Mechanical force-driven TNFαendocytosis governs stem cell homeostasis

Wenjing Yu,Chider Chen,Xiaoxing Kou,3,Bingdong Sui,4,Tingting Yu,5,Dawei Liu,5,Runci Wang,Jun Wang and Songtao Shi,3

INTRODUCTION

Mesenchymal stem cells(MSCs),heterogeneous primitive cells that were originally discovered in the adult bone marrow stroma,possess self-renewal and multidifferentiation potential and are critical for maintaining homeostasis in multiple tissues/organs.1Understanding the regulatory mechanisms of MSC behaviors under physiological and pathological conditions represents a fundamental scientific issue with implications for deciphering disease pathogenesis and establishing appropriate stem cellbased therapeutics.1In particular,the notion that MSCs closely interact with the immune system has increasingly been recognized,and the function of MSCs has been shown to be tightly governed by multiple inflammatory cytokines.2–3We have previously reported that tumor necrosis factor alpha(TNFα)and interferon gamma(IFNγ)synergistically impair MSC stemness in the context of osteopenia and diminished tissue regeneration,3–4whereas others have documented that the immunomodulatory effect of MSCs is mediated by IFNγcombined with TNFαand interleukins(ILs).5–6Interestingly,in addition to the various immunosuppressive factors they produce,2MSCs are also known to express and secrete certain inflammatory cytokines,which have been reported to respond to extrinsic stresses,such as hypoxia and immune mediators.7–10However,whether MSCs are capable of releasing inflammatory cytokines under physiological conditions and what the functional purposes of MSC-derived inflammatory cytokines might be remain unclear.

The production,secretion,and functional execution of inflammatory cytokines involve a series of fine-tuned molecular mechanisms.11These mechanism are particularly elaborate for TNFα,the prototypic member of the TNF superfamily ligands;TNFαplays a crucial role in regulating a variety of biological processes,not only inflammation.11–12Initially expressed after translation in T lymphocytes,macrophages and other types of cells,TNFαexists as a 26 kD membrane-anchored precursor(pro-TNFα),the extracellular domain of which is cleaved by metalloproteinases,specifically the TNFα-converting enzyme(TACE),to yield the 17 kD mature TNFα.13Biologically active TNFαtransduces signals in an autocrine/paracrine manner through two receptors,TNFR1 and TNFR2,and activates downstream cascades,including nuclear factor kappaB(NFκB)transcriptional signaling.14Intriguingly,the effects of TNFαare dependent on the recipient cell types and the dosage and duration of TNFαapplication.For MSCs,it has been reported that high concentrations of TNFα(over 1 ng·mL-1)in culture suppress osteogenic potential via NFκB activation,while lower doses of TNFαshow benefits to some extent.4,15Furthermore,in aging-and disease-associated microenvironments in vivo upregulated TNFαlevels are detected in the circulation and in the bone marrow space,resulting in resident MSC impairment and osteopenic phenotypes.4,16Therefore,it is safe to assume that TNFαneeds to reach pathological levels to exert detrimental effects on MSCs.Whether and how physiological levels of TNFα might regulate MSCs via potentially distinct mechanisms is an interesting and significant question that warrants investigation.

Mechanical stimuli,including physical cues from the matrix and applied forces,represent another critical extrinsic factor that controls MSC fate,particularly in bone,where the response to mechanical loading is constantly adapted and coordinated.17In this regard,bone marrow MSCs(BMMSCs)are especially mechanosensitive and mechanoresponsive to support their functional regulation and bone maintenance.18Importantly,the mechanical unloading experienced by astronauts in spaceflight and bedridden patients leads to progressive bone loss,but therapeutics have yet to be established.19Accordingly,hindlimb suspension in rodents recapitulates the osteopenic phenotype seen in human unloading,in which BMMSCs exhibit osteogenic inhibition due to coordinated responses of multiple signaling pathways.20Notably,mechanical unloading also induces general humoral alterations,including the onset of systemic inflammation,with increased susceptibility to autoimmune disorders.21Further dissecting the mechanisms underlying the mechanical regulation of MSCs via potential inflammatory cytokine reactions would be helpful for identifying novel molecular targets for counteracting unloadinginduced osteopenia.

In this study,we aimed to investigate the physiology and mechanisms of the interaction between MSCs and inflammatory cytokines and to examine the potential pathophysiological and therapeutic contributions of this process to mechanical regulation.Surprisingly,we reveal that MSCs are capable of producing and releasing a relatively low level of TNFαunder physiological conditions and that physiological levels of TNFαare indispensable for the functional maintenance of MSCs both in vitro and in vivo.Mechanistically,we further discovered a receptor-independent mechanism by which TNFαexerts its effects through endocytosis,which is driven by mechanical force and safeguards MSC homeostasis based on the uptake of TNFα,restricting the activation of mammalian target of rapamycin(mTOR)signaling.Importantly,inhibition of mTOR signaling by rapamycin serves as an effective osteoanabolic therapeutic strategy to protect against TNFαdeficiency and mechanical unloading.Collectively,these findings unravel the physiological framework by which the dynamic TNFαshuttle-based mTOR equilibrium governs MSC and bone homeostasis.

RESULTS

MSCs produce and release TNFαfor functional maintenance

To investigate whether and to what extent MSCs are capable of producing inflammatory cytokines under physiological conditions,we cultured mouse BMMSCs and analyzed their production of TNFα,which is one of the most widely studied inflammatory cytokines with a variety of biological functions.11A combination of quantitative real-time polymerase chain reaction(qRT-PCR),Western blot analysis,and enzyme-linked immunosorbent assay(ELISA)was applied to detect the mRNA expression,protein expression,and secretion of TNFα(Fig.1a–c).TNFαproduction by MSCs was evaluated in comparison with the levels produced by naïve and activated T cells,T helper 1(Th1)cells,and macrophages with and without lipopolysaccharide(LPS)treatment(Fig.1a–c).The data demonstrated that MSCs indeed expressed both pro-TNFαand TNFα,and they released a certain amount of TNFαinto the culture medium,which was more than the amount of TNFαreleased by naïve T cells but much less than the amount of TNFαreleased by Th1 cells and LPS-activated macrophages during the same period(Fig.1a–c).

To address whether physiological TNFαis required for the functional maintenance of MSCs,we first examined bone formation and bone marrow adiposity,which indicate the osteogenic and adipogenic behaviors of tissue-resident MSCs,22–23in wild-type(WT)and TNFα-/-mice.Calcein labeling analysis showed that TNFα-/-mice suffered from a decrease in bone formation,as demonstrated by a lower mineral apposition rate(MAR)and bone formation rate(BFR)compared to those of WT mice(Fig.1d).On the other hand,TNFα-/-mice exhibited higher percentages of fatty marrow area than WT mice(Fig.1e).To directly evaluate whether MSC function was dysregulated under TNFαdeficiency,we isolated BMMSCs from WT and TNFα-/-mice(Fig.S1A,B)and analyzed their behaviors ex vivo in terms of colony formation,proliferation,and osteogenic and adipogenic differentiation(Fig.1f–i).22,24The data confirmed that TNFα-/-MSCs were less capable of forming fibroblastic colonies(Fig.1f),incorporating bromodeoxyuridine(BrdU)(Fig.1g)during proliferation and differentiating into mineralizing osteoblasts than were WT MSCs(Fig.1h),while they were misdirected toward adipocytes during differentiation(Fig.1i).Interestingly,MSCs derived from mice lacking IFNγand IL-6,two additional potentially important inflammatory regulators of MSC function,10,25did not show functional impairments(Fig.S1C–G),suggesting that TNFαdependent MSC regulation occurs without general effects of inflammatory cytokines.

