mTOR inhibition rescues osteopenia in mice with systemic sclerosis

Because Fbn1 gene mutation leads to significant loss of bone volume and increase in BM adipocytes in B6.Cg-Fbn1^Tsk/J (Fbn1^+/−) mice (Barisic-Dujmovic et al., 2007), we hypothesize that Fbn1 deficiency may reduce osteogenic differentiation of BMMSCs and elevate their adipogenic differentiation. To test this hypothesis, we confirmed that Fbn1 deficiency resulted in an osteopenia phenotype in Fbn1^+/− mice. MicroCT and histological analysis showed that BMD, bone volume versus total volume (BV/TV), and distal femoral trabecular bone structure of Fbn1^+/− mice were markedly decreased compared with the WT littermates (Fig. 1, A–D). Histomorphometric analysis revealed that the numbers of both osteoblasts and osteoclasts in the femur of Fbn1^+/− mice were significantly reduced in comparison with the WT group by alkaline phosphatase (ALP) immunohistochemical (IHC) staining and tartrate-resistant acid phosphate (TRAP) staining, respectively (Fig. 1, E and F). The level of soluble receptor activator of nuclear factor κB ligand (sRANKL), but not osteoprotegerin (OPG), was significantly reduced (Fig. 1, G and H). In addition, we examined the in vivo function of osteoclasts and found that the serum type I collagen cross-linked telopeptide (CTX) level was significantly decreased in Fbn1^+/− mice (Fig. 1 I). These data imply that the loss of bone volume in Fbn1^+/− mice may be mainly associated with an insufficient bone formation. To examine whether Fbn1 deficiency affects the stem cell properties of BMMSCs, we isolated BMMSCs (Fig. 2, A–C) to show that the number of colony forming unit-fibroblasts (CFU-Fs) was significantly reduced and the expression level of FBN1 was decreased in Fbn1^+/− BMMSCs compared with the WT control (Fig. 1, J and L). The osteogenic differentiation capacity of Fbn1^+/− BMMSCs was decreased, as indicated by reduced mineralized nodule formation and expression of the osteogenic regulators runt-related transcription factor 2 (RUNX2), ALP, and osteocalcin (OCN; Fig. 1, K and L). Using an established in vivo BMMSC implantation assay, in which 4 × 10⁶ BMMSCs with carrier hydroxyapatite tricalcium phosphate (HA/TCP) particles were subcutaneously implanted into immunocompromised mice, we showed that BMMSCs derived from Fbn1^+/− mice generated less bone structure than the WT group at 8 wk after implantation (Fig. 1 M). To further confirm the role of Fbn1 in BMMSC differentiation, we used siRNA to silence Fbn1 gene expression in BMMSCs (Fig. 2 D) and found a significantly reduced osteogenic differentiation, as assessed by alizarin red staining to show reduced mineralized nodule formation, immunoblotting analysis to show decreased expression of the osteogenic markers RUNX2, ALP, and OCN, and in vivo BMMSC implantation assay to show reduction in new bone formation (Fig. 1, N–P). To further confirm reduced osteogenesis in Fbn1^+/− mice, we performed a calcein labeling assay to show that the bone turnover rate was decreased in Fbn1^+/− mice (Fig. 1 Q). In contrast, Oil red O staining showed the number of adipocytes in Fbn1^+/− BM was markedly increased when compared with the WT littermates (Fig. 2 E). To confirm that Fbn1 deficiency results in elevated adipogenesis, Fbn1^+/− BMMSCs were found to exhibit a significantly elevated number of Oil red O–positive cells and up-regulation of the adipogenic regulators peroxisome proliferator-activated receptor γ2 (PPARγ2) and lipoprotein lipase (LPL) under adipogenic culture conditions (Fig. 2, F and G). Furthermore, we showed that Fbn1-silenced BMMSCs by siRNA had a significantly increased number of Oil red O–positive cells and up-regulation of PPARγ2 and LPL (Fig. 2, H and I). These data suggest that Fbn1 governs osteogenic/adipogenic lineage differentiation in BMMSCs.

