MSC osteogenic induction medium (OIM), adipogenic induction medium (AIM) and chondrogenic induction medium (CIM) were procured from Cyagen Bioscience. For Western blot analysis, antibodies against ATG5, ATG12, Beclin1, LC3A/B, SOX2, P-PI3K, RUNX2 were obtained from Cell Signaling Technology (Beverly, MA, United States). Antibodies specific for P62, P53, AKT, P-AKT, PI3K, mTOR, P-mTOR, GAPDH were supplied from Proteintech (Wuhan, China). Antibodies against NANOG, OCT4, ACAN, COL2, AIPOQ, PPARγ2 were purchased from Abcam (Cambridge, United States). Antibodies specific for P16 and P21 were obtained from Finetest (Wuhan, China). Antibodies specific for ALP and OCN were procured from Affinity (Cincinnati, United States). For flow cytometry analysis, the antibodies against CD34, CD44, CD45, CD29, CD90, CD105 were obtained from Elabscience Biotechnology (Wuhan, China). For immunofluorescence staining, antibodies specific for P16 and P53 were purchased from Proteintech. Antibody against P21 were obtained from ABclonal (Wuhan, China).

Isolation of BMSCs was performed according to previous study (Tu et al., 2020). Briefly, whole bone marrow was collected from femurs and tibias of male Sprague-Dawley rats (6–8 weeks old). The bone marrow was suspended using growth medium (GM) which is DMEM/F12 medium (HyClone, USA) containing 10% FBS (Gibco, USA), 1% penicillin and streptomycin (Sigma-Aldrich, USA). The medium was changed every 2 days, and adherent cells were kept. Cells were passaged using 0.25% trypsin-EDTA solution (Gibco, USA) When they reached approximately 80% confluence. The fourth or thirteenth passage was used for following experiments.

For BMSCs identification, specific cell surface markers and BMSCs-multipotent potential for osteogenic, adipogenic and chondrogenic differentiation were detected.

The EMF device (Figure 1) was designed and manufactured by the CH-Hall electronic devices corporation (Beijing, China). Concisely, the device comprises a waveform generator, a gauss meter, an amplifier, a hall probe, Helmholtz coils and a control computer. After finishing the parameters setting on the computer, the signals were created by the waveform generator, being amplified and transferred to the coils. The real-time frequency as well as the strength of the EMF could be measured by the gauss meter and hall probe. The coils generating EMF were placed in an incubator with a humidified atmosphere (37°C, 5% CO₂). In this study, we used sinusoidal EMF with a parameter of 0.4 mT, 50 Hz. During exposure, cultured BMSCs or rats were placed in the center of the coils.

BMSCs at a density of 1,000 cells/well were seeded in 96-well plates. The cell proliferation was analyzed using a CCK-8 kit following the standard protocol (Boster, Wuhan, China). Concisely, 10 μL CCK-8 solution mixed with 100 μL GM was added to each well. After 2 h incubation, a microplate reader (Bio-TEK Instruments, United States) was used to measure the OD value per well at 450 nm. CCK8 assays were conducted from day o to day 6. All OD values were normalized to those of day 0.

BMSCs at a density of 400 cells/dish were seeded in a 3.5-cm dish. The cells were cultured in GM which was half exchanged every 3 days. After 7 days of culture, cells were washed with PBS and fixed using 4% paraformaldehyde (Sigma-Aldrich, United States). Subsequently, cells were stained using 0.1% toluidine blue solution (Sigma-Aldrich, United States). Colonies including more than 50 cells were counted and the colony-forming ratio was acquired by colony numbers/400 (%).

BMSCs at a density of 2,000 cells/well were seeded in 96-well plates. The cells were labeled with EdU solution (Beyotime, Shanghai, China) and continued culturing for 6 h. Then, cells were fixed using 4% paraformaldehyde and treated with 0.3% Triton X-100 for 15 min. Afterwards, the cells in each well were added with click reaction mixture and incubated in the dark for half an hour. The nucleus was dyed by DAPI for 10 min and cells were observed using a fluorescence microscope.

BMSCs at a density of 5,000 cells/well were seeded in 24-well plates. The cells were cultured in GM before they reached approximately 70% confluence. Then the cells were cultured with osteogenic induction medium (Cyagen Bioscience, United States) for next 2 weeks. Afterwards, cells were washed with PBS and fixed using 4% paraformaldehyde. Finally, the induced cells were stained using 40 mM Alizarin Red S (Sigma-Aldrich).

BMSCs at a density of 5,000 cells/well were seeded in 24-well plates. The cells were cultured in GM before they reached approximately 100% confluence. Then the cells were cultured with adipogenic induction medium (Cyagen Bioscience, United States) for next 2 weeks. Afterwards, cells were washed with PBS and fixed using 4% paraformaldehyde. Finally, the induced cells were stained using filtered Oil Red O (Sigma-Aldrich) at a concentration of 60%.

BMSCs at a density of 5,000 cells/well were seeded in 24-well plates. The cells were cultured in GM before they reached approximately 100% confluence. Then the cells were cultured with chondrogenic induction medium (Cyagen Bioscience, United States) for next 2 weeks. Afterwards, cells were washed with PBS and fixed using 4% paraformaldehyde. Finally, the induced cells were stained using Alcian Blue solution (pH 2.5, Sigma-Aldrich).

Cells were harvested and lysed using RIPA buffer (Boster, Wuhan). Then 25 μg protein samples were transferred using PVDF membrane (Millipore, United States) with a Bio-rad blotting system. The membranes were blocked using 5% bovine serum albumin (Sigma-Aldrich) and incubated with corresponding primary and secondary antibodies. All bands were visualized using Western ECL Substrate Kit (Thermo, United States) with a Bio-rad image capture system. For signal pathways detection, cells were first serum-starved overnight and treated with or without EMF for 30 min.

To determine the senescence of BMSCs, SA-β-gal staining was employed. Briefly, BMSCs at a density of 5 × 10⁴ cells/well were seeded in 6-well plates. The cells were washed with PBS and fixed using 4% paraformaldehyde. Then cells were stained using SA-β-gal staining solution (Beyotime) according to the manufacturer’s protocol. Aged BMSCs stained in blue were photographed with an inverted microscope.

BMSCs at a density of 5,000 cells/well were seeded in 24-well plates. Cells were washed with PBS and fixed using 4% paraformaldehyde at room temperature. Then cells were permeabilized using 0.2% Triton X-100 (Sigma-Aldrich) and blocked using 5% bovine serum albumin (Sigma-Aldrich). Next, BMSCs were incubated with primary antibodies overnight at 4°C. Afterwards, cells were incubated with appropriate fluorescent secondary antibodies in the dark. Finally, the cell nuclei were stained using DAPI for 10 min. After rinsing, cells were observed using a fluorescence microscope.

The RNA sequences of ATG5 siRNAs were designed and synthesized by HanBio (Shanghai, China). Briefly, BMSCs were cultured in a 6-well plate (2 × 10⁵ cells per well). Lipofectamine 2000 (Invitrogen) was used to deliver siRNA into BMSCs following the manufacturer’s instructions. 24 h post transfection, the inhibitory efficiency of specific silencing was examined by the Western blot and qPCR. For the experiment with ATG5 overexpression adenovirus infection, the BMSCs were infected with adenovirus (HanBio) at a multiplicity of infection (MOI) of 60 for 4 h and the efficiency of ATG5 overexpression was observed 24 h post infection. Overexpression was also assessed using Western blot analysis and qPCR.

All animal experiments were in accordance with Wuhan University Committee on the Use and Care of Animals. The operation procedure was carried out as described before (Dubus et al., 2022). Briefly, male SD rats (weigh 280–320 g) were anesthetized through intraperitoneal injection of 3% pentobarbital. A 1 cm sagittal incision down to the periosteum was made on the skull and a 6-mm diameter (critical-size) defect was created on single side of the parietal bone using a micro-trephine. The 10% gelatin methacryloyl (GelMA) hydrogels (Sigma-Aldrich) was used as a carrier of BMSCs. All uniform defects were randomly allocated into four groups: 1) Control group including acellular GelMA hydrogel; 2) P4 group including GelMA hydrogel seeded with BMSCs at passage 4; 3) P13 group including GelMA hydrogel seeded with BMSCs at passage 13; 4) P13+EMF group including P13 BMSCs-laden GelMA hydrogel combined with EMF treatment (50 Hz, 0.4 mT, 4 h/day). All rats fully recovered 24 h after surgery. Efforts were made to minimize animal sufferings.

Six rats per group were sacrificed by injection of an overdose of anesthetics at 8 weeks after surgery. The harvested rat craniums were fixed using 4% paraformaldehyde for 3 days. Then the fixed samples were scanned using a micro-CT (μCT 40, Scanco 274 Medical, Switzerland). The defect region as well as the surrounding bone was reconstructed using VGStudio software at a constant threshold (a voltage of 80 kV, a current of 145 mA, and a resolution of 9 µm pixel⁻¹). The bone volume/total volume (BV/TV) as well as the bone mineral density (BMD) within the bone defect of each group were calculated and analyzed.

Following CT scanning, all specimens were decalcified using 10% EDTA solution for 1 month. Then samples were dehydrated in graded ethanols, embedded within paraffin and cut into 5-μm slices. The slices were stained with HE as well as Masson’s trichrome. New bone area fraction was calculated as new bone area/defect area within the defect of each section. Immunohistological staining was also performed using antibodies against ALP.

Data were presented as mean ± SD. Statistical analysis were evaluated by Student’s t test or one-way ANOVA followed by Tukey post hoc test. For quantitative analysis of Western blot, Immunofluorescence staining and Immunohistological staining, ImageJ software was employed. p < 0.05 was recognized as statistically significant.

Harvested primary BMSCs were firstly identified via detecting cell surface markers and multipotent potential for differentiation. As shown in Supplementary Figure S1, primary cells were positive for CD29, CD44, CD90, CD105 while negative for CD34 and CD45. Meanwhile, under specific induction medium, cells were able to differentiate to the osteogenic, adipogenic and chondrogenic lineages.

To determine the effects of EMF on proliferation capacity of BMSCs during long-term in vitro expansion, we chose rat BMSCs cultured for four passage (P4) and 13 passages (P13) for analysis. According to the EdU incorporation analysis, lower percentage of proliferating cells was observed in P13 BMSCs compared with P4 BMSCs, while 24 h EMF treatment could improve the decreased proliferation of P13 BMSCs (Figures 2A, B). We also performed the CCK-8 (Figure 2C) and fibroblastic colony-forming (Figures 2D, E) assay, BMSCs showed decreased proliferation rate as well as lower formation of colony forming units during the long-term in vitro culture. However, EMF could partly reverse this change. NANOG, OCT4 and SOX2 are three key transcription factors for maintaining cell stemness. From the Western blot analysis, P13 BMSCs exhibited reduced expression of NANOG, OCT4 and SOX2 compared with P4 BMSCs. Interestingly, 24 h EMF exposure alleviated the stemness loss of long-passaged BMSCs (Figures 2F, G).

We then explored the effects of EMF on differentiation potential of BMSCs during long-term in vitro culture. P4 BMSCs, P13 BMSCs with or without EMF exposure were cultured in osteogenic induction medium, adipogenic induction medium and chondrogenic induction medium respectively. After 7 days culture in inductive medium, Western blot analysis indicated that P13 BMSCs exhibited lower expression of RUNX2, OCN, ALP (Figures 3A, B), ACAN, COL2 (Figures 3E, F) and higher expression of ADIPOQ and PPARγ2 (Figures 3C, D) compared with P4 BMSCs. However, EMF treatment could significantly partly reverse these changes. After 14 days culture in inductive medium, we next performed the Alizarin Red staining (Figures 3G, H), Oil Red O (Figures 3I, J) staining and Alcian Blue staining (Figures 3K, L) in three groups, results showed the same tendency. Taken together, EMF exposure effectively rescued the decline of differentiation potential of BMSCs during long-term in vitro expansion.

