKO mice at this age (Fig. 7A). As expected from the smaller size, strain gauge analyses revealed that the mutant tibia required less force to produce the same strain at the mid-diaphyseal surface (Fig. S3). In this study, we selected a peak compression strain of -2200 µå, which required an axial loading of 5.6 N in the RiCKO mice but 8 N in the wild-type littermates. Dynamic histomorphometry showed that loading similarly increased MS/BS at the periosteal surface in both genotypes (minimal changes at the endosteal surface) (Fig. 7B). However, loading induced a significantly lesser MAR response at the periosteal surface in the RiCKO mice than the wild-type littermates (Fig. 7C). Consequently, loading-induced BFR at the periosteal surface was significantly reduced in the RiCKO mice (Fig. 7D). Consistent with the lesser BFR response, post-loading ìCT scan revealed that the loading-induced increase in cortical tissue volume or bone volume was reduced in the RiCKO mice, although the latter parameter did not reach statistical significance (data not shown). Overall, mTORC2 activity in bone is necessary for the normal anabolic response to mechanical stimuli.

Discussion We have uncovered rictor as an important regulator of limb skeletal growth and bone formation in mice. Targeted deletion of rictor, an essential component of mTORC2, in the limb mesenchyme, delays chondrocyte hypertrophy and shortens the long bones. Moreover, rictor is necessary for optimal cortical bone accrual in postnatal life, both under basal conditions and in response to experimental mechanical loading. Thus, rictor-mediated mTORC2 signaling plays distinct roles in both chondrocytes and osteoblasts.

It is worth noting that rictor deletion also reduces the width of the long bones. The width deficit is noticeable in the cartilage anlage prior to the onset of hypertrophy, and persists throughout the life of RiCKO mice. How mTORC2 controls cartilage width is not understood at present, but it does not seem to involve changes in the overall chondrocyte proliferation or survival. It is possible that mTORC2

affects the dimension of the initial mesenchymal condensations, but future studies are necessary to test this possibility. Moreover, we cannot rule out that rictor may perform other functions beyond mTORC2 to contribute to the phenotype.

The study identifies distinct roles for mTORC2 from mTORC1 in the skeleton. Previously, deletion of the mTORC1-specific raptor with Prx1-Cre caused severe skeletal dysplasia, mainly due to impaired protein synthesis in chondrocytes (39). The severity of cartilage phenotype and neonatal lethality in those mice precluded a detailed analysis of bone formation. Here we show that loss of rictor and mTORC2 signaling only modestly affects cartilage development. Studies of the postnatal mice allowed us to uncover the role of mTORC2 in osteoblasts. The precise role of mTORC1 in bone formation warrants further study. We previously showed that mTORC1 contributed to the hyperactive osteoblast activity in response to Wnt overexpression, but appeared to be dispensable for basal bone formation when raptor was deleted for three weeks beginning at one month of age (40). While this paper was under preparation, others reported that osteoblast-specific deletion of mTOR decreased trabecular bone mass, whereas hyperactivation of mTORC1 through TSC2 deletion led to excessive production of immature osteoblasts (41). Future studies to delete mTORC1 specifically in the osteoblast lineage will provide additional insights about its role in bone formation.

We find that rictor is necessary for normal osteoblast activity. Although the bone formation rate is lower in both cortical and trabecular compartments, reduced bone mass is only seen in the cortex of the mutant mice. This is likely due to the concurrent decrease in osteoclast number in the trabecular bone. The osteoclast defect appears to be secondary to rictor deficiency in the osteoblast lineage, as the rictor-deficient cells express less Rankl. Thus, rictor in osteoblast-lineage cells not only enhances osteoblast activity directly, but also promotes osteoclastogenesis indirectly.

It is somewhat surprising that osteoblast numbers were not reduced in the RiCKO mice, given that rictor-deficient BMSC were impaired in osteoblast differentiation in vitro. It is possible that other

mechanisms may compensate for mTORC2 deficiency in vivo. Our in vitro results are consistent with those from a previous study where rictor was knocked down with siRNA in marrow-derived mesenchymal stem cells (9). In that study, the authors suggested that mTORC2 regulates cell-lineage selection in favor of osteoblasts over adipocytes. Such a model would predict greater marrow adiposity in the RiCKO mice, but we observed fewer bone marrow adipocytes instead. Thus, mTORC2 appears to be required for proper bone marrow adipogenesis in vivo.

A major finding of the present work is mTORC2 as a necessary mediator for the anabolic response to mechanical loading. The adaptive response of bone to mechanical loading has been long recognized. Loss of skeletal loading due to intrauterine neuromuscular diseases in humans is known to diminish bone size and strength at birth (42). Conversely, the playing arms of tennis players develop larger bones than the nonplaying arms (43). More recent studies in rodents have identified Wnt signaling as an important mechano-responsive mechanism (44). Mice lacking the Wnt co-receptor Lrp5 are severely impaired in bone anabolic response to loading (45). Furthermore, loading markedly suppresses production of the secreted Wnt antagonist Sost by osteocytes (21). Finally, we have recently shown that Wnt-Lrp5 signaling activates mTORC2 to promote glycolysis (8). Thus, the Sost-Wnt-mTORC2 signaling axis may be a major mechanism responsible for the bone anabolic response to loading. Further elucidation of the molecular events downstream of mTORC2 may open avenues for developing novel anabolic strategies mimicking mechanical loading.

Acknowledgement We thank Dr. Jeffery Arbeit (Washington University) for the Rictorf/f mice. The work is supported by NIH grants R01 AR060456, R01 AR055923 (FL), R01 047867 (MS), T32 AR060719 (NH) and P30 AR057235 (Washington University Musculoskeletal Research Center).

Figure legends Figure 1. mTORC2 promotes embryonic skeletal growth (A) Western blot analyses of E12.5 forelimbs. (B) Limb skeleton at E18.5. Cartilage stained blue; bone collar stained red. FL: forelimb; HL: hindlimb. (C) Relative lengths of long bones. BL: bone collar length; CL: cartilage length; TL: total length; BW: bone width. N=4 for each genotype; *: p

mTORC2 signaling promotes skeletal growth and bone formation in mice.

Mammalian target of rapamycin (mTOR) is an evolutionarily conserved serine/threonine kinase controlling many physiological processes in mammals. mTOR ...
1009KB Sizes 0 Downloads 6 Views