Exosomes Derived from Bone Marrow Mesenchymal Stem Cell Preconditioned by Low-Intensity Pulsed Ultrasound Stimulation Promote Bone-Tendon Interface Fibrocartilage Regeneration and Ameliorate Rotator Cuff Fatty Inltration

Fibrovascular scar healing of bone-tendon interface (BTI) instead of functional brocartilage regeneration is the main concern associated with unsatisfactory prognosis in rotator cuff repair. Mesenchymal stem cells exosomes have been reported to be a new promising cell-free approach for rotator cuff healing. Whereas, controvercies abound in whether exosomes of native MSCs alone can effectively induce chondrogenesis. In this study, we aimed to explore the effect of Exosomes derived from low-intensity pulsed ultrasound stimulation (LIPUS)-preconditioned bone marrow mesenchymal stem cells (LIPUS-BMSC-Exos) or un-preconditioned BMSCs (BMSC-Exos) on rotator cuff healing and the underlying mechanism. Specically, C57BL/6 mice underwent unilateral supraspinatus tendon detachment and repair were randomly assigned to saline, BMSCs-Exos or LIPUS-BMSC-Exos injection therapy. The results indicated that the biomechanical properties of the supraspinatus tendon-humeral junction were signicantly improved in the LIPUS-BMSC-Exos group than that of the BMSCs-Exos group. The LIPUS-BMSC-Exos group also exhibited a higher histological score and more newly regenerated brocartilage at the repair site at postoperative 2 and 4 weeks and less fatty inltration at 4 weeks than the BMSCs-Exos group. In vitro, co-culture of BMSCs with LIPUS-BMSC-Exos could signicantly promote BMSCs chondrogenic differentiation and inhibit adipogenic differentiation than the BMSCs and BMSC-Exos co-cultured group did. Subsequently, quantitative real-time polymerase chain reaction revealed signicantly higher enrichment of chondrogenic miRNAs and less enrichment of adipogenic miRNAs in LIPUS-BMSC-Exos compared with BMSC-Exos. Moreover, we demonstrated that this chondrogenesis-inducing potential was primarily attributed to miR-140, one of the most abundant miRNAs in LIPUS-BMSC-Exos. Collectively, our results highlight the regenerative potential of LIPUS-BMSC-Exos to promote BTI brocartilage regeneration and ameliorate supraspinatus fatty inltration by positive regulation of pro-chondrogenetic and anti-adipogenetic of BMSCs differentiation which was primarily through delivering miR-140.


