Introduction

Tendon injuries have high incidence due to vigorous exercises, inappropriate actions, accidents, and population aging1,2. Unfortunately, tendon-related injuries are often result in inefficient repair, and up to 40% of patients will suffer from tendon rupture again3. Available treatments fail to restore the structural integrity and biomechanical functions of injured tendons due to the limited endogenous regeneration capacity4. In recent years, with the vigorous development of stem cells, stem cell-based therapeutics have been demonstrated as a promising approach for tendon repair5,6 Among these various types of stem cells, mesenchymal stem cells (MSCs), are widely used in tissue engineering because of their proliferation ability, multi-lineage differentiation potential and immunomodulatory ability7. Therapeutic effects of MSCs are affected both by cell quantity and the direct differentiation of MSCs at injection site. However, the direct injection of stem cell suspension raises safety and effectiveness concerns due to the cell dispersion, uncontrolled cell behaviors, and the lack of support to guide the cells8. Cell sheet engineering provides a scaffold-free 3D extracellular matrix (ECM) enriched environment, which can locally delivery cells with preservation of the cell-cell junctions upon implantation9. This technique combines the benefits of both stem cells and cell sheet engineering, making it a highly effective strategy. Due to these advantages, cell sheets have already been applied as a treatment for tissue/organ damage, such as corneal injury10, heart/liver diseases11,12, and tendon injury13.

In light of the clinical applications, adipose-derived mesenchymal stem cells (AD-MSCs) have drawn increasing attention benefit by being easily harvested with low invasive procedure, and providing larger number cells in comparison to bone marrow stem cells (BMSCs)14. AD-MSCs derived cell sheets had been proved to be multipotent2, and by configuring the culture medium constituents, AD-MSCs cell sheets can be inducted into a tendon-like lineage15,16. Despite their abundant source, AD-MSCs often required to be expanded in vitro for several population doublings to achieve a sufficient number of cells before cell sheet formation and transplantation. Ex vivo expanded MSCs also have finite lifespans, tissue regeneration by AD-MSCs is affected by their gradual decline in regenerative capacity after reaching a certain number of cell divisions17. Culture-associated senescence has been considered as an important instigator of AD-MSCs dysfunction during the process of ex vivo expansion. Therefore, it is imperative to understand the mechanisms underlying MSCs senescence and explore potential techniques to delay aging. This knowledge is not only vital for basic research but also holds significant implications for practical applications.

Like many other pathological process, MSCs senescence is a complex phenomenon caused by various factors including both genetic and environmental factors. Shortening of telomere, DNA damage, accumulation of chromosomal aberrations, and oxidative stress have all been well characterized and documented as contributing factors in this process18,19. However, the mechanisms underlying MSCs senescence remain incompletely understood. In recent years, there have been extensive studies revealing that culture expansion of stem cells evokes significant epigenetic alterations20. Among these, cellular senescence is associated with significant changes in histone methylation marks, which play a crucial role in controlling gene expression profiles18,21,22. Fernández et al. previously found that the active chromatin mark H3K4me1 is related to DNA hypomethylation, and the H3K4me1-associated DNA hypomethylation could play a role in the deregulation of gene expression during aging23. Replicative cellular senescence is considered a form of in vitro aging sharing many of the molecular events associated with physiological aging24. The association between histone methylation and lifespan was demonstrated by deletion of the H3K4 methyltransferases in C.elegans, which extended the worms lifespan25. Setd7 was initially described as a histone lysine methyltransferase capable of catalyzing the monomethylation of H3K4me1 in vitro26. It has been found to regulate inflammation-induced cell proliferation, migration, and the oxidative stress response27, which also play important roles during cellular senescence. Judson et al. reported that inhibition of cytoplasmic Setd7, potently expands naive, undifferentiated mouse and human muscle stem cells (MuSCs)28. However, whether Setd7 can regulate the senescence of MSC remains unknown.

Hence, the present study aims to explore the potential role of Setd7 in the replicative senescence of AD-MSCs, with a focus on enhancing their differentiation capacity. Additionally, we aimed to enhance the tenogenic capacity of AD-MSCs derived cell sheets, with the goal of facilitating the regeneration of patellar tendons. To achieve this, we analyzed data from Gene Expression Omnibus (GEO) database and found a positive correlation between SETD7 expression and the replicative senescence of hMSCs. Subsequently, we utilized continuous in vitro subculture of human AD-MSCs (hAD-MSCs) as the cell replicative senescence model and confirmed the strong induction of SETD7 in the in vitro cell subculture model. We then validated that PFI-2, a pharmacological agent, could target SETD7 to retard the culture-associated senescent phenotype of hAD-MSCs, leading to enhanced self-renewal and multi-lineage differentiation ability. Finally, we confirmed the therapeutic efficiency of PFI-2-treated hAD-MSCs derived cell sheets in a rat patellar tendon injury model. Collectively, the findings of this study provide a new molecular target for antagonizing cellular senescence of MSCs during prolonged passages, thus maximum the supply of therapeutic-grade MSC products.

