A deep learning-predicted POG maintains rat TSPC stemness and functions

In search of specific small molecules that could restore the age-related decline in tendon healing capacity, we employed DLEPS to predict stemness-targeting small molecules. Basically, DLEPS utilizes a two-step approach to predict the pharmacological effects of small molecules. DLEPS first applies deep neural networks to predict the quantitative relationship between small molecules and gene expression changes in cells after small molecule treatment. DLEPS then uses the GSEA algorithm to evaluate whether the gene expression changes caused by small molecules reverse disease-associated gene expression changes to predict the pharmacological effects of small molecules. The utility of DLEPS has been demonstrated across various diseases, including inducing adipose browning, treating NASH and hyperuricemia, and reversing aging phenotypes.23 Compared with the latest version of DLEPS, we upgraded the statistical analysis of how well each gene was predicted in all 12,328 genes using the Pearson correlation coefficient (Fig. 1a, b). In neonatal tendons, TSPCs are highly enriched and have robust tenogenic and self-renewal potentials.21,22 Very recently, we generated two groups of expression signatures from neonatal and adult rats and found a distinct decline in the stemness and regenerative capacity of rat TSPCs (rTSPCs) from the neonatal to adult stage.22 Here, we employed two groups of gene signatures to calculate the efficacy scores, namely, a Yamanaka factor score defined by four Yamanaka factors and a tendon score, using the top-ranked DEGs from a comparison of the transcription of tendon tissue from neonatal rats and adult rats (Fig. 1c). Finally, we selected the intersection of top-ranked candidates in both scores (12 molecules in total) to be tested in the cell culture or directly in the animal models (Fig. 1d, Table. S1). POG, a chromone extracted from Saposhnikovia root, was top ranked with good pharmacological properties (Fig. 1e). Although studies have demonstrated the functional roles of POG as an anticonvulsant, anti-inflammatory, and anticancer molecule,24,25 reports on its activities in the regulation of aging-related stem cell functions and tissue maintenance are currently lacking.

Fig. 1

A deep learning-predicted POG maintains rTSPC stemness and functions. a Schematic of the deep neural network employed to predict stemness enhancement. b Distribution of Pearson correlation coefficient r of predicted and empirical profiles for all 12 328 genes over 3 000 molecules in the test set. c Scatter plot of the Yamanaka factor score versus the tendon score. d Highlight of positively predicted molecules in the t-SNE plot of the molecular library. e Schematic of the POG molecular formula. f Schematic of serial passaging of rTSPCs from 3-month-old rats treated with POG. g (i) CFU-F assay of rTSPCs at P12 with DMSO or POG treatment during serial passaging. (n = 3). h Immunofluorescence staining (i) and semiquantification (ii) of Ki67 in rTSPCs at P12 (n = 3). i Immunofluorescence staining of SOX2 and OCT4 in rTSPCs at P12 (n = 3). j (i) SAβ-gal staining (left panel) and immunofluorescence staining of the DNA injury-related protein γ-H2AX (right panel) in p12 rTSPCs (n = 3). k Western blotting of the senescence-related proteins P21 and P53 in rTSPCs at P12. l RT‒qPCR of Il-6 and Tnf-a gene expression in rTSPCs at P12 (n = 3). m Heatmap of the differentially expressed gene profiles of rTSPCs at P12 n = 3. n (i) Sirius Red staining of rTSPCs at P12 (n = 3). o Immunofluorescence staining of the tenogenic markers TNMD and TNC in rTSPCs at P12 (n = 3). p Schematic of subcutaneous transplantation of collagen sponges with rTSPCs at P12. q HE and immunofluorescence staining of the tenogenic markers TNC and TNMD n = 5. Data are represented as the mean ± SD. (*P < 0.05; **P < 0.01; ***P < 0.001)