Background
Tendon integrates with bone through a specialized brocartilaginous tissue. 5 The bone-tendon insertion (BTI) plays a critical role in mediating load transmission and minimizing concentrated stress between tendon and bone. 6, 38, 49 Surgical reconstruction are standard treatments for rotator cuff injuries, with over 270,000 shoulder rotator cuff repairs performed annually in the United States alone. 27 However, surgical repair is simply a physical re-attachment of tendon to bone and valid evidence demonstrate that BTI healing occurs through formation of disorganized brovascular scar tissue with high levels of type III collagen rather than regeneration of a ordered transitional brocartilaginous tissue which consists of high levels of collagen type II, proteoglycans, 3,19,63,64 which means the brocartilage layer is regenerated slowly and incompletely at the BTI. Therefore, achievement of BTI healing, speci cally brocartilage regeneration at the BTI is the key for a successful clinical outcome. 20,30,53 Meanwhile, atrophy and fatty in ltration of the repaired supraspinatus is another adverse factor that impair the function of the repaired rotator cuff. 7,15 Hence, it is also important to alleviate or halt fatty in ltration after the repair of rotator cuff tears. 1,34 Exosomes (Exos) released from Mesenchymal stem cells (MSCs) can exert biological activities similar to those of the MSCs by transferring information to damaged tissue. 32 MSCs can regulate the differentiation of other neighboring stem cells, such as osteogenesis, adipogenesis and angiogenesis. 26,28,55 However, there's a lot of controversy as to whether the Exos of native MSCs can effectively induce chondrogenesis. 28,41,51,77 Actually, few literatures demonstrate that the Exos of native MSCs alone can effectively induce chondrogenesis. In recent years, increasing evidence indicates that the biological functions of the generated Exos were determined by the state of the Exos donor cells. 21 Most recently, Exos derived from Kartogenin preconditioned MSCs were proven to be able to promote chondrogenesis of MSCs. 28 Besides, it was found that sEVs derived from miR-92a-3p-overexpressing MSCs could facilitate cartilage proliferation and suppress cartilage degradation. 51 All these aspects show promising prospect that Exos released from specially treated MSCs can trigger the stable lineage-speci c chondrogenic differentiation of native MSCs. 28 LIPUS is an established, widely applied and Food and the Drug Administration (FDA) approved intervention for enhancing bone healing in fractures and non-unions. 2 LIPUS is a classical nondestructive biophysical therapy in which mechanical energy is transmitted transcutaneously as highfrequency acoustical pressure waves into biological tissues. 60 Over the last two decades, LIPUS therapy has been proved to be an effective strategy to enhance repair of articular cartilage and stimulate tendonbone junction healing. 8, 43-48, 50, 60 Meanwhile, LIPUS can induce chondrogenesis of MSCs in vitro by enhancing the synthesis of matrix proteins such as collagen type II and proteoglycans and expression of chondrogenic markers such as Sox9 and TIMP-2. 37 However, whether LIPUS achive its chondrogenic function through paracrine factors, especially Exos, remain unclear. All these data provide us with valuable sight of LIPUS preconditioning of MSCs-derived Exos as a promising new cell-free biotherapy agent to trigger the chondrogenesis of MSCs in cartilage repair. To the best of our knowledge, there was no previous study report MSCs Exos derived from LIPUS intervention.
In present research, we investigated the feasibility of applying Exos derived from LIPUS preconditioned BMSCs (LIPUS-BMSC-Exos) as a novel biomimetic tool to promote brocartilage regeneration and alleviate fatty in ltration in a murine rotator cuff repair model. We also assessed the effects of LIPUS-BMSC-Exos on chondrogenesis and adipogenesis, in vitro and preliminarily clari ed the underlying mechanism.

Study Design
In vivo study All animal protocols were approved by the Animal Committee of Xiangya Hospital of Central South University (No. 2019030490). A total of 120 mature (12-week-old) male C57BL/6 mice underwent unilateral supraspinatus tendon (ST) detachment and repair procedure following Rodeo animal model of rotator cuff tear. 36 The mice were randomly assigned to 1 of 3 groups according to different treatment.

Isolation and Identi cation of BMSCs
Six 4-week-old mice were euthanized and their femurs and tibia were obtained immediately. Bone marrow were ushed out from marrow cavity and were blowed evenly. Suspension were incubated with standard media comprising α-MEM (Gibco, Grand Island, USA) supplemented with 10% fetal bovine serum (FBS; Gibco), 100 U/mL penicillin and 100 µg/mL streptomycin (Gibco) at 37°C in a humidi ed atmosphere of 5% CO2 and 95% air. The medium was replaced after colonies of broblast-like cells appeared. After 3-4 days of incubation (at 80% con uence), cells were re-plated in 10-cm Petri dishes. Early-passage BMSCs (passages 2-4) were used for the follow-up experiments.
An inverted microscope (Leica DMI6000B, Solms, Germany) was used to observe the morphology of BMSCs. Immunophenotype of BMSCs was identi ed by ow cytometry with the characteristic surface markers of multipotential stem cells (CD29, CD34, CD44, CD45, and CD90). All antibodies were purchased from BD Biosciences (San Jose, USA). The adipogenic, osteogenic and chondrogenic differentiation potentials of BMSCs were determined by using the respective differentiation medium (Cyagen Biosciences Inc, Guangzhou, China). Adipogenesis of BMSCs was evaluated using Oil Red O (ORO) staining to observe lipid droplets at day 21 of induction, osteogenesis of BMSCs was examined using Alizarin Red S (ARS) staining to test calcium depositions at day 14 of induction, and chondrocyte differentiation of BMSCs was assessed by Toluidine Blue staining of extracellular matrix at day 21 of induction.