Because previous studies showed that Fbn1 mutation induced T_H2 infiltration with a high level of IL4 (Gabrielli et al., 2009), we used IHC staining and ELISA to confirm a higher number of IL4-positive cells in Fbn1^+/− mouse BM and an elevated IL4 level in the peripheral blood (Fig. 3, A and B). In addition, we confirmed that Fbn1^+/− mice showed significantly increased levels of T_H2 cells (Fig. 3, C-F). Quantitative PCR and ELISA showed that WT and Fbn1^+/− BMMSCs expressed similar low levels of IL4 (Fig. 3, G and H). Next, we examined whether activation of IL4 downstream signaling contributed to reduced osteogenesis and elevated adipogenesis in Fbn1^+/− mice. We used immunofluorescence staining to confirm that BMMSCs coexpressed the MSC marker CD73 with IL4Rα (Fig. 3 I). Compared with control BMMSCs, Fbn1^+/− BMMSCs show up-regulated expression of IL4Rα (Fig. 3 J), which was further confirmed by qPCR analysis (Fig. 3 K). Western blotting also showed that Fbn1 deficiency up-regulated IL4Rα downstream signaling phosphoinositide 3-kinase (PI3K)–p110, phosphorylated Akt (p-Akt), and phosphorylated mTOR (p-mTOR) in Fbn^+/− BMMSCs (Fig. 3 J). To verify that IL4Rα/mTOR signaling contributes to BMMSC osteo/adipo-lineage selection, we examined whether knockdown of the IL4Rα/mTOR cascade could rescue the altered lineage differentiation in Fbn1^+/− BMMSCs. Having confirmed that shRNAs could knock down Il4rα and mTOR (Fig. 3 L) and that rapamycin treatment could inhibit mTOR signaling in a dose-dependent manner (Fig. 3 M), we showed that Il4rα and mTOR shRNAs, as well as rapamycin treatment, significantly improved osteogenic differentiation of Fbn1^+/− BMMSCs, as indicated by alizarin red staining to show increased mineralized nodule formation, Western blotting to show elevated expression of the RUNX2, ALP, and OCN, and in vivo BMMSC implantation assay to show increased new bone formation at 8 wk after implantation (Fig. 4, A–C). Conversely, the elevated adipogenic differentiation in Fbn1^+/− BMMSCs was significantly reduced in the shRNAs or rapamycin treatment groups, as indicated by a decreased number of Oil red O–positive cells and down-regulation of the PPARγ2 and LPL (Fig. 4 D). To avoid off-target effect, we used IL4Rα neutralizing antibody to block IL4 signaling and found decreased expression of p-mTOR and elevated expression of RUNX2 in Fbn1^+/− BMMSCs when compared with control group (Fig. 4 E). We next revealed that mTOR regulated osteogenic differentiation of Fbn1^+/− BMMSCs through its downstream signaling phosphorylated P70 ribosomal S6 protein kinase (P70S6K). Western blot analysis showed that p-P70S6K expression level was increased in Fbn1^+/− BMMSCs (Fig. 4 F). When p-mTOR expression was knocked down by Il4rα and mTOR shRNAs, or rapamycin treatment, in Fbn1^+/− BMMSCs, a corresponding down-regulation of p-mTOR and p-P70S6K was revealed by Western blot analysis (Fig. 3 N). Although knockdown of P70s6k expression by siRNA in Fbn1^+/− BMMSCs resulted in up-regulation of RUNX2 expression (Fig. 4 G), the expression levels of mTOR was unaffected (Fig. 4 H). To avoid off-target effects, a second P70s6k siRNA was used to show elevated expression of RUNX2 (Fig. 4 I). In addition, the osteogenic differentiation was markedly increased in P70s6k knockdown Fbn1^+/− BMMSCs, as indicated by elevated mineralized nodule formation (Fig. 4 J) and expression of the RUNX2, ALP, and OCN (Fig. 4 K). In contrast, the number of Oil red O–positive cells (Fig. 4 L) and the expression levels of the adipogenic genes were significantly down-regulated (Fig. 4 M), suggesting that mTOR regulates P70S6K to control RUNX2-mediated osteogenic differentiation and PPARγ2-mediated adipogenic differentiation.

To further confirm that the IL4–IL4Rα pathway activates mTOR signaling, we performed a dosage and kinetics analysis to show that IL4 at a concentration of 50–100 ng/ml is able to activate mTOR signaling in WT BMMSCs and the activation of mTOR signaling can occur as earlier as 6 h after IL4 treatment (Fig. 5, A and B). When treated with IL4, Fbn1^+/− BMMSCs exhibited a more significant increase in the expression of p-mTOR than the control BMMSCs (Fig. 5 A). Western blot analysis confirmed up-regulation of mTOR downstream signaling including PI3K, p-Akt, p-mTOR, and p-P70S6K (Fig. 5 C). Furthermore, we showed that IL4 treatment–activated mTOR signaling inhibited osteogenic differentiation of WT BMMSCs, as indicated by reduced mineralized nodule formation, decreased expression of the RUNX2, ALP, and OCN, and reduced bone formation when subcutaneously implanted into immunocompromised mice (Fig. 5, D–F), along with elevated adipogenic differentiation, as indicated by the increased number of Oil red O–positive cells and the expression of PPARγ2 and LPL (Fig. 5, G and H). These data suggest that activation of the IL4/IL4Rα/mTOR/P70S6K cascade contributes to osteogenic deficiency of Fbn1^+/− BMMSCs and that blockade of mTOR signaling may be a promising therapeutic approach for rescuing osteogenic deficiency in Fbn1^+/− mice.

Because mTOR signaling involves two complexes, namely mTOR complex 1 (TORC1) and complex 2 (TORC2), we next examined which mTOR complex plays a functional role in regulating Fbn1^+/− BMMSC differentiation. We found that Fbn1^+/− BMMSCs expressed elevated levels of mTOR/Raptor/TORC1 and mTOR/Rictor/TORC2 when compared with the control group (Fig. 5 I). When using siRNA to knockdown either TORC1 or TORC2, the expression of the osteogenic marker Runx2 was up-regulated in Fbn1^+/− BMMSCs (Fig. 5 J). Knockdown of TORC2 resulted in down-regulation of both Raptor and Rictor, whereas knockdown of TORC1 only resulted in down-regulation of Raptor (Fig. 5 K), suggesting that TORC2 may act upstream of TORC1. To avoid off-target effects, we used secondary TORC1 and TORC2 siRNAs to show they are able to elevate expression of RUNX2 in Fbn1^+/− BMMSCs (Fig. 5 L). Because Akt is a downstream target of TORC2, we examined the expression levels of p-Akt after the knockdown of either TORC1 or TORC2 and confirmed that TORC2 acted upstream of TORC1 through the Akt cascade in Fbn1^+/− BMMSCs (Fig. 5 M).

