Melatonin-induced Cell Rejuvenation from Long-term Ex-vivo passaging Functioned by its Restoration of Damaged Cell Autophagic Processes

Background: Stem cells undergone long-term ex-vivo expansion are most likely functionally compromised (namely cellular senescence) in terms of their stem cell properties and therapeutic potentials. Due to the ability to attenuate cellular senescence, melatonin (MLT) has been proposed as an adjuvant across long-term cell expansion protocols, but the underlying mechanism remains largely unknown. Methods: Human periodontal ligament stem cells (PDLSCs) were isolated and cultured ex-vivo for 15 passages, and passage 2, 7 and 15 cells were used to interrogate the cellular senescence and alteration in cell autophagy during long-term expansion. The cellular senescence features were evidenced by senescence-associated β-galacotosidase (SA-β-gal) activity and the expression of senescence-related proteins including p53, p21, p16 and γ-H2AX. Electronic microscope was used to observe the autophagic vesicles. Adenovirus mRFP-GFP-LC3 was transfected to indicate the alteration of autophagic ux during long-term expansion, and the autophagy-associated proteins Atg7, Beclin-1, LC3-II and p62 were evaluated by Western blot. Results: It was found that long-term in-vitro passaging led to an accumulated SA-β-gal, elevated expressions of p53, p21, p16 and γ-H2AX, along with downregulated autophagy-associated proteins Atg7, Beclin-1 and LC3 as well as a mounting autophagy substrate p62. In accordance with expectation, supplemented with MLT not only ameliorated cells to a younger state but also restored the impaired autophagy level in senescent cells. Additionally, we demonstrated that autophagy inhibitor could block such MLT-induced cell rejuvenation. When the underlying signaling pathways involved was interrogated, we found that MLT receptor (MT) participated in mediating MLT-related autophagy restoration by regulating PI3K/AKT/mTOR signaling pathway. Conclusions: The present study suggests that MLT may rejuvenate long-term expansion-caused cellular senescence by restoring autophagy, more likely via PI3K/AKT/mTOR signaling pathway in an MT-dependent manner. This is the rst report identifying the MT-dependent PI3K/AKT/mTOR signaling involved in MLT-induced autophagy alteration, pointing to a potential target for using autophagy-restoring agents such as MLT to develop optimized clinical-scale cell production protocols. investigations indicated that the MT-dependent PI3K/AKT/mTOR signaling was involved in MLT-induced alteration in cell autophagy. This is the rst report demonstrates that MLT-induced cell rejuvenation functioned by the regulation and restoration of damaged cell autophagic processes, pointing to a potential target for using autophagy-restoring agents to develop optimized clinical-scale cell production protocols to product adequate cellular materials for cellular therapy and regenerative medicine.

its regulation of cell autophagy, and the MT/PI3K/AKT/mTOR signaling is likely to be an important pathway involved into this regulatory process.
To test our hypothesis, we isolated human periodontal ligament (PDL) stem cells (PDLSCs) in the present study using previous established protocol [32,33]. The obtained PDLSCs were cultured in-vitro for 15 passages, and passage 2 (P2), passage 7 (P7) and passage 15 (P15) cells were used to investigate cellular senescence and autophagy level caused by long-term expansion. Next, cellular responses of P2, P7 and P15 cells with MLT supplement were evaluated to test the biological function of MLT in maintaining cell stemness and indeed its ability to restore autophagy. Finally, we screened and veri ed the potential signaling pathways involved in the MLT-induced cell rejuvenation.

| Isolation and cultivation of PDLSCs
Human PDLSCs were isolated from extracted third molars or orthodontic teeth from 6 donors (2 males, 4 females, aged 26.5 ± 6.3) admitted to Department of Oral and Maxillofacial Surgery, School of Stomatology, Fourth Military Medical University. All donors signed the informed consent form in advance.
Extracted teeth were examined in advance to exclude periodontitis, and were immediately transferred into a centrifuge tube contained with complete cultural medium (α-minimum essential medium (α-MEM, Invitrogen) containing 100 U/mL penicillin (Invitrogen), 100 µg/mL streptomycin (Invitrogen), and 10% fetal bovine serum (FBS; Sijiqing, Hangzhou, China) to laboratory. After repeatedly rinsing with sterile phosphate-buffered saline (PBS; Corning, New York, USA), PDL tissue was gently scraped from the middle of the root surface. The scraped tissue was collected and digested in 3 mg/mL type I collagenase (Sigma-Aldrich, St. Louis, USA) in 37 ℃ for 1 h. After digestion, tissues were seeded in 6-well plates (Invitrogen, Carlsbad, CA, USA) with complete cultural medium. The medium was changed every 2 days until cells migrated from the tissue and reached 80% con uence. We here recognized those primary cells as passage 0 cells (P0). Then the P0 cells were cultured in-vitro for 15 passages, while P2, P7 and P15 cells were labeled and cryopreserved at liquid nitrogen using CELLSAVING (New Cell & Molecular Biotech, Suzhou, China). Cryopreserved P2, P7 and P15 cells from the same donor were recovered at the same time, grouped for the identi cation of MSC properties and further investigation.