LIPUS Intervention of BMSCs
An ST-SONIC LIPUS exposure apparatus from ITO Corporation Ltd. was used (Tokyo, Japan) to apply LIPUS intervention on BMSCs. BMSCs at passge 2 were cultured in six-well plates. When cells grew to about 80% fusion, the culture medium was replaced by an Exos-depleted FBS-containing medium (EXO-FBS-250A-1; System Biosciences, Mountain View, USA). After BMSCs reached con uence, the dish was put on the transducer of the LIPUS apparatus which located at the cell incubator. The cells were exposed to LIPUS from the bottom of the culture dish with a coupling gel placed between the LIPUS transducer and culture dish. The distance between the transducer and the cells was approximately 5-6mm. The ultrasound exposure was conducted for 20 minutes daily at a frequency of 1.5 MHz in a pulsed-wave mode (200 µs pulse burst width with repetitive frequency of 1 kHz at the intensity of 30 mW/cm 2 ) 24, 67 .

Exos Extraction and Labeling
After received LIPUS treatment or natural culture (also cultured in Exos-depleted FBS-containing medium) for 3 days, the conditioned medium were collected and sequentially centrifuged at 300 ×g for 10min, Page 6/33 2000 ×g for 30 min, 10,000 ×g for 30 min and then ltered through a 0.22 mm lter (BD Biosciences) to eliminate dead cells and cellular debris. The supernatant was then subjected to ultracentrifugation at 100,000 ×g for 4h at 4°C. After washing with PBS (100,000 ×g for 20 min), the Exos-containing pellet was resuspended in a right amount of PBS. The protein concentration of the Exos suspension was quanti ed by BCA protein assay kit (Thermo Fisher Scienti c, Waltham, USA). Exos were stored at − 80°C for later use. The morphology of Exos was observed by transmission electron microscope (Hitachi, Tokyo, Japan). Exosomes particle size distribution was determined by Nanoparticle tracking analysis (NTA) using ZetaView PMX 120 (Particle Metrix, Inning am Ammersee, Germany). The expression of speci c surface markers (CD9, CD63, Calnexin and TSG101) was identi ed by Western Blot assay.
To track the location and internalization of the Exos, they were pre-stained with DiR stains (40757ES25, Yeasen, Shanghai, China) for in vivo study and with PKH26 green uorescent cell linker kit (Sigma-Aldrich, St. Louis, USA) for in vitro study according to the manufacturer's instructions. After labeling, the Exos were washed in PBS and collected by ultracentrifugation (100,000 ×g for 30 min) at 4 ℃. Then, the labeled Exos were resuspended in serum-free α-MEM as previously described. 16 Animal model and Exos Treatment The mice underwent unilateral ST detachment and repair in the left shoulder according to previous protocols. 4,39,78 In brief, after anesthetized with intraperitoneal injection of 0.3% pentobarbital sodium (0.6 mL/20 g; Sigma-Aldrich), a longitudinal incision was made on the lateral side of the left shoulder. The deltoid muscle was then minimally detached at the surgical neck of humerus. External rotation of left shoulder was made and the acromion was elevated to adequate expose the supraspinatus tendonhumeral. A 6 − 0 polypropylene suture needle (Double armed prolene; Ethicon, San Lorenzo, USA) was sutured through ST in an "8" gure fashion. Then, the ST was sharply transected at its insertion site on the greater tuberosity, and the remaining tendon and brocartilage layer of the footprint were gently abraded with a No.15 blade to expose the bellowing spongy bone. A 30-gauge needle was used to carefully drill a transosseous bone tunnel traversing the humeral head under the footprint. Finally, the ST was tied to its original insertion site with the two suture limbs shuttled through the tunnel, one from the interior to the lateral, and the other from the lateral to the interior. 39 After repair, BMSC-Exos and LIPUS-BMSC-Exos (10 11 Exos suspended in 20 µL of PBS 71 ) were injected at ST and supraspinatus muscle (SM). The mice in the control group were injected with an equal volume of saline at the time of repair. After injection, the wounds were closed in layers. All the mice were allowed free cage activity after surgery.