It was reported that FBN1 interacts with LTBP-1 to bind TGF-β (Charbonneau et al., 2004; Ramirez and Sakai, 2010). As such, Fbn1 deficiency causes an increased release of TGF-β to the skin and lung (Gabrielli et al., 2009; Holm et al., 2011). We examined the level of TGF-β in Fbn1^+/− mouse BM and Fbn1^+/− BMMSC culture medium and found that both BM and medium contained elevated levels of TGF-β, as assessed by ELISA assay (Fig. 6 A). Next, we used Western blotting to confirm that TGF-β downstream signaling p-SMAD3 and SMAD-associated transcription factor SP1 (Docagne et al., 2004; Jungert et al., 2006) were significantly increased in Fbn1^+/− BMMSCs compared with the control BMMSCs (Fig. 6 B). To further confirm that Fbn1 deficiency-mediated IL4Rα expression occurs through TGF-β release, we treated BMMSCs with recombinant TGF-β1 or Fbn1 siRNA, followed by TGF-β neutralizing antibody blocking. Western blot analysis revealed that both TGF-β1 and Fbn1 siRNA treatment increased IL4Rα levels, which could be blocked by TGF-β neutralizing antibody treatment (Fig. 6, C–F). To further confirm that TGF-β signaling activates IL4Rα expression, we used siRNA to block canonical TGF-β pathway and then examine the expression level of IL4Rα in Fbn1^+/− BMMSCs. Western blot analysis showed that IL4Rα was significantly decreased in Fbn1^+/− BMMSCs after Smad2/3 siRNA treatment (Fig. 6 G), indicating that TGF-β/SMAD3 signaling may directly regulate IL4Rα expression. To determine whether increased expression of IL4Rα in Fbn1^+/− BMMSCs is attributed to the TGF-β/SMAD3 signaling, we generated an Il4rα promoter reporter construct in which the defined region of the Il4rα promoter and flanking region were placed upstream of a reporter gene encoding firefly luciferase. We used BIOBASE biological databases to search the Il4rα promoter sequence and found two SP1 and one STAT6 candidate sites, which were closely matching with the SMAD3-associated SP1 and IL4Rα downstream STAT6 transcription factors consensus targets (Fig. 6 H). We next demonstrated that Il4rα promoter activity was markedly induced in both Fbn1^+/− and TGF-β–treated BMMSCs. When Fbn1^+/− BMMSCs were transfected with reporter vector, the luciferase assay showed a significantly increased Il4rα promoter activity. Introduction of SP1 mutated reporter vector markedly diminished the expression of the Il4rα-luciferase reporter in both Fbn1^+/− and TGF-β–treated BMMSCs, indicating the direct initiation of IL4Rα expression by TGF-β–SMAD3–SP1 cascades (Fig. 6 H). We next determined whether SP1 regulates Il4rα promoter in BMMSCs. Using chromatin immunoprecipitation (ChIP), the SP1 binding consensus sequence within the promoter region was examined to confirm its ability to recruit SP1. As expected, SP1-bound DNA at the candidate site was significantly enriched in TGF-β–treated BMMSCs and Fbn1^+/− BMMSCs (Fig. 6 I), suggesting that TGF-β may act as an initiator to elevate IL4Rα expression. In parallel, we used a ChIP-qPCR assay and found that STAT6 directly bound to the predicted promoter region of Il4rα and that STAT6 recruitment was enriched at the Il4rα promoter in Fbn1^+/− or TGF-β–treated BMMSCs (Fig. 6 I). Western blot analysis further confirmed that an IL4Rα-JAK1-STAT6 cascade was activated in Fbn1^+/− BMMSCs, as shown by up-regulation of p-JAK1 and p-STAT6 expression (Fig. 6 J). These results prompted us to examine whether STAT6 induced IL4Rα expression at the transcriptional level using a promoter reporter assay. Control BMMSCs, TGF-β–treated BMMSCs, and Fbn1^+/− BMMSCs were transfected with a reporter vector or STAT6 candidate site mutated reporter vector, followed by luciferase assay. TGF-β–treated and Fbn1^+/− BMMSCs showed increased promoter activity in comparison to control BMMSCs. Introduction of STAT6 mutated reporter vector markedly diminished the expression of the Il4rα-luciferase reporter in Fbn1^+/− BMMSCs, but not TGF-β–treated BMMSCs. Thus, TGF-β may act as an initiator to elevate IL4Rα expression, and TGF-β–activated IL4Rα is independent from IL4Rα-STAT6 cascades (Fig. 6 H). These results prompted us to examine whether TGF-β and IL4 synergistically activate the mTOR signaling in BMMSCs. We found that TGF-β, but not IL4, treatment elevated the expression levels of IL4Rα in BMMSCs by Western blot analysis (Fig. 6, K and L). Although TGF-β and IL4 treatments each elevated expression levels of p-mTOR in BMMSCs, combinatorial treatment of TGF-β and IL4 resulted in a marked increase in p-mTOR expression (Fig. 6, K and L). These results indicate that the Fbn1 deficiency-induced TGF-β–SMAD3–SP1/IL4Rα cascade synergistically enhances mTOR signaling with STAT6 signaling in Fbn1^+/− BMMSCs at the transcriptional level, which leads to lineage alteration of Fbn1^+/− BMMSC.

Because IL4Rα/mTOR alters BMMSC lineage commitment, depletion of mTOR expression in BMMSC lineage may serve as a key approach to recover Fbn1 deficiency-induced osteopenia. To achieve such a tissue-specific KO, we first generated Prrx1-Cre;Il4rα^f/-;Fbn1^+/− triple conditional KO (TKO) mice to KO IL4Rα in Fbn1^+/− BMMSCs (Fig. 7 A). Il4rα^f/− littermates were used as normal control, and Il4rα^f/−;Fbn1^+/− (DKO) littermates were used as osteopenia models (Fig. 7 A). The efficacy of Cre-medicated deletion of floxed alleles was shown by qPCR analysis (Fig. 7 B). MicroCT analysis showed that TKO mice have significantly increased BMD compared with DKO mice, indicating that Il4rα conditional KO rescued BMD in Fbn1 deficiency–induced osteopenia (Fig. 7 C). Histological analysis showed that trabecular bone volume in TKO mice was increased compared with the DKO mice (Fig. 7 D). The numbers of adipocytes in TKO mouse BM were significantly reduced when compared with the DKO mice, as assessed by Oil red O staining (Fig. 7 D). To investigate the role of IL4Rα in the differentiation of Fbn1^+/− BMMSCs, we isolated BMMSCs from DKO mice and knocked down Il4rα using an adenovirus expressing Cre with GFP (Ad-Cre-DKO-BMMSCs) or an adenovirus expressing GFP only (Ad-GFP-DKO-BMMSCs) as a control. Ad-Cre-DKO-BMMSCs showed significantly increased osteogenic differentiation, as assessed by alizarin red staining indicating elevated mineralized nodule formation (Fig. 7 E); Western blotting indicating rescued expression of the mTOR/osteogenic cascade genes p-mTOR, RUNX2, ALP, and OCN (Fig. 7 F); and in vivo implantation indicating elevated bone formation capacity compared with the Ad-GFP-DKO-BMMSCs (Fig. 7 H). To avoid off-target effect, we used IL4Rα neutralizing antibody to block IL4 signaling and found elevated expression of RUNX2 in DKO BMMSCs (Fig. 7 G). In contrast, Ad-Cre-DKO-BMMSCs showed a significant decrease in adipogenic differentiation compared with Ad-GFP-DKO-BMMSCs, as shown by a decreased number of Oil red O–positive cells and down-regulation of the adipogenic genes PPARγ2 and LPL (Fig. 7, I and J).