To determine whether POG rescues the phenotypic characteristics of rTSPCs, we first adopted a replicative cellular senescence model, which has been widely utilized to investigate cell stemness and senescence.26,27,28 Flow cytometry showed that rTSPCs were positive for the mesenchymal stem cell markers CD105, CD90, and C29 and negative for CD34 (a hematopoietic stem cell marker) and CD45 (a leukocyte marker) (Fig. S1a). Our preliminary experiments also confirmed that replicative senescence caused by serial passaging from passage 3 (P3) to passage 12 (P12) disrupted the stemness and differentiation capacities of rTSPCs isolated from 3-month-old (3M) rats (young rats; Fig. S2a–h). Next, according to previous studies and preliminary experiments,29 POG at a relatively moderate concentration of 20 μmol·L−1 displayed a stemness-promoting effect, which was reflected by RT‒qPCR results (Fig. S1b). Therefore, in follow-up experiments, 20 μmol·L−1 POG was added to the culture medium of rTSPCs in each passage from P3 to P12. An equal amount of the vehicle control dimethyl sulfoxide (DMSO) was added to the control group to exclude the influence of solvent (Fig. 1f). Compared with the DMSO-treated cells, the POG-treated rTSPCs showed markedly elevated self-renewal and proliferation, as evidenced by the larger numbers of colonies and a marked increase in Ki67+ cells (Fig. 1g, h). Immunofluorescence staining and RT‒qPCR demonstrated that POG treatment significantly enhanced the expression of SOX2 and OCT4 in rTSPCs at P12, indicating that POG helps to maintain rTSPC stemness during long-term passage (Fig. 1i, S3a). Moreover, POG treatment ameliorated multiple aging hallmarks, as evidenced by the following findings: (i) significantly decreased SAβ-gal-positive cells; (ii) repressed DNA damage response, as indicated by fewer γ-H2AX+ cells (Fig. 1j); (iii) markedly reduced expression of the aging-associated factors P53 and P21 (Fig. 1k); and (iv) reduced mRNA expression of Il-6, Il-1β, and Tnf-α, which indicated a decline in SASP (Fig. 1l). Consistent with these results, microarray also indicated that the expression of SASP-related genes was significantly decreased in the POG-treated rTSPCs at P12. Additionally, GSEA showed that POG-downregulated genes were associated with the TNF-α/NF-κB signaling pathway, further confirming the profound effect in delaying senescence (Fig. 1m).

Next, we evaluated whether sustained exposure to POG helped to maintain the tenogenic differentiation potential of rTSPCs after prolonged passages. Before tenogenic induction, we stopped POG treatment in rTSPCs at P12 and then transferred the DMSO- and POG-treated passaged rTSPCs into tenogenic differentiation medium without POG. As a result, the POG-treated passaged rTSPCs showed strong Sirius Red staining after 14 d of tenogenic induction, as well as marked increases in the protein expression of the key tenogenic markers TNMD, TNC, scleraxis (SCX), and fibromodulin (FMOD), compared with the DMSO-treated passaged rTSPCs, indicative of the functional rejuvenation of senescent rTSPCs (Fig. 1n, o, S3b). Furthermore, we subcutaneously transplanted DMSO- and POG-treated passaged rTSPCs at P12 with collagen sponges into nude mice and harvested tissues 8 weeks after surgery. Hematoxylin and eosin (HE) and immunofluorescence staining showed that POG-treated passaged rTSPCs maintained their intrinsic regenerative capacity, allowing the formation of dense tendon collagen fibers with higher levels of TNC and TNMD expression (Fig. 1p, q).

The spheroid-forming capacity of stem cells could represent their own stemness property.30,31 Using our previously reported methods,22 we constructed multicellular spheroids of DMSO- and POG-treated rTSPCs at P12 by replacing standard culture dishes with low-adhesion culture dishes for a 7-d culture period (Fig. S3c). Long-term passage led to a poor spheroid-forming capacity in the DMSO-treated passaged rTSPCs (~6.0 ± 4.0 total); in contrast, the POG-treated passaged rTSPCs were found to assemble into a higher number of spheroids (~38.5 ± 6.5 total) (Fig. S3d). Immunofluorescence analysis demonstrated that spheroids formed by the POG-treated passaged rTSPCs exhibited higher expression levels of SOX2 colocalized with OCT4 (Fig. S3e). Next, the tenogenic differentiation capacity of these spheroids was evaluated by transferring spheroids onto normal-adhesion culture dishes coated with Matrigel supplemented with the tenogenic induction medium for 14 d. Scanning electron microscopy (SEM) revealed that spheroids formed by the POG-treated passaged rTSPCs stretched out more longitudinally aligned collagen fibers (Fig. S3f). Immunofluorescence analysis confirmed that the POG-restored passaged rTSPCs expressed the tendon-specific factor TNMD at higher levels after 21 d of induction (Fig. S3g). Collectively, these findings indicate that pharmacological manipulation with POG during in vitro long-term passage rescued the intrinsic tenogenic differentiation potential of rTSPCs by both maintaining stemness and inhibiting senescence.