Exos Tracking in vivo
To track the DiR labeled Exos in vivo, a non-invasive tracking system (IVIS Spectrum, PerkinElmer, USA) was used to image the DiR distribution and intensity on the mice shoulder at 3 days and 7 days postoperation. Furthermore, the SMSTH complexes were harvested and imaged again from lateral view using this system, and then sectioned to further determine the number of labeled cells present in the healing tissue with DAPI (0.5 µg/mL; Invitrogen, Carlsbad, USA) staining the cell nuclei.

Histological and Immuno uorescent Analyses
The SMSTH complex specimens were harvested from left shoulder joint at postoperative 2 and 4 weeks and divided into two parts: the SM and the supraspinatus tendon-humeral (STH) complex. The STH samples were xed in 4% paraformaldehyde for 24 h and then decalci ed in ethylene diamine tetraacetic acid(EDTA) decalcifying uid (Servicebio, Wuhan, China) for 7 days. The SM specimens were routinely xed in GD solution for 24h.
For histological analysis, after the STH specimens were xed, decalci ed and embedded in para ne, 5 mm-thick sections from the mid-coronal plane of each specimen, which through the supraspinatus tendon and the greater tuberosity, were sliced by a microtome (Leica RM2125; Reichert-Jung GmbH). The SM specimens were embedded in para ne, then the transverse sections of myo bers of SM were obtained. The STH sections were stained with hematoxylin and eosin (H&E, Solarbio Co Ltd, Beijing, China) and toluidine blue O/fast green (Sigma-Aldrich). The SM sections were stained with ORO (Solarbio Co Ltd). H&E Sections were used for histological description of the repaired BTI junction, and semiquantitative analysis for cell density around the the BTI. 72 Toluidine blue O/fast green-stained slices were used to calculate the area, thickness 73 , and proteoglycan content 10, 45 of the regenerated brocartilage layer using ImageJ software (Version 1.32, National Institutes of Health; Bethesda, MD, USA). All sections were observed using a transmitted light microscope (Olympus CX31; Olympus Inc, Hamburg, Germany) under the same conditions. To assay the area percentage of fatty in ltration, images of the entire cross-section of ORO-stained SM were scanned with a digital slide scanner (Pannoramic MIDI; 3DHISTECH Ltd, Budapest, Hungary) and then measured by ImageJ software (National Institutes of Health) for quanti cation. All of the data analyses were performed by the same person on the investigational team (B.W.).
For immuno uorescence staining for chondrogenic-related gene of collagen II and Sox9, the STH specimens were embedded in optimal cutting temperature compound (OCT, Sakura netec USA inc, Torrance, USA) and then 5 mm-thick frozen sections were acquired. After washed out for OCT and blocked with 3% BSA at room temperature for 30 min, The frozen sections were incubated with the primary antibody collagen II (1:100; A nity Biosciences, Changzhou, China) and Sox9 (1:200; Abcam, Cambridge, Britain) overnight at 4°C. After washing three times with PBS, sections were incubated with the secondary antibody (1:250; Abcam) at room temperature for 2h in a dark environment. Nuclei were stained with DAPI (0.5µg/mL; Invitrogen). The signals were examined with a Zeiss AxioImager. M2 uorescence microscope (Zeiss, Solms, Germany) equipped with an Apotome.2 System. Mean uorescence intensity (MFI) determination were made using ImageJ software and the MFI values were normalized to 100 cells. 68 Biomechanical Testing For biomechanical testing, each STH complex was carefully dissected from surrounding tissues under a surgical microscope and stored at -80 ℃ immediately. The specimens were thawed at room temperature before testing and were keep moist in 0.9% saline solution at 37 ℃ during the test. Biomechanical testing was conducted with investigators blinded to groups. Before testing, all sutures in STH complex were con rmed to be carefully removed. The humerus was mounted and the supraspinatus tendon was xed in custom grip with sandpaper and ethylcyanoacrylate. The STH complex was placed into a microcomputer controlled electronic material testing system (WD-T, Zhuoji Instrument Equipment Co., Ltd. Shanghai, China) which allow a uniaxial tensile at an abduction angle of approximately 60 degrees. The specimens were preconditioned under a preload between 0.0 and 0.5 N for 3 times and then loaded to failure at a rate of 1 mm/min. The failure load was recorded, and the stiffness was calculated from the curve of load to deformation. 9 qRT-PCR Analysis For miRNA detection, total Exos-derived miRNAs were isolated using the miRNeasy Micro Kit (Qiagen, Hilden, Germany). Then the Mir-X™ miRNA First-Strand Synthesis Kit (Takara Biotechnology, Shiga, Japan) were used to synthesize cDNA for miRNAs according to the manufacturer's protocol. Ampli cation reactions were performed using a miRNA SYBR Green qRT-PCR Kit (Takara) with the provided universal reverse primer and miRNA reference gene U6. The miRNA-speci c forward primers were synthesized by RiboBio (Guangzhou, China).