| Flow cytometry analysis
P2, P7 and P15 cells were digested, collected, and washed twice with PBS. Then cells were transferred into sterile Eppendorf tubes (Eppendorf, Hamburg, Germany) with at least 1 × 10 5 cells/tube. After respectively incubated at 4 °C in the dark for 1 h with monoclonal antibodies against human CD90, CD105, CD146, CD34, CD45, or CD31 (all from eBioscience, San Diego, CA, USA) at a 1:1000 dilution, while cells incubated with PBS were used as the negative control. The immunophenotypes of the cells were tested with a Beckman Coulter Epics XL cytometer (Beckman Coulter, Fullerton, CA, USA).

| Colony formation assay
P2, P7 and P15 cells were seeded into 100-mm culture dishes (Invitrogen) at the density of 1 × 10 3 cells/well. The medium was refreshed every 2 days for 10 days. Cells were then rinsed twice and were xed using 4% paraformaldehyde (Servicebio, Wuhan, China) for 20 min and stained with 0.1% toluidine blue (Sigma-Aldrich, St. Louis, MO, USA) for 15 min. Stained cells were rinsed to wash extra dye and were observed under stereomicroscope (Olympus Optical, Tokyo, Japan). Cells aggregate containing more than 50 cells were recognized as a colony.

| Cell Counting Kit-8 (CCK-8) assay
P2, P7 and P15 cells were seeded into 96-well plates (Invitrogen) at the density of 1 × 10 3 cells/well. The medium was refreshed every 2 days for 8 days. During the 8-day culture, 200 µL medium with 20 µL CCK-8 reagent (Dojindo Corporation, Tokyo, Japan) was added to each well at an identical time every day, and the plate was incubated 37 ℃ for 1 h before absorbance at 450 nm was detected with a microplate reader (TECAN, Männedorf, Switzerland) to measure the proliferation ability of cells.

| Cell differentiation assay
P2, P7 and P15 cells were seeded into 6-well dishes at the density of 5 × 10 5 cells/well. After cells reached 80% con uence, medium was changed into osteo-inductive medium, adipo-inductive medium or chondrogenic inductive medium (all from Cyagen, Guangzhou, China) to ascertain their multilineage differentiation capacity. Medium was refreshed every 2 days until 21 days for osteo-induction or adipoinduction, or 28 days for chondrogenic induction. After induction, mineralized nodes indicating osteogenesis, lipid droplets indicating adipogenesis or acidic poly-saccharides indicating chondrogenesis were visualized by Alizarin red staining (Sigma-Aldrich), Oil red O staining (Cyagen), and Alcian blue staining (Cyagen) respectively. To quantitatively analyse the mineralized nodules after Alizarin red staining, the mineralized nodules were dissolved with 6% cetyl-pyridine for 15 min. Then the absorbance at 570 nm was tested by a microplate reader (TECAN).

| Alkaline phosphatase (ALP) staining
ALP activity in PDLSCs was analyzed by ALP staining. In brief, P2, P7 and P15 cells were seeded into 6well dishes at the density of 5 × 10 5 cells/well in complete medium. After cells reached 80% con uence, medium was changed into osteo-inductive medium for 7 days. ALP staining was performed afterwards using a BCIP/NBT ALP staining kit (Beyotime Institute of Biotechnology, Nantong, China) according to the manufacturer's instructions. ALP quantitative analysis was performed using an alkaline phosphatase assay kit (Nanjing Jiancheng Bioengineering Institute, Nanjing China) according to the manufacturer's instructions.