Materials and methods

Datasets analysis

The datasets discussed here have been deposited in NCBI’s Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/). Datasets of SETD7, TWIST1, WRAP53, p16INK4A, p15, and ARHGAP 29 expression during in vitro senescence of human bone marrow mesenchymal stem cells (hBMSCs) were obtained from GEO Series accession number GSE9593 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE9593). Expression data were available for three samples. Expression profiles of SETD7, TWIST1, TWIST2, and ARHGAP 29 in hBMSCs cultured in a 3-dimensional (3D) system or in conventional petri dishes (2D) for 19 days were downloaded from GEO Series accession number GSE52896 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE52896). Expression data were available for five samples.

Cell culture

hAD-MSCs was provided by Cyagen Biosciences (HUXMD-01001). Cells were cultured in low glucose DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (all from Gibco) at 37 °C with 5% CO2 and saturated humidity. For the PFI control, hAD-MSCs were cultured in standard medium supplemented with PFI (10 µM, Selleck). Fresh medium was replaced every 3 days. Cells were passaged when grown to 80–90% confluency. Cells from passages 2 (P2) to 10 were used for further studies.

Senescence-associated β-galactosidase (SA-β-gal) staining

SA-β-gal staining was performed as described previously29. Briefly, cultured cells were washed with PBS and fixed in 4% paraformaldehyde for 15 min at room temperature. Next, fixed cells were incubated with fresh SA-β-gal staining solution overnight at 37 °C (Beyotime Biotechnology). Images were captured using an inverted phase contrast microscope (Nikon). The percentage of positive cells was calculated by counting at least 200 cells in five random fields of view.

RNA isolation, reverse transcription, and qRT-PCR

Total cellular RNA was isolated by lysis in TRIzol (Beijing Solarbio Science & Technology Co., Ltd.) and one microgram of total RNA was used for cDNA synthesis. PCR was performed using a Brilliant SYBR Green QPCR Master Mix (Takara) on a Light Cycler apparatus (Bio-Rad CFX96). The PCR cycling consisted of 40 cycles of amplification of the template DNA with primer annealing at 60 °C. The relative level of expression of each target gene was then calculated using the 2 − ΔΔCt method30. Primer sequences and products size were listed in Table S1 and S2. Each real-time PCR was performed on at least 3 different experimental samples; representative results are shown as target gene expression normalized to reference gene GAPDH. Error bars reflect one standard deviation from the mean of technical replicates as described.

Western blot analysis

Proteins were extracted by using RIPA lysis buffer containing protease inhibitors cocktail (Beyotime Biotechnology). Protein lysates were then centrifuged to pellet cell debris, and the supernatant was removed and quantified by BCA Protein Assay Kit (Thermo). Protein samples (20 mg) were run using SDS-PAGE and transferred to polyvinylidene fluoride membranes. Membranes were blocked in 5% BSA and incubated with primary antibodies at 4 °C overnight. Primary antibodies include anti-SETD7 (abcam), and p16INK4A (abcam), anti-p21 (abcam), anti-EPHA4 (abcam), anti-EYA2 (Sigma), anti-DCN (biobision), anti-MKX (LifeSpan BioSciences), and anti-GAPDH (Proteintech). HRP-conjugated IgG was used as a secondary antibody (Beyotime Biotechnology). Protein bands were visualized using WesternBright sirius HRP Substrate (Advansta). Protein expression levels were calculated by Image J.

Population doubling time (PDT)

To examine cell growth rate, the time required by cells for each population doubling (PDT) were determined during successive subcultures using the formula: PDT = tplg2/(lgNH − lgNI), where NI is the inoculums cell number; NH is the cell harvest number; and ‘‘t’’ is the time of the culture (in hours)31.

Differentiation assays

The osteogenic, adipogenic, and chondrogenic differentiation capacity of hAD-MSCs at P4 and 8 with or without PFI-2 pretreatment were assessed as described previously32. To promote adipogenic differentiation, the cells were cultured in DMEM supplemented with adipogenic stimulatory supplements (StemCell Technologies) for 2 weeks. Positive induction was detected by oil red staining. For quantification, the stain was solubilized with 60% isopropyl alcohol, then absorbance was measured at 450 nm. To promote osteogenic differentiation, the cells were induced at low density (103 cells/cm2) in the presence of 10 mM β-glycerol phosphate, 0.1 µM dexamethasone, and 50 µg/mL ascorbic acid (all from Sigma-Aldrich) cocktails for 3 weeks. To quantify the stained nodules, the stain was solubilized with 0.5 mL 5% SDS in 0.5 N HCl for 30 min at room temperature. Absorbance was measured at 405 nm. Calcium deposits were detected by alizarin red staining. To promote chondrogenic differentiation, confluent monolayer cultures of hAD-MSCs were released using trypsin-EDTA and suspended in growth medium at a density of 2.5 × 107 cells/mL. Micro-masses were obtained by pipetting 20 µL of cell suspension into individual wells of 24-well plates. After 3 h, the growth medium was gently added and the cultures were left to rest for an additional of 24 h. Differentiation was promoted by 10 ng/mL TGF-β3 supplementation. After 14 days, micro-masses were harvested for Alcian blue staining.

Colony forming ability assays

A colony formation assay was used to characterize the self-renewal potential of hAD-MSCs. Briefly, hAD-MSCs (P4 and 8) with or without PFI-2 treatment were replating at 10 and 20 cells/cm2 in low DMEM plus 10% FBS for 14 days at 37 °C. The colonies formed were stained with 1% crystal violet (Sigma) in methanol for 10 min.