We next generated Prrx1-Cre;Mtor^f/+;Fbn1^+/− TKO mice to KO mTOR in Fbn1^+/− BMMSCs (Fig. 8 A). Mtor^f/+ littermates were used as normal control, and Mtor^f/+;Fbn1^+/− (DKO) littermates were used as osteopenia models (Fig. 8 A). The efficacy of Cre-medicated deletion of floxed allele was shown by qPCR analysis (Fig. 8 B). MicroCT analysis showed that Mtor conditional KO rescued BMD in Fbn1 deficiency–induced osteopenia (Fig. 8 C). Histological analysis showed that trabecular bone volume in TKO mice was increased compared with the DKO mice (Fig. 8 D). The numbers of adipocytes in TKO mouse BM were significantly reduced when compared with the DKO mice, as assessed by Oil red O staining (Fig. 8 D). To investigate the role of mTOR in the differentiation of Fbn1 deficient BMMSCs, we isolated BMMSCs from DKO mice and knocked down Mtor using an adenovirus expressing Cre with GFP or an adenovirus expressing GFP only as a control. Ad-Cre-DKO-BMMSCs showed significantly increased osteogenic differentiation, as assessed by alizarin red staining indicating elevated mineralized nodule formation (Fig. 8 E); Western blot analysis indicating rescued expression of the osteogenic cascade genes RUNX2, ALP, and OCN (Fig. 8 F); and in vivo implantation indicating elevated bone formation capacity compared with the Ad-GFP-DKO-BMMSCs (Fig. 8 G). In contrast, Ad-Cre-DKO-BMMSCs showed a significant decrease in adipogenic differentiation compared with Ad-GFP-DKO-BMMSCs, as shown by a decreased number of Oil red O–positive cells and down-regulation of the adipogenic genes PPARγ2 and LPL (Fig. 8, H and I).

It has been shown that SP7-expressing osteoblastic progenitors have the ability to control osteoblastic/adipogenic lineage switch (Zhang et al., 2011; Song et al., 2012). To examine whether knocking down mTOR signaling in the preosteoblastic stage can recover the lineage alteration and osteopenia phenotype in Fbn1^+/− mice, we generated a tissue-specific KO, SP7-Cre;Mtor^f/+;Fbn1^+/− (TKO) conditional Mtor KO mice (Fig. 9 A). Floxed Mtor littermates (Mtor^f/+) were used as normal control and Mtor^f/+;Fbn1^+/− (DKO) littermates were used as osteopenia models (Fig. 9 A). The efficacy of Cre-medicated deletion of floxed allele was shown by qPCR analysis (Fig. 9 B). MicroCT analysis and histological analyses showed that TKO mice have significantly increased BMD and trabecular bone volume compared with DKO mice (Fig. 9 C), suggesting that Mtor conditional KO in the preosteoblastic lineage rescued osteopenia in Fbn1-deficient mice. The number of adipocytes in TKO mouse BM was significantly reduced when compared with the DKO mice, as assessed by Oil red O staining (Fig. 9 C). These data prompted us to use a drug, specifically rapamycin, to attempt to ameliorate the disease phenotype in Fbn1^+/− mice.

Next, we examined whether rapamycin treatment rescued the osteopenia phenotype in Fbn1^+/− mice. Rapamycin was intraperitoneally administered to Fbn1^+/− mice at 8 or 6 wk of age for 14 or 28 consecutive days, respectively. The samples were harvested at 10 wk of age for further evaluation (Fig. 10 A). Histological and microCT analysis showed that rapamycin treatment elevated BMD and trabecular bone volume in Fbn1^+/− mice (Fig. 10 B). The number of adipocytes in BM of rapamycin-treated Fbn1^+/− mice was significantly reduced when compared with untreated Fbn1^+/− mouse BM, as assessed by Oil red O staining (Fig. 10 B). BMMSCs isolated from rapamycin-treated Fbn1^+/− mice showed significantly increased CFU-F number compared with the untreated group (Fig. 10 C). Osteogenic differentiation of BMMSCs from rapamycin-treated Fbn1^+/− mice was markedly improved when compared with the untreated Fbn1^+/− BMMSCs, as analyzed by alizarin red staining to show elevated mineralized nodule formation and Western blot analysis to show decreased expression of p-mTOR and p-P70S6K and increased expression of the RUNX2, ALP, and OCN (Fig. 10, D and E). In contrast, BMMSCs isolated from rapamycin-treated Fbn1^+/− mice showed a significant decrease in adipogenic differentiation compared with the untreated Fbn1^+/− BMMSCs, as shown by decreased number of Oil red O–positive cells and down-regulation of the PPARγ2 and LPL (Fig. 10, F and G).