Exos Uptake Assay in vitro
To observe Exos internalization by BMSCs, native BMSCs were incubated with serum-free α-MEM containing LIPUS-BMSC-Exos or BMSC-Exos at a nal concentration of 20 µg/mL for 6h and 12h respectively. 69 Then, cells were stained with CytoPainter Phalloidin-iFluor 488 reagent (ab176753, abcam, Cambridge, MA) for 30 min and washed with PBS. Nuclei were stained with DAPI (0.5 µg/mL; Invitrogen, Carlsbad, USA). The cellular uptake of Exos was examined by Zeiss AxioImager.M2 microscope (Zeiss). To determine the uptake of Exos-derived miRNAs, BMSCs cells were treated with Exos for 12h and then collected to analyze the expression of miRNAs by qRT-PCR.

Chondrogenic Differentiation Assay
The 3D pellet culture was performed to induce chondrogenic differentiation. In brief, native BMSCs were incubated with 1 mL of growth medium containing 20 µg/mL blank liposomes, BMSC-Exos, LIPUS-BMSC-Exos or liposomes. 16, 28, 69 BMSCs transfected with miR-140 mimic or mimic NC were cultured in 1 mL of growth medium, while BMSCs transfected with miR-140 inhibitor or inhibitor NC were cultured in 1 mL of growth medium containing 20 µg/mL of LIPUS-BMSC-Exos components. 28 After 48h culture, a suspension containing 5×10 5 cells were collected into a 15ml centrifuge tube, centrifuged to form pellets, and cultured in growth medium. The medium was replaced every 3 days for all groups (n = 6). 3 weeks later, The pellets were xed, embedded in para ne, sectioned, and stained with H&E, Toluidine blue, and Safranine O. Chondrogenesis of different groups in vitro were quantitatively compared using the Bern score system. 22 Brie y, the higher total score means the better degree of neocartilage formation. The levels of Chondrogenesis-speci c marker of collagen type II (1:100, 4°C overnight) were assessed by immunohistochemistry in different groups. Brie y, the pellets were xed, sectioned, stained, counterstained, dehydrated, hyalinized, and mounted. What's more, the expression of chondrogenicrelated gene of Sox9, Aggrecan, collagen II was analyzed by qRT-PCR.