| MLT concentration selection and supplement
MLT (Sigma-Aldrich) was screened for the best concentration before supplemented into culture. MLT was added into basal medium at the concentration of 0, 10 nM, 100 nM, 1 µM, 10 µM respectively 24 h before testing. To examine the viability of P15 cells with MLT supplement at different concentrations, CCK-8 assay was performed. Cells were seeded into 96-well plate (Invitrogen) at the density of 5.0 × 10 3 cells/well. After MLT supplement for 24 h, the CCK-8 reagent (Dojindo Corporation) was added to each culture well and incubated at 37 °C for 1 h. Absorbance at 450 nm was then detected with a microplate reader (TECAN). To detect the in uence of different MLT concentration to cellular senescence and autophagy, the expression of senescence-related proteins including p53, p21, p16 and γ-H2AX, and the expression of autophagy-related proteins including Atg7, p62, Beclin-1 and LC3-II were evaluated by qRT-PCR and Western blot (see the sections of qRT-PCR and Western blot analysis). After nding out the optimum concentration, MLT was added into basal medium 24 h before tests.

| Chemicals and reagents
To inhibit the MLT-mediated autophagy elevation, 5 mM 3-methyladenine ( Each reagent was rstly dissolved in DMSO (Sigma-Aldrich) to form storing solution at the concentration of 10 mM then diluted into nal applied concentration by basal medium before use.
2.6 | SA-β-gal activity analysis SA-β-gal activity is widely recognized during the detection of cellular senescence. SA-β-gal activity was detected using the senescence β-galactosidase staining kit (Cell Signaling T echnology (CST), MA, USA) as manufacturer's instruction. SA-β-gal positive cells were dyed in blue, observed and calculated based on three randomly selected bright elds. The percentage of SA-β-gal positive cells out of the total number of cells was analyzed.

| Transmission electron microscopy
To detect the autophagic vesicles, cells were xed using 3% (w/v) glutaraldehyde for at least 48 h at 4 °C and post-xed with osmium tetroxide. Samples were dehydrated in a graded series of alcohol concentrations, embedded in epoxy resins, and sectioned. The autophagic vesicles in cells were observed with transmission electron microscopy (Hitachi, Tokyo, Japan)

| Indirect immune uorescent labeling and ow cytometry analysis
The amount of LC3-II was evaluated by indirect immune uorescent labeling for GFP-LC3 followed by ow cytometry analysis. Cells were xed and permeabilized with commercial Fixation/Permeabilization Solution Kit (BD Biosciences Pharmingen, San Diego, USA) for 20 min at 4 °C. After rinsed twice, cells were incubated with the LC3B-speci c antibody (rabbit polyclonal, ProteinTech, Illinois, USA, # 18725-1-AP) at a 1:200 dilution in 4 °C overnight. Then cells were rinsed three times to remove extra primary antibody before incubated at room temperature in the dark for 2 h with the Dylight 488, goat anti-rabbit lgG (1:200, AntiProtech Inc, California, USA). The labeled cells were analyzed using a Beckman Coulter Epics XL cytometer (Beckman Coulter).

| Detection of autophagic ux with adenovirus mRFP-GFP-LC3
To detect autophagic ux, adenovirus mRFP-GFP-LC3 (HanBio Technology, Shanghai, China) were transfected into PDLSCs. Cells were seeded into confocal 24-well dishes (Xinyou Biotechnology, Hangzhou, China) at a density of 1 × 10 5 cells/well. After cells adhered and reached 50% con uence, adenovirus mRFP-GFP-LC3 was transfected according to the manufacturer's instructions. Medium was renewed 2 h after transfection to remove extra adenovirus. Pharmaceutical treatments were conducted 12 h after transfection. After respective treatment, transfected cells were xed by 4% paraformaldehyde (Servicebio) and observed under the Olympus FV1000 laser confocal microscope (Tokyo, Japan). Both red and yellow puncta were counted and analyzed in the merged image based on three randomly selected elds.

| Western blot analysis
Prepared cells were lysed in RIPA buffer (Beyotime) added with phosphatase inhibitors (Sigma-Aldrich). Protein concentration was measured using the bicinchoninic acid (BCA) method kit (Solarbio, Beijing, China). Protein samples were separated by sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Beyotime). The concentration of applied SDS-PAGE gel was determined by molecule weight of proteins.
That was, 15% for LC3, p16, γ-H2AX and p21; 10% for p-PI3K, PI3K, p-AKT, AKT, Atg7, Beclin-1, p53 and GADPH; and 6% for p-mTOR and mTOR. The proteins were transferred onto PVDF membranes (Millipore, Billerica, MA, USA). After blocking with 5% non-fat milk at room temperature for 1 h, the membrane was incubated with primary antibodies at 4 °C overnight and with the corresponding horse radish peroxidase (HRP)-conjugated secondary antibody (1:2000; goat anti-rabbit IgG, CST, #7074; goat anti-mouse IgG, CST, #7076) for 2 h at room temperature the next day. Finally, the blots were detected using enhanced chemiluminescence substrate (ECL kit, Millipore). The phosphorylated protein was normalized to the corresponding total protein. GAPDH was used as the housekeeping gene for internal normalization.
ImageJ software was used to analyze protein bands. The primary antibodies used for Western blot were against p16 (

| Statistical Analysis
The PDLSCs before and after long-term passaging for studies were from the identical donors. All assays were performed at least three times independently. GraphPad Prism 7 software was employed for statistical analysis. Differences between groups were assessed by one-way analysis of variance (Oneway ANOVA) followed by Tukey's multiple comparisons tests, Sidak's multiple comparisons tests, or Dunnett's multiple comparisons. The results were presented as the mean and standard deviation (mean ± SD). All the experiments were repeated in PDLSCs from at least 3 different donors. Statistical signi cance was expressed as p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), or p < 0.0001 (****).

| Long-term passaging causes cellular senescence
PDLSCs were successfully isolated from PDL tissues from 6 donors (2 males, 4 females, aged 26.5 ± 6.3), and were expanded in-vitro for 15 passages. P2, P7 and P15 cells from the same donor were grouped for investigation. To determine whether the isolated and passaged cells can be used as PDLSCs, we detected cell surface markers of MSCs by ow cytometry analysis using previous established protocol. We found that all the P2, P7 and P15 cells positively expressed MSC markers including CD90, CD105, and CD146, while the endothelial cell marker (CD31) and hematopoiesis-related markers (CD34 and CD45) were negative (Fig. S1a). Besides, to con rm the differentiation potential towards cartilage, adipose and osseous tissue, we conducted the Alcian blue, Oil red O, Alizarin red and ALP staining. The stained acidic poly-saccharides, lipid droplets and mineralized nodes respectively suggested that the P2, P7 and P15 cells all had multi-linage differentiation potential including chondrogenesis, adipogenesis, and osteogenesis ( Fig. S1b-d, f), while quantitative analysis of Alizarin red staining and ALP activity suggested the osteogenesis potential of PDLSCs in P15 signi cantly decreased after long-term passaging ( Fig. S1e, g). Furthermore, as evaluated by colony-forming assay and CCK-8 assay, the cells from all passages were proved to possess the ability of colony-forming and proliferation. However, the ability of colony-forming and proliferation signi cantly declined along with long-term passaging (Fig. 1a,   b). In parallel, the SA-β-gal activity, which re ects cellular senescence extent [34], was found markedly higher in P15 cells (Fig. 1c). To further verify cellular senescence after in-vitro expansion, we chose several senescence-associated genes including p16, p21, p53 and γ-H2AX for investigation. p16, p21 and p53 are recognized as cell cycle regulatory protein, whose increased expression often indicates disordered cell cycle. And γ-H2AX, manifesting discrete nuclear foci, could sensitively indicate DNA double-strand break damage. As evaluated by Western blot, the expression of these senescence-related proteins was found to increase after long-term passaging (Fig. 1d). Taken together, the senescenceassociated features including inhibited proliferation capacity, hindered osteogenesis potential, increased SA-β-gal activity, arrested cell cycle and DNA damage were observed in P7 cells and more prominently in P15 cells compared to P2 cells, suggesting long-term passaging caused cellular senescence.

| Autophagy is impaired after long-term passaging
Autophagy is a consecutive process including autophagosomes formation, maturation, fusion with lysosomes to form autolysosome, and degradation. The process from the formation to the degradation of autophagosomes is termed as autophagic ux. We rst observed the autophagic vesicles in P2, P7 and P15 cells with electron microscopy. Accompanied by swollen mitochondria shown in P15 cells, the autophagic vesicles were scarcely observed (Fig. 2a). In order to detect the proceeding autophagic ux, adenovirus mRFP-GFP-LC3 (a tandem uorescent-tagged microtubule-associated protein light chain 3 (LC3) reporter containing monomeric red uorescent protein (mRFP) and GFP) was transfected into P2, P7 and P15 cells. Given that the uorescence of GFP could be quenched due to low pH in lysosome, the red puncta indicate mature autolysosomes, while yellow puncta formed by the overlay of red and green puncta indicate non-fused autophagosomes. After long-term passaging, the number of red puncta (autolysosomes) prominently declined, suggesting a possible blockage of the formation of autophagosomes (Fig. 2b). To further verify whether the decline of autolysosomes in senescent cells was caused by inhibited formation or increased degradation, Baf, an inhibitor for lysosome formation was added 4 h before tests. After Baf supplement, yellow puncta (non-fused autophagosomes) were saturated in P2 cells because of the failure of autolysosome formation, while the accumulation of yellow puncta were blunted in P7 and P15 cells, suggesting the saturation of yellow puncta after Baf treatment was not resulted from increasing degradation but the blockage of autophagic ux (Fig. 2b). The formation of autophagosomes could also be indicated by the LC3-II, given that its localization on both outer and inner membranes of autophagosomes could accumulate after prohibiting autophagosomes degradation by Baf treatment. The amount of LC3-II showed no signi cant differences before and after Baf treatment in P7 and P15 cells as evaluated by ow cytometry analysis and Western blot, while LC3-II signi cantly accumulated in P2 cells after Baf treatment ( Fig. 2c and 2d). These results all indicated the inhibited formation of autophagosomes in the cells after long-term expansion. As evaluated by Western blot, the expression of autophagy-related proteins including Atg7 and Beclin-1 decreased after long-term expansion, while the expression of autophagy substrates indicator p62 increased along with PDLSC senescence (Fig. 2e), indicating the impaired autophagy in senescent cells caused by long-term expansion.