POG nanoparticles rescue aged tendon self-healing capacity by enhancing TSPC stemness and rejuvenating senescent phenotypes

Given the beneficial therapeutic effects of POG on the in vitro replicative senescence model of rTSPCs, we next investigated whether POG treatment could directly rescue impaired biological functions of rTSPCs induced by in vivo natural aging. First, tendons from 18-month-old (18M) rats (aged rats) displayed typical aging characteristics, including collagen degradation, more lipid deposition in TEM images and expression of the senescence markers P21 and γ-H2AX (Fig. S4a–c). Second, tendon injury in 18-month-old aged rats displayed increased lipid deposition, accumulation of senescent cells and secretion of inflammatory factors compared to that of 3-month-old rats (Fig. S4d–k). Third, rTSPCs isolated from aged rats displayed seriously impaired colony-forming, cell proliferation, and tenogenic differentiation capacities, as demonstrated by a larger number of SAβ-gal-positive cells, higher levels of P53 and P21 expression, and elevated mRNA expression of Il-6 and Tnf-α (Fig. S5a–f). These results indicate that the 18M aged rat model is a reliable animal model to investigate the antiaging functions of POG. Then, aged rTSPCs isolated from the aged rats were treated with 20 μmol·L−1 POG (18M + POG group) or DMSO (18M + DMSO group) as a control once every 2 d for 7 d (Fig. 2a). CFU-F assays showed that POG treatment markedly increased the number of CFU-Fs in aged rTSPCs compared to that in the 18M + DMSO group (Fig. 2b). In vitro serial passaging-induced cell senescence is significantly distinct from in vivo natural aging-induced senescence, which may explain some differences between the results here and those in Fig. 1g. RT‒qPCR and immunofluorescence staining revealed that POG treatment significantly enhanced the protein expression of Ki67, SOX2, and OCT4 in aged rTSPCs (Fig. 2c, d). In addition, the percentage of SA-β-gal-positive senescent cells was markedly decreased in the 18M + POG group compared to the 18M + DMSO group (Fig. 2e). This finding was coupled with a significant suppression of the senescence-related proteins P21 and P53 (Fig. 2f). POG treatment also decreased the gene expression of several SASP factors, including Tnf-α and Il-6, in aged rTSPCs (Fig. 2g).

Fig. 2

Systemic delivery of POG rescues the aged tendon self-healing capacity by enhancing rTSPC stemness and rejuvenating senescent phenotypes. a Schematic outlining the design of POG intervention in senescent rTSPCs from aged rats. b CFU-F assay of DMSO- and POG-treated aged rTSPCs (n = 3). c Immunofluorescence staining of Ki67 in DMSO- and POG-treated aged rTSPCs (n = 3). d RT‒qPCR of the stemness proteins SOX2 and OCT4 in DMSO- and POG-treated aged rTSPCs (n = 3). e SAβ-gal staining of DMSO- and POG-treated aged rTSPCs (n = 3). f Western blotting of senescence-related proteins P53 and P21 in DMSO- and POG-treated aged rTSPCs (n = 3). g RT‒qPCR of Il-6 and Tnf-a gene expression in DMSO- and POG-treated aged rTSPCs (n = 4). h (i) Sirius Red staining of DMSO- and POG-treated aged rTSPCs (n = 3). i (i) Immunofluorescence staining of the tenogenic markers TNC, TNMD and FMOD in DMSO- and POG-treated aged rTSPCs (n = 3). j Schematic outlining systematic administration of POG on the partial transection tendon injury in 18-month-old rats. k HE staining of injured Achilles tendons n = 5. l (i) Oil Red O staining of injured Achilles tendons (n = 5). m (i) Immunofluorescence costaining of CD146 and BRDU in injured Achilles tendons (n = 5). n HE staining of injured Achilles tendons n = 5. Data are represented as the mean ± SD. (*P < 0.05; **P < 0.01; ***P < 0.001)

Prior to tenogenic induction, we stimulated the senescent rTSPCs with POG once every 2 d for 5–7 d before transferring them into differentiation medium for 10–14 d. Immunofluorescence showed that the senescent TSPCs from the 18M + POG group possessed markedly increased positive areas of Sirius Red staining and higher expression levels of the key tenogenic markers TNC, TNMD, and FMOD than those from the 18M + DMSO group (Fig. 2h, i). Combining the results from the in vitro serial passage-inducing senescence model with the in vivo natural aging model, we found that the DLEPS screening system could efficiently and precisely identify a small-molecule compound, POG, that maintains rTSPC stemness and inhibits senescence.