Cell sheets culture

As described previously, hAD-MSCs (P8) with or without PFI-2 treatment were evenly seeded in well plates. Upon reaching 80–90% confluence, culture medium was changed to high DMEM supplemented with 10% FBS and 50 mg/mL ascorbic acid (Sigma). A multilayered cell sheet was detached from the substratum by applying a small roll-up force after 2 weeks in culture and harvested for the subsequent RNA isolation, western blot, and transplantation.

Animal model of patellar tendon injury and repair

Elven female Sprague-Dawley adult rats (bodyweight 200–220 g) were purchased from Hunan Silaikejingda Experimental Animal Co.,Ltd. All experimental procedures involving animals were conducted in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and were approved by the Animal Ethics Committee of Hunan University (HNUBIO202101002). The study was also conducted in compliance with the ARRIVE guidelines, ensuring that the methods and results sections fully adhere to the recommendations for reporting in vivo experiments. Briefly, the rats were treated with cyclophosphamide (150 mg/kg) 24 h before the operation. Animals were anesthetized by intraperitoneal injection of 1.25% tribromethyl alcohol (10 mL/kg), then a gap wound model (1 mm in width and 4 mm in length) of the patellar tendon was created. PFI-2 treated cell sheets (PFI-CS) or control cell sheets (Con-CS) were then implanted into the left and the right defect. The wounds were then irrigated, the fascia and skin were closed. The animals were allowed free cage activity until euthanasia. At 2 and 4 weeks after implantation, the animals were euthanized using a method that ensures rapid and painless death. Specifically, an overdose of a sodium pentobarbital-based euthanasia solution was administered via intraperitoneal injection at a dosage of 120 mg/kg body weight. The repaired patellar tendons were collected and processed for histology, transmission electron microscopy (TEM), gene analysis, and mechanical testing.

Histologic evaluation

Specimens were immediately fixed in 10% neutral buffered formalin, dehydrated using graded ethanol, cleaned, and embedded in paraffin. Histologic sections (7 μm) were prepared using a microtome (Leica RM2245) and stained with hematoxylin and eosin (H&E). Images were captured using a digital slide scanner system (3DHISTECH). General histological scores were performed using a blinded semi-quantitative scoring system based on six parameters (fiber structure, fiber arrangement, rounding of nuclei, inflammation, vascularity, and cell population) of H&E staining by three independent observers who were blinded to the annotated histological type6. In addition, Masson trichrome and Sirius red staining were performed according to standard procedures to examine the general appearance of the collagen fibers. Image J software was applied to semi-quantitative analyze the average optical density of the repaired area, based on Masson staining. Briefly, the integrated optical density (IOD) and area of the collagen-stained regions to determine the mean optical density (MOD = IOD / area SUM). Polarizing microscopy was used to identify mature and alignment of collagen fibrils.

Immunohistochemistry

To analyze protein expression levels within tendon-like tissue, immunohistochemical staining was performed on paraffin sections. The primary antibodies used in this study were anti-human nuclei antibody (Abcam), anti-MKX antibody (LifeSpan BioSciences), anti-Decorin antibody (Biovision). The specimens were then incubated with goat anti-mouse (Beyotime Biotechnology) or goat anti-rabbit (Beyotime Biotechnology) second antibodies conjugated with HRP. Then, 3, 3’-diaminobenzidine substrate system (DAB, Advansta) was used for color development according to the manufacturer’s protocol. Hematoxylin staining was used to reveal the nuclei.

Collagen content assay

The amount of deposited collagen was quantified using a hydroxyproline assay kit (Jiancheng, Ltd., Nanjing, China) following the manufacturer’s protocol. The repaired patellar tendons were prepared from five independent samples.

TEM

Tissue specimens were freshly fixed follow the standard procedures for TEM to assess collagen fibril diameter. Approximately 500 collagen fibrils of each sample were measured to obtain an accurate representation of the fibril diameter distribution. Image-ProPlus software was used to measure the collagen fibril diameters.

Mechanical testing

Mechanical testing was performed using a material testing machine (HYC-2011, Hongjin) equipped with a 50 N load cell. The hind limbs were collected and all soft tissue spanning the knee, except for the center of the patellar tendon, were transected. Measurement of the tendon cross-sectional area was performed, and the femur-patellar tendon-tibia complex (FPTC) was then rigidly fixed to custom-made clamps. After applying a preload of 0.1 N, each FPTC underwent preconditioning by cyclic elongation of 0–0.5 mm for 20 cycles at 5 mm/min. This was followed by a load-to-failure test at an elongation rate of 5 mm/min. The structural properties of the FPTC were represented by stiffness (N/mm), failure force (N), modulus (MPa) and stress at failure (MPa).

Statistical analysis

All quantitative data are presented as mean ± SD and checked by normality tests. Student’s paired t test was performed for comparison of data of paired samples and ANOVA was used for multiple group comparisons. p < 0.05 was considered statistically significant. The significance level is presented as either *p < 0.05 or **p < 0.01.