Next,we intended to determine whether TNFαper se maintains MSC function by treating cultured TNFα-/-MSCs with TNFα.Doseeffect analyses demonstrated that a low dose of 1 ng·mL-1TNFα mimicked the physiological concentration of TNFαsecreted by WT MSCs(Fig.S2A).Furthermore,1 ng·mL-1and lower doses of TNFα exhibited a limited ability to activate TNFR downstream signaling pathways,which were previously revealed to be detrimental for MSCs3–4(Fig.S2B–D).Therefore,we applied TNFαat a concentration of 1 ng·mL-1and explored whether this low“physiological”level of TNFαwould be beneficial for MSCs.We further used receptor activator of nuclear factor kappaB ligand(RANKL),another member of the TNF superfamily of cytokines that is important for bone homeostasis,26at the same dose of 1 ng·mL-1to treat MSCs as a control for TNFα.We discovered that 1 ng·mL-1TNFα,but not RANKL,was effective in rescuing the functional decline of MSCs derived from TNFα-/-mice,as proven by increases in colony formation,proliferation,and osteogenic differentiation with inhibition of adipogenesis(Fig.1f–i).Additional in vitro experiments using genetic overexpression of TNFα in TNFα-/-MSCs confirmed that TNFαis functionally important for preserving homeostasis of MSCs(Fig.S3A–D).Further knockdown of TNFαusing a small interfering RNA(siRNA)followed by extrinsic TNFαrescue also revealed that TNFαis indispensable for MSC homeostasis(Fig.S3E–H).

The initially expressed transmembrane pro-TNFαcan be specifically cleaved by TACE to release soluble,biologically active,mature TNFα.13Intriguingly,pro-TNFαfailed to rescue the deficient function of MSCs derived from TNFα-/-mice,suggesting that TNFαregulation of MSCs was based on the mature form(Fig.S4A–D).Therefore,we further investigated whether the release of mature TNFαis required for the functional regulation of MSCs by using TNF protease inhibitor-2(TAPI-2),a hydroxamate-based inhibitor of TACE.27We confirmed that TAPI-2 application elevated the intracellular protein level of pro-TNFαwhile remarkably suppressing MSC release of TNFαinto the culture medium(Fig.1j).We further revealed that TAPI-2 treatment in WT MSCs induced functional decline,mimicking the phenotype of TNFα-/-MSCs(Fig.1k–n).Taken together,the above results indicated that MSCs produced and released TNFαfor their functional maintenance(Fig.1o).

TNFαsafeguards MSC homeostasis in a receptor-independent manner through endocytosis

Next,we explored how TNFαregulates MSC homeostasis.Exogenously added TNFαat a concentration of 1 ng·mL-1was able to establish a physiological TNFαmicroenvironment to maintain MSC function without activating canonical TNFR downstream signaling in cultured MSCs(Fig.S2A,B).To further evaluate whether TNFR contributes to the regulation of MSCs,we isolated MSCs from TNFR-/-mice and analyzed their function compared to that of WT and TNFα-/-MSCs(Figs.2a–d and S1A,B).The data demonstrated that the colony-forming and proliferative capacities of TNFR-/-MSCs were not significantly decreased compared to those of WT MSCs(Fig.2a,b).Moreover,although TNFR-/-MSCs were less capable of osteogenesis than WT MSCs,TNFα-/-MSCs were deficient in osteogenic differentiation(Fig.2c).In addition,TNFR-/-MSCs showed similar adipogenic differentiation to WT MSCs(Fig.2d).These results indicated limited effects of TNFR on MSC function,which were further confirmed by in vivo analyses of bone formation and bone marrow adiposity,revealing that WT and TNFR-/-mice were comparable(Fig.2e,f).

Fig.1 MSCs produce and release TNFαfor functional maintenance.a–c Examination of mRNA expression levels,protein expression levels,and secreted concentrations of TNFαin different cell types.N=3.For quantification of Western blotting,two-tailed Student’s t test was used for comparisons between different cell types and MSCs.d Calcein labeling for bone formation analysis in WT and TNFα-/-mice(N=5).Scale bar=50μm.e Oli red O staining for bone marrow adiposity in WT and TNFα-/-mice(N=5).Scale bar=150μm.f–i Functional analyses of MSCs according to CFU,BrdU labeling,osteogenic,and adipogenic differentiation.TNFαand RANKL were added at 1 ng·mL-1.N=5.Scale bars=100μm.j Examination of protein expression levels and secreted concentrations of TNFα.N=3.TAPI-2 was used to inhibit TACE at 120 nmol·L-1.k–n Functional analyses of MSCs according to CFU,BrdU labeling,and osteogenic and adipogenic differentiation.N=3.Scale bars=100μm.o Diagram showing the production and release of TNFαfor the functional maintenance of MSCs.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and WT groups.*P＜0.05.N.S.not significant.Data represent the mean±SD

Fig.2 TNFαsafeguards MSC homeostasis in a receptor-independent manner through endocytosis.a–d Functional analyses of MSCs according to CFU,BrdU labeling,and osteogenic and adipogenic differentiation.MSCs were derived from WT,TNFα-/-or TNFR-/-mice(N=5).Scale bars=100μm.e Calcein labeling for bone formation analysis in WT and TNFR-/-mice(N=5).Scale bar=50μm.f Oli red O staining for bone marrow adiposity in WT and TNFR-/-mice(N=5).Scale bar=150μm.g Endocytosis analysis of FITC-labeled TNFαuptake by MSCs for 24 h in vitro.Pitstop® 2 was used to inhibit clathrin-mediated endocytosis at 12μmol·L-1.N=3.Scale bars=20μm(top)and 7μm(bottom).h Western blot analysis of protein expression levels(N=3).i–k Functional analyses of MSCs according to BrdU labeling and osteogenic and adipogenic differentiation.N=3.Scale bars=100μm.l Diagram showing that TNFαregulates MSC homeostasis in a receptorindependent manner through endocytosis.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and WT groups.*P＜0.05.N.S.not significant.Data represent the mean±SD

In deciphering how TNFαregulates MSC homeostasis in a receptor-independent manner,we noticed that when TNFα-/-MSCs were treated with TNFα,the levels of TNFαdetected in the culture medium were always much lower than what we originally added(Fig.S2A)and that intracellular TNFαcould be found(Fig.S2B).These data indicated that TNFαmight be taken up by deficient MSCs despite the lack of receptor activation.To address this hypothesis,we labeled TNFαwith fluorescein isothiocyanate(FITC)for MSC treatments and discovered that FITC-TNFαwas indeed taken up by TNFα-/-MSCs(Fig.2g).To examine whether TNFαuptake was mediated by endocytotic processes,we used Pitstop®2,an inhibitor of clathrin-mediated endocytosis.28We identified that Pitstop®2 effectively inhibited,but did not block,TNFαuptake by TNFα-/-MSCs,suggesting that TNFα uptake occurred at least partially through clathrin-mediated endocytosis(Fig.2g).We further found that TNFαuptake was not affected by TNFR deficiency,indicating a receptorindependent mechanism(Fig.2g).

Fig.3 Endocytosed TNFαbinds to mTORC2 and restricts activation of mTOR signaling for functional regulation of MSCs.a–c Western blot analyses of mTOR signaling in response to TNFR or TNFαdeficiency.N=3.TNFαwas added at 1 ng·mL-1.d,e Co-IP analyses of the binding of TNFαto mTOR complex components(N=3).f Analysis of mTORC1 activity using an ELISA-based assay(N=3).g–i Western blot analyses of mTOR signaling.N=3.TAPI-2,TNFα,Pitstop®2,and rapamycin(Rapa)were added at 120 nmol·L-1,1 ng·mL-1,12μmol·L-1,and 50 nmol·L-1,respectively,in vitro.j–m Functional analyses of MSCs according to CFU,BrdU labeling,and osteogenic and adipogenic differentiation(N=3).Scale bars=100μm.n Diagram showing TNFαbinding to mTORC2 and restricting activation of mTOR signaling for functional regulation of MSCs.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and WT groups.*P＜0.05.N.S.not significant.Data represent the mean±SD

Next,we examined whether TNFαendocytosis regulates MSC function.Western blot analysis confirmed that Pitstop®2 treatment efficiently suppressed the uptake of TNFαinto TNFα-/-MSCs(Fig.2h).As expected,Pitstop®2 treatment further significantly diminished the ability of TNFαto rescue TNFα-/-MSCs,leading to dysregulation of proliferation and differentiation(Fig.2i–k).The above results collectively suggested that TNFα safeguarded MSC homeostasis in a receptor-independent manner through endocytosis(Fig.2l).

Endocytosed TNFαbinds to mTORC2 and restricts the activation of mTOR signaling

Next,we investigated how endocytosed TNFαregulates MSC ho

meostasis.In screening the multiple signaling pathways that modulate MSC behaviors,we discovered that mTOR signaling changed in response to TNFαdeficiency and replenishment(Fig.3a),whereas the canonical Wnt pathway,the transforming growth factor-beta(TGF-β)pathway and the extracellular signalregulated kinase(ERK)pathway were not influenced by TNFα(Fig.S5A).In detail,phosphorylation of mTOR(p-mTOR),which indicates pathway activation,29was restricted by TNFαbut was not affected by TNFR deficiency in MSCs(Fig.3a).mTOR functions downstream of phosphoinositide 3-kinase(PI3K)and phosphatase and tensin homolog(PTEN)in two distinct kinase complexes,termed mTOR complex 1(mTORC1)and complex 2(mTORC2),and Akt is a downstream target of mTORC2 that regulates mTORC1 activity.29We therefore analyzed the different components of mTOR signaling and found that TNFαinhibited phosphorylation-mediated activation of key mTORC2 subunits,Rictor and Sin1(Figs.3b and S3I),without affecting PI3K and PTEN expression in MSCs(Fig.S5B).TNFαalso suppressed the phosphorylation of Akt and the key mTORC1 subunit Raptor,as well as the downstream ribosomal protein S6 kinase(P70S6K),in MSCs29(Figs.3c and S3J).

Fig.4 Rapamycin infusion improves MSC function in TNFα-/-mice in vivo.a Schematic diagram showing the experimental design to investigate the effect of rapamycin treatment on TNFα-/-mice in vivo.b–e Functional analyses of MSCs according to CFU,BrdU labeling,and osteogenic and adipogenic differentiation(N=5).Scale bars=100μm.f Western blot analyses of mTOR signaling(N=3).g Calcein labeling for bone formation analysis.N=5.Scale bar=50μm.h Oli red O staining for bone marrow adiposity.N=5.Scale bar=150μm.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and WT groups.*P＜0.05.Data represent the mean±SD

To decipher how TNFαregulates mTOR signaling in MSCs,we hypothesized that TNFαinteracted with mTOR complexes given that the upstream factors PI3K and PTEN were unaffected(Fig.S5B).To address this hypothesis,we performed a protein coimmunoprecipitation(co-IP)assay to examine the potential binding of TNFαwith mTOR complexes.We discovered that when using an anti-TNFαantibody for co-IP,mTOR,Rictor and Sin1 could be captured,but Raptor failed to be detected,suggesting that TNFαbinds with mTORC2 but not mTORC1 in MSCs(Fig.3d).Further co-IP analyses using mTOR,Rictor and Sin1 antibodies also captured TNFαin MSCs(Fig.3e).Interestingly,although TNFαdid not directly interact with mTORC1,mTORC1 kinase activity was repressed by TNFαin MSCs,possibly through the restricted activation of Akt(Fig.3f).We also confirmed that the release and uptake cycle of TNFαcontributed to mTOR suppression in MSCs,as inhibiting TACE with TAPI-2 and inhibiting clathrin-mediated endocytosis with Pitstop®2 resulted in mTOR signaling activation,mimicking the effects of TNFαdeficiency(Fig.3g,h).Additionally,although pro-TNFαcould be taken up by MSCs,mTOR activation was not repressed by pro-TNFα,collectively proving that only the mature form of TNFαwas capable of regulating mTOR signaling in MSCs(Fig.S4E–H).These data indicated that endocytosed TNFα bound to mTORC2 and restricted the activation of mTOR signaling in MSCs,which might be responsible for the observed functional regulation.

Rapamycin protects MSC function against TNFαdeficiency

The above results prompted us to investigate whether inhibition of mTOR signaling indeed contributes to the functional regulation of MSCs by TNFαand whether pharmaceutical mTOR suppression serves as an effective approach to rescue the functional decline of MSCs under TNFαdeficiency.In this regard,we tested rapamycin(Rapa),a potent and specific chemical inhibitor of mTOR complexes and an approved drug in the clinic,30on TNFα-/-MSCs.We confirmed that rapamycin application suppressed activated mTOR signaling in TNFα-/-MSCs in vitro(Fig.3i).Functional analyses of MSCs further identified that rapamycin treatment significantly improved the colony-forming capability,proliferation,and osteogenic differentiation of TNFα-/-MSCs,while rapamycin inhibited the adipogenic differentiation of TNFα-/-MSCs,suggesting restoration of the homeostatic condition of MSCs in the context of TNFα deficiency(Fig.3j–m).These data indicated that mTOR signaling was a critical downstream target of TNFαin regulating MSC homeostasis and a potential therapeutic target for counteracting MSC disorders in the absence of TNFα(Fig.3n).