Statistical analysis
Data are shown as means ± standard deviation (SD). Student's t test was used to analyze differences between two groups. For multiple group comparisons, one-way analysis of variance (ANOVA) was utilized, and Tukey's multiple comparisons test was used to analyze signi cant differences between groups. Statistical analysis was conducted using GraphPad Prism software (Version 8, San Diego, USA) and P < 0.05 was considered statistically signi cant.

Characterization of BMSCs and BMSCs Derived Exos
BMSCs isolated from mice bone marrow exhibited a typical spindle broblast-like morphology under light microscopy ( Fig. 1A) and were able to differentiate into adipocytes, osteoblasts or chondrocytes after adipogenic, osteogenic or chondrogenic medium induction (Fig. 1B). Flow cytometric analysis showed that BMSCs were positive for CD29, CD44, and CD90, and negative for CD34 and CD45 (Fig. 1C). The Exos (BMSC-Exos and LIPUS-BMSC-Exos) we obtained presented a typical cup-like appearance with double membrane structures ranging from 50 to 200 nm under transmission electron microscope (Fig.  1D), and the NTA indicated that the size distributions of Exos derived from un-preconditioned BMSCs (BMSC-Exos) and LIPUS-BMSC-Exos were homogeneous (Fig. 1E). Furthermore, Western Blot analysis demonstrated that these nanovesicles expressed exosomal marker proteins including CD9, CD63, and TSG101 and were negative for Calnexin (Fig. 1F). The data indicate that these nanoparticles are Exos.

Exosomes Tracking in vivo
After injection of DiR labeled BMSC-Exos or LIPUS-BMSC-Exos at SMSTH complex for 3 and 7 days, the non-invasive tracking system (IVIS Spectrum) was used to image the DiR distribution and intensity. IVIS images showed that positive DiR signal was detected on the mice shoulder in the BMSC-Exos group and LIPUS-BMSC-Exos group, which were proved to exist at SMSTH complex by the lateral view of the SMSTH tissues ( Fig. 2A). Additionally, uorescence section also demonstrated the presence of DiR spots in SMSTH complex ( Fig. 2A). Semi-quanti cation analysis found no signi cant difference of mean uorescence positive cell ratio and uorescence intensity between the 2 groups (P = 0.737, 0.947, respectively) (Fig. 2B, 2C). The data suggested that the labeled Exos were delivered to the targeted area.

LIPUS-BMSC-Exos Enhance BTI Healing by Promoting Regeneration of Fibrocartilage Zone
For H&E staining, supraspinatus tendon (ST) was healed to the humeral head and the enthesis was gradually regenerated with time. At postoperative 2 weeks, brovascular tissue was found at the BTI which was poorly organized with abundant in ammatory cells and some cartilage-like cells (Fig. 3A). The mean cell density in the BTI was lower in the LIPUS-BMSC-Exos groups (4744 ± 401 cells/mm 2 ) compared with BMSC-Exos group (6058 ± 526 cells/mm 2 ) and control group (6117 ± 555 cells/mm 2 ) (P < 0.05, Fig. 3B). At postoperative 4 weeks, BTI regeneration and remolding were observed in the attachment, which was characterized with prevalent brocartilage cells characterized with inconspicuous nucleus and larger cartilage lacuna than those chondroid cells appeared at week 2 (Fig. 3A). Mean cell density in the BTI was lower in both LIPUS-BMSC-Exos group (3884 ± 386 cells/mm 2 ) and BMSC-Exos group (4429 ± 425 cells/mm 2 ) compared with the control group (5853 ± 604 cells/mm 2 ) (P < 0.01). In addition, the mean cell density in the BMSC-Exos group was still signi cantly higher than that in the LIPUS-BMSC-Exos group (P < 0.01, Fig. 3C).
For Toluidine blue O/fast green staining, brovascular tissue was found at the BTI interface, without obvious gaps in all groups at postoperative 2 weeks. At postoperative 4 weeks, varying degrees of hypertrophic brocartilage-like cells embedded by a characteristic matrix rich in proteoglycans were observed in the 3 groups (Fig. 3D). Obvious more brocartilage formation was noted in the LIPUS-BMSC-Exos group at both postoperative 2 and 4 weeks. Quantitatively, the area, thickness, and proteoglycan content of brocartilage layer in the LIPUS-BMSC-Exos group were signi cantly higher than those of the other 2 groups at both postoperative 2 and 4 weeks (P < 0.05 for all, Fig. 3E). Additionally, the BMSC-Exos group showed more regenerated brocartilage than the control group at postoperative 4 weeks (P < 0.05), while there was no signi cant difference between the 2 groups at postoperative 2 weeks (P > 0.05 for all, Fig. 3E).