Next, we examined the potential of the systemic delivery of POG to improve endogenous rTSPC functions and self-healing capacity in aged rats (Fig. 2j). The aged rats were first orally administered POG (18M + POG group) or DMSO as a vehicle (18M+vehicle group) once a day for 2 month. Oral administration ended 2 d prior to the surgery. Then, a partial incision was made in the Achilles tendon, as previously described.22 After 7 d, the injured tendons from the 18M + POG group displayed reparative effects that were comparable to those in the injured tendons from the young rats (Fig. S4d–f), including relatively complete tendon structure with continuous thin fibers and a significant reduction in lipid droplets and fatty-acid-binding protein 4 (FABP4)-positive adipose cells (Fig. 2k, l). Moreover, a greater percentage of CD146+BrdU+ endogenous rTSPCs accumulated at sites of injury in the 18M + POG group than in the 18M+vehicle group (Fig. 2m). After 4 weeks, the injured tendons with POG pretreatment showed more ordered and thicker collagen and reduced inflammatory infiltration, as evidenced by the significantly decreased proportion of IL-6+ and MMP9+ cells, while fragmentary fibers and persistent CD68+ inflammatory cells were observed in injured areas in the 18M+vehicle group (Fig. 2n, S6a, b). Moreover, despite the existence of a small gap in reparative histological effects compared to those of uninjured tendons from young rats (Fig. S4a), two months of POG pretreatment was found to delay the progression of collagen disorder and relieve cellular senescence in aged, uninjured tendons, as evidenced by the declining ratio of γ-H2AX- and P21-positive cells and immunofluorescence staining (Fig. S6c–e). Hence, the oral administration of POG effectively relieved multiple phenomena of tendon aging, including cellular senescence, lipid droplet deposition, inflammatory infiltration, and impaired recruitment capacity of endogenous stem cells.

To provide a local sustained release of POG at the sites of tendon injury, we applied FDA-approved polylactic glycolic acid (PLGA) nanoparticles to load POG (POG-nps; Fig. 3a). SEM showed that both POG-nps and PLGA nanoparticles were spherically shaped, with similar diameters of 100–300 nm in the dry state (Fig. 3b). POG was gradually released over a 14-d period, as determined by high-performance liquid chromatography (HPLC‒MS; Fig. 3c). To determine whether the drug-releasing particles had antisenescent and protenogenic functions, we treated aged rTSPCs from aged rats with POG-nps at a concentration of 20 μmol·L−1 after dilution that was consistent with the previous dosage in vitro. POG-nps effectively rescued aged rTSPC functions, as evidenced by increased Ki67+ cells and reduced protein expression levels of P53 and P21 in aged rTSPCs in vitro (Fig. 3d, e). Moreover, pretreatment with POG-nps for 4 d prior to tenogenic induction promoted the expression of the tenogenic markers FMOD and TNC, with reduced activation of P65 and decreased secretion of the inflammatory factor IL-6 in aged rTSPCs (Fig. 3f, S7a, b), thereby recapitulating the antisenescent effect of POG.

Fig. 3

Local delivery of POG nanoparticles effectively promotes endogenous tendon healing by inhibiting lipid droplet deposition and inflammatory progression. a Schematic outlining local delivery of POG-nps to partial transection tendon injuries in aged rats. b (i) SEM images of PLGA (vehicle) and PLGA nanoparticles loaded with POG (POG-nps). (ii) Size distribution determined by dynamic light scattering (n = 4). c Cumulative release curves of POG from the POG-nps obtained using HPLC‒MS analysis. d (i) Immunofluorescence staining of Ki67 (n = 3). e Western blotting of the senescence-related proteins P53 and P21 (n = 3). f (i) Immunofluorescence staining of the tenogenic marker TNC (n = 3). g HE staining of neotendons from the aged rats n = 5. h (i) Immunofluorescence staining of CD146 in sections (n = 5). i Immunofluorescence staining of FABP4 (n = 5). j HE and Masson trichrome staining of neotendons n = 5. k (i) Immunofluorescence staining of COL1 and FMOD (n = 5). l (i) Immunofluorescence staining of γ-H2AX (n = 5). m (i) Immunofluorescence staining of IL-6 and MMP9 (n = 5). Data are represented as the mean ± SD. (**P < 0.01; ***P < 0.001)