| MLT ameliorates PDLSC senescence and restores autophagy
To investigate the effect of MLT to autophagy in senescent PDLSCs, MLT was supplemented to basal medium. Firstly, MLT concentration was screened for optimal application. P15 cells were treated with 0, 10 nM, 100 nM, 1 µM, 10 µM MLT respectively for 24 h. According to CCK-8 assay, cells treated with 1 µM MLT showed the highest viability (Fig. S2a). Consistently, as evaluated by Western blot, the expression of autophagy-related proteins including Atg7, Beclin-1 and LC3-II were signi cantly higher while p62 was lower in cells treated with 1 µM MLT, collectively suggesting a higher autophagy level in the 1 µM MLTtreated group (Fig. S2b, c). The expression of senescence-related proteins was evaluated by qRT-PCR. The expression of p16, p21 and γ-H2AX was lower when treated with 1 µM MLT, while the expression of p53 showed no signi cant difference in cells treated with 100 nM and 1 µM MLT (Fig. S2d). Taken together, in order to achieve preeminent cell viability, better autophagy restoration and rejuvenation effect, we here chose 1 µM MLT for further studies.
After supplemented with MLT, SA-β-gal staining was employed to investigate the effect of MLT in cellular senescence. The increased SA-β-gal positive cells in P7 and P15 cells compared to P2 cells drastically decreased after MLT treatment (Fig. 3a, b). In consist with SA-β-gal staining, increased expression of p16, p21 and p53 in P7 and P15 PDLCSs declined after MLT treatment as evaluated by Western blot (Fig. 3c, d), collectively suggesting the rejuvenation effect of MLT in senescent PDLSCs.
To detect the alteration in autophagic ux after MLT treatment, adenovirus mRFP-GFP-LC3 was transfected into P15 cells 24 h before MLT treatment. MLT was supplemented into basal medium for 24 h while Baf was added 4 h before tests. As shown in the merged image, the number of both red puncta and total puncta prominently elevated after MLT treatment, while the Baf treatment could further increase the autophagosomes accumulation (yellow puncta) in senescent cells with MLT treatment (Fig. 4a). These observations suggested promoted fusion of autophagosomes and lysosomes, inferring propelled lysosome-based autophagosome degradation. As for the amount of LC3-II evidenced by Western blot and ow cytometry, MLT could increase the expression of LC3-II in P15 cells, while the amount of LC3-II in MLT treated cells further increased after incubated with Baf, suggesting that the former impeded autophagic ux in P15 cells could be restored after MLT treatment (Fig. 4b, c). The expression of Atg7, Beclin-1 and p62 was evaluated by Western blot. As shown in the blots, the defective expression of Atg7 and Beclin-1 in P7 and P15 cells signi cantly elevated after MLT treatment, while the accumulated p62 in P7 and P15 cells decreased after MLT treatment (Fig. 4d). These results collectively suggested that MLT treatment could restore autophagy in senescent cells, which may consequently ameliorate the cellular senescence. In short, MLT treatment could restore autophagy and attenuate senescence features in senescent cells.