Fig.5 Mechanical force drives TNFαendocytosis and promotes MSC function.a,b Examination of secreted concentrations and protein expression levels of TNFα.MSCs underwent cyclic loading of stretch force at 15%elongation,0.5 Hz.N=3.c Western blot analyses of mTOR signaling(N=3).d–f Functional analyses of MSCs according to BrdU labeling and osteogenic and adipogenic differentiation.N=3.Scale bars=100μm.Stretch force was applied twice per week for 2 h per session.g qRT-PCR analysis of the mRNA expression levels of TNFα(N=3).h,i Endocytosis analysis of FITC-labeled TNFαuptake by MSCs for 2 h in vitro.Pitstop®2 was used to inhibit clathrin-mediated endocytosis at 12μmol·L-1.Stretch force was applied for a 2 h period.N=3.Scale bars=20μm.j,k Endocytosis analysis of Alexa FluorTM 488-conjugated bovine serum albumin(488-BSA)taken up by MSCs for 2 h in vitro.Pitstop® 2 was used to inhibit clathrin-mediated endocytosis at 12μmol·L-1.Stretch force was applied for a 2 h period.N=3.Scale bars=20μm.l Diagram showing mechanical force-driven TNFα endocytosis.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and stretch 0 h/control groups.*P＜0.05.N.S.not significant.Data represent the mean±SD

To further evaluate whether inhibition of mTOR signaling rescues the functional decline of MSCs in TNFα-/-mice in vivo,we infused rapamycin into TNFα-/-mice intraperitoneally at a dosage of 1.5 mg·kg-1per day(Fig.4a);this approach was based on our previous study using rapamycin to ameliorate osteopenia.30After 2 weeks of rapamycin administration,MSCs demonstrated remarkable recovery despite TNFαdeficiency,as indicated by the formation of more fibroblastic colonies ex vivo(Fig.4b).Furthermore,MSCs derived from rapamycintreated TNFα-/-mice showed higher proliferative capacity than TNFα-/-MSCs(Fig.4c).Regarding differentiation,MSCs derived from rapamycin-treated TNFα-/-mice showed a preference for osteogenesis rather than adipogenesis(Fig.4d,e).Molecular examinations of ex vivo MSCs confirmed that rapamycin treatment repressed mTOR signaling under TNFαdeficiency(Fig.4f).Functional recovery of MSCs further contributed to improved bone formation with decreased bone marrow adiposity in rapamycin-treated TNFα-/-mice(Fig.4g,h).These results collectively suggested that rapamycin was an effective osteoanabolic therapeutic that protected MSC function against TNFα deficiency.

Mechanical force drives TNFαendocytosis and promotes MSC function

Next,we aimed to investigate how the release and uptake of TNFα are coupled for MSC regulation.MSCs are exposed to a variety of mechanical stimuli in vivo,18which have been reported to influence cellular endocytotic processes.31–32To investigate whether mechanical force regulates TNFαshuttling across the MSC membrane,we applied an intermittent stretch-loading model to cultured MSCs(Fig.5a).Intriguingly,we discovered that the TNFαconcentration in the culture medium of MSCs dropped dramatically during each of the 2 h stretching periods,whereas it gradually increased during mechanical unloading and generally increased with advancing time(Fig.5a).Correspondingly,intracellular protein levels of TNFαwere upregulated after stretching but diminished upon unloading(Fig.5b).These data suggested constitutive production and release of TNFαby MSCs with inducible internalization triggered by mechanical force.

To determine whether the force-driven dynamics of TNFα contribute to MSC regulation,we examined and revealed changes in mTOR signaling activation in accordance with stretch-provoked TNFαfluctuations(Fig.5c).Further cellular functional analyses demonstrated that intermittent stretching for 2 h per session twice per week significantly promoted the proliferation and osteogenic differentiation of MSCs,with inhibition of adipogenesis(Fig.5d–f).As expected,stretch-loading did not induce changes in TNFαmRNA expression(Fig.5g).Instead,mechanical stretching promoted TNFαuptake in both WT and TNFα-/-MSCs(Fig.5h,i).We used Alexa FluorTM488-conjugated bovine serum albumin(488-BSA)to confirm that the general endocytosis rate was also boosted by mechanical stretching(Fig.5j,k).Importantly,when Pitstop®2 was used to inhibit clathrin-mediated endocytosis,stretch-induced TNFαand 488-BSA internalization was blocked(Fig.5h–k).To further confirm that mechanical stretching restricted mTOR signaling via TNFαendocytosis,we applied mechanical stretching to TNFα-/-MSCs and then examined mTOR signaling by Western blot analysis.The results showed that there was no significant alteration in mTOR signaling between the stretch-loading and control groups(Fig.S6A).TNFαtreatment followed by mechanical stretching restricted mTOR signaling,whereas this effect was diminished in the presence of Pitstop®2(Fig.S6B).Collectively,these results suggested that mechanical force drove TNFαendocytosis and promoted MSC function(Fig.5l).

Mechanical unloading induces MSC functional decline and bone loss that are counteracted by rapamycin

Inspired by the above data,we intended to investigate the pathophysiological implications of mechanical force-induced TNFαendocytosis.Hindlimb unloading(HU)of mice has been widely used as a ground-based model to simulate the effects of spaceflight and microgravity environments,in which osteopenia and MSC dysfunction have been reported.20Interestingly,we discovered that HU mice were characterized by an elevated level of TNFαin the bone marrow(Fig.6a).To determine whether TNFα uptake by MSCs was impaired after hindlimb suspension,we applied Prx1-Cre;tdTomato mice,in which mesenchymal cells are labeled specifically in vivo.We found that among the bone marrow tdTomato+cells,TNFα+cells were indeed less abundant in HU mice than in the control mice,but there was a higher prevalence of p-mTOR+cells,indicating diminished TNFα dynamics with mTOR signaling activation under mechanical unloading(Fig.6b).

Considering the above findings,we continued to evaluate whether inhibition of mTOR signaling by rapamycin infusion counteracts the putative MSC functional decline in HU mice(Fig.6c).As expected,rapamycin administration prevented the activation of mTOR signaling in MSCs after hindlimb suspension(Fig.6d).Furthermore,rapamycin treatment in vivo protected MSCs against HU-induced functional impairments,as indicated by improved colony formation,proliferation,and osteogenic differentiation in MSCs derived from rapamycin-treated mice with suppressed adipogenesis(Fig.6e–h).Importantly,we revealed that rapamycin infusion substantially prevented the decline of bone formation in HU mice,and rapamycin also rescued bone marrow adiposity(Fig.6i,j).Finally,we identified rapamycin as an effective therapy for preserving bone mass despite mechanical unloading(Fig.6k).Collectively,these data indicated that mechanical unloading induced MSC functional decline and bone loss based on TNFαand mTOR signaling dysregulation,and these effects were counteracted by rapamycin(Fig.6l).