FBN1 is essential for the formation of elastic fibers or microfibrils that provide strength and flexibility to connective tissue. Normally, FBN1 is abundant in the connective tissue of the aorta, ligature of the human eye lens, bones, and lungs. Mutation of the fibrillin gene causes SSc and Marfan’s syndrome, in which bone and connective tissue disorders are observed. The bone defects in SSc patients were considered to be an insignificant clinical symptom; however, recent emerging evidence shows that bone resorption may be a common clinical manifestation (Atteritano et al., 2013; Kilic et al., 2013; Omair et al., 2013). In the present study, we identified a marked osteopenia phenotype in Fbn1 mutant-induced SSc mice. Recently, it was suggested that FBN1 and FBN2 might differentially regulate endogenous BMP and TGF-β activity in osteoblasts to affect bone growth and development (Nistala et al., 2010). In this study, we reveal that Fbn1 deficiency results in an increased level of TGF-β in BM, which, in turn, initiates a cascade in which IL4Rα is up-regulated in BMMSCs via SMAD3/SP1 binding to Il4rα promoter, leading, as a consequence, to the activation of mTOR-P70S6K signaling to directly suppress osteogenesis via RUNX2 and promote adipogenesis via PPARγ2, respectively. Osteogenic differentiation deficiency in Fbn1^+/− BMMSCs may contribute to an osteoporotic phenotype, such as low BMD or reduced trabecular bone structure, in Fbn1^+/− SSc mice. However, inhibition of mTOR signaling using rapamycin treatment appears to be a promising approach to rescue osteoporotic disorders in the SSc mice.

It is known that mTOR signaling positively regulates expression of the adipogenic gene PPARγ2, which is critical for adipogenesis of BMMSCs (Kim and Chen, 2004; Yu et al., 2008; Zhang et al., 2009). However, controversial findings were reported regarding mTOR signaling in osteogenesis in that mTOR may be required for osteoblast proliferation, which is involved in several signaling pathways through IL6 (Kozawa et al., 2001; Takai et al., 2007; Takai et al., 2008). In contrast, rapamycin may either inhibit (Shoba and Lee, 2003) or stimulate (Lee et al., 2010b; Martin et al., 2010) osteogenesis, depending on cell type or differentiation stages. Recently, it has been reported that bone matrix secretes IGF-1 (insulin-like growth factor-1) to recruit normal BMMSCs for bone remodeling by activating mTOR signaling (Xian et al., 2012), indicating that mTOR may contribute to the activation of different sets of biological regulatory controls under certain conditions. In this study, we reveal that elevated IL4Rα expression leads to the activation of mTOR-P70S6K signaling in Fbn1^+/− BMMSCs, which directly inhibits the osteogenic gene RUNX2 and suppresses bone regeneration in vitro and in vivo. In contrast, rapamycin represses mTOR/P70S6K signaling, which results in rescuing BMMSC deficiency and ameliorating the osteoporotic phenotype in Fbn1^+/− mice, implicating that IL4Rα/mTOR is, in fact, a major signaling pathway contributing to the osteopenia phenotype in Fbn1^+/− mice. mTORC1 controls multiple cellular regulation processes, such as protein synthesis, cell proliferation, metabolism, and autophagy (Huang and Fingar, 2014). It is possible that activation of the mTORC1 pathway increases biosynthetic activity to improve osteogenesis rather than decrease bone formation. Thus, the mechanistic details may not totally explain the phenotype and further study is required to explore the detailed relationship between the mechanism and the phenotype. Because Fbn1^−/− mice are embryonic lethal, we cannot examine whether Fbn1 null shows a phenotype similar to that observed in Fbn1-deficient conditions.

T_H2 cells are a major source of IL4 production in SSc (Kodera et al., 2002; Gabrielli et al., 2009). Fbn1^+/− mice show T_H2 infiltration and high levels of IL4. Disruption of IL4 can rescue the fibrotic phenotype, suggesting that a high level of IL4 may also play a critical role in the disease pathogenesis (Kodera et al., 2002; Wynn, 2004; Gabrielli et al., 2009). Recent studies show that TGF-β contributes to the fibrotic phenotype in SSc through activation of extracellular matrix protein expression, suggesting that TGF-β and IL4 signaling may synergistically activate downstream targets to induce fibrosis in SSc (Varga and Abraham, 2007; Gabrielli et al., 2009; Gerber et al., 2013). Although overproduction of IL4 can induce osteoporotic phenotype in mice (Lewis et al., 1993), activation of IL4Rα may inhibit osteoblast proliferation (Frost et al., 2001), indicating that IL4 may affect BMMSC function. Conversely, it was reported that local delivery of IL4, using a gene therapy approach, prevented bone erosion in arthritis animal models via abrogation of osteoclastogenesis (Lubberts et al., 2000; Woods et al., 2001; Saidenberg-Kermanac’h et al., 2004). These findings indicate that elevated IL4 caused by Fbn1 deficiency may contribute to a low osteoblast/osteoclast turnover rate, as observed in Fbn1^+/− mice. It was reported that TGF-β stimulates osteoclastogenesis via the TAK1–NF-κB pathway (Mizukami et al., 2002; Gingery et al., 2008). IL4 is a potent inhibitor of osteoclastogenesis that operates through various mechanisms, such as rapid suppression of NFATc1 expression (Wei et al., 2002; Cheng et al., 2011) and reduction of RANKL expression (Fujii et al., 2012) via activation of the STAT6 pathway. Our results showed that Prrx1-Cre–mediated knockdown of IL4Rα rescued the osteopenia phenotype and altered BMMSC lineage differentiation in Fbn1^+/− mice; however, elevated serum TGF-β levels were unchanged. Therefore, IL4 signaling may play a critical role in determining osteoclastogenesis in Fbn1^+/− mice. Our data also suggest that IL4/IL4Rα signaling activates mTOR signaling in Fbn1^+/− BMMSCs, which alters BMMSC osteogenic/adipogenic lineage differentiation, as a consequence of reduced osteogenesis and elevated adipogenesis in BM.

The lineage switch between osteoblasts and adipocytes may be regulated by multiple mechanisms, including at least IL4R/mTOR signaling at the early BMMSC (Prrx1-Cre) and osteoblastic progenitor stages (SP7-Cre). Previous studies showed that SP7-expressing osteoblastic progenitors have the ability to control the osteoblastic/adipogenic lineage switch (Zhang et al., 2011; Song et al., 2012) and that SP7 can be regulated by small microRNA (Zhang et al., 2011). Our data showed that SP7-Cre–mediated KO of IL4Rα/mTOR cascades can rescue the BMMSC lineage alteration and osteopenia phenotype in Fbn1^+/− mice. These data may encourage us to use a drug such as rapamycin to ameliorate the disease phenotype in Fbn1^+/− mice. Rapamycin, a specific mTOR inhibitor, is a novel nonsteroidal anticancer and immunosuppressive drug (Abraham and Wiederrecht, 1996; Wang et al., 2003; Shaw et al., 2004). Several preclinical studies showed that rapamycin is effective in treating autoimmune diseases, such as systemic lupus erythematosus and rheumatoid arthritis (Yoon, 2009; Islander et al., 2011). These diseases are usually associated with bone disorders (Weinstein, 2010). Although efficacy of rapamycin treatment in SSc patients must be assessed in clinical studies, phase I clinical trials showed optimal safety for its use in treating scleroderma patients (Su et al., 2009). Added experimental evidence showed that rapamycin treatment resulted in a significant improvement of skin phenotype with regulation of immune index, such as reducing the levels of IL4 and IL17 in Fbn1^+/− mice (Yoshizaki et al., 2010). These data suggest that the underlying mechanism of rapamycin-mediated treatment may be associated with immunomodulation, leading to the down-regulation of T_H2 cells (Lee et al., 2010a). Up-regulation of osteogenesis and down-regulation of adipogenesis in rapamycin-treated Fbn1^+/− mice via inhibition of mTOR signaling in BMMSCs may help to rebuild functional osteoblasts to improve niche microenvironment for immune cells and rebuild homeostasis for the immune system. Because of the extensive functional roles of rapamycin in immune, anticancer, and osteoporosis therapies, it is necessary to explore detailed mechanisms by which rapamycin may target several signaling pathways simultaneously.

Female C57BL/6J, B6.Cg-Fbn1^Tsk/J, B6.Cg-Tg(Prrx1-cre)1Cjt/J, B6.Cg-Tg (Sp7-tTA, tetO-EGFP/cre)1Amc/J, and B6.129S4-Mtor^tm1.2koz/J mice were purchased from The Jackson Laboratory and maintained in C57BL/6J background in at least 10 backcrosses. Age-matched female littermates were used in all experiments. Female immunocompromised nude mice (Beige nu/nu XIDIII) were purchased from Harlan. IL4Rα null (Il4rα^−/−) and floxed (Il4rα^ϕ/ϕ; Herbert et al., 2004) mice were gifts from F. Brombacher (University of Cape Town, Cape Town, South Africa). To generate tissue-specific Cre-mediated KO models, Cre-, floxed-, and Fbn1^+/− mice were intercrossed, and age-matched female littermates were used as WT controls. All animal experiments were performed under the institutionally approved protocols for the use of animal research (University of Southern California protocol numbers 11141, 11953, and 11327).

Femurs were harvested and analyzed by Inveon micro-CT system (Siemens AG). Cross-sectional volumetric BMD was measured at right femur mid-diaphysis with a density phantom. Using 3-dimensional images, a region of interest in secondary spongiosa was manually drawn near the endocortical surface, and BV/TV was assessed as a cancellous bone morphometric parameter.

Approximately 4.0 × 10⁶ BMMSCs were mixed with HA/TCP ceramic powders (40 mg; Zimmer Inc.) and subcutaneously implanted into 8-wk-old immunocompromised mice (Gronthos et al., 2000). At 8 wk after implantation, the transplants were harvested, fixed in 4% PFA, and decalcified with 5% EDTA, pH 7.4, followed by paraffin embedding. The 6-µm paraffin sections were stained with H&E chemical staining. Total BV/TV was quantified by ImageJ software (National Institutes of Health).

To assess the adipose tissue in trabecular areas, femurs were fixed in 4% paraformaldehyde and decalcified with 5% EDTA, pH 7.4, followed by cryosection. Sections were stained with Oil Red O, and positive areas were quantified under microscopy and shown as a percentage of total areas.

Rapamycin (LC Laboratories) was in the vehicle containing 0.2% sodium carboxymethylcellulose (Sigma-Aldrich) and 0.25% polysorbate 80 (Sigma-Aldrich). Rapamycin with vehicle was intraperitoneally administered to Fbn1-deficient mice at a dose of 1.5 mg/kg/d for 14 and 28 d, respectively (Blazar et al., 1994). The disease group and control mice were treated with vehicle only. Treatment was started at age 6 or 8 wk, and all groups of mice were healthy.

To assess trabecular bone and BM areas, femurs were fixed in 4% paraformaldehyde (Sigma-Aldrich) and then decalcified with 5% EDTA, pH 7.4, followed by paraffin embedding. Paraffin sections (6 µm) were stained with hematoxylin and eosin (H&E) and analyzed by ImageJ software. To perform immunohistochemistry staining, the paraffin-embedded sections were blocked with serum matched to secondary antibodies, incubated with the ALP or IL4 specific antibodies (1:400; Santa Cruz Biotechnology, Inc.) at 4°C for overnight, and then stained using VECTASTAIN UNIVERSAL elite ABC kit and ImmPACT VIP Peroxidase Substrate kit (Vector Laboratories) according to the manufacturer’s instruction. To quantify osteoclast activity, mature osteoclasts were determined by TRAP-positive cells on the bone surface. Deparaffinized sections were refixed with a mixture of 50% ethanol and 50% acetone for 10 min. TRAP staining solutions (1.6% naphthol AS-BI phosphate in N, N-dimethylformamide, 0.14% fast red-violet LB diazonium salt, 0.097% tartaric acid, and 0.04% MgCl₂ in 0.2 M sodium acetate buffer at pH 5.0) were freshly made. The sections were incubated in the solution for 10 min at 37°C under shield and counterstained with toluidine blue. All reagents for TRAP staining were purchased from Sigma-Aldrich. To label the mineralization fronts, the mice were given subcutaneous injections of calcein (15 mg/kg body weight; Sigma-Aldrich) in 2% sodium bicarbonate solution 10 and 3 d before sacrifice. Bone dynamic histomorphometric analyses for MAR, MS/BS, and BFR/BS were performed according to the standardized nomenclature for bone histomorphometry (Parfitt et al., 1987).

Peripheral blood serum and cell culture medium were collected, and sRANKL, OPG, IL4, and TGF-β protein levels were analyzed using mouse ELISA Ready-SET-GO kits (eBioscience), according to the manufacturer’s instructions. CTX levels were analyzed with a mouse C-telopeptide of type I collagen ELISA kit (Novateinbio), according to the manufacturer’s instructions. Human IL4 and TGF-β protein levels were analyzed using ELISA Ready-SET-GO kits (eBioscience), according to the manufacturer’s instructions. Human sRANKL protein levels were analyzed using sRANKL Super X ELISA kit (Antigenix America Inc.) and OPG levels were analyzed by OPG instant ELISA kit (eBioscience), according to the manufacturer’s instructions. For measurement of anti-dsDNA antibodies and ANA, peripheral blood serum samples were collected from all experimental mice and analyzed by commercially available ELISA kits (Alpha Diagnostics) according to the manufacturer’s instructions.

Single suspension of BM-derived all-nuclear cells (ANCs; 15 × 10⁶) from femurs and tibias was seeded into 100 mm culture dishes (Corning) and cultured at 37°C with 5% CO₂. After 2 d, nonadherent cells were removed, and attached cells were cultured for 16 d in α-MEM (Invitrogen) supplemented with 20% FBS, 2 mM l-glutamine (Invitrogen), 55 µM 2-mercaptoethanol (Invitrogen), 100 U/ml penicillin, and 100 µg/ml streptomycin (Invitrogen). These plastic attached single colonies were passaged with frequent medium changes to eliminate potential hematopoietic cells contamination (Soleimani and Nadri, 2009; Zhu et al., 2010). Flow cytometric analysis was performed to show a surface epitope profile of isolated BMMSCs (Fig. 2 A). To further confirm that the single colony-derived plastic adherent BMMSCs were not contaminated with hematopoietic cells, we depleted T cells in BMMSC culture using CD3 antibody treatment. In brief, 1 ml CD3 antibody was added to 10⁷ BMMSCs and the mixture was incubated at 4°C for 30 min. The cells were centrifuged for 5 min to remove unbound antibody, resuspended in 1 ml diluted (1:6) rabbit complement (Pel-Freez Biologicals) and incubated at 37°C for 30 min. Subsequently, CD3 antibody/complement treatment was repeated once (Gilmore et al., 1986). In addition, we sorted BMMSCs from BM cells according to their MSC-specific surface markers using flow cytometry, as reported previously (Houlihan et al., 2012). BM cells were selected by lineage markers (Lin⁻) and then sorted by PDGFRα and Sca-1 (PαS). Alizarin Red staining and Western blot analysis showed that BMMSCs derived from plastic collected single colonies, CD3 antibody/complement depletion, and flow cytometric sorting had similar osteogenic abilities (Fig. 2, B and C). For CFU-F assay, 10⁶ ANCs from BM were seeded in T25 cell culture flasks (Corning). After 16 d, the cultures were washed by PBS and stained with 1% toluidine blue solution with 2% paraformaldehyde. The cell clusters with more than 50 cells were counted as colonies under microscopy.

BMMSCs were cultured under osteogenic culture condition, containing 2 mM β-glycerophosphate (Sigma-Aldrich), 100 µM l-ascorbic acid 2-phosphate (Wako), and 10 nM dexamethasone (Sigma-Aldrich) in the growth medium. After 3 wk induction, 1% Alizarin Red S (Sigma-Aldrich) staining was performed to detect matrix mineralization, and the stained areas were quantified by ImageJ software and shown as a percentage of the total area.

Cells were lysed in M-PER mammalian protein extraction reagent (Thermo Fisher Scientific) with protease and phosphatase inhibitors (Roche), and proteins were quantified using protein concentration assay (Bio-Rad Laboratories). 20 µg of proteins were separated by SDS-PAGE and transferred to 0.2 µm nitrocellulose membranes (Millipore). The membranes were blocked with 5% nonfat dry milk and 0.1% Tween-20 for 1 h, followed by incubation overnight with the primary antibodies diluted in blocking solution according to manufacturer’s instructions. Antibodies to mouse ALP, LPL, PPARγ2, IL4Rα, PI3K-p110, phosphor-SMAD2/3, and SP1 were purchased from Santa Cruz Biotechnology, Inc. Antibodies to human RUNX2, ALP, and OCN were purchased from Santa Cruz Biotechnology, Inc. Antibodies to phosphor-mTOR (Ser2448), mTOR, phospho-Akt (Ser473), Akt, phosphor-P70S6K (T389; S371), phosphor-Rictor, phosphor-Raptor, and SMAD2/3 were obtained from Cell Signaling Technology. Antibodies to mouse RUNX2, FBN1, OCN, and β-actin were purchased from Abcam, GeneTex, Millipore, and Sigma-Aldrich. The membranes were then incubated for 1 h in HRP-conjugated secondary antibody (Santa Cruz Biotechnology, Inc.) diluted at 1:10,000 in blocking solution. Immunoreactive proteins were detected using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific) and BioMax film (Kodak). The intensity of bands was measured by using ImageJ software and normalized to β-actin.

For adipogenic induction, 500 nM isobutylmethylxanthine (Sigma-Aldrich), 60 µM indomethacin (Sigma-Aldrich), 500 nM hydrocortisone (Sigma-Aldrich), 10 µg/ml insulin (Sigma-Aldrich), and 100 nM l-ascorbic acid phosphate were added into the growth medium. After 7 d, the cultured cells were stained with Oil red O (Sigma-Aldrich), and positive cells were quantified under microscopy and shown as a number to total cells.