To further verify the in vivo effect of the local delivery of drug-loaded nanoparticles on small-range tendon repair in aged rats, we injected POG-nps (18M + POG-nps) or PLGA nanoparticles alone as a vehicle (18M+vehicle) into the partially transected Achilles tendons at Days 0 and 7. Fluorescence staining showed retention of Cy5 in the tendon tissue at Days 0 and 7, indicating that the PLGA nanoparticles were an effective drug delivery vehicle in injured tendons (Fig. S7c). Similar to the therapeutic effects of the oral administration of POG (Fig. 2k–n), the local injection of POG-nps largely attenuated the formation of vacuole-like structures and lipid droplet deposition and substantially recruited larger amounts of endogenous CD146+ cells to the injured tendons after 7 d compared to those of the 18M+vehicle group (Fig. 3g–i). Four weeks after the incision injury, densely packed collagen fibers formed, with higher expression levels of the tenogenic markers COL1 and FMOD in the 18M + POG-nps group, indicating that POG-nps maintained the tenogenic differentiation of endogenous rTSPCs (Fig. 3j, k). Moreover, the delivery of POG-nps alleviated the expression of the senescence marker γ-H2AX (Fig. 3l) and the inflammatory markers MMP9 and IL-6, which are characteristics of age-related tendon diseases (Fig. 3m).

Both the systemic administration of POG and the local delivery of POG nanoparticles effectively enhanced the healing capacity of partial transection in aged rat tendons by enhancing rTSPC stemness, rejuvenating senescent phenotypes, and simultaneously suppressing lipid droplet deposition and inflammatory progression.

POG prevents rTSPC senescence by suppressing NF-κB

To explore how POG prevents aged rTSPC senescence, we first performed RNA sequencing (RNA-seq) to compare the transcriptome profiles of senescent rTSPCs from aged rats treated with POG or DMSO once every 2 d for 7 d. GO enrichment and pathway enrichment analysis demonstrated that aging-related genes and several aging-related signaling pathways, including p53 and AGE-RAGE signaling, were decreased in the POG-treated aged rTSPCs compared to the DMSO-treated aged rTSPCs (Fig. 4a). Moreover, multiple beneficial biological processes (e.g., animal organ regeneration, telomere maintenance, and mitochondrial DNA repair) exhibited an elevated tendency, indicative of the functional recovery of POG-treated aged rTSPCs (Fig. 4b). Notably, a consistent downregulation was observed in senescence-associated pathways, including NF-κB and mTOR, which have been previously reported to regulate the senescence of various adult stem cells.32,33 Recent studies have shown that the inhibition of mTOR or NF-κB could reduce inflammation or promote tendon repair.34,35 However, it was unclear whether the NF-κB pathway was also overactivated in the progression of aged tendons or TSPC senescence. Therefore, we performed immunofluorescence staining of the active form of the NF-κB subunit p65 on uninjured tendon sections from young and aged rats, which has been widely used as an indicator of the activation of the NF-κB pathway.36,37,38 The results revealed that the uninjured tendons from the aged rats had a higher ratio of cells expressing p65 than those from the young rats, and POG administration markedly reduced the percentage of p65+ cells in the aged rats (Fig. 4c). After tendon injury, different types of cells begin to expand or are recruited to the injured areas.39 NF-κB is often activated in response to injury stimuli and is beneficial for the start-up of the body response in the early phase. However, the chronic activation of NF-κB could delay tissue healing.40 Four weeks after the incision injury, we found that distinct p65-positive expression was sustained at the injury sites of aged rats compared to young rats, after which the administration of POG markedly repressed its expression (Fig. 4d). NF-κB can be activated by proinflammatory cytokines, such as TNF-α and IL-17, in inflammatory diseases and tissue injuries.41 Moreover, chronic inflammatory insult could induce cell senescence in vivo.42 To determine whether POG could inhibit TNF-induced NF-κB in rTSPCs isolated from young rats, we continuously stimulated rTSPCs treated with POG or DMSO with 10 ng·mL−1 TNF-α for 0, 15, and 30 min. As shown in Fig. 4e, POG inhibited TNF-α-induced p65 phosphorylation in rTSPCs. Immunofluorescence staining confirmed that POG inhibited p65 activation in the POG-pretreated rTSPCs compared to the DMSO-pretreated rTSPCs (Fig. 4f). Colony-forming experiments revealed that POG pretreatment once every 2 d for 14 d largely rescued the impaired CFU-F formation capacity that was caused by persistent TNF-α stimulation (Fig. 4g). Notably, TNF-α stimulation induced distinct cellular senescence, which was significantly suppressed by POG treatment, as evidenced by the results of SA-β-gal staining (Fig. 4h). Previous studies showed that chronic inflammation could induce senescence in mesenchymal stem cells, which may explain the poor tissue healing occurring in aging and the chronic disease of multiple tissues.43,44 Sirius Red staining showed that POG treatment significantly attenuated the inhibitory effect of TNF-α on collagen matrix formation by rTSPCs (Fig. 4i). Furthermore, immunofluorescence staining revealed that POG treatment significantly reduced the inhibitory effect of TNF-α on TNC secretion (Fig. 4j).