DISCUSSION

MSCs interact with the immune system for functional modulation and maintenance of tissue/organ homeostasis,1,5while they are known to secrete inflammatory cytokines with unclear implications under physiological conditions.7–9Inflammatory cytokines play crucial roles in regulating various basic biological processes in organisms,whereas the effects and mechanisms of inflammatory cytokines in the context of the stem cell pool remain largely unexplored.3–5In the present study,we revealed that physiological levels of TNFαare needed for functional homeostasis of MSCs via autocrine/paracrine signaling,which is of particular significance for bone maintenance in vivo.We further uncovered a new mechanism mediating the beneficial effects of low levels of TNFαon MSCs,which was proven to be receptor-independent and mediated by mechanical force-driven endocytosis to maintain mTOR signaling equilibrium.Accordingly,to counteract the MSC dysfunction and osteopenia induced by mechanical unloading,rapamycin inhibition of mTOR signaling functions as a downstream therapeutic to protect against TNFαdeficiency.Taken together,these findings suggest that dynamic TNFαrelease and endocytosis safeguard MSC homeostasis and identify an osteoanabolic approach for mechanical unloading.

The regulatory framework of MSC function involves dynamic interactions of MSCs with the surrounding niche,in which MSCs secrete a broad spectrum of cytokines to modulate the niche and respond to various environmental cues for behavioral adaptation.1,16Among the niche factors,immune components,particularly inflammatory cytokines,strongly regulate MSC behaviors in resident and transplanted tissues,and they diminish MSC function in pathophysiological processes to induce tissue deterioration,such as osteoporosis.3–4Interestingly,inflammatory cytokines also prime MSCs to function when MSCs are infused for immunomodulation.5–6In this regard,despite evidence suggesting that MSCs secrete inflammatory cytokines in a proinflammatory functional state,7it remains unclear whether MSC-derived inflammatory cytokines influence their own behaviors under physiological conditions.In this study,we reveal for the first time that MSCs produce and release low levels of TNFα,which is indispensable for the functional maintenance of MSCs.The autocrine/paracrine mode of TNFαregulation of MSCs indicates that MSCs create a beneficial physiological niche for their own homeostasis.Studies have also reported that other inflammatory cytokines,such as IFNγ and IL-6,serve as autocrine regulators of MSCs,but the observations were made during the differentiation process or under inflammatory conditions.10,25Here,we identified that genetic depletion of only TNFα,but not IFNγor IL-6,results in MSC impairment,suggesting that TNFαhas a unique function among inflammatory cytokines to safeguard the stemness of MSCs.Accordingly,our mechanistic experiments reveal that TNFαfunctions through endocytosis to inhibit mTOR signaling,which is independent of TNFR and TNFR downstream pathways.Our findings indicate for the first time a noninflammatory role of TNFαin governing stem cell fate,paving the way for future studies on physiological TNFα-mediated tissue/organ development and homeostatic control.

Fig.6 Mechanical unloading induces MSC functional decline and bone loss that are counteracted by rapamycin.a ELISA analysis of TNFα levels in the bone marrow of control and hindlimb unloading(HU)mice(N=5).b Flow cytometric analyses of TNFα+cells and p-mTOR+cells in bone marrow mesenchymal cells.Prx1-cre:tdTomato mice(N=5)were used to label bone marrow mesenchymal cells in vivo.c Schematic diagram showing the experimental design used to investigate the effect of rapamycin treatment on HU mice in vivo.d Western blot analyses of mTOR signaling(N=3).e–h Functional analyses of MSCs according to CFU,BrdU labeling,and osteogenic and adipogenic differentiation.N=5.Scale bars=100μm.i Calcein labeling for bone formation analysis.N=5.Scale bar=50μm.j Oli red O staining for bone marrow adiposity.N=5.Scale bar=150μm.k Micro-CT analysis of trabecular bone mass(N=5).l Diagram showing mechanical unloading-induced MSC functional decline and bone loss counteracted by rapamycin.For quantification of Western blotting,a two-tailed Student’s t test was used for the comparison between the treatment and control groups.*P＜0.05.Data represent the mean±SD

To our knowledge,the receptor-independent mechanism(s)mediating the effects of inflammatory cytokines have not been revealed previously.Here,inspired by simple observations of the differential influences of TNFαand TNFR deficiency on MSC homeostasis,as well as the decreased levels of TNFαcompared to the concentrations originally added to the MSC culture medium,we discovered that physiological TNFαis endocytosed into MSCs without stimulating TNFR downstream signaling.This novel endocytotic mechanism mediating the effects of TNFαwould therefore be considered irrelevant to the canonical role of TNFαas an inflammatory cytokine.Accordingly,at low and high levels,TNFαfunctions through differential mechanisms to modulate different molecular targets,contributing to stem cell regulation under physiological and pathological conditions,as also indicated in previous studies.4,15Endocytosis integrates nutrient internalization,diverse signal transduction pathways and components in the plasma membrane,thus contributing to cellular and organismal function as a fundamental life process.33For stem cells,it has been reported that clathrin-mediated endocytosis is critical for balancing TGF-βand ERK signaling outputs to regulate embryonic stem cell pluripotency and fate choices and that endocytosis is linked to autophagy to restrict intestinal stem cell proliferation.34–35We have further documented that MSCs use endocytosis to take up apoptotic extracellular vesicles for functional maintenance.24Together with the findings of the present study,these works establish endocytosis as a dynamic yet constitutive mechanism that is indispensable for orchestrating stem cell behaviors.Whether there is a noncanonical mechanism underlying TNFα endocytosis,such as a mechanism associated with inward bending of the membrane to form the endocytic vesicle,remains to be elucidated in future studies.

Rapamycin was first isolated in 1972 from the bacterium Streptomyces hygroscopicus;it was initially identified as an antifungal agent and an immunosuppressant and was later discovered to possess antitumor properties.29Subsequently,mTOR signaling was uncovered as the molecular target of rapamycin,with multiple studies revealing mTOR as a master regulator of cells that senses nutrients and growth factors and integrates downstream cascades to coordinate cell growth,metabolism,and autophagy.29It has been increasingly reported that mTOR signaling is involved in the development and maintenance of various stem and progenitor cell populations.29In particular,mTOR regulation of MSCs has yielded conflicting results,with distinct roles of mTORC1 and mTORC2 in MSC lineage commitment.36In this regard,we have previously reported that activation of mTORC1 suppresses RUNX2 expression and promotes PPARγexpression in MSCs,underlying the shift in MSC differentiation from osteogenesis to adipogenesis in osteopenia.30,37–38In the present study,we further discovered that mTORC2 activation in the context of TNFαdeficiency and mechanical unloading accompanies mTORC1 activation to result in MSC dysfunction,indicating synergistic roles of mTORC2 and mTORC1 in regulating MSC fate.Importantly,we reveal that endocytosed TNFαbinds to mTORC2 to suppress signaling,which represents a novel mechanism of mTOR regulation independent of PI3K,the currently recognized upstream signal of mTOR.29It would be interesting to investigate in future studies whether TNFα control of mTOR equilibrium affects MSC behavioral adaptation to other environmental cues,such as metabolic and autophagic changes in responses to nutrient signals,29thereby coordinating MSC function to maintain tissue/organ homeostasis.