| Inhibited autophagy deteriorates cellular senescence
In order to explore if the rejuvenation effect by MLT was through autophagy regulation, we supplemented a widely used autophagy inhibitor 3-MA to reverse the rescued autophagy in MLT-treated P15 cells. As veri ed by adenovirus mRFP-GFP-LC3 transfection, 3-MA could further inhibit impaired autophagy in senescent cells. What's more, the red puncta (autolysosomes) in MLT-treated P15 cells signi cantly decreased after 3-MA treatments (Fig. 5a, b). As shown in Western blot analysis, the expression of autophagy-related proteins including Atg7, Beclin-1, LC3-II declined, while p62 was stacked after 3-MA single treatment compared to the control group. 3-MA treatment also abrogated the restored autophagy mediated by MLT, evidenced by Western blot analysis (Fig. 5c, d). These results collectively indicated that 3-MA could successfully impede the MLT-restored autophagy.
With impeded autophagy caused by 3-MA, the senescence-associated alteration was deteriorated. With the existence of MLT, 3-MA treatment increased the number of SA-β-gal positive cells (Fig. 5e).
Accordingly, the expression of senescence-related proteins p16, p21, p53 and γ-H2AX further increased with 3-MA treatment, even with the MLT treatment ( Fig. 5f, g), indicating aggravated arrest of cell cycle and DNA damage along with impaired autophagy. Taken together, loss of autophagy could further exacerbate the senescence-associated features and deter the MLT-induced amelioration of senescence.

| MLT receptor (MT) is involved in MLT-mediated autophagy regulation
To determine whether MLT-mediated autophagy regulation was in a receptor-dependent pathway, luzindole (LUZ) was supplemented as a non-selective antagonist of MT 24 h ahead of tests. As shown in the adenovirus mRFP-GFP-LC3 transfected cells, red puncta in LUZ-treated cells decreased compared to the control group, while the increased red puncta mediated by MLT was also hindered by LUZ treatment (Fig. 6a, b), suggesting the restored autophagic ux was re-suppressed by LUZ treatment. Consistently, the reverted expression of autophagy-related proteins including Atg7, Beclin-1, LC3-II and p62 with MLT and LUZ simultaneous treatment compared to MLT treated group suggested that MLT restored autophagy in senescent PDLSCs in an MT-dependent way (Fig. 6c, d). As evaluated by SA-β-gal staining ( Fig. 6e) and Western blot for senescence-associated proteins including p16, p21, p53 and γ-H2AX ( Fig. 6f), the senescence features of P15 cells were deteriorated when LUZ blocked MT with or without MLT treatment.
Next, we explored the possible downstream signaling pathway in this receptor-involved MLT-mediated autophagy regulation. As suggested in previous study, MT is involved in the down-regulation of the PI3K/AKT by MLT, while the PI3K/AKT/mTOR pathway is recognized as a main signaling of autophagy regulation. Hence we here tested the expression of phosphorylated form of PI3K/AKT/mTOR in cells treated with LUZ. As evaluated by Western blot, the expression of p-PI3K, p-AKT and p-mTOR in LUZ-treated group signi cantly increased compared to the control group, while it had little effect on the total protein expression of PI3K, AKT and mTOR. It also suggested that MLT treatment could elevate the expression of p-PI3K, p-AKT and p-mTOR while the treatment had no effect on the total protein expression. What's more, the expression of both phosphorylated and total proteins of PI3K, AKT and mTOR was comparable between the control group and MLT and LUZ co-treated group (Fig. S3a-d) (Fig. 7a). The consistent result was also shown in the expression of autophagy-associated proteins including Atg7, Beclin-1, LC3-II and p62 evaluated by Western blot (Fig. 7b, c). These results all suggested that the inducible effect of MLT for autophagy in senescent cells was regressively abrogated when using the MT blocker, PI3K/Akt agonist and mTOR agonist respectively, indicating the involvement of MT/PI3K/Akt/mTOR in MLT-mediated autophagy restoration. Along with the regressively inhibited autophagy, the attenuated senescence features mediated by MLT were consequently deteriorated in MLT + LUZ group, MLT + SC79 group and MLT + MHY1485 group, as evidenced by the increased number of SA-β-gal + cells (Fig. 8a, b), and augmented expression of senescence associated proteins including p16, p21, p53 and γ-H2AX (Fig. 8c,   d), indicating the down-regulation of autophagy via MT/PI3K/AKT/mTOR signaling deteriorated senescence in cells undergone long-term ex-vivo passaging.