Fig. 4

POG prevents rTSPC senescence by suppressing NF-κB signaling. a Gene set enrichment analysis of downregulated pathways of aged rTSPCs isolated from aged rats with or without POG treatment. b GO term analysis of related biological processes of aged rTSPCs isolated from aged rats with or without POG treatment. c (i) Immunofluorescence staining of P65 in uninjured tendons (n = 5). d (i) Immunofluorescence staining of P65 in injured tendons (n = 5). e Western blotting of the phosphorylation of P65. f (i) Immunofluorescence staining of P65. (ii) Semiquantification of (i) (n = 3). g (i) CFU-F assay of DMSO- and POG-treated rTSPCs from young rats after 14 d of TNF-α stimulation (n = 3). h (i) SAβ-gal staining of DMSO- and POG-treated rTSPCs from young rats after 14 d of TNF-α stimulation (n = 3). i (i) Sirius Red staining of DMSO- and POG-treated rTSPCs from young rats after 14 d of TNF-α stimulation. (ii) Semiquantification of (i) (n = 3). j (i) Immunofluorescence staining of TNC in DMSO- and POG-treated rTSPCs from young rats after 14 d of TNF-α stimulation. (ii) Semiquantification of (i) (n = 3). Data are represented as the mean ± SD. (*P < 0.05; **P < 0.01; ***P < 0.001)

POG maintains aged rTSPC function by inhibiting mTOR signaling and activating autophagy

Given that the inhibition of the mTOR pathway promotes lifespan and its role in the aging of adult stem cells,45,46 we sought to investigate whether POG also inhibited rTSPC senescence during in vitro serial passaging and in vivo natural aging by inhibiting the activation of mTOR. Senescent rTSPCs from the aged rats with POG stimulation (18M + POG) showed inhibited phosphorylated S6 (pS6) expression compared to the DMSO-treated controls (18M + DMSO) (Fig. 5a). Similarly, the POG-treated passaged rTSPCs at P12 (P12 + POG) displayed distinctly reduced S6 phosphorylation compared to the DMSO-treated passaged rTSPCs from the young rats at P12 (P12 + DMSO) (Fig. 5b). The immunofluorescence of uninjured tendon sections in the 18M + POG group showed significantly lower levels of pS6 expression than those in the 18 M+vehicle group (Fig. 5c). As reported previously, mTOR is a negative regulator of autophagy, and its activities are gradually increased in multiple tissues during aging. The utility of mTOR inhibition as an autophagy-inducing therapy for age-related diseases has been explored widely.47,48 The slow rate of autophagy and the accumulation of autophagic vesicles and lysosomes are characteristics of senescent phenotypes.49 Here, we visualized lysosomal trafficking using LysoTracker and found that aged rTSPCs from the 18+ POG group displayed markedly decreased lysosomes, close to the normal level observed in young rTSPCs from young rats (3M + DMSO) (Fig. 5d). Senescent cells were often ineffective in eliminating abnormal proteins,50,51 as reflected by the increased protein content in aged rTSPCs from the 18 + DMSO group and its recovery to normal levels in aged rTSPCs after POG treatment (Fig. 5e). In fact, S-βgal is an intracellular lysosomal enzyme. When stem cells become senescent, excessive staining of S-βgal will occur in the cells, which also reflects the abnormal metabolic condition of lysosomes.52 Therefore, our results demonstrate that POG is able to improve the condition of lysosomes in senescent cells. Next, we examined whether POG treatment could restore normal autophagic flux in aged rTSPCs from aged rats. Immunofluorescence staining of lysosomes and microscopic detection of cytoplasmic GFP-LC3 dots revealed that POG induced autophagic flux in aged rTSPCs (Fig. 5f). Furthermore, supplementation with POG significantly reduced the p62 protein level and increased the microtubule-associated protein 1 A/1B light chain 3B (LCB II) expression level in aged rTSPCs (Fig. 5g). Similarly, serial passaging also increased the expression of the typical autophagy target protein SQSTM1/p62 and decreased the LCB II level, which was reversed by POG treatment (Fig. 5h). Notably, P62 can serve as a receptor for vesicles to be degraded by autophagy, as well as for ubiquitinated protein aggregates to be cleared. The p62 protein can thus target autophagosomes and promote the clearance of ubiquitinated proteins.53 Therefore, the reduction in P62 protein levels here also reflects the reduction in abnormal proteins to some extent.

Fig. 5

POG maintains aged rTSPC functions by inhibiting mTOR signaling and activating autophagy. a Western blotting of pS6 and P21 in rTSPCs. b Western blotting of pS6 and P21 in rTSPCs. c (i) Immunofluorescence staining of pS6 in uninjured tendons (n = 5). d (i) LysoTracker staining showing lysosomes of TSPCs. (ii) Semiquantification of (i). (n = 30 different cells). e Total intracellular protein content of rTSPCs (n = 3). f (i) Immunofluorescence staining of the autophagy protein LC3 in rTSPCs. (ii) Semiquantification of (i) (n = 30 different cells). g Western blotting of LC3I/II and P62 in rTSPCs. h Western blotting of the autophagy proteins LC3I/II and P62 in rTSPCs. i Western blotting of ATG7, P21 and γ-H2AX in rTSPCs n = 3. j (i) CFU-F assay of rTSPCs from aged rats. (ii) Semiquantification of (i) (n = 3). k (i) SAβ-gal staining of rTSPCs. (ii) Semiquantification of (i) (n = 3). l (i) Immunofluorescence staining of the DNA injury-related protein γ-H2AX in rTSPCs. (ii) Semiquantification of (i) (n = 3). m (i) Sirius Red staining (left panel) and immunofluorescence staining (right panel) of the tenogenic markers TNC and TNMD. (ii) Semiquantification of (i) (n = 3). n HE staining of parallel-aligned collagen structures n = 3. Data are represented as the mean ± SD. (ns: not significant; *P < 0.05; **P < 0.01; ***P < 0.001)

These results indicate that senescent rTSPCs induced by in vitro serial passaging and in vivo natural aging displayed elevated phosphorylated S6 expression and impaired autophagic activity, which could both be reversed by in vitro POG treatment. This finding further confirms that the in vitro long-term passage model was also a reliable cell senescence model that could partially reflect the functional changes in natural aging cells.

ATG5, ATG7, and BECN1 (encoding Beclin-1) code for essential components of the autophagic machinery and are downregulated during normal aging.47,54 We thus hypothesized that the beneficial biological effects of POG on aged rTSPCs from 18-month-old rats depend on autophagic regulation. For analysis of this hypothesis, aged rTSPCs were stably infected with a lentiviral vector expressing shRNA targeting the autophagy gene ATG7 (sh-ATG7). Western blotting showed that the expression of P21 and γ-H2AX was significantly elevated after the knockdown of ATG7 (Fig. 5i). Importantly, ATG7 knockdown abrogated the functions of POG in maintaining the colony-forming capacity of aged rTSPCs (Fig. 5j), inhibiting cellular senescence (Fig.