The mechanical properties of the stem cell niche critically influence cell shape and fate decisions,which are mediated by the cytoskeleton and the related mechanosensing pathway.39Changes in the cytoskeleton also contribute to the endocytotic process,which requires dynamic adaptations of the plasma membrane and the transportation of endosomal vesicles.33Not surprisingly,endocytosis has been reported to be influenced by applied forces,31–32which was confirmed in this study for the uptake of TNFα.However,further work is needed to elucidate how mechanical force promotes the endocytosis process and to specify the underlying molecular machinery that responds to force stimuli.Accordingly,it is speculated that loss of mechanical stimuli causes endocytosis failure and MSC dysfunction,indicating that force-driven membrane dynamics are a basic characteristic of living cells.At the organismal level,severe symptoms of skeletal muscle atrophy and bone loss have been observed when individuals are exposed to microgravity during space missions.19,40However,approaches to counteracting mechanical unloading-induced tissue alterations are currently limited.In this study,we used rapamycin,which has been widely applied in the clinic,to effectively protect MSCs and bone homeostasis despite mechanical unloading,highlighting a pharmaceutical therapeutic with osteoanabolic benefits in vivo.Notably,although rapamycin may exert immunosuppressive effects when infused systemically,29we did not observe detrimental side effects during rapamycin application.Instead,the feasibility and effectiveness of using rapamycin to rescue the function of resident MSCs and treat multiple osteopenias have been established by our group,providing a promising therapeutic strategy.30,37–38

In summary,we unraveled a previously unrecognized role and mechanism of physiological TNFαthat involves dynamic release and endocytosis to safeguard the functional homeostasis of MSCs,and we revealed downstream mTOR signaling as a therapeutic target for counteracting mechanical unloading-induced MSC dysfunction and osteopenia.

MATERIALS AND METHODS

Animals

Female C57BL/6J(WT,No.000664),B6.129S-Tnftm1Gkl/J(TNFα-/-,No.003008),B6.129S-Tnfrsf1atm1ImxTnfrsf1btm1Imx/J(TNFR-/-,No.003243),B6.129S7-Ifngtm1Ts/J(IFNγ-/-,No.002287),B6.129S2-Il6tm1Kopf/J(IL-6-/-,No.002650),B6.Cg-Tg(Prrx1-cre)1Cjt/J(Prx1-Cre,No.005584),and B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J(tdTomato,No.007909)mice were purchased from Jackson Laboratory and maintained on a C57BL/6J background for at least ten backcrosses.Prx1-Cre and tdTomato transgenics were interbred to obtain Prx1-Cre;tdTomato mice for tracking mesenchymal cells in vivo.Genotyping was performed by PCR using tail samples from mice and primer sequences provided by Jackson Laboratory.Agematched female littermates were used in all experiments.Female immunocompromised nude mice(Beige nu/nu XIDIII)were purchased from Harlan.Mice were housed under pathogen-free conditions,maintained on a standard 12-h light-dark cycle,and given food and water ad libitum.All animal experiments were performed under institutionally approved protocols for animal research(University of Pennsylvania,Protocol No.805478).

Hindlimb unloading

WT and Prx1-Cre;tdTomato mice at 8 weeks of age were subjected to continuous tail suspension for 2 weeks,which was performed according to previous studies.20Briefly,the mice were individually caged and suspended by the tail using a strip of adhesive surgical tape attached to a chain hanging from a pulley.The mice were suspended at a 30° angle to the floor with only the forelimbs touching the floor,allowing the mice to move and access food and water freely.At sacrifice,the hindlimb bones were sampled for the indicated analyses.

Isolation and culture of mouse MSCs

Isolation and culture of MSCs from mouse bone marrow were performed according to our previous protocol.22,24Briefly,whole bone marrow cells from femora and tibia were seeded,incubated overnight,and rinsed with phosphate-buffered saline(PBS)to remove the nonadherent cells.The adherent cells were cultured with alpha-minimum essential medium supplemented with 20%fetal bovine serum(FBS),2 mmol·L-1L-glutamine,55μmol·L-12-mercaptoethanol,100 U·mL-1penicillin,and 100μg·mL-1streptomycin(all from Invitrogen,USA)at 37°C in a humidified atmosphere of 5% CO2.MSCs were digested with 0.25% trypsin(Invitrogen,USA)and passaged for functional experiments after seeding at appropriate densities.

Isolation and culture of mouse T cells and macrophages

For culture of mouse T cells,splenocytes were collected and treated with ACK lysis buffer(Lonza,Switzerland)to remove red blood cells.Naïve T cells were isolated with a plate-bound antimouse CD3 antibody at 5μg·mL-1(eBioscience,USA).Activated T cells were obtained by stimulating naïve T cells for 48 h with a soluble anti-mouse CD28 antibody at 2μg·mL-1(eBioscience,USA)in Dulbecco’s modified Eagle’s medium(DMEM)supplemented with 10% heat-inactivated FBS,2 mmol·L-1L-glutamine,55μmol·L-12-mercaptoethanol,10 mmol·L-1hydroxyethyl piperazine ethanesulfonic acid(HEPES),1 mmol·L-1sodium pyruvate,100 U·mL-1penicillin,and 100μg·mL-1streptomycin(all from Invitrogen,USA).Th1 cells were obtained by treating activated T cells with 20 ng·mL-1IL-2(PeproTech,USA),20 ng·mL-1IL-12(PeproTech,USA),and 10μg·mL-1anti-IL-4 blocking antibody(BioLegend,USA).

Mouse macrophages were isolated by seeding 2×106femoral ANCs in 6-well plates with 20 ng·mL-1macrophagecolony stimulating factor(M-CSF;PeproTech,USA)in DMEM supplemented with 15% heat-inactivated FBS,100 U·mL-1penicillin and 100μg·mL-1streptomycin(all from Invitrogen,USA).After 48 h,nonadherent cells were removed,and the adherent macrophages were cultured for another 48 h with 20 ng·mL-1M-CSF in the presence or absence of 1 ng·mL-1LPS(Millipore,USA).

Chemical treatments

Chemical reagents and treatments were as follows:recombinant mouse TNFα(PeproTech,USA),mouse RANKL(PeproTech,USA),human TNFα(R&D Systems,USA),and human pro-TNFα(R&D Systems,USA)were added at the indicated concentrations,with a physiological concentration of 1 ng·mL-1used in most experiments.TAPI-2(Millipore,USA)was used at 120 nmol·L-1.Pitstop®2(Abcam,UK)was added at 12μmol·L-1.Rapamycin(Abcam,UK)was intraperitoneally administered at 50 nmol·L-1in vitro and 1.5 mg·kg-1per day in vivo for 2 weeks.30

TNFαoverexpression and knockdown in vitro

For knockdown of TNFαin vitro,serum-starved MSCs were treated with TNFαsiRNA or vehicle siRNA control(Santa Cruz Biotechnology,USA)transfected with the Lipofectamine RNAi-MAX reagent(Invitrogen,USA)according to the manufacturers’instructions.For overexpression of TNFα,green fluorescent protein(GFP)-TNFαfusion protein expression plasmids were kindly provided by Dr.Jennifer Stow(Addgene plasmid No.28089).Empty plasmids with the same backbone were used as the control.Plasmids were transduced using Lipofectamine LTX with Plus reagent(Life Technologies,USA)according to the manufacturer’s instructions.

In vitro cyclic stretching

MSC monolayers grown on collagen-coated Flexcell plates were subjected to cyclic stretching(15% elongation at 0.5 Hz)using a Flexcell Fx-4000T tension unit(Flexcell International,USA)for 2 h periods.Control wells were plugged at the bottom by rubber capping without application of any stretches.

Colony-forming unit(CFU)assay

The formation of fibroblastic colonies(CFU-F)by MSCs was evaluated according to a previous study.22Briefly,a total of 1.5×106all nuclear cells(ANCs)from the bone marrow were seeded in 60 mm culture dishes and cultured for 16 days.The colonies were washed with PBS,fixed with 2% paraformaldehyde(PFA;Sigma-Aldrich,USA),and stained with 0.5% toluidine blue solution(Sigma-Aldrich,USA).The number of cell colonies was counted under a microscope,and groups of more than 50 cells were considered colonies.

BrdU cell proliferation assay

Analysis of MSC proliferation was performed using BrdU labeling.24MSCs were seeded onto 8-well chamber slides(Thermo Fisher Scientific,USA)at a concentration of 2×104cells per well.After adherence,BrdU labeling reagent(Invitrogen,USA)was added to the medium at 1:100 for 48 h.The cells were then fixed with 70%ethanol,denatured with 2 N HCl,and stained with a BrdU Staining Kit(Invitrogen,USA)according to the manufacturer’s instructions.Fluoroshield mounting medium with DAPI(Abcam,UK)was used for counterstaining and mounting.The BrdU-positive cells in five fields of view from each sample were quantified using ImageJ software(National Institute of Health,USA)and are presented relative to the total number of cells.

Osteogenic differentiation

For analysis of osteogenic differentiation capability,22MSCs were cultured in osteogenic inductive medium including 2 mmol·L-1β-glycerophosphate(Sigma-Aldrich,USA),100μmol·L-1L-ascorbic acid phosphate(Wako,Japan),and 10 nmol·L-1dexamethasone(Sigma-Aldrich,USA).After 4 weeks of induction,in vitro mineralization was detected by 1% Alizarin Red S(Sigma-Aldrich,USA)staining,and the positively stained areas were quantified as percentages of the total area using ImageJ software(National Institute of Health,USA).Protein expression levels of osteogenic marker genes were also examined by a Western blot assay,as described below.

Adipogenic differentiation

For adipogenic differentiation,24MSCs were cultured in adipogenic inductive medium containing 500 nmol·L-1isobutylmethylxanthine(Sigma-Aldrich,USA),60μmol·L-1indomethacin(Sigma-Aldrich,USA),500 nmol·L-1hydrocortisone(Sigma-Aldrich,USA),10μg·mL-1insulin(Sigma-Aldrich,USA),and 100 nmol·L-1Lascorbic acid phosphate(Wako,Japan).After 7 days of induction,lipid droplets were stained with Oil Red O(Sigma-Aldrich,USA),and the positively stained cells were quantified as percentages of the total cells using ImageJ software(National Institute of Health,USA).Protein expression levels of adipogenic marker genes were also examined by a Western blot assay,as described below.

Ectopic tissue formation

Ectopic tissue formation was performed to analyze MSC function in vivo,as previously stated.24MSCs(4×106)were mixed with 40 mg of hydroxyapatite/tricalcium phosphate ceramic powder(Zimmer Inc.,USA)as a carrier and were subcutaneously implanted into 8-week-old immunocompromised mice.At 8 weeks after implantation,the implants were harvested,fixed in 4% PFA,decalcified with 10% ethylenediaminetetraacetic acid(EDTA,pH 7.4),embedded in paraffin,and sectioned into 6μm slices.The sections were then stained with hematoxylin and eosin,and de novo formed bone areas were analyzed using ImageJ software(National Institute of Health,USA)and shown as percentages of the total area.

Endocytosis analysis

Recombinant mouse TNFα(PeproTech,USA),human TNFα(R&D Systems,USA),and human pro-TNFα(R&D Systems,USA)were labeled with FITC using the ProtOnTMFluorescein Labeling Kit(Vector Laboratories,Italy)according to the manufacturer’s instructions.FITC-labeled TNFαand pro-TNFαwere then used to treat MSCs on coverslips at 1 ng·mL-1for 24 h.The cells were then fixed in 4% PFA,and the coverslips were mounted using Fluoroshield Mounting Medium with DAPI(Abcam,UK).The FITCpositive cells from five fields per group were quantified and are shown as percentages of the total cell number.FITC intake rates were analyzed using flow cytometric analysis,as described below.

Flow cytometric analysis

For analysis of in situ MSCs,ANCs from the hindlimbs of Prx1-Cre;tdTomato mice were collected and stained with a FITC anti-mouse TNFαantibody(BioLegend,USA)at 1:100 or a PerCP anti-mouse p-mTOR antibody(Thermo Fisher Scientific,USA)at 1:100 for 60 min on ice using Intracellular Staining Permeabilization Wash Buffer(BioLegend,USA).For analysis of MSC endocytosis,MSCs were collected after FITC-labeled TNFαand pro-TNFαtreatments and fixed in 2%PFA.For analysis of MSC surface markers,cultured MSCs were collected and stained with PE-conjugated antibodies against CD73(Thermo Fisher Scientific,USA),CD90(BioLegend,USA),CD105(Thermo Fisher Scientific,USA),CD146(Thermo Fisher Scientific,USA),CD166(Thermo Fisher Scientific,USA),Stem cell antigen 1(Sca1;BD Biosciences,USA),and CD45(Thermo Fisher Scientific,USA)and a PerCP-conjugated antibody against CD34(BioLegend,USA)at 1:100 for 60 min on ice.All samples were analyzed using FACSCaliburwith CellQuest software(BD Bioscience,USA).

qRT-PCR

Total RNA was isolated from cultured cells using the miRNeasy Mini Kit(Qiagen,Germany)according to the manufacturer’s instructions.cDNA was synthesized using SuperScript III(Life Technologies,USA).Real-time PCR was performed using SYBR Green Supermix(Bio-Rad,USA)on a CFX96TMReal-Time PCR System(Bio-Rad,USA).The primers for mouse TNFαwere as follows:forward,5′-CCTGTAGCCCACGTCGTAG-3′;reverse,5′-G GGAGTAGACAAGGTACAACCC-3′.The primers for mouse GAPDH were as follows:forward,5′-TGTGTCCGTCGTGGATCTGA-3′;reverse,5′-TTGCTGTTGAAGTCGCAGGAG-3′.

Western blot

Western blot assays were performed according to previous studies.22,24Cultured MSCs were lysed in RIPA Lysis Buffer with protease and phosphatase inhibitors(Santa Cruz Biotechnology,USA).Protein levels were quantified using the PierceTMBCA Protein Assay Kit(Thermo Fisher Scientific,USA).A total of 20μg of protein was separated by SDS-PAGE(Invitrogen,USA)and transferred to 0.2μm nitrocellulose membranes(Millipore,USA).The membranes were then blocked with 5%nonfat dry milk and 0.1%Tween-20 for 1 h,followed by incubation overnight at 4°C with the following primary antibodies:antibodies against cJun,p-p53,p53,p-NFκB(p50),NFκB(p50),osteocalcin(OCN),poly(ADP-ribose)polymerase(PARP),and peroxisome proliferator-activated receptor gamma(PPARγ)were purchased from Santa Cruz Biotechnology,USA,and were used at concentrations of 1:200;antibodies against TNFR2,runt-related transcription factor 2(RUNX2),PI3K,PTEN,p-mTOR,mTOR,p-Akt,Akt,Caspase 3,p-cJun,cJun N-terminal kinase(JNK),p-JNK,p-NFκB(p65),NFκB(p65),p-p38,p38,active-β-catenin,β-catenin,p-ERK1/2,ERK1/2,p-P70S6K,P70S6K,p-Rictor,p-Raptor,Raptor and p-Sin1 were obtained from Cell Signaling Technology,USA,and were used at concentrations of 1:1 000;the antibody against lipoprotein lipase was purchased from Thermo Fisher Scientific,USA,and was used at a concentration of 1:1 000;antibodies against TNFR1,alkaline phosphatase,p-Smad3,Smad3,and GFP were purchased from Abcam,UK,and were used at concentrations of 1:1 000;antibodies against Rictor and Sin1 were purchased from Bethyl Laboratories,USA,and were used at concentrations of 1:1 000;and the antibody againstβ-Actin was purchased from Sigma-Aldrich,USA,and was used at a concentration of 1:1 000.The membranes were then washed and incubated at room temperature for 1 h with horseradish peroxidase-conjugated secondary antibodies(Santa Cruz Biotechnology,USA).Immunoreactive proteins were detected using SuperSignalTMWest Pico PLUS Chemiluminescent Substrate,SuperSignalTMWest Femto Maximum Sensitivity Substrate(Thermo Fisher Scientific,USA)and Autoradiography Film(Labscientific Inc.,USA).All Western blot data were verified by more than three independent experiments.

ELISA

Supernatants of cell culture media and bone marrow were collected after centrifugation at 12 000 r·min-1for 15 min.Concentrations of TNFαwere analyzed using a Mouse ELISA MAXTMDeluxe kit(BioLegend,USA)according to the manufacturer’s instructions.

Co-IP assay

Cultured MSCs were lysed in RIPA Lysis Buffer with protease and phosphatase inhibitors(Santa Cruz Biotechnology,USA).Protein levels were quantified using the PierceTMBCA Protein Assay Kit(Thermo Fisher Scientific,USA).One microgram of the control IgG,together with 20μL of resuspended protein A/G PLUS-agarose,was added to protein lysates and incubated at 4°C for 30 min.The beads were pelleted by centrifugation at 2 500 r·min-1for 5 min at 4°C,and the supernatants containing total cellular protein(450μg)were collected.Primary antibodies were added at 1:100 and incubated overnight at 4°C followed by incubation with 20μL of resuspended protein A/G PLUS-agarose for 2 h.Immunoprecipitates were then collected after centrifuging at 2 500 r·min-1for 5 min at 4°C and were resuspended in 30μL of electrophoresis sample buffer.Ten-microliter aliquots were subjected to Western blot analysis.All Co-IP data were verified by more than three independent experiments.

mTORC1 activity analysis

Whole-cell lysates were collected from MSCs followed by co-IP with an anti-mTOR antibody.Purified and enriched mTOR complexes were used to measure kinase activity with the KLISATMmTOR Activity Kit(Millipore,USA)according to the manufacturer’s instructions.

Calcein labeling assay

Calcein labeling histomorphometric analysis was performed as reported previously.22Mice were intraperitoneally injected with calcein(Sigma-Aldrich,USA)at 15 mg·kg-1prepared in 2%sodium bicarbonate solution at 10 days and 3 days before sacrifice.Bone formation analyses using MAR and BFR were performed according to the standardized nomenclature for bone histomorphometry under a fluorescence microscope(IX71;Olympus,Japan).

Bone marrow adiposity analysis

Femora were fixed in 4%PFA and decalcified with 10%EDTA(pH 7.4),followed by cryosectioning and staining with Oil Red O solution.Bone marrow adipocytes surrounding the trabecular areas were analyzed.Positively stained areas were quantified using ImageJ software(National Institute of Health,USA)and are shown as percentages of the total area.

Micro-CT analysis

At sacrifice,the femora were fixed in 4%PFA and scanned using a high-resolution ScancoμCT35 scanner(Scanco Medical AG,Switzerland)with a voxel size of 20μm at 70 kV and 200μA.Images were then reconstructed,and data were analyzed by measuring BMD and BV/TV.

Statistics

All data are represented as the mean±standard deviation.Comparisons between two groups were analyzed using independent unpaired two-tailed Student’s t tests,and comparisons among multiple groups were analyzed using one-way analysis of variance with Bonferroni post hoc tests.P values＜0.05 were considered statistically significant.

ACKNOWLEDGEMENTS

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(K99E025915 to C.C.),a Schoenleber Pilot Research Grant(to S.S.)from the University of Pennsylvania School of Dental Medicine,the Guangdong Financial Fund for High-Caliber Hospital Construction,the Postdoctoral Innovative Talents Support Program of China(BX20190380 to B.S.),and the General Program of the China Postdoctoral Science Foundation(2019M663986 to B.S.).

AUTHOR CONTRIBUTIONS

W.Y.designed and performed the experiments,analyzed the data,and wrote the manuscript.C.C.,X.K.,B.S.,T.Y.,D.L.,and R.W.performed the experiments,analyzed the data,and revised the manuscript.J.W.designed the experiments and analyzed and interpreted the data.S.S.conceived the project,supervised the work,analyzed the data,and revised the manuscript.

ADDITIONAL INFORMATION

The online version of this article(https://doi.org/10.1038/s41413-020-00117-x)contains supplementary material,which is available to authorized users.

Competing interests:The authors declare no competing interests.