BMMSCs (0.5 × 10⁶) were seeded to a 6-well culture plate and treated with Fbn1 siRNA (Santa Cruz Biotechnology, Inc.), P70s6k siRNA (Santa Cruz Biotechnology, Inc.), Il4rα shRNA (Santa Cruz Biotechnology, Inc.), or Mtor shRNA (Addgene), according to the manufacturers’ instructions. After transfection, cells were either used for protein extraction for Western immunoblotting or for differentiation induction. For chemical reagent treatments, serum-starved MSCs were treated with 10–100 ng/ml rIL-4 (R&D Systems), 50 nM rapamycin (LC Laboratories), or 1 µg/ml TGF-β neutralized antibody (R&D Systems). For Western immunoblotting, MSCs were cultured under growth medium with drugs for 24 h, and protein was extracted by using M-PER mammalian protein extraction reagent. For differentiation induction, MSCs were cultured under inductive conditions in the presence of drugs (added every 3 d) for 3 wk, followed by staining and gene expression analysis.

To detect T_H2 percentage of CD4⁺ cells in BM and splenocyte (Ansel et al., 2004), BM or splenocyte-derived single-suspension ANCs were incubated with anti-CD4-PerCP (BioLegend), followed by staining with anti–IFN-γ-APC (BioLegend) and anti-IL4-PE (eBioscience), anti–IL5-PE (eBioscience), or anti–IL13-eFluor 660 (eBioscience) using an intracellular antigen staining kit (eBioscience). To further identify T_H2 cells infiltration, ANCs were incubated with anti-CD4-PerCP, followed by staining with combined T_H2 cytokines, such as anti–IL4-PE/anti–IL13-eFluor 660 and anti–IL5-PE/anti–IL13-eFluor 660 antibodies, using an intracellular antigen staining kit. Because CC-chemokine receptor 3 (CCR3) has been reported as a marker for T_H2 cells (Sallusto et al., 1997), we detected CD4⁺CCR3⁺, CD4⁺CCR3⁺IL4⁺, and CD4⁺CCR3⁺IL13⁺ subpopulations using anti-CCR3-FITC antibody (BioLegend). Cells were analyzed by FACS^Calibur with CellQuest software (BD).

BMMSCs were cultured on 4-well chamber slides (Nunc; 2 × 10³/well) and then fixed with 4% paraformaldehyde. The Chamber slides are incubated with primary anti CD73 antibody (1:400; BD) and anti-IL4Rα (1:400; BD) at 4°C for overnight and then treated with Rhodamine-conjugated secondary antibody (1:400; SouthernBiotech) or Alexa Fluor 488–conjugated secondary antibody (1:200, Invitrogen) for 30 min at room temperature. Finally, slides were mounted with VECTASHIELD mounting medium (Vector Laboratories).

BMMSCs were mixed with HA/TCP ceramic powders, and Extracel-HP hydrogel (Glycosan Biosystems), containing 200 ng IL4, was covered on the surface of the implants for slow release of the cytokine (Liu et al., 2011). At 8 wk after implantation, the implants were harvested and decalcified. Newly formed mineralized tissue areas were analyzed by paraffin section and H&E staining.

Promoter reporters were prepared as our previous description (Chen et al., 2014). Il4rα-luciferase promoter reporter constructs were generated by PCR using Pfu polymerase and mouse genomic DNA as a template. Primers containing upstream XhoI and HindIII downstream restriction sites were used to generate Il4rα promoter fragment (forward, 5′-CTCGAGGAATTCATGCTGCTTTCTCG-3′, and reverse, 5′-AAGCTTGGCTTTCCCACCGCCCGTT-3′). Restriction-digested PCR products were subcloned into pGL3-Basic vector (Promega). Point mutants were introduced into the reporter by the Pfu–DpnI method. All clones were confirmed by sequencing on both strands. BMMSCs cultured in 6-well plates were cotransfected with 2 µg of luciferase reporter and 100 ng Renilla luciferase expression vector to control for transfection efficiency. 48 h after transfection, cells were lysed in 1× passive lysis buffer, and luciferase activity was measured using the Dual-Glo Luciferase System (Promega) and a luminometer (Turner Biosystems).

BMMSCs grown in 10-cm cell culture dishes were fixed for 10 min at room temperature by addition of 1% paraformaldehyde to the growth medium. Cells were washed twice in cold PBS supplemented with complete protease inhibitor cocktail and gently scraped from the plate. Cell lysis and ChIP were performed using the ChIP Assay kit (Millipore). For chromatin fragmentation, cells were sonicated using a Branson Sonifier 450 on power setting 4 in 30-s bursts with 1 min cooling on ice for a total sonication time of 4 min. For immunoprecipitations, 1:100 dilution of SP1 antibody was used to capture protein–DNA complexes, and isotype-matched IgG was used as negative control. All resulting precipitated DNA samples were quantified with real-time PCR and expressed as the percentage of input DNA.

All experimental group sizes were chosen to ensure adequate statistical power despite the highly variable nature of the studies performed. No animal excluded, and animals were randomly assigned groups for the studies. All experiments were not performed in a blinded fashion. Data were assessed for normal distribution and similar variance between groups. Comparisons between two groups were analyzed by independent unpaired two-tailed Student’s t test, and the comparisons between more than two groups were analyzed by one-way ANOVA with the Bonferroni adjustment. The p-values <0.05 were considered statistically significant.

We thank Dr. Xiaohong Duan and Mr. Tao Zhou from the Fourth Military Medical University for generating the microCT images.

This work was supported by grants from the National Institute of Dental and Craniofacial Research, National Institutes of Health, Department of Health and Human Services (R01DE017449 and R01DE019932 to S. Shi), and a grant from the California Institute for Regenerative Medicine (RN1-00572 for S. Shi).

The authors declare no competing financial interests.