| Discussion
To bridge the gap between limited number of acquirable stem cells from donors and the large quantity of stem cells that are needed for tissue engineering, it is necessary to apply long-term in-vitro passaging for cell large-scale production, especially in autologous MSC transplantation [3]. However, accumulating evidence suggested that long-term passaging might cause cellular senescence, with unexpected alterations including anomalous morphology [8], hindered proliferation capacity [35], decreased differentiation potential including chondrogenesis and osteogenesis [36]. To avoid the senescenceassociated alterations, the passage 2 to passage 5 cells, which were believed to be possessed with analogical capacity in self-renewal and differentiation, were often chosen as objective of study in numbers of studies. What's more, one research suggested that the stemness of MSCs could only be retained up to 6 passages, and signi cantly declined after 10 passages [37]. Another study suggested that the proliferation, differentiation and immunomodulation functions of mouse bone marrow-derived MSCs decreased after 15 passages [26]. Here, we chose P2 cells as young group, P7 and P15 cells as senescent group to study the senescence-associated alteration along with long-term in-vitro expansion. In consistent with previous investigations, our study reported that in PDLSCs, long-term in-vitro expansion led to senescence evidenced by increased SA-β-gal activity, elevated expression of senescence-associated proteins p16, p21 p53 and γ-H2AX, which symbolized arrested cell cycle and DNA damage.
In order to overcome the challenge of accompanied senescence during in-vitro passaging, it's prerequisite to nd the mechanism in lone-term expansion-caused senescence. It has been revealed that hindered autophagy was concomitant with age-related disorders [38, 39], organismal aging [40], and also senescent stem cells [12][13]. Furthermore, inhibition of autophagy could accelerate neurodegenerative disease by accumulating disease-causing aggregate-prone proteins [41] and also induce kidney damage in aging mice [42]. These results indicted indivisible relationship between autophagy and senescence. To our knowledge, the autophagy level of senescent stem cells after long-term passaging has hardly been tested before. In consistent with physiological or disease-related organismal and cellular senescence, a decline in basal autophagy was rstly found in long-term in-vitro expansion-caused cellular senescence in our study, as evidenced by impeded autophagic ux, prohibited expression of autophagy-related proteins including Atg7, Beclin-1 and LC3-II and accumulated autophagy substrate p62. To further verify the connection between autophagy alteration and cellular senescence, autophagy inhibitor 3-MA was supplemented. We found that 3-MA-mediated autophagy inhibition could deteriorate the accumulation of senescence, evidenced by elevated SA-β-gal capacity, worsened DNA damage and arrested cell cycle, suggesting hindered autophagy could deteriorate cellular senescence. On the contrary, previous study has found that the restored autophagy could alleviate the senescent status and reach functionally improvement. Pharmaceutical treatment with urolithin A to induce autophagy could improve the agerelated declined muscle function and enhance exercise capacity in old rodents [43]. The restored autophagy mediated by caloric restriction could decrease the cellular ROS level and enhance the regenerative ability of aging stem cells [44]. Stimulation of lysosomal activity to promote autophagy in the aged neural stem cells signi cantly enhanced their ability for neural regeneration [45]. These results collectively indicated that the regulation of autophagy could be a potential strategy for functional maintenance of senescent stem cells.
Up to now, researchers have made great effort to restore the functionality of senescent stem cells. The strategies to reach optimum cultivation for restoring or promoting capacity of stem cells can be generally divided into three categories. One is to mimic the in-vivo microenvironment and induce prior expression of cytoprotective and stemness-associated genes, including but not limited to providing an extracellular matrix (ECM)-simulated environment, and a short-term hypoxia preconditioning. Another is to utilize exogenous molecules to activate extracellular or intracellular signaling for resisting the harsh in situ microenvironment caused by ischemia, nutrition-deprivation and oxidative damage, such as pretreatments using exogenous cytokines, growth factors and other small molecule drugs. In addition, genetic strategy modulates the stem cells to acquire superior capacity in self-protection, proliferation and differentiation. Although progress has been made using former strategies, further problems have arisen.
Hypoxic preconditioning was reported to give rise to arrested cell cycle and prohibited cell proliferation [46,47]. The utilization of exogenous signaling proteins seemed to have relative low e ciency and greater expense on account of their short half-life time [48]. Genetic engineering may cause the risk of insertional mutagenesis and oncogene transaction. Thus, other strategies that simultaneously prossess features of high e ciency and secured application are needed in restoring the functionality of senescent stem cells. As con rmed by recent studies, MLT exhibits unique superiority in tissue engineering. As a pleiotropic endogenous hormone secreted by pineal and almost every other tissue, MLT was con rmed to have low toxicity during treatment by both animal and clinic studies [49,50]. In several studies, MLT could enhance the survival stability by attenuating apoptosis, resisting oxidative stress and in ammation damage [51], improve the chondrogenesis and osteogenesis of MSCs [53], and also maintain the stemness of long-term passaged stem cells, resulting in preserved proliferation, osteogenic differentiation and immunomodulation capacity [26]. Except from the well-known role in stem cell-based therapy as a probable free radical scavenger with excellent anti-in ammation and differentiation enhancement properties, recently, researchers also found that MLT could function as a regulator for autophagy. In some studies, MLT could promote the basal levels of autophagy under physiological conditions and maintain neuronal homeostasis and survival from a subarachnoid hemorrhage followed by brain injury [53], and also modulates autophagy to attenuate the cardiac ischemia/reperfusion injury [54,55]. The property of MLT-mediated regulation of autophagy could also bene t in some age-related diseases. It was found that MLT in-vivo application ameliorated AD-induced cardiac atrophy [28], and enhanced the impeded cognitive function in tau-related AD rats [27] by restoring the autophagic ux. What's more, MLT could improve degradation of damaged mitochondria by mitophagy during aging and under neurodegenerative conditions [56]. Similar to the studies mentioned above, our study also con rmed that MLT could effectively restore the cellular autophagy not only by elevating the basal autophagy level in senescent P15 cells, but also by reinstating the autophagic level by enhancing the lysosome-dependent degradation of autophagosomes. Our study also found that restored autophagy was further accompanied with ameliorated senescence. To further clarify if it was the restored autophagy that bene ted the rejuvenation effects, we found that the defective autophagy caused by the autophagy inhibitor 3-MA could consequently abrogate the MLT-mediated rejuvenation. Therefore, our study suggests that the restored autophagy alleviated cellular senescence, and the manipulation of autophagy could be an effective strategy to maintain functionality of senescent cells.
To investigate the underlying mechanism of MLT-mediated autophagy restoration, we rst explored if the effect was dependent on MT. We supplemented MT-speci ed inhibitor LUZ into cultured cells. We found that LUZ could further exacerbate the impeded autophagy in senescent cells without MLT treatment, giving rise to aggravated cellular senescence. Even in senescent cells treated with MLT, LUZ could abrogate the inducible effect of autophagy mediated by MLT, and further inhibited the rejuvenation mediated by MLT. These results indicated that the MLT-mediated autophagy restoration was in an MTdependent manner in senescent cells. What's more, mTOR plays a role as central switch to autophagy.
Accordingly, our study demonstrated that the expression of p-PI3K, p-AKT and p-mTOR signi cantly decreased after the treatment of MLT but was reversely induced to increase by LUZ treatment. We thus concluded that the MT-dependent MLT-mediated autophagy was possibly through the inhibitory regulation of the PI3K/AKT/mTOR pathway. This is in consistence with a current study, which con rmed that MLT could regulate the PI3K/AKT pathway in an inhibitory way [29].
To precisely con rm the MLT-mediated autophagy through MT/PI3K/AKT/mTOR pathway, we respectively applied LUZ, PI3K/AKT speci c agonist SC79 and mTOR agonist MHY1485 with the MLT treatment, and found that the autophagy restoration and senescence attenuation effect mediated by MLT could be abrogated with blocked MT and activation of PI3K/AKT/mTOR pathway. However, our study didn't determine the exact type of MT involving in the MLT-mediated autophagy restoration, which should be our main pursue in the future.
To sum up, our study revealed that impaired autophagy was one of the vital mechanisms of cellular senescence induced by long-term ex-vivo expansion, which can be restored by MLT through MT/ PI3K/AKT/mTOR pathway. This work innovatively purposes a strategy that MLT treatment targeting impeded autophagy can be utilized to rejuvenate stem cells during long-term passaging. The investigation focused on the mechanism that inhibiting the PI3K/AKT/mTOR pathway could enhance autophagy to attenuate cellular senescence, which may provide possible target for the research and even clinical application aiming at the rejuvenation of stem cells for cellular therapy.

| Conclusion
Large-scale production of cellular materials for use in a clinical setting necessitates safe and costeffective agents to rescue cell dysfunction and therapeutic failure of stem cells following long-term exvivo expansion. Although use of MLT as a cell-medium adjuvant has been proposed to prevent cells from senescence following long-term passaging, the underlying mechanism remains largely unexplored. In this study, we found that the function of MLT in combating cell aging was related to its role as an autophagy regulator. Further investigations indicated that the MT-dependent PI3K/AKT/mTOR signaling was involved in MLT-induced alteration in cell autophagy. This is the rst report demonstrates that MLTinduced cell rejuvenation functioned by the regulation and restoration of damaged cell autophagic processes, pointing to a potential target for using autophagy-restoring agents to develop optimized clinical-scale cell production protocols to product adequate cellular materials for cellular therapy and regenerative medicine. Availability of data and materials All data generated or analyzed during this study are included in this published article.

Code availability
All software used in this study was obtained legally, and custom code is available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The experimental protocol of this study was approved by the Ethics Committee of the Stomatological Hospital of FMMU (201203), and informed consent was signed by all the subjects who donated their extracted teeth for cell isolation.