Introduction

Bone marrow stromal/stem cells (BMSCs) are generally considered as the common progenitors for both osteoblasts and adipocytes in bone marrow, and have been greatly shown for clinical application potential.1 However, their number and function decline with aging, especially the preferential differentiation of aged BMSCs into adipocytes rather than osteoblasts is reasonably accepted as a leading cause of senile osteoporosis (SOP), which is characterized by increased bone marrow fat accumulation and decreased bone loss.2,3,4 Thus, the balance between osteogenic and adipogenic lineage commitment of BMSCs is essential for bone homeostasis. Despite the mechanisms under which the lineage shift occurs in aged BMSCs are not fully clear, accumulated studies have showed that diversity strategies for BMSCs rejuvenation are of benefits for bone quality and even healthspan improvement.5,6,7,8,9,10,11,12,13 For example, modification of transcription factors,6,7 epigenetics10,11,12,13,14 and autophagy5,15,16 that enhanced osteogenesis and decreased adipogenesis of BMSCs alleviated SOP in mice. The latest evidence uncovered that premature aging of skeletal stem/progenitor cells caused bone loss.17 Therefore, it is assumed that stimulation of bone formation by BMSCs rejuvenation in vivo is an effective and attractive strategy for age-associated bone loss.18 Although the only currently approved anabolic agent-teriparatide-a recombinant parathyroid hormone (rPTH), has been shown to stimulate BMSCs osteogenesis and suppress adipogenesis in a variety of manners,19,20,21,22,23,24 several adverse factors, such as progressive blunting of bone formation after 18–24 months treatment, decreased osteoblast precursors and increased osteosarcoma risks, limit its widespread use.18,23

Autophagy is a complex dynamic process involved in the degradation and correction of intracellular damaged proteins and organelles, playing a vital role in bone and other tissue homeostasis.16,25,26 Increasing evidence showed that compromised autophagy is considered as one hallmark of aging and associated with aged-related diseases, including osteoporosis.16,25,26,27,28,29 A previous study also demonstrated that suppression of autophagy in osteocytes resulted in skeletal aging.30 In the bone marrow, autophagy is required for BMSCs homeostasis and plays a fundamental role in the onset and progression of pathological osteoporosis.26 Decreased autophagy has been shown in aged BMSCs, while enhanced autophagy by rapamycin could direct osteogenic lineage of aged BMSCs and rescued bone loss.31,32 Therefore, targeting autophagy that can rejuvenate stem cells is one of the most promising strategies for geroprotection.33 Although rapamycin has been served as a golden standard for autophagy induction and shown exciting results in clinical trials, the adverse effects limit its widely used in the treatment of SOP and other age-related diseases.34,35,36 Hence, more safer autophagy agonists are needed to be explored.

Transcription factor EB (TFEB) is a key transcriptional regulator of autophagy and lysosomal biogenesis. Emerging discoveries demonstrated that TFEB overexpression promoted longevity and reduced the burden of diseases, holding great promise as a therapeutic strategy for multiple age-associated diseases.37,38 Regulation of TFEB has been shown to control the activities of osteoblasts and osteoclasts,39,40 the two main cells playing in the coupling of bone remodeling for homeostasis, implying its potential use in osteoporosis prevention. However, little is known about the relationship between TFEB activities and osteoporosis. As precursors of bone lineage cells, BMSCs directly contribute to bone remodeling by differentiating into osteoblasts, but how and to what extend TFEB regulates fate decision of aged BMSCs in bone marrow is still unclear. Beyond its photodynamic therapy, verteporfin also shows autophagy inhibition,41 but little is known about the detailed effects on different steps involved in autophagy and fate decision of BMSCs.

In this study, we synthesized a novel small molecule compound (named “CXM102”) that could promote autophagy activities in aged BMSCs via enhancement of TFEB nuclear translocation, leading to senescence rejuvenation and bone anabolic effects in middle age mice. Additionally, low dose and long-term administration of CXM102 showed better benefits for healthspan than rapamycin in mice, including extended lifespan, reduced serum levels of inflammation, less lipid droplets and fibrosis in organs. Our results demonstrated that CXM102 could significantly counteract aberrant lineage allocations of aged BMSCs, alleviate osteoporotic bone loss, increase healthspan and longevity of middle age mice.

Results

CXM102 induces autophagic influx in hBMSCs

As AMPK activators, arctigenin and its derivatives show remarkable therapeutic potential against a wide range of human diseases.42,43 But from an unsuccessful experience of synthesis of arctigenin derivatives, we tried to redesign and synthesize a novel compound (Figs. S1a and 1a, hydrogen spectrum in Fig. S1b). Then the bioactivities were screened by cell viability. CXM102 had no cytotoxic effects on several normal cells, including BMSCs, liver cells (L02), synoviocytes and skin fibroblasts (FEK4) (Fig. S1c), but exhibited significant inhibition of cell growth in many cancer cell lines (Hep3B, HCCLM3, HCT116, DU145, A549 and THP-1) and rheumatoid arthritis fibroblast-like synoviocytes (RA-FLS) in a dose-dependent manner (Fig. S1d), suggesting its bidirectional effects on cell growth in human normal cells and cancer cells. Given the roles of arctigenin in activation of AMPK pathway and mesenchymal stem cell fate regulation,43,44 we tried to study the effects of CXM102 on autophagy activities in young and old hBMSCs by comparing with rapamycin. Interestingly, CXM102 significantly increased autophagic activities in both young and old hMSCs, as revealed by elevated levels of LC3II and decreased p62 (Fig. 1b, c). Considering that activation of autophagy can delay cellular senescence in BMSCs,32 we analyzed the effect of CXM102 on autophagic flux in old hBMSCs treated with chloroquine (CQ). Compared with rapamycin, CXM102 increased higher autophagic influx in old hBMSCs (Fig. 1b–e). Additionally, autophagy related genes (Atg5, Atg7) and lysosome related genes (Lamp1, Ctsb, Ctsd) were also upregulated by CXM102, accompanied with the elevated expression of upstream regulatory gene Tfeb (Fig. 1f). These results demonstrated that CXM102 could increase autophagy in young and old hBMSCs.

Fig. 1
figure 1

CXM102 increases autophagic influx in hBMSCs. a Structure of CXM102. Western blotting (b) and quantitative (c) analysis of autophagy influx in young and old hBMSCs stimulated by DMSO (Control), CXM102 or rapamycin with CQ or not. Represent images (d) and relative intensity (e) of LC3II puncta detection in young and old hBMSCs treated with DMSO (Control), CXM102, rapamycin or CQ. f Autophagy and lysosome related gene analysis in old BMSCs treated with DMSO (Control), CXM102 or rapamycin. Results are presented as means ± SD, n ≥ 3, *P < 0.05, ** P < 0.01

Full size image

CXM102 ameliorates cellular senescence of hBMSCs in vitro

It is widely accepted that autophagy declines with age in BMSCs and activation of autophagy could alleviate cellular senescence.15,32 Then we explored the inducible autophagic effects of CXM102 on senescent hBMSCs. Aged BMSCs are characterized by elevated senescence-associated β-galactosidase (SA-β-gal) activities, levels of cyclin-dependent kinase 2a (CDK2a/p16INK4a), phosphorylation of Ser-139 of histone H2A.X (γH2A.X), reactive oxygen species (ROS) and adipogenic bias, but accompanied with declined autophagy, mitochondrial membrane potential (MMP) and osteogenic bias (Fig. S2), which were consistent with others’ works.3,45 However, CXM102 increased autophagosomes, but reduced SA-β-gal activities, levels of p16INK4a, γH2A.X, ROS and adipogenesis, while restored the MMP and osteogenesis in old hBMSCs (Fig. S2). To confirm the effects of CXM102 on senescence alleviation, we used hydrogen peroxide (H2O2) to induce cellular senescence in a hBMSCs cell line. As revealed by SA-β-gal staining (Fig. 2a), 2-h exposure of 200 μmol/L H2O2 was successfully to induce cellular senescence, but pretreatment with CXM102 led to significantly decreased SA-β-gal activities in a dose-dependent manner (Fig. 2a, b). Besides, CXM102 also downregulated levels of p16INK4a, γH2A.X, ROS (Fig. 2c–e). In addition, CXM102 promoted osteogenic and reduced adipogenic differentiation of hBMSCs (Fig. 2f). To detect the effect of CXM102 on mitochondrial integrity, we conducted MMP detection based on JC-1 staining46,47 and found that CXM102 could remarkably restored the lost MMP (Fig. 2g, h). However, pretreatment with 3-methyladenine (3-MA) blocked the CXM102-induced autophagy assembly and failed to rejuvenate old hBMSCs, as charactered by elevated SA-β-gal activities, levels of p16INK4a and γH2A.X, ROS and decreased MMP and osteogenic differentiation (Fig. S2). Taken together, these data demonstrated that CXM102-induced autophagy could rejuvenate aged hBMSCs in vitro.

Fig. 2
figure 2

CXM102 ameliorates the senescence-associated phenotypes of hBMSCs. Representative images (a) and quantitative analysis (b) of SA-β-gal and DAPI staining. Representative images (c) and quantitative analysis (d) of double immunofluorescent staining of p16INK4a and γH2AX. e Representative images and quantitative analysis of intracellular ROS level. f Representative images and quantitative analysis of relative gene levels of Runx2 and Pparγ for osteogenic and adipogenic differentiation with or without CXM102 treatment. Representative images (g) of JC-1 staining and quantitative analysis and intensity analysis of mitochondrial membrane potential (MMP) (h) in g. Results are presented as means ± SD, n ≥ 3. **P < 0.01

Full size image

CXM102 induces autolysosomes formation via TFEB nuclear translocation in aged hBMSCs

Subsequently, we investigated the upstream regulators that control the autophagic activation by CXM102. As CXM102 could upregulate Tfeb and autophagy and lysosome related genes (Fig. 1f), we tried to investigate the protein level of TFEB, which has been considered as a master regulator of lysosomal biogenesis and autophagy.38,48,49 As expected, CXM102 upregulated total TFEB expression in aged hBMSCs (Fig. 3a). Given that the transcriptional activity of TFEB mainly depends on its cellular localization, we found that CXM102 remarkedly increased nuclear translocation of TFEB (Fig. 3b–d). To confirm the role of TFEB in CXM102-induced autophagy, we conducted siRNA and pharmacological inhibitors to explore the activities of TFEB. Transfection with siTFEB could downregulate Tfeb and its downstream genes expression within 24 h, but not in 72 h (Fig. S3a). This data was consistent with the results from immunoblotting and immunofluorescent experiments (Fig. S3b, c). These results suggested that siTFEB transfection failed to inhibit nuclear translocation of TFEB for long-term experiments. However, we screened an inhibitor, verteporfin (VP), for the special regulation of nuclear translocation of TFEB. VP didn’t change the global gene and protein levels of TFEB (Fig. S3a, b), but inhibit its nuclear translocation (Fig. 3b–d) and downstream genes for 72 h (Fig. S3a). Interestingly, CXM102 partially rescued the VP-induced inhibition of TFEB translocation into nucleus (Fig. 3c, d). In accordance with the TFEB activities regulated by CXM102 and VP, markers for autophagosomes (LC3II) and lysosomes (LAMP1) exhibited similar expression (Fig. 3e–g). Correspondingly, VP treatment led to deteriorated senescence of old hBMSCs, including higher SA-β-Gal activities, γH2AX and p16INK4a positive cells and lower MMP (Fig. S3d–f). These results demonstrated that CXM102 induced autophagy via TFEB nuclear translocation to suppress hBMSCs senescence.

Fig. 3
figure 3

CXM102 induces autolysosomes formation via TFEB nuclear translocation in hBMSCs. a Western blotting analysis of total TFEB expression treated with or without CXM102. Representative cytosol and nucleus TFEB analysis (b), immunofluorescent staining for sublocation (c) and relative quantitative analysis (d) of TFEB treated with or without CXM102. e Representative fluorescent images of double staining of LC3II and LAMP1. f, g Quantitative analysis of average LC3II and LAMP1 per cell in c Results are presented as means ± SD, n = 3. *P < 0.05, **P < 0.01

Full size image

CXM102 regulates lineage choice of hBMSCs through nuclear translocation of TFEB

To explore the roles of TFEB in osteogenic and adipogenic commitment of hBMSCs regulated by CXM102, we induced aged hBMSCs to differentiation medium and detected the relationships between TFEB and specific transcription factors for osteogenesis and adipogenesis, Runx2 and PPARγ, respectively. When exposure to osteogenic induction, CXM102 increased osteogenesis of young hBMSCs revealed by ARS staining (Fig. 4a-i). At the molecular level, CXM102 promoted TFEB translocated into nucleus (Fig. S4a) and upregulated the gene and protein expression of Runx2 (Figs. 4b, S4b). However, VP inhibited the nuclear fraction of TFEB (Fig. S4a), downregulated both gene and protein levels of Runx2 (Figs. 4b, S4b) and decreased osteogenesis (Fig. 4a-i). Nevertheless, CXM102 could restore the VP-induced decline in osteogenesis. By contrast, CXM102 significantly induced TFEB translocation into nucleus (Fig. 4a-ii), but downregulated PPARγ expression (Figs. 4b and S4b) and inhibited adipogenesis of young hBMSCs while exposure to adipogenic induction (Fig. 4a-ii). Interestingly, VP also inhibited adipogenesis of young hBMSCs, and showed synergistic inhibition of adipogenesis with CXM102. These results suggested that CXM102 favored osteogenic differentiation of young hBMSCs.

Fig. 4
figure 4

CXM102 remodels the bone-fat differentiation of young and aged hBMSCs in vitro. a Representative images of young hBMSCs responded to (i) osteogenic (ARS staining) or (ii) adipogenic (ORO staining) induction treated with or without verteporfin (VP) and CXM102. TFEB activities and specific transcription factors for osteogenesis (Runx2) and adipogenesis (PPARγ) were also detected by immunofluorescent staining. Arrows indicate the sublocation of TFEB. b Quantitative gene analysis of Runx2 and PPARγ in a. c Representative images of aged hBMSCs treated with or without verteporfin (VP) and CXM102 in a bivalent differentiation system. Arrowheads indicated adipocytes stained with ARS. d Quantitative analysis of relative (i) mineralized nodule formation and lipid droplets in C. Results are presented as means ± SD, n = 3. *P < 0.05, **P < 0.01, n.s. means no significance

Full size image

The preferential shift of lineage commitment into adipocytes of aged BMSCs contributes to increased marrow adiposity and in turn impairs osteogenic regeneration and hematopoiesis. Recently, we have invented a new method that permitted both osteogenic and adipogenic differentiation of BMSCs in the same culture system.50 To better understand the effects of CXM102 on the fate determination of aged hBMSCs, we performed this bivalent differentiation assay to observed the final osteogenesis and adipogenesis. In control group, little mineralized nodules were formed as revealed by ARS staining, in which the lipid droplets were also distinguished (Fig. 4c, arrowheads). Meanwhile, more lipid droplets were detected by ORO staining (Fig. 4d), confirming the preferential differentiation into adipocytes of aged BMSCs. However, CXM102 remarkably reduced lipid droplets and increased mineralized nodules (Fig. 4c, d). Despite VP suppressed osteogenesis and promoted adipogenesis of aged hBMSCs, which was contrary to its effects on young hBMSCs, supplement of CXM102 reduced adipocytes and promoted osteogenesis. Furthermore, these results were further confirmed by gene and protein analysis of Runx2 and PPARγ (Fig. S4c, d), indicating that CXM102 could remarkably direct aged hBMSCs differentiation into osteoblasts at the expense of adipocytes. Taken together, these results suggested that CXM102 could remodel the bone-fat imbalance of aged hBMSCs in vitro.

CXM102 promotes bone formation and ameliorates age-related bone loss in mice

Considering the specifical osteogenesis of CXM102 controlled by VP, whether there were the same effects to regulate bone metabolism in vivo was still unknown. As bone mineral density (BMD) declined with ageing in all groups of mice, CXM102 significantly delayed the loss of BMD value compared with other groups, except for the young mice (Fig. 5a). After a 4-month-long treatment with the drugs, serum levels of the bone formation marker (N-terminal propeptide of type I procollagen, PINP) and bone resorption marker (C-telopeptide of type I collagen, CTX-I) were also analyzed. Compared with the 3-month-old mice (Young), serum PINP was remarkedly decreased in old mice, but was significantly upregulated by administration of CXM102 (Fig. 5b). Conversely, serum CTX-1 increased with aging, but didn’t significantly change upon CXM102 stimulation (Fig. 5c), suggesting that CXM102 may mainly regulate bone formation in vivo.

Fig. 5
figure 5

CXM102 ameliorates aged-related bone loss in male mice. a Representative curves of BMD value and the relative changes of BMD to the starting point followed by 4-month treatment with CXM102 (OC), verteporfin (OV), or both (OVC). Red asterisks indicate the comparing with the Old group, while blue asterisks indicate the comparing with OV group. Serum levels of P1NP (b) and CTX-1 (c) from 3-month (young), 14-month (old) and 18-month-old mice treated with CXM102 (OC), or verteporfin (OV), or both (OVC). d Representative 2D and 3D micro-CT images of the distal femoral trabecular bone microarchitecture. e Quantitative analysis of distal femoral parameters including bone volume to tissue volume (BV/TV), trabecular number (Tb. N), trabecular thickness (Tb. Th), and trabecular separation (Tb. Sp). Quantitative analysis of maximum loading (f) and maximum stiffiness (g) of femurs by a three-point bending assay. h Quantitative analysis of Young’s modulus of femurs. Results are presented as means ± SD, n ≥3. *P < 0.05, **P < 0.01

Full size image

Subsequently, we assessed the effects of CXM102 on bone microarchitecture of the distal femurs by micro-CT scanning. As shown in 2D and 3D images, CXM102 significantly reduced total bone loss compared with the old mice (Fig. 5d, e). Meanwhile, a significant increase of total bone volume (BV/TV), number (Tb. N) and thickness (Tb. Th) of trabecular bone, as well as a decrease of trabecular separation (Tb.Sp.) were presented in OC mice compared with the old mice (Fig. 5e). Furthermore, the maximum loading of bone fracture (Fig. 5f), stress (Fig. 5g) and Young’s modulus (Fig. 5h) in CXM102-treated mice were also significantly increased compared with the old mice, suggesting that CXM102 could effectively improve bone strength and prevent fragility fracture of senile osteoporosis.

However, pretreatment with VP severely attenuated the bone anabolic effects of CXM102 in vivo, leading to lower BMD and serum P1NP, less bone mass and mechanical properties (Fig. 5). In addition, it was also important to note that long-term injection of VP may accelerate the bone loss and worsened senile osteoporosis, but the mechanisms remained unclear.

CXM102 remodels the bone-fat imbalance of aged marrow stromal cells in aged mice

Aged-related osteoporosis is featured with low bone mass and excessive accumulation of adipocytes in bone marrow. Whether the bone anabolic effects of CXM102 under aging were attributed to the lineage shift of aged BMSCs remained unclear. Next, we focused on the effects of CXM102 on bone marrow cells. In contrary to young mice, there were more adipocytes (Fig. 6a-ii, arrowheads) and less osteoblasts in old marrow, but CXM102 administration significantly increased osteoblasts and reduced adipocytes accumulation (Fig. 6a, e, f). These results were further confirmed by specific staining of Runx2 and perilipin (Fig. 6b, g, h). Then we immediately analyzed the effects of CXM102 on TFEB expression of bone marrow cells. Consistently, ageing led to decreased the nuclear expression of TFEB in marrow cells, while CXM102 dramatically increased the nuclear fraction in old mice (Fig. 6c, j). Meanwhile, CXM102 also elevated Runx2 level in marrow cells as revealed by colocalization staining with TFEB (Fig. S4a, c), suggesting the enhanced nuclear level of TFEB triggered by CXM102 promoted osteogenesis in old mice. Similar to the aged hBMSCs, pretreatment with VP blocked the bone-forming effects of CXM102 and reduced the nuclear TFEB positive stromal cells.

Fig. 6
figure 6

CXM102 remodels the bone-fat imbalance of aged marrow stromal cells in middle aged mice. a Representative images of H&E staining for histological analysis of distal femurs from the indicated mice. (i-v) Enlarged areas from each section. b Double-staining of perilipin and Runx2 in bone marrow sections. c Immunohistologic analysis of TFEB in marrow cells. d Immunofluorescent double-staining of LepR and LC3 in bone marrow sections. Quantitative analysis of osteoblasts (e), adipocytes (f), Runx2 positive cells (g), perilipin positive cells (h), osteoclasts (i), nuclear TFEB positive cells (j), relative LepR positive cells (k) and LC3II intensity (l) to old mice. Results are presented as means ± SD, n = 5. *P < 0.05, **P < 0.01

Full size image

Recent studies have demonstrated that BMSCs expressing leptin receptor (LepR+) were major source of bone and marrow adipocytes.24,51,52 To investigate if CXM102 could enhance the autophagy activities of LepR+ BMSCs in vivo, we examined LepR and LC3II expression in the trabecular bone area of distal femur. Our results revealed that the double-positive cells in OC mice were significantly outnumbered by the old group, but was distinctly diminished by VP (Fig. 6d, k, l). These results indicated that CXM102 could promote autophagy activities of LepR+ BMSCs in vivo, but was also inhibited by VP.

Notably, myelofibrosis was also observed in bone marrow of old mice (Fig. S5b, green arrowheads), and VP treatment exacerbated the myelofibrosis. To our surprise, CXM102 significantly suppressed the myelofibrosis in old mice (Fig. S5b, d). Consequently, these results indicated that CXM102 was not only able to remold bone-fat imbalance but also create a healthier environment for fate choice of aged marrow cells under aging.

CXM102 reduces serum inflammation and increases healthspan and longevity in male mice

Given that rejuvenation of BMSCs and autophagic promotion have shown great benefits for decreased inflammation and healthspan improvement, we next determined the long-term effects of CXM102 starting at middle age (14-month-old) in male C57BL/6 mice on aged-related inflammation and lifespan. As low-dose (8 mg/kg weight) of rapamycin is considered as gold standard for healthy aging,34 we compared the long-term effects of the same dose of CXM102 on survival and health parameters with rapamycin (Rapa). The survival curves of control (DMSO), CXM102- and Rapa-treated mice began to separate at 3 months of administration (Fig. 7a). While rapamycin injection led to a 5.68% extension of median lifespan, CXM102 increased up to 22.3%. Chronic inflammation has been considered as a hallmark of aging. In this study, the serum levels of TNF-α, IL-1β and IL-6 in CXM102-treated mice were significantly lower than old and rapamycin-treated mice (Fig. 7b–d). However, our work also showed that rapamycin failed to reduce serum levels of IL-1β and IL-6 compared with control mice (Fig. 7c, d), despite it diminished serum level of TNF-α (Fig. 7b). This result suggested that CXM102 may had a better effect on inflammaging than rapamycin.

Fig. 7
figure 7

Long-term treatment with CXM102 increases healthspan and lifespan in male mice. a Kaplan–Meier survival curve for mice treated either with DMSO, or rapamycin, or CXM102. Gray dashed lines indicated median lifespan. Statistic analysis of serum levels of TNF-α (b), IL-1β (c) and IL-6 (d) at 14-month (before injection) and 20-month (6-month after injection with DMSO, or rapamycin, or CXM102). e Representative images of histological analysis from liver (i, Oiled red O staining), kidney (ii, Masson’s staining) and heart (iii, Masson’s staining). Quantitative analysis of lipid droplets in liver (f), numbers of fibers in glomeruli (g) and diameter of renal tubules (h) in kidney and numbers of fibers (i) in heart. Results are presented as means ± SD, n = 5. *P < 0.05, **P < 0.01, n.s. means no significance

Full size image

Aging is also characterized by multi-organ dysfunction, including but not limited to non-alcoholic fatty liver disease (NAFLD), renal fibrosis and myocardial fibrosis, and rapamycin has been shown to extend lifespan and healthspan effectively.34,35 In this work, we meanwhile compared the histomorphology changes of these main organs treated with DMSO, CXM102 or rapamycin. As fat droplets occurred in liver with aging, 6-month treatment with CXM102 diminished a larger number of fat accumulation than rapamycin did (Fig. 7e–i, f). While rapamycin had little and no significant impacts on renal fibrosis and diameter of renal tubules, CXM102 largely reduced renal fibrosis and increased the diameter of renal tubules, respectively (Fig. 7e-ii, g, h). In parallel to the changes occurred in kidney, heart fibrosis was also substantially decreased in CXM102-treated mice (Fig. 7e-iii and l). Taken together, these results suggested the more beneficial effects of CXM102 on healthy aging than rapamycin.

Discussion

Autophagy plays fundamental roles in the homeostasis and fate determination of BMSCs,15 and an age-related decline in autophagy is widely accepted as one hallmark of aging.25,27,53 Impaired autophagy resulted in adipogenic differentiation,54 while increased autophagy led to osteogenic differentiation of BMSCs32 and improve the therapeutic application of BMSCs.55 In this study, we provided a novel autophagy inducer, CXM102, showing better benefits than rapamycin, to rejuvenate aged hBMSCs through the enhanced nuclear translocation of TFEB. It is noteworthy that CXM102 had no effects on normal cell growth, but exhibited inhibitory effects on multiple cancer cells (Fig. 1d). As autophagy may act as a double-edged sword that both helps tissue homeostasis and facilitates cancer development,28,56 further studies are required to explore the unknown roles of CXM102 in cancers.

TFEB is a master transcriptional regulator of lysosomal biogenesis and autophagy, but little is known about the relationship between TFEB activities and fate decision of aged hBMSCs. In this work, we found that total protein level of TFEB and the targeting genes were downregulated by siTFEB within 24 h, but restored to the initial levels at 72 h after transfection (Fig. S3a, b). This phenomenon could be explained by the immediate disruption of total tfeb transcription caused by siTFEB transfection, thus leading to downregulated TFEB level. But when cultured cells were proliferating, the dose of siTFEB was not enough to disrupt the RNA strands. Considering autophagy is an ongoing and dynamic process, it was not suitable for siRNA to perform long term experiments like differentiation. Fortunately, we found VP had a specific role in prevention of TFEB nuclear translocation (Fig. 3c). VP is an FDA-approved drug that normally used for photodynamic therapy in clinical, but it also inhibits autophagy without light activation.41 Nevertheless, little is known about the detailed effects involved in autophagy. Here we demonstrated that VP could inhibit nuclear translocation of TFEB to suppress the formation of autophagolysosome, decrease osteogenesis in aged hBMSCs. Interestingly, VP has been shown to attenuate heterotopic ossification by inhibiting osteogenesis of tendon stem cells via YAP/β-catenin pathway.57 Consistent with 3-MA (Fig. S2), VP also inhibited the cytoprotective effects of CXM102 on senescence of hBMSCs (Fig. S3d–f). Recent studies showed that the transcriptional activities of TFEB were mainly controlled by mTORC1 and AMPK kinases,37,38,49,58 but how CXM102 regulated the intracellular sublocation of TFEB is still unclear.

Another hallmark of aged BMSCs is the skewed differentiation potency from osteoblast to adipocyte,7 featured by increased lipid droplets formation and less mineralized nodules when exposure to differentiation medium. We found that CXM102 not only enhanced the nuclear translocation of TFEB, but also promoted Runx2 expression in nucleus in respond to osteogenic induction (Fig. 4a-i). Hence, there may be a synergetic effect of CXM102 with Runx2 to regulate the osteogenic differentiation of BMSCs. By contrast, CXM102 specifically inhibited PPARγ expression, thus leading to less lipid droplets formation (Fig. 4a-ii). Additionally, we conducted a bivalent differentiation system to observe the effects of CXM102 on lineage changes of aged hBMSCs. Our results indicated that CXM102 remarkably promoted osteogenesis and suppress adipogenesis of aged BMSCs (Fig. 4c, d), indicating the powerful bone-forming effects of CXM102 that directed osteogenic differentiation of aged BMSCs.

SOP is featured with increased marrow adipocytes, decreased BMD, bone mass and strength, leading to an increased risk of fractures.59 Here we demonstrated that CXM102 significantly delayed the loss of BMD and bone mass, but augmented bone strength (Fig. 5), suggesting its potentially use for osteoporosis. As the only tissue where bone and fat coexist in the same microenvironment, the bone marrow offers a unique window into the investigation of molecular events governing the lineage commitment of BMSCs.6 Considering that CXM102 promoted the nuclear translocation of TFEB, autophagy and preferential osteogenesis of aged hBMSCs in vitro, the in vivo anti-osteoporosis effects could be reasonably explained by the lineage shift of aged BMSCs conducted by CXM102, which was consistent with reduced marrow adipocytes, increased osteoblasts and autophagy activities (Fig. 6).

Besides, myelofibrosis is another challenge of aging. We found that myelofibrosis increased with age, but CXM102 significantly inhibited the myelofibrosis (Fig. S4b). Since LepR+ BMSCs have been identified as key components of hematopoietic microenvironment52,60,61 and responsible for myelofibrosis,62,63 it was reasonable that enhancement of TFEB-mediated autophagy by CXM102 treatment in LepR+ BMSCs could reduce the myelofibrosis, but further studies are needed to illustrate the detailed mechanisms.

In addition, we also found that long-term injection of VP accelerated bone loss and weaken the strength, accompanied with less osteoblasts, more adipocytes and osteoclasts, reduced nuclear TFEB activities in marrow cells, reminding that more attention should be paid for clinical use, despite CXM102 could partially alleviate the adverse effects (Figs. 5 and 6). Taken together, these results suggested that CXM102 could remodel a healthier osteogenic microenvironment in aged bone marrow for lineage allocation of BMSCs and hematopoiesis in vivo.64

Aging is characterized by accumulation of chronic inflammation and progressive loss of physiological integrity, leading to multiorgan dysfunction and reduced quality of life.18,65,66,67 Accumulation of marrow adipocytes results in increased inflammation, and osteoclast formation and bone resorption by secreting (pro)inflammatory factors. Compared with control and rapamycin-treated mice, our results revealed that CXM102 significantly reduced serum levels of (pro)inflammatory factors, including TNF-α, IL-1β and IL-6 (Fig. 7b–d). Given that bone remodeling plays key roles in homeostasis of bone marrow microenvironment,64 we proposed that the increased bone anabolism and attenuated serum inflammation may contribute to healthier hematopoiesis for bone and other organs that led to better survival in CXM102-treated mice. Osteoimmunology is termed to study the interactions between skeletal and immune systems,68,69,70 but whether CXM102 play roles in marrow inflammation and immune development is of great interest in future studies.

In summary, we clearly demonstrated the novel compound CXM102 could rejuvenate and remodel bone-fat imbalance of aged BMSCs via enhanced nuclear translocation of TFEB, leading to alleviated bone loss and increased healthspan and longevity in middle age male mice. Our results will shed new lights on therapeutic strategies for senile osteoporosis and healthy aging.

Limitation of this study

We acknowledge several limitations of this study. First, we did not determine the effects of CXM102 on female mice, mainly because of the possible perturbation of estrogen deficiency in female mice for senile osteoporosis. Second, only 4 and 5 mice were analyzed in the study of osteoporotic features and longevity, respectively. Despite the data within each group are relatively centralized, larger number of both male and female mice will be required to confirm the anti-osteoporosis effects of CXM102 in vivo. Third, whether the LepR+ BMSCs are truly responsive to CXM102 and how the lineage choice changes in mice are unclear. Maybe more sophisticated experimental techniques like lineage-tracing models, will address these urgent issues. Finally, the post-transcriptional modification of TFEB has been extensively elucidated, yet it is important to uncover the mechanism that how CXM102 regulates the TFEB activities in subsequent studies.

Materials and methods

Materials and reagents

All reagents used for cell culture, including low-glucose Dulbecco’s Modified Eagle’s Media (DMEM; Gibco, 11054, 020), fetal bovine serum (FBS; Gibco, 10099, 141C), non-essential amino acids (Gibco, 11140, 050), antibiotics (P/S; Gibco, 15070, 063) and trypsin (Gibco, 15400, 054) were purchased from Thermo Fisher Scientific Incorporation. Recombinant basic fibroblasts growth factor (bFGF; 100-18B), epidermal growth factor (EGF; AF-100-15) were purchased from PeproTech Inc.. PBS (P1020), Chloroquine (CQ; IC4440), rapamycin (R8140) and 3-Methyladenine (3-MA; IM0190) were purchased from Beijing Solarbio Science & Technology Co., Ltd (Solarbio Inc.). Hydrogen peroxide solution (H2O2; Sigma, 88597) was from Merck group. All materials and regents for western blotting were purchased from Beyotime Institute of Biotechnology (Beyotime), including polyvinylidene difluoride membranes (FFP19), Bovine Serum Albumin (BSA; ST023), TBST buffer (ST673).

All drugs used for in vitro differentiation of hBMSCs, including dexamethasone (Sigma; D4902), β-glycerophosphate (Sigma; G9422), L-ascorbic acid (Sigma; A4544), indomethacin (Sigma; I7378), isobutyl methylxanthine (Sigma; IBMX, I7018) were purchased from Merck. Insulin (I8830) was from Solarbio Inc. Verteporfin was from MedChemExpress LLC (MCE; CL318952). CXM102 was synthesized in our lab as described in a patent (CN23NN14118I).

Primary and secondary antibodies used in this study were shown in Table S1. Cytochemical staining dyes, including Alizarin red S (G8550) and Oil red O (O8010) were from Solarbio Inc. The working solution was prepared based on the manufacture’s instruction.

Animal approval

All animal experiments were performed in accordance with the protocol approved by Guangxi University of Chinese Medicine Institutional Welfare and Ethical Committee (Approval No. DW2022). C57/BL6 mice were reared in a pathogen-free facility at a controlled temperature (23–25 °C), under a 12 h light and dark cycle with food and water provided ad libitum. The cage and bedding were changed once a week. 3-month-old and 14-month-old mice were used in this study.

Cell culture

Young, old and immortal hBMSCs, normal synoviocytes, FEK4 and RA-FLS were kindly gift from Dr. Li Yang (Chongqing University) and maintained as described before.50,71 Human normal liver cells (L02) and cancer cell lines, including Hep3B, HCCLM3, HCT116, DU145, A549 and THP-1 were from Dr. Erwei Hao (Guangxi University of Chinese Medicine). Briefly, 5 × 105 cells/mL were plated in a 100-mm diameter cell culture dish containing low-glucose DMEM supplemented with 10% FBS, 100 U/mL P/S, 0.1 mmol/L non-essential amino acids, 20 ng/mL bFGF and 20 ng/mL EGF. Cells were maintained in a humidified incubator at 37 °C with 5% CO2. When 70% to 80% confluent, adherent cells were trypsinized with 0.05% trypsin-1mmol/L EDTA at 37 °C for 2 min, harvested, and expanded for future use.

Effects of CXM102 on cell growth

To study the effects of CXM102 on cell growth, a density of 2000 cells per well of indicated cell types were seeded into a 96-well plate as described before.71 cells were treated with different concentrations of CXM102 (0, 5, 10, 20, 30, 50 μmol/L) for different time points (12, 24 or 36 h). Cell activities were determined by Cell Counting Kit-8 kit (Beyotime, C0038) following the manufacturer’s instruction.

Effects of CXM102 on hBMSCs senescence

To study the effects of CXM102 on cellular senescence, hBMSCs were pretreated with different concentrations of CXM102 (0, 5, 10, 20 μmol/L) 2 h before 200 μmol/L H2O2 addition into culture medium for 2 h at 37 °C, as described by others,72 followed by fresh medium change containing 20 μmol/L H2O2 for 4 h. Then hBMSCs were fixed and analyzed the levels of senescence β-Galactosidase activities (SA-β-Gal; Beyotime, C0602), reactive oxygen species (ROS; Beyotime, S0033S), mitochondrial membrane potential (MMP; JC-1, Beyotime, C2006) according to the manufacturer’s instruction. Molecules levels of p16INK4a and γH2AX, osteogenic and adipogenic differentiation were also detected as described below.

Effects of CXM102 on autophagy

To investigate the roles of autophagy and transcription factor EB (TFEB) in hBMSCs senescence, 2.5 mmol/L 3-MA or 50 μmol/L CQ was added into the culture medium for 4 h before CXM102 treatment. Then cells were fixed and detected by immunofluorescence (IF) or western blotting for the analysis of autophagic influx, autophagosomes, autolysosomes and TFEB activities as described below.

Immunofluorescent (IF) /Immunohistological (IHC) analysis

For immunofluorescent/immunohistochemical staining, cells or deparaffined sections of femurs were fixed with 4% paraformaldehyde (Beyotime, P0099), permeatilized with 0.2% Triton X-100 (Solarbio, T8200), and blocked with 1% bovine serum albumin (Solarbio, SW3015). Then samples were incubated with primary antibodes (Table S1) or non-immune antiserum as a negative control at 4°C overnight. After washing with PBS, sections were incubated with HRP-conjugated or fluorescent secondary antibodies (Table S1) for 1 h, and counterstained with DAPI for 10 min at room temperature. Images were obtained by upright microscope (BX53, Olympus), or fluorescence microscopy (DMI8, Leica). For quantitative analysis, at least four independent samples of each group were analyzed by Image-Pro Plus 6.0 software.

RNA interference and transfection

To study the role of TFEB in CXM102-mediated autophagy, the following sequence of siRNA was used for TFEB downregulation. siRNA scramble sense, 5’-UAGCGACUAAACACAUCAAU‑3’ and antisense, 5’‑UUGAUGUGUUAGUCGCUAU‑3’; siTFEB sense, 5’-GGAUCAAGGAGCUGGGAAUUU‑3’ and antisense, 5’- AUUCCCAGCUCCUUGAUCCUU-3’ (GenePharma, Suzhou, China). The transfection of siRNA was conducted as we did before.73 Briefly, hBMSCs (2 × 105 cells per well) were seeded into a 24-well plate, then 50 nmol/L siRNA scramble or siTFEB was transfected into hBMSCs with Lipofectamine 2000 (Invitrogen, 11668, 019) according to the manufacturer’s instructions. Finally, the protein level was assessed by IF staining or western blotting.

Quantitative real-time PCR

Total cellular RNAs were isolated using RNA Extraction Kits (Solarbio, R1200) according to the manufacturer’s instructions and quantified by Nanodrop One C (Thermo Fisher Scientific, Madison, USA). 1 μg RNA was analyzed by using a TaqMan One Step RT-qPCR Kit (Solarbio, T2210) on LightCycle 96 Instrument (Roche, Mannhein, Germany) according to the manufacturer’s instructions. The primers used in this study are shown in Table S2. The relative expression of target genes was normalized to Gapdh and calculated using the 2−∆∆CT method.

Western blotting

Total protein was extracted using a protein extraction kit (Solarbio, BC3710) and the protein concentration was determined by BCA (Beyotime, P0012) assay according to the manufacturer’s instructions. Next, 25 μg of protein sample was subjected to 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, followed by transfer to polyvinylidene difluoride membranes. Then the membranes were blocked with 3% BSA in TBST buffer for 2 h at room temperature, followed by incubation with the indicated primary antibodies (Table S1) at 4 °C overnight and horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Targeted proteins were visualized with enhanced chemiluminescence solution (Beyotime, P0018S) and detected using a ChemiDoc Touch Imaging system (Bio-Rad, CA, USA).

Effects of CXM102 on differentiation induction

Osteogenic, adipogenic, and bivalent differentiation and detection of BMSCs were strictly conducted as we described in a recent method article.50

Animal experiments

To study the effects of CXM102 on aged-related bone metabolism, 14-month-old mice were divided into four groups, followed by a period of 4-month-long treatment with DMSO (Indicated as “Old”), CXM102 alone (OC; 20 mg/kg body weight), verteporfin alone (OV; 10 mg/kg) and CXM102 combined with verteporfin (OVC). In OVC group, verteporfin was pretreated 2 weeks before CXM102 injection. All drugs were dissolved into 50 μL DMSO and injected subcutaneously twice a week. Mice were sacrificed 1 month after the final injection, serum and femurs were collected for analysis.

BMD measurement

To assess the effects of CXM102 on BMD changes in vivo, the BMD values of 14-month-old mice were measured as the initial time and then followed by a period of 4-month injection with DMSO (Marked as “Old”), CXM102 (OC), verteporfin (OV) alone or both (OVC). In addition, 3-month-old untreated mice (Marked as “Young”) were used as positive control in this experiment. BMD was measured once a month by dual energy x-ray absorptiometry (DXA) as described by others.74 Briefly, mice were anesthetized with isoflurane (Yuyan Corporation, Shanghai, China) in a chamber, then were placed on the Lunar PIXImus densitometer platform (GE Medical-Lunar, Madison, WI, USA). Half of the distal femur was selected as area of interest and perform a scout scan. BMD value was recorded by the PIXImus software (Lunar Piximus 2.0).

ELISA assay

Mouse blood serum was diluted 1:1 using PBS and the diluted serum levels of P1NP (Sangon Biotech, D721053), CTX-1; (Sangon Biotech, D721204), TNF-α (Sangon Biotech, D721217), IL-1β (Sangon Biotech, D721017) and IL-6 (Sangon Biotech, D721022) were measured by ELISA kits according to the manufacture’s instruction.

MicroCT analysis

MicroCT analysis was performed using the settings according to the manufacture’s instruction. Generally, Mouse femurs were dissected, fixed with 4% paraformaldehyde (Beyotime, P0099) for 48 h and stored in 70% ethanol, loaded into 10-mm diameter scanning tubes, and scanned with a peak tube voltage of 50 kV and current of 0.145 mA at 18 μm resolution, (Latheta LCT200, Hitachi Aloka, Japan). The 2D and 3D model visualization software (BeeViewer v3.4.1) and data analysis software (VGStudioMAX v2.2) were applied to analyze the distal femoral metaphyseal trabecular bone and the parameters of the diaphyseal cortical bone. We defined a region of interest (ROI) as the area between 2%–6% proximal to the growth plate in the distal femora. Cross-sectional images of the femora were established for 3D histomorphometry analysis of the trabecular bone. The percentage of bone volume (BV/TV), the number (Tb. N), thickness (Tb. Th) and space (Tb. Sp) of trabecular bone were analyzed.

Histological analysis

Histological analysis of femurs was performed as we did before.71 Briefly, the dissected distal femurs were fixed in 4% paraformaldehyde (Beyotime, P0099) for 48 h, decalcified with 15% EDTA (pH7.2-7.4) and were paraffin embedded. 5mm-thickness sections were cut and subjected to haematoxylin and eosin (H&E; Solarbio, G1120) or Masson’s trichrome staining (Solarbio, G1340). Images were taken by a microscope (BX53, Olympus) and analyzed by Image-Pro Plus 6.0 (Media Cybernetics, MD, USA) software.

Biomechanical testing

To investigate mechanical strength of femurs, an electronic universal material testing machine (AGS-10kNJ, Shimadzu, Japan) was used for the three-point bending test as described in other’s experiment.31 Before the test, the long and short diameters of the femoral shaft were measured with a Vernier caliper, then femurs were placed between two support abutments. The load was applied to middle of the femoral shaft through a touching probe. During this test, the posterior surface was in tension, and the anterior surface was in compression. The strength variables (bending modulus of elasticity, bending energy, maximum bending stress, and bending rigidity coefficient) were calculated with a standard engineering formula.

Lifespan analysis

To compared the effects of CXM102 and rapamycin on lifespan, middle-aged mice were divided into three groups, which were administered with rapamycin (8 mg/kg weight), CXM102 (8 mg/kg weight) or DMSO twice a week for a period of 6-month-long term. 1 month after the final injection, three mice from each group were sacrificed for assessment of histological changes of liver, kidney and heart. The other mice were freely fed until natural death. Survival curves were calculated by Origin 8.0 (OriginLab, Guangzhou, China).

Statistical analysis

Results are represented as means ± standard deviations. Statistical analysis was performed using Student’s t-test as well as one-way analysis of variance (ANOVA) followed by the Tukey HSD test for post hoc comparison (Origin 8.0, OriginLab). Difference was considered significant when P < 0.05 indicated as *, while more significant when P < 0.01 indicated as **.