Adipose-derived Mesenchymal Stem Cell-derived Exosomes Promote Tendon Healing by Activating Both SMAD1/5/9 and SMAD2/3


 Background: The use of adipose-derived mesenchymal stem cell-derived exosomes (ADSC-Exos) may become a new therapeutic method in biomedicine owing to their important role in regenerative medicine. However, the role of ADSC-Exos in tendon repair has not yet been evaluated. Therefore, we aimed to clarify the healing effects of ADSC-Exos on tendon injury.Methods: The adipose-derived mesenchymal stem cells (ADSCs) and tendon stem cells (TSCs) were isolated from subcutaneous fat and tendon tissues of Sprague Dawley rats, respectively, and exosomes were isolated from ADSCs. The proliferation and migration of TSCs induced by ADSC-Exos were analyzed by EdU, cell scratch and transwell assays. We used western blot to analyze tenogenic differentiation of TSCs and the role of the SMAD signaling pathways. Then we explored a new treatment method for tendon injury, combining exosome therapy with local targeting using a biohydrogel. Immunofluorescence and immunohistochemistry were used to detect the expression of inflammatory and tenogenic differentiation after tendon injury, respectively. The quality of tendon healing was evaluated by Hematoxylin-eosin (H&E) staining and biomechanical testing.Results: ADSC-Exos could be absorbed by TSCs, and promoted the proliferation, migration, and tenogenic differentiation of these cells. This effect may have depended on activation of the SMAD2/3 and SMAD1/5/9 pathways. Furthermore, ADSC-Exos inhibited the early inflammatory reaction and promoted tendon healing in vivo.Conclusions: Overall, we demonstrated that ADSC-Exos contributed to tendon regeneration and provided proof of concept of a new approach for treating tendon injuries.


Introduction
Tendon is a dense connective tissue consisting of limited tendon cells and abundant extracellular matrix (ECM). Tendon injuries are of signi cant concern worldwide, with more than 30 million affected patients annually [1]. Tendon healing is slow as a result of its hypocellularity and hypovascularity, and involves three overlapping phases: in ammation, proliferation, and remodeling [2,3]. Furthermore, the self-healing potential of any tissue depends, in part, on its endogenous resident stem cells. The viability and tenogenic differentiation of TSCs are the main mechanisms of tendon repair [4]. However, in ammation during the healing phase may compromise biomechanical function [5][6][7]. Therefore, it is important to enhance tendon healing by promoting anti-in ammation and the proliferation of TSCs.
Mesenchymal stem cells have demonstrated great potential in tissue healing [8]. Speci cally, ADSCs are highly bene cial for clinical applications because of their abundant and conveniently accessible sources [9]. When transplanted, ADSCs are able to modulate the in ammatory environment and ECM balance to stimulate tendon regeneration [10][11][12]. Recent studies have demonstrated that the effectiveness of ADSCs in regenerative medicine is due to their paracrine effects [13]. Thus, ADSCs have been identi ed as new therapeutic agents in biomedicine [14].
Exosomes are membrane-bound extracellular vesicles that target cells by endocytosis or membrane fusion, and are important paracrine factors for stem cells [15]. In addition, exosomes play important roles in immune regulation, apoptosis, and tissue regeneration [16]. The therapeutic effect of ADSC-Exos has been demonstrated in multiple diseases. This is of great signi cance in the future development of tissue repair and regeneration engineering [17].
We hypothesize that ADSC-Exos promote tendon repair by regulating the biological characteristics of TSCs as well as the extracellular microenvironment. Speci cally, in this study, we investigated the effects of ADSC-Exos on the proliferation, migration, and differentiation of TSCs in vitro, and during in ammation and regeneration situations in vivo.

Animals
Male Sprague Dawley rats weighing 180-230 g at 8-10 weeks of age were provided by the Experiment Center of Harbin Medical University (Harbin, Heilongjiang, China). All animals were treated according to the United States National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the protocol was approved by the corresponding ethics committee (no. Ky2018-135).

Isolation and identi cation of TSCs and ADSCs
The isolation methods of ADSCs and TSCs were as performed in previous studies [18,19]. In brief, TSCs were isolated from rat tendon and cultured in Dulbecco's modi ed Eagle's medium (DMEM) (Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS) (Biological Industries, Kibbutz Beit-Haemek, Israel) and 1% penicillin-streptomycin (Beyotime, Haimen, China). The multilineage differentiation potential of TSCs, as well as the identi cation of surface markers (CD90-and CD105-positive, and CD106-and CD11b-negative), were demonstrated in our previous study [19]. ADSCs were isolated from subcutaneous fat of rats and cultured in DMEM/F12 (Invitrogen) containing 10% FBS and 1% penicillinstreptomycin. Flow cytometry was used to identify surface markers. The adipogenic, osteogenic and chondrogenic differentiation of ADSCs was induced in differentiation medium (Cyagen, Santa Clara, CA, USA) to identify their differentiation potential.

Isolation and identi cation of ADSC-Exos
At 80% con uence, the culture medium of the ADSCs was changed to exosome-depleted medium (DMEM/F12 containing 10% exosome-depleted FBS (Biological Industries) and 1% penicillinstreptomycin) and incubated for 24 h. Then, the culture medium was collected without ADSCs and centrifuged at 300 × g for 10 min, 3000 × g for 10 min, 10,000 × g for 30 min, and 100,000 × g for 2 h to isolate exosomes. Exosomes attached to the bottom of the centrifuge tube were diluted with phosphate-buffered saline. Nanoparticle tracking analysis (NTA), transmission electron microscopy (TEM), and western blotting were used to identify and evaluate the collected exosomes.

Cellular internalization of ADSC-Exos
ADSC-Exos were incubated with 1 µM PKH26 (Sigma-Aldrich, St. Louis, MO, USA) in Diluent C (Sigma-Aldrich) for 5 min and excess dye was removed by ultracentrifugation. The labeled exosomes were subsequently added to serum-free medium of TSC cultures and incubated overnight. The nuclei were labeled with Hoechst 33342 (UE, China) and photos were taken with an inverted uorescence microscope (Leica, Wetzlar, Germany).

Treatment of TSCs with ADSC-Exos
For determining the effect of ADSC-Exo treatment, 1 × 10 6 TSCs were seeded into six-well culture plates for 24 h and divided randomly into four groups. ADSC-Exos were added to exosome-free medium at 0, 25, 50, or 100 μg/mL and used to replace the TSC culture medium. For additional analyses, 10 nM of the TGF-β/SMAD2/3 inhibitor, SB431542, or 10 nM of the BMP/SMAD1/5/9 inhibitor, dorsomorphin, (MedChemExpress, Monmouth Junction, NJ, USA) were added to the TSCs 30 min prior to addition of the ADSC-Exos. TSCs treated with various concentrations of ADSC-Exos, with or without inhibitors, were incubated for 30 min or 24 h, then collected for analyses.

EdU assay
For the cell proliferation analysis, TSCs were incubated with 50 μM 5-ethynyl-2′-deoxyuridine (EdU) from an EdU Assay Kit (UE) for 4 h. The TSCs were then xed with 4% paraformaldehyde and stained using the same EdU assay kit. Nuclei were labeled with Hoechst 33342 and photos were taken with an inverted uorescence microscope.

Scratch assay
TSCs at 2 × 10 5 cells/well were inoculated into a 6-well plate for overnight culture. A straight line wound was made in the cultured cells using a sterile 200-μL pipette tip. Serum-free medium with ADSC-Exos was then added into each well. Images were obtained at 0 and 24 h after ADSC-Exo treatment using an inverted microscope with an Axiocam 506 camera and ZEN 2011 software (Zeiss, Oberkochen, Germany).
Transwell assay TSCs at 1 × 10 5 cell/well were inoculated into the Transwell upper chamber, and ADSC-Exos were added into the lower compartment. After culturing for 24 h, the TSCs were xed with absolute ethanol, then stained with crystal violet. Images were obtained under a light microscope.

Western blot analyses
Experimental protocols and surgical procedures A total of 63 Sprague-Dawley rats were divided into three groups of 21: (1) Control: animals that underwent surgery for partial resection of the patellar tendon; (2) Gelatin methacryloyl (GelMA): animals that underwent surgery for patellar tendon partial resection and were inoculated with 30 μL GelMA (EFL-GM-60, 10% w/v) over the tendon defect; and (3) ADSC-Exos: animals for which the injured patellar tendon was treated with 30 μL GelMA containing 200 μg of ADSC-Exos. The exosome content was determined according to previous studies [19]. Rats were anesthetized with 0.3% sodium pentobarbital (30 mg/kg). The right patellar tendon was surgically exposed and the central 1/3 of the tendon tissue removed as in previous studies [20]. GelMA was then inoculated into the lesion and cross-linked into a gel state by ultraviolet light. The skin incision was closed using 4-0 sutures. The modeling process is shown in Figure 1. Animals were sacri ced on days 7, 14, or 28 post-surgery and the tendons were collected.

Histopathological and immunohistochemical analyses
Para n-embedded tendon tissues were sectioned at a thickness of 4 μm. The tissues were then stained with H&E for histopathological analysis. The stained patellar tendons were evaluated using light microscopy.
For immunohistochemical analyses, para n sections of tendon tissues were incubated with Immuno-Block reagent for 30 min after being depara nized and rehydrated.

Biomechanical testing
Two bony ends of a healing tendon were xed on a universal material testing machine (Zwick, Roell, Germany). The tissues were investigated using a standard failure test with a testing speed of 5 mm/min. Failure load (N) and stiffness (N/mm) were obtained by the software of the testing machine. Young's modulus (N × 10 3 /mm 2 ) was calculated after measuring the cross-sectional area (mm 2 ) of the tendon with a vernier caliper.

Statistical analyses
All values are expressed as means ± standard deviation. Quantitative data for each group were analyzed by a one-way analysis of variance followed by the Tukey-Kramer test. P < 0.05 was considered statistically signi cant.

Characterization and internalization of ADSC-Exos
Transmission electron microscopy showed that ADSC-Exos were round or elliptical vesicular structures ( Figure 3A). The NTA revealed the mean diameter of ADSC-Exos to be 109.6 nm ( Figure 3B). Western blot analyses con rmed that the ADSC-Exo surface markers, CD9, TSG101, and HSP70, were positively expressed ( Figure 3C). Finally, ADSC-Exos were internalized by TSCs and showed red uorescence ( Figure  3D).

ADSC-Exos promoted the proliferation, migration, and tenogenic differentiation of TSCs
We rst measured the effect of ADSC-Exos on the proliferation and migration of TSCs. The EdU assay showed that ADSC-Exos promoted TSC proliferation ( Figure 4A, B). The Transwell assay con rmed that ADSC-Exos promoted TSC migration with increasing concentrations of exosomes ( Figure 4C, D). The scratch test showed results consistent with these ndings (Figure 4E, F). Then, we investigated whether ADSC-Exos affected the differentiation of TSCs. Western blot analyses showed ADSC-Exos signi cantly increased the protein expression of TNMD, collagen I, and SCX, but had no effect on ALP or Runx2 ( Figure  4G-L). These results suggest that ADSC-Exos promote the tenogenic differentiation ability of TSCs but have no effect on osteogenic differentiation.
ADSC-Exos regulated TSC proliferation, migration, and tenogenic differentiation by activating TGF-β/SMAD2/3 and BMP/SMAD1/5/9 signaling pathways To investigate the regulatory effect of ADSC-Exos, we evaluated their effects on the proliferation, migration, and tendon differentiation of TSCs by pretreating them with SB431542 or dorsomorphin. As expected, the proliferation ( Figure 5G, H) and migration ( Figure 5I-L) of TSCs were signi cantly decreased in the ADSC-Exos + SB431542 and ADSC-Exos + dorsomorphin groups compared with that in the ADSC-Exos only group. Similarly, western blot analyses showed that pretreatment with SB431542 or dorsomorphin signi cantly decreased expression of the tenogenic differentiation genes, TNMD, collagen I, and SCX, in TSCs ( Figure 5M-P).

ADSC-Exos regulated the early in ammatory response during tendon healing
We investigated the in vivo effect of ADSC-Exos on early healing of tendon injury. At week 1 after injury, the level of CCR7 (M1 macrophage marker) decreased in the ADSC-Exo group while the level of CD163 (M2 macrophage marker) increased ( Figure 6A, B). Furthermore, IL-10 (an anti-in ammatory factor) increased, and IL-6 (a pro-in ammatory factor) decreased ( Figure 6C, D). Quantitative analyses showed there were more CD163 + and IL-10 + cells in the ADSC-Exo group, while CCR7 + and IL-6 + cells predominated in the control and GelMA groups ( Figure 6E).

ADSC-Exos improved the healing of tendon injury
We next assessed whether ADSC-Exos contributed to the healing of patellar tendon injury in rats. H&E staining showed the ADSC-Exo group had signi cantly higher cell density and more longitudinal brous tissue in the defect area at week 2 compared with the other groups ( Figure 7A). At week 4, the collagen ber alignment in the ADSC-Exo group was more compact than other groups ( Figure 7H).
Immunohistochemical analyses showed higher expression of TNMD, collagen I, and SCX in the ADSC-Exo group at week 2 than in the control and GelMA groups ( Figure 7B-D). At week 4, the expression of these three genes remained high in the ADSC-Exo group ( Figure 7I-K). Furthermore, ALP and Runx2 expression was unchanged among the three groups at both week 2 and 4 ( Figure 7E, F, L, M). The results of the quantitative analyses are shown in Figure 7G, N.
Biomechanical testing showed that the failure load, stiffness, and Young's modulus of the patellar tendon in the ADSC-Exo group were signi cantly increased compared with the control and GelMA groups ( Figure   7O-R).

ADSC-Exos promoted TSC proliferation during tendon healing
To investigate the mechanism by which ADSC-Exos promoted tendon healing in vivo, we measured the number of TSCs in tendon tissue during early healing. CD146 was used as a marker of TSCs [21].
Immunohistochemical staining showed that the number of CD146 + TSCs in the injured tendon increased with extension of the healing time. Meanwhile, as expected, the number of CD146 + TSCs increased signi cantly in the ADSC-Exo group ( Figure 8A, B).

Discussion
Improving the quality of healing after tendon injury remains a major medical challenge. TSCs play an important role in tendon healing [22]. However, Zhang et al. reported that culture-expanded TSCs were prone to lose their phenotypic characteristics and exhibited reduced regeneration ability [23]. Therefore, activating the proliferation and differentiation of TSCs is key to improving tendon healing.
We rst studied the in uence of ADSC-Exos on TSCs in vitro. The results revealed that ADSC-Exos were internalized into TSCs and promoted their proliferation, migration, and tenogenic differentiation. Implantation of TSCs improves tendon healing in rats [24][25][26][27] and the activity of TSCs determines the quality of this healing. It is well-known that the SMAD family of signaling pathways play important roles in regulating stem cell functions, with two typical SMAD signaling pathways, TGF-β/SMAD2/3 and BMP/SMAD1/5/9, having potential signi cance in regulating the activity of TSCs [28][29][30]. Accordingly, we hypothesized that ADSC-Exos promoted the proliferation, migration, and tenogenic differentiation of TSCs by activating SMAD family signaling pathways. As expected, ADSC-Exos increased the phosphorylation of SMAD2/3 and SMAD1/5/9 in TSCs, which was later found to be attenuated by the inhibitors, SB431542 and dorsomorphin, respectively. We also found that application of these two inhibitors blocked the effects of ADSC-Exos on the activity of TSCs. These results support the hypothesis that ADSC-Exos enhanced the proliferation and migration of TSCs by promoting activation of the TGFβ/SMAD2/3 and BMP/SMAD1/5/9 signaling pathways.
Tenogenic differentiation is a complex process. SCX is a key molecule in the early development of tendons. It is responsible for the differentiation of TSCs into tenocytes and the positive regulation of TNMD expression [31,32]. Subsequently, the TNMD gene is necessary for tendon maturation and has a positive effect on the self-renewal of TSCs [33]. In addition, the expression of collagen I determines the strength of tendons [34]. Because abnormal ossi cation during tendon healing affects normal tendon functions, we hypothesized that ADSC-Exos would be able to promote tenogenic differentiation and inhibit osteogenic differentiation of TSCs. The results showed that, indeed, ADSC-Exos increased TNMD, collagen , and SCX expression in TSCs via activation of the TGF-β/SMAD2/3 and BMP/SMAD1/5/9 pathways. However, ADSC-Exos did not affect the expression of ALP or Runx2 in TSCs. This suggests that ADSC-Exos could effectively promote tenogenic differentiation of TSCs, but not inhibit osteogenic differentiation.
Scar formation caused by in ammation after tendon injury is a major cause of histological changes affecting tendon healing prognosis [35]. Therefore, inhibiting the early in ammatory response of tendon injury is bene cial to early healing. Macrophage in ltration and the release of pro-in ammatory cytokines are characteristics of early in ammation after tendon injury [36,37]. Shen et al. found that ADSC-Exos reduced the early in ammatory response after tendon injury by regulating macrophages, while Heo et al. con rmed that ADSC-Exos were able to modulate macrophages from an M1 to M2 phenotype [10,38]. In the current study, we found that CD163 + M2 macrophages increased signi cantly in the ADSC-Exo group.
In addition, the M2-stimulating factor, IL-10, was increased in the ADSC-Exo group. Therefore we suggest that ADSC-Exos were able to alleviate early in ammation after tendon injury by modulating macrophages.
Tissue integrity is the standard for evaluating the quality of tendon healing. We used the central 1/3 patellar tendon injury rat model to evaluate tendon healing. H&E staining showed that collagen bers in the ADSC-Exo group were more regular compared to those in the control and GelMA groups. In addition, the biomechanical properties of tendon tissues in the ADSC-Exo group were signi cantly improved at 4 weeks. We also investigated the regulatory effect of ADSC-Exos on TSCs in vivo. Immunohistochemical analyses showed that ADSC-Exos promoted the expression of tenogenic differentiation genes in vivo but did not inhibit the expression of osteogenic differentiation genes in the injured area.
In previous reports, CD146 has been used as a surface marker of TSCs; CD146 + TSCs switch to an activated state during tendon-injury healing and increase their proliferation, migration, and tenogenic differentiation ability [39]. Our results showed that the expression of CD146 + TSCs in the ADSC-Exo group was the highest among three groups. This indicated that ADSC-Exos promoted the proliferation ability of CD146 + TSCs.
Exosomes are generally used to repair tissues by intravenous or local injection. However, due to di culty in their local retention, exosomes are unable to exert their full biological e cacy. GelMA is a photosensitive biohydrogel with excellent biocompatibility and degradability, and is widely used in various tissue engineering applications [40,41]. GelMA exists in a liquid state at 37°C and becomes cross-linked under ultraviolet light to form a gel state with ECM properties. Because of its mild response to environmental conditions, GelMA has great advantages for use in biomedicine and is expected to be applicable for various clinical treatments [42]. For instance, Aubin et al. attempted to change the proliferative arrangement of different cells using micropatterned GelMA to provide a theoretical basis for constructing functional tissues in vitro [43]. Zou et al. used GelMA to construct biomimetic bone with a trabecular bone structure and Hu et al. used GelMA microspheres loaded with small extracellular vesicles to promote cartilage regeneration [44,45]. In the current study, GelMA was used as a carrier of ADSC-Exos to provide a good microenvironment for exosome storage and their gradual absorption. Therefore, ADSC-Exos loaded into GelMA is a promising treatment for tendon injury.
The current study does have some limitations. First, we selected only one time point to analyze phosphorylation in TSCs. Phosphorylation is a continuous process and the 30-min time point selected may not be optimal to detect TSC phosphorylation. Second, we only evaluated short-term tendon healing. The long-term therapeutic effect of ADSC-Exos on tendon healing (scar formation) requires further study.
In addition, which speci c substance in exosomes activates the SMAD pathways still needs further exploration.