Results

SETD7 expression is positively correlated with the replicative senescence of hMSCs

To identify whether SETD7 regulates the senescence of MSCs, we examined the RNA-sequence data of early and late passages of MSCs from the publicly available gene expression datasets. They are accessible through GEO series accession number GSE9593, GSE52896, and GSE59966. To determine gene expression changes during in vitro senescence of hBMSCs, we analyzed global gene expression profiles from early passage (P2) and senescent passage (PX) samples, as reported in GSE9593 datasets by Wagner et al.33. We selected the P2 and P7 cells for gene expression analysis. As expected, with the increase of passage, the expression of stemness TWIST1 decreased34,35, while senescence-related genes including p16INK4A, p15, WRAP53, and ARHGAP 29, were significantly upregulated at passage 2 and passage 7 from three MSC donors. Among the genes upregulated in senescent hBMSCs, SETD7 has shown significantly increased (Fig. 1A). More importantly, TWIST1 mRNA expression demonstrated a negative correlation with the mRNA expression of SETD7 (Fig. 1B, left panel, R2 = 0.5254). In contrast, senescence-associated genes including p15 (Fig. 1B, middle panel, R2 = 0.9912) and ARHGAP 29 (Fig. 1B, right panel, R2 = 0.6683) observed a strong positive correlation with the expression of SETD7. The data from GSE59966, contributed by Hanzelmann et al., which compared gene profile of MSCs at early (P4) and late passages (P13), also indicated an upregulated expression of SETD7 after culture expansion (Figure S1)36. Conventional monolayer culture of MSCs on plastic dishes (2D) is associated with a progressive reduction of functionality, and is always accompanied with cellular senescence. 3D culture niche, which provides a tissue-specific microenvironment maintains MSCs function37,38. Microarray analysis (GSE52896) conducted by Papadimitropoulos et al. compared 2D and 3D expanded MSCs identified significantly upregulated of stemness marker TWIST1 and TWIST2 in the 3D expansion when compared with the 2D cultures. Suppression of SETD7 expression was found in the 3D-perfusion vs. the 2D-expanded MSCs (Fig. 1C), indicating that the expression of SETD7 under 2D culture conditions maybe not conducive to the maintenance of cell function39.

Fig. 1
figure 1

Changes of SETD7 and cellular senescence genes expression in hMSCs duringex vivo culture, as analysis of from GEO data. (A) Expression values of SETD7, TWIST1, and senescence-related gene in hBMSCs at P2 and P7 from dataset GSE9593, n = 3. (B) The correlation between SETD7 and TWIST1, p15, and ARHGAP29 (GSE9593). (C) Expression values of SETD7, TWIST1/2, and ARHGAP29 in hBMSCs cultured in a 3D perfusion bioreactor system or in conventional petri dishes (2D) from dataset GSE52896, n = 5. Data shown from representative sample and represented as mean ± SD. Two-tailed Student’s test, *p < 0.05, **p < 0.01. Abbreviations: P, passage; 3D, 3 dimensional; 2D, 3 dimensional.

Full size image

In attempt to confirm the relationship between SETD7 and MSCs senescence, here replicative senescent MSCs was achieved by long-term cultivation until the signatures of aging were detected. The cellular senescence markers in hAD-MSCs at P4, P8, and P10 was further investigated. SA-β-gal activity is the most common biomarker for evaluation of the senescent state of continuous cell cultures under in vitro condition29. As expected, a progressively increased SA-β-gal activity in hAD-MSCs was observed between P4 and P10 (Fig. 2A, B). In agreement with the staining results, analysis of stemness and proliferation genes levels of TWIST1, KI67, and WRAP53 in hAD-MSCs showed sharply decline during serial expansion, whereas cell senescent molecular markers including p16INK4A, p21, and ARHGAP29 were consistently increased along with prolonged passages. In accordance with the previous GEO analysis, SETD7 mRNA expression level was readily upregulated from P4 to P10 (Fig. 2C). The protein expression level of p16INK4A and p21 in hAD-MSCs were consistent with the genes, which remarkably increased with serial passaging (Fig. 2D, E). Moreover, the protein level of SETD7 also progressively increased with in senescent hAD-MSCs. Collectively, these results confirm the GEO data, and suggest that SETD7 is potentially involved in hAD-MSCs senescence.

Fig. 2
figure 2

A gradual increase of SETD7 was observed during successive passages of hAD-MSCsin vitro. (A) SA-β-gal staining in serial passage hAD-MSCs (red arrows: senescent cells). Scale bar = 50 μm. (B) The percentages of senescent cells were determined. (C) SETD7 and senescence-associated gene expression were assessed by qRT-PCR in hAD-MSCs at P4, P8, and P10, n = 3. (D, E) Western blot and quantitative analysis of SETD7, p21, and p16INK4A protein expression in hAD-MSCs at P4, P8, and P10, n = 3. Data shown from representative sample and represented as mean ± SD. One-way analysis of variance (ANOVA), *p < 0.05, **p < 0.01. Abbreviation: P, passage.

Full size image

Inhibition of SETD7 attenuates the replicative senescent changes of hAD-MSCs

Given that continuous passage led to cell senescence and increasing expression of SETD7. We next tried to investigate the specific role of SETD7 in hAD-MSCs senescence. hAD-MSCs cultures were exposed to the inhibitor of SETD7 (PFI-2), cells treated with DMSO were considered as control. A gradual senescent process in hAD-MSCs was demonstrated by progressively increased SA-β-gal-positive cells. Notably, PFI-2 treatment showed an attenuated senescent phenotype in the late passages (P8 and P10) compared to the untreated cells, as indicated by reduced β-gal-positive senescent cells (Fig. 3A, B), which suggested its senescence-delaying potential. This was also coupled with a significant repression of p16INK4A and p21 protein expression in hAD-MSCs at late passages with PFI-2 treatment (Fig. 3C and Figure S2). Besides, mRNA levels of SETD7 and cell senescent markers (p21 and ARHGAP29) in hAD-MSCs were highly upregulated from P4 to P10, while PFI-2 treatment reduced their expression (Fig. 3D). More importantly, PFI-2 treatment rescued the expression of KI67, TWIST1, and WRAP53, which were pivotal genes for the cellular proliferation and stemness. This gradual decline in the proliferative ability of hAD-MSCs with concomitant raise in the passage number was further confirmed by calculating PDT (Figure S3). Taken together, the above findings demonstrate that inhibiting the activity of SETD7 by PFI-2 can delay the senescent process of ex vivo cultured hAD-MSCs.

Fig. 3
figure 3

Inhibition of histone methyltransferase SETD7 delayed cellular senescence in hAD-MSCs. (A, B) Representative images of SA-β-gal staining (A) and quantitative analysis of SA-β-gal-positive cells (B) in hAD-MSCs at definite passages with or without PFI-2 treatment. Scale bar = 100 μm. (C) Western blot analysis of SETD7, p21, and p16INK4A protein expression in hAD-MSCs at P4, P8, and P10 with or without PFI-2 treatment. (D) The expression of SETD7 and senescence-related genes in hAD-MSCs at indicated passages with or without PFI-2 treatment. Data shown from representative sample and represented as mean ± SD. One-way analysis of variance (ANOVA), *p < 0.05, **p < 0.01. Abbreviation: P, passage.

Full size image

PFI-2 restores the self-renewal deficit and promotes the differentiation capacity of senescent hAD-MSCs

Replicative senescence influence both proliferation and differentiation potential of MSC. We then investigated the effects of PFI-2 on the functional capacity of hAD-MSCs between early and late passage. One of the earliest, indirect methods of assessing MSCs senescence is the colony-forming assay. Under standard culture conditions, the CFU capacity of hAD-MSCs was markedly decreased with increasing passage numbers. PFI-2 treated hAD-MSCs exhibited significantly enhanced CFU efficiency in the late passages compared with the control cells (Fig. 4A, B). The hallmark of MSC is its ability to differentiate into mesoderm lineages. The adipogenic, osteogenic, and chondrogenic differentiation capacity of hAD-MSCs was verified by Oil Red O staining of lipid vesicles, Alizarin red staining of calcium deposits, and Alcian blue staining of sulfated proteoglycans, respectively. Unsurprisingly, hAD-MSCs showed dramatically decayed in adipogenic and chondrogenic differentiation potential in both control and PFI-2 groups with the increase of cell passage (Fig. 4C, D, and Figure S4), but the adipogenic and chongrogeneic capacity were better retained in PFI-2 group as compared to control group. But for osteogenic differentiation, hAD-MSCs even exhibited slightly increased in osteogenic differentiation potential over continuous subculture, but no significant difference existed between the control and PFI-2 cultured MSCs (Fig. 4E, F). This phenomenon had also been reported in previous studies that the expression of osteogenic genes such as collagen 1, alkaline phosphatase (ALP), Runx2, bone sialoprotein, osteocalcin and osteopontin increase over continuous subculture of MSCs33,40,41,42.

Fig. 4
figure 4

Suppression of histone methyltransferase SETD7 enhanced differentiation capacity of hAD-MSCsin vitro. (A, B) Clonogenic activity and quantitative analysis of hAD-MSCs in earlier and senescent cultures with or without PFI-2 treatment. (C) Representative images of oil red staining in different passage of hAD-MSCs (P4 and P8) under induction condition with or without PFI-2 treatment, n = 3. Scale bar = 100 μm. (D) Alcian blue staining of hAD-MSCs in earlier and senescent under induction condition with or without FPI-2 treatment. (E, F) Alizarin red staining (F) and quantitative analysis (E) of earlier and senescent hAD-MSCs under induction condition with or without PFI-2 treatment, n = 3. Scale bar = 100 μm. (G, H) qRT-PCR (H) and western blot (G) analysis of tendon-associated genes and proteins in cell sheets from hAD-MSCs at P8, n = 3. Data shown from representative sample and represented as mean ± SD. Two-tailed Student’s test, *p < 0.05, **p < 0.01. Abbreviation: P, passage. hAD-MSCs used for cell sheets induction were at P8.

Full size image

To further evaluate the effect of PFI-2 on tenogenic differentiation potential of hAD-MSCs at late passage, we examined the mRNA and protein levels of the key tendon-related markers using cell sheet model. Gene expression profile showed that PFI-2 treatment led to a strong induction of tendon-related genes, including SCX (2.65-fold, p < 0.01), SIX1 (2.94-fold, p < 0.01), EGR1, MKX, HOXA11 (4.88-fold, p < 0.01), EPHA4 (2.91-fold, p < 0.01), TNMD (3.25-fold, p < 0.05), FMOD (1.94-fold, p < 0.05), BGN (2.39-fold, p < 0.01), TNC (2.46-fold, p < 0.01), DCN, and COL1 (Fig. 4G). Consistent with the gene expression results, PFI-2 treatment also increased the protein levels of EPAH4, EYA2, DCN, and MKX in the cell sheets (Fig. 4H and Figure S5). Collectively, these data demonstrate that inhibition of SETD7 can maintain self-renewal and the differentiation capacity of senescent hAD-MSCs improve the tendon differentiation ability of hAD-MSCs.

Transplantation of cell sheets from PFI-2 treated hAD-MSCs with improved tenogenic potential in a rat patellar tendon window defect model

Having established the role of SETD7 in hAD-MSCs senescence in vitro, we moved forward to examined the efficacy of hAD-MSCs with or without PFI-2 treatment for tendon regeneration, for which a rat patellar tendon window defect model was adopted (Fig. 5A). In consideration of stem cell injection causes massive cell loss and a low rate of survival, hAD-MSCs derived cell sheets with (PFI-CS) or without (Con-CS) PFI-2 treatment will overcome these limitations. Briefly, a 1 × 4 mm2 window defect in the middle of the patellar tendon was created (Fig. 5B, upper panel), after that PFI-CS or Con-CS was implanted in the defect (Fig. 5B, lower panel). Samples were collected for further analysis at 2 or 4 weeks after implantation. In the view of gross morphology, the regenerated tendons in both groups integrated well with the surrounding tendon tissues. The window gap could be observed in both groups (Fig. 5C) but became less distinctive with time in both groups (Fig. 5D). Notably, at both time points, the window defect was more visible in the Con-CS group, when compared to that in the PFI-CS group. And the repaired tendon looked white in PFI-CS-treated group, while it looked translucent, faint yellow in the Con-CS-treated group. Furthermore, immunohistochemistry with anti-human nuclei primary antibodies confirmed the survival of implanted hAD-MSCs in vivo (Figure S6).

Fig. 5
figure 5

Cell sheets with PFI-2 treatment for tendon regeneration in a rat patellar tendon window defect mode. (A) Experimental design for the in vivo animal experiment. (B) A window defect (1 mm × 4 mm) was created in the patellar tendon of rats (top panel). Then the cell sheets were placed within the tendon defect and sutured to the tendon (bottom panel). (C, D) Gross observation of repaired tendons at 2 and 4 weeks after implantation. (E, F) The expression levels of tendon-related genes in repaired tendons at 2 weeks (E) and 4 weeks (F) post-surgery, n = 4. Data shown from representative sample and represented as mean ± SD. Two-tailed Student’s test, *p < 0.05, **p < 0.01. Abbreviation: CS, cell sheet; Con-CS, cell sheet without PFI-2 treatment; PFI-CS, cell sheet with PFI-2 treatment. hAD-MSCs used here were at P8.

Full size image

To evaluate the effects at molecular level, we detected the expression differentiation of tendon-related genes, including transcription factor genes (Scx, Mkx, and Egr 1), ECM molecules (Col 14 and Dcn), Epha 4 receptor, Eya 2, and Tnmd at 2 and 4 weeks after implantation. The results indicated that PFI-CS treatment significantly enhanced the tendon-related genes expression in comparison with the Con-CS-treated group, at both time point (Fig. 5E, F). In detail, the expression levels of Scx, Egr 1, Col 14, and Dcn were 3.11-fold (p < 0.01), 1.95-fold (p < 0.05), 2.90-fold (p < 0.01), and 3.21-fold (p < 0.01) than that in the control group 2 weeks post-surgery (Fig. 5E). The levels of all the tendon-related genes detected here were increased over time in both groups. To be specific, Scx, Mkx, Egr 1, Epha4, Eya2, Col 14, and Dcn were 6.50-fold (p < 0.01), 3.37-fold (p < 0.05), 8.60-fold (p < 0.01), 7.31-fold (p < 0.01), 6.36-fold (p < 0.01), 7.36-fold (p < 0.01), and 3.53-fold (p < 0.05) higher in PFI-CS group than that of the Con-CS group (Fig. 5F). Tnmd is indeed a highly specific marker for mature tendons and is crucial for understanding tendon development and maturation. Our results showed a modest increase in Tnmd expression within the PFI-CS group relative to the Con-CS group. Although, this difference did not reach statistical significance.

After that, histological staining of tendon section was used to observe the tissue structure of repaired tendons. At 2 weeks post-surgery, H&E and Masson trichrome staining showed that PFI-CS-treated tendons repaired with denser connective tissue and more ECM deposition than that of Con-CS-treated tendons (Figure S8A), with a greater number of cells exhibiting a spindle-shaped morphology aligned/organized along the longitudinal (tensile) axis of the tendon. The histological score of the PFI-CS group was significantly lower than that of the Con-CS group (6.44 ± 0.88 vs. 8.26 ± 0.65) (Fig. 6C), particularly in fiber arrangement and density of cells (Figure S7A). But there was still a clear gap between host tendon and neo-tendon at both groups at 2 weeks, which was also confirmed by the polarized light observation (Fig. 6A). The tendons repaired with better structure at 4 weeks when compared with 2 weeks, as indicated by increased matrix intensity and parallelism in both groups (Fig. 6B, Figure S7B and S8B). In addition, the histological score of PFI-CS group was lower than that of the Con-CS group (5.56 ± 0.91 vs. 7.52 ± 1.41) (Fig. 6C), which suggested PFI-CS promoted tendon repair with better tissue structure. Polarized light detection also showed that the formation of continuous collagen fibers at the junction site in PFI-CS-treated tendon after 4 weeks. Meanwhile, the collagen content evaluation confirmed that more collagen was deposited in the PFI-CS group than that in the Con-CS group (Fig. 6D). In addition, the expression levels of tendon-related proteins were assayed with immunohistochemistry. Enhanced expression of MKX and DCN was found in the PFI-CS group when compared with the Con-CS group at 4 weeks post-surgery (Fig. 6E).

Fig. 6
figure 6

Histological analysis and ECM deposition at the repair site after 2 and 4 weeks of implantation. (A, B) Representative H&E, Masson’s trichrome, and polarized light images of Con-CS and PFI-CS treated tendons at 2 weeks (A) and 4 weeks (B) post-surgery. Scale bars = 100 μm (H&E, Masson, and polar light). (C) Histological score of repaired tendons, n = 4. (D) Collagen content analysis of repaired tendons, n = 5. (E) The expression of special proteins in Con-CS and PFI-CS repaired tendons at 4 weeks post-injury. Scale bars = 50 μm. Data shown from representative sample and represented as mean ± SD. Two-tailed Student’s test, *p < 0.05, **p < 0.01. Abbreviation: CS, cell sheet; Con-CS, cell sheet without PFI-2 treatment; PFI-CS, cell sheet with PFI-2 treatment; H&E, hematoxylin and eosin; N, neo-tendon; T, host tendon.

Full size image

TEM was then used to observe the ultrastructure of repaired tendons in each group. As shown in Fig. 7A, both groups exhibited uniform and smaller collagen fibrils in comparison to native tendon. However, the PFI-CS group illustrated larger fibril diameter than in the Con-CS group, as shown by the distribution of fibrils diameters. Approximately 74.8% of the fibril diameters in the PFI-CS group ranged from 40 to 55 nm, while most collagen fibrils in the Con-CS group ranged from 35 to 45 nm at 4 weeks postoperatively. Additionally, the mechanical properties (stiffness, failure force, modulus, and stress at failure) were the most convincing evidence for tendon functional repair. The repaired tendons were then subjected to mechanical tests, all specimens failed at the midsubstance of the repaired tendon. The results showed that, the PFI-CS group had better biomechanical properties than Con-CS group (Fig. 7B). In detail, the stiffness, failure at force, modulus, and stress at failure were, respectively, 1.82-fold, 1.85-fold (p < 0.05), 1.78-fold, and 2.98-fold (p < 0.05) higher in the PFI-CS treated tendon compared to the Con-CS repaired tendon. In general, the above data demonstrated that the PFI-CS provides repaired tendons with superior structural and functional properties.

Fig. 7
figure 7

Ultrastructure and mechanical properties of repaired tendons at 4 weeks post-surgery. (A) TEM images (cross and longitude view) and histogram of collagen fiber diameters of Con-CS and PFI-CS tendons, n = 4. Scale bars = 50 nm. (B) The mechanical properties (stiffness, stress at failure, modulus, and failure force) of tendons after repair were evaluated, n = 6. Data shown from representative sample and represented as mean ± SD. Two-tailed Student’s test, * p < 0.05. Abbreviation: CS, cell sheet; Con-CS, cell sheet without PFI-2 treatment; PFI-CS, cell sheet with PFI-2 treatment.

Full size image

Discussion

Stem cell-based cell sheet engineering retains the benefits of an intact ECM, allows the development of scaffold-free constructs, and reduces the amount of cell loss that typically occurs during traditional cell seeding processes. However, the therapeutic outcomes remain variable due to the senescent of stem cells that occurs after prolonged culture expansion in vitro. In this study, we demonstrated that SETD7 is involved in the replicative senescence of hAD-MSCs. Pharmacological inhibition of SETD7 by PFI-2 boosts in vitro expansion of hAD-MSCs, as well as maintains their multi-lineage differentiation ability, while delays the progress toward senescence. Additionally, our study demonstrated that the in situ transplantation of hAD-MSCs cell sheets treated with PFI-2 in a rat patellar tendon injury model resulted in enhanced therapeutic effects for tendon repair. In summary, this study provides a novel target and approach to prolong the onset of senescence in hAD-MSCs, as well as enhance their therapeutic potential.

Cell sheet technology preserves cell-cell junctions and ECM integrity. Cell sheets derived from AD-MSCs have generated much research interests given the typical abundance of adipose tissue. However, it still face certain challenges, such as potential hampering of therapeutic efficacy due to intrinsic factors like senescence42,43. Epigenetic changes had been considered as a contributor to trigger and maintain functional differences of stem cells during ex vivo expansion. Recent investigations have attempted to map the epigenetic changes associated with stem cells senescence, however with this area still in its infancy. For example, Koch et al. have identified changes in DNA-methylation at six specific CpG sites as a marker to predict the senescent of cultured MSCs44. In another study, inhibition of DNMT1 and DNMT3B, induced cellular senescence, which accompanied by hypomethylation of the p16Ink4A locus and p21 promoter45. Noteworthy, gene-specific methylation always tissue-specific, which cause the method to be time/cost consuming and hard to intervene. Direct involvement of histones in the process of aging is illustrated in budding yeast. When yeast cells deficient in the histone chaperone (Asf1), they displayed lower histone levels correlating to a shorter lifespan46. This result implies that improper chromatin structure is a pivotal factor of causing aging. Indeed, recent studies have provided experimental evidence of the participate of histone methyltransferase in cellular senescence. For instance, SUV39H1, which is a H3K9 methyltransferase, has been found to involve in oncogene-induced senescence and pathological aging process47. SETD7, which is also a methyltransferase of H3K4, its potential role in stem cell senescence has not been characterized yet. Previous studies shown that the low DNA methylation sites associated with MSCs derived from aged donors are generally located in H3K4me1 enriched region, suggesting that SETD7 may be involved in the regulation of MSCs senescence23.

Here, GEO data analysis and cell in vitro senescence model suggested that SETD7 concomitantly upregulated along with ex vivo expanded. In line with this, a strong positive correlation of SETD7 mRNA levels with p15 expression was found in the course of in vitro culture of MSCs, and similar trend was observed for ARHGAP29. These data make it reasonable to explore the role of SETD7 in cellular senescence. Several attempts have been made to prevent MSCs from senescence in vitro. For example, a genetic engineering based method has been developed by introducing a retroviral vector which contains the catalytic subunit of telomerase gene to BMSCs48. By using this strategy, MSCs can maintain normal proliferative lifespan and differentiation capacity throughout the subculture process in vitro. Although effective, the risk of genetic changes is the major concern of this method. As epigenetic changes are reversible, epigenetic interventions are more transient, reversible, and potentially less harmful to reprogram senescent stem cells into youthful functional stem cells when compared to gene therapy. In our study, the SETD7 inhibitor PFI-2 inhibited the activity of SETD7 in hAD-MSCs, as well as the expression of senescence-related genes and proteins. As a result, the proliferation ability of hAD-MSCs was significantly enhanced evidently by CFU assay. This was consistent with a previous study, which Schellenberg et al. showed that in hAD-MSCs, 18.4% of passage 1 cells generated CFU colonies, and decreased significantly over subsequent passages49. More importantly, the PFI-2 treatment had a certain maintenance effect on the adipogenic and chondrogenic differentiation ability of hAD-MSCs when compared with the control group. While there was no obvious change in the osteogenic differentiation of hAD-MSCs with prolonged culture. These results were in line with other independent studies which reported that osteogenic differentiation potential of MSCs increased up to passage 10 and decreased thereafter. However, the adipogenic and chondrogenic differentiation potential decreased passage-dependently from the starting50,51. As a seed cell for tendon repair, our results also confirmed that ex vivo cultured hAD-MSCs with the treatment of PFI-2 could significantly improve the levels of tendon-related genes and proteins in a cell sheet model. Therefore, our in vitro results indicated that PFI-2 could delay the senescence process of serial passaged hAD-MSCs, thus maintain the stemness and differentiation capacity of hAD-MSCs. This also has a certain positive effect on solving the functional loss of stem cells expanded in vitro. These conclusions are basically consistent with a previous study, which inhibition of SETD7 promotes the in vitro expansion of MuSCs and enhance the transplantation potential of MuSCs28.

In an animal model of patellar tendon injury, hAD-MSCs derived cell sheets in presence or absence of PFI-2 treatment were rolling up into cylindrical and transplanted to the injured sites. To determine the effect of PIF-2 on delay cellular senescence, hAD-MSCs used here were at late passage. As a result, the cell sheets derived from PFI-2 pretreated hAD-MSCs could better induce the secretion of collagen and restore the arrangement of collagen fibers in vivo, thereby promoting the expression of tendon-related genes and improving the mechanical properties of the repaired tendon. However, compared with normal tendon tissue, there is still a large gap in terms of collagen fiber arrangement, fiber diameter, and mechanical parameters.

Some limitations also exist in our current study, inhibition of SETD7 maintained the biological functions of continuous passaged hAD-MSCs, but the underlying mechanism remains to be fully delineated. We will aim in follow-up studies to determine how SETD7 works during the process of cellular senescence. For instance, although SETD7 was initially found as a histone methyltransferase which mediate H3K4me1 in vitro. Some studies reported that SETD7 can also methylate non-histone proteins, such as p5352, pRB53, and Yes-associated protein (YAP)54, thus regulate various cellular functions. SETD7 could monomethylate YAP and promote its cytoplasmic retention28. We found that YAP was gradually translocated into the plasma during the serial passages of hAD-MSCs (data not shown), suggesting YAP methylation may be one of the mechanisms by which SETD7 regulates the hAD-MSCs replicative senescence. While further research is needed to confirm this hypothesis. Moreover, as tissue-specific differences exist, further study is also required to determine whether the anti-aging effects of PFI-2 is hAD-MSCs specific or it is also applicable to stem cells from other sources, such as bone marrow.

Conclusions

Overall, this study suggested SETD7 is an important contributor to prolonged cultured hAD-MSCs senescence. Pharmaceutical inhibition of SETD7 in hAD-MSCs maintained their stemness with a significant postponement in the onset of senescence, thus improving their therapeutic capacity for tendon repair. Therefore, this study provides a new target for chemical intervention to enhance the therapeutic potential of ex vivo expanded hAD-MSCs and accelerate tendon repair in vivo.