To verify the effects of EMF exposure on BMSCs cell senescence, we firstly conducted the SA-β-gal staining. Results indicated P13 BMSCs showed notably higher SA-β-galactosidase activity compared with P4 BMSCs, while 24 h EMF treatment could significantly ameliorate this process (Figures 4A, B). Furthermore, By Western blot analysis, senescent BMSCs (P13 BMSCs) showed lower expression of P16, P53 and P21 after 24 h EMF exposure (Figures 4C, D). We also performed the immunofluorescence staining of P16, P21 and P53 in three groups, results exhibited the same trend (Figures 4E–G).

To determine the role of autophagy in cell senescence during long-term in vitro passaging, we firstly examined the autophagy level in P4 and P13 BMSCs. As shown in Supplementary Figures S2A, B, P13 BMSCs exhibited lower expression of ATG5, ATG12, Beclin-1, LC3B/A and higher expression of P62. Moreover, increased levels of P-PI3K, P-AKT and P-mTOR were observed in P4 BMSCs compared with those of P13 BMSCs (Supplementary Figures S2C, D). Above results suggested long-term in vitro expansion of BMSCs was accompanied with reduced autophagy. We then used autophagy inducer rapamycin in P13 BMSCs, cells showed decreased SA-β-galactosidase activity after rapamycin treatment (Supplementary Figure S3A, B). Meanwhile, the Western blot and immunofluorescence staining detection suggested aging-associated markers including P16, P53 and P21 were significantly decreased in P13 BMSCs after rapamycin stimulation (Supplementary Figures S3C–G). Additionally, over-expression of ATG5 in P13 BMSCs could inhibit the SA-β-galactosidase activity (Figures 5A, B) and the level of P16, P53 and P21 (Figures 5C–G). We also examined the cell senescence in P4 BMSCs by silencing ATG5. Higher levels of SA-β-galactosidase activity (Figures 6A, B) as well as increased expression of P16, P53 and P21(Figures 6C–G) were observed in P4 BMSCs after ATG5 silencing. Taken together, our results suggested impaired autophagy contributed to the BMSCs aging during the long-term in vitro expansion.

To illustrate whether EMF exposure inhibited BMSCs cell senescence by regulating autophagy, we used siRNA to knock down ATG5 as well as autophagy inhibitor chloroquine (CQ). As shown in Supplementary Figures S4, 24 h EMF exposure triggered the increased autophagy level in long passaged BMSCs (P13 BMSCs). P13 BMSCs showed decreased levels of SA-β-galactosidase activity after 24 h EMF treatment, while knockdown of ATG5 (Figures 7A, B) or using CQ (Supplementary Figures S5A, B) could antagonize this change. Cell senescence-associated proteins including P16, P53 and P21 were decreased in P13 BMSCs after 24 h EMF exposure. However, the Western blot assay (Figures 7C, D; Supplementary Figures S5C, D) and immunofluorescence staining (Figures 7E–G; Supplementary Figures S5E–G) revealed EMF lost its cell rejuvenation effects after knockdown of ATG5 or treatment of CQ in P13 BMSCs. Taken together, EMF delayed long passaged BMSCs senescence partly via promoting autophagy.

To determine the effects of EMF on the regenerative capacity of long-term passaged BMSCs in vivo, we created a rat critical-sized calvarial defect model. Acellular GelMA hydrogel or GelMA hydrogels mixed with P4, P13 BMSCs were transplanted into the calvarial defects of rat for 8 weeks. For P13 BMSCs transplant, rats were randomly selected to be treated with EMF (50Hz, 0.4mT, 4h/day) or not. Micro-CT imaging indicated that the restorative effects of P13 BMSCs was significantly declined compared with P4 BMSCs, while EMF exposure could effectively ameliorate this tendency manifested by higher values of BMD and BV/TV (Figures 8A–C). Moreover, HE and Masson’s trichrome staining were carried out to evaluate the new bone formation of each group. New bone area fraction comparisons of each group at 8 weeks revealed that EMF exposure significantly enhanced the bone regeneration of long passaged BMSCs in vivo (Figures 8D–F). We also performed the immunohistochemistry staining of ALP in four groups. P13 BMSCs exhibited decreased osteogenic capacity compared with P4 BMSCs in vivo. However, EMF exposure could partly reverse this change (Figures 8G, H).