LIPUS-BMSC-Exos Decrease Fatty In ltration of the Repaired Rotator Cuff
Fatty in ltration of supraspinatus increased progressively with time and obvious fatty in ltration could be in the saline group see at postoperative 4 weeks (Fig. 5A). No statistical difference was noted compared with the values of area percentage of fatty in ltration between groups at postoperative 2 weeks (P = 0.200, Fig. 5B). At week 4, the LIPUS-BMSC-Exos group showed signi cantly lower fatty in ltration (9.33% ± 1.10%) compared with the BMSC-Exos group (13.59% ± 1.21%) and the saline group (19.22%± 2.02%) (both P < 0.001, Fig. 5C), which indicate that local injection of LIPUS-BMSC-Exos into the SM at the time repair showed marked inhibition of fatty in ltration. Notablely, LIPUS-BMSC-Exos therapy also decreased supraspinatus fatty in ltration at week 4 (P < 0.001, Fig. 5C).

LIPUS-BMSC-Exos Enhance Biomechanical Properties of the Repaired Rotator Cuff
In biomechanical testing, all STH complex failed at the tendon-bone attachment. The mean ultimate load to failure of the LIPUS-BMSC-Exos group at postoperative 2 weeks (2.66 ± 0.13 N) and at postoperative 4 weeks (4.68 ± 0.17 N) was signi cantly greater than that in the control group (2.17 ± 0.12 N at 2 weeks, 3.52 ± 0.20 N at 4 weeks, respectively) and in the BMSC-Exos group (2.28 ± 0.11 N at 2 weeks, 4.13 ± 0.14 N at 4 weeks, respectively) (P < 0.001, Fig. 5D). Meanwhile, stiffness was also signi cantly greater in the LIPUS-BMSC-Exos group than in the control group and BMSC-Exos group (P < 0.05, Fig. 5E). Although obvious difference of the failure load and stiffness between BMSC-Exos group and the control group were not noted at postoperative 2 weeks (P > 0.05 for both), these index were signi cantly improved in the BMSC-Exos group at postoperative 4 weeks (P < 0.01, Fig. 5D, 5E).

LIPUS-BMSC-Exos Promote Chondrogenic Differentiation of MSCs
After incubation with different Exos components, recipient cells were subjected to pellet culture, and chondrogenesis was evaluated by histological examination, qRT-PCR 21 days later. Cartilage lacuna-like structures and extracellular matrix deposition were observed in the LIPUS-BMSC-Exos group (Fig. 6A), whereas no obvious chondrogenesis was observed in the Blank group and BMSC-Exos group. In addition, the Bern scores and expression of chondrogenesis-related genes (SOX9, aggrecan, and collagen II) in LIPUS-BMSC-Exos group were remarkably elevated compared to the BMSC-Exos group and the blank group (P < 0.01, Fig. 6B-10E). Notablely, BMSC-Exos group showed no signi cant difference of Bern scores and expression of chondrogenesis-related genes compared with the Blank group (P > 0.01 for all, Fig. 6B-10E).

LIUPUS-BMSC-Exos deliver miR-140 into MSCs
To detect whether LIUPUS-BMSC-Exos or BMSC-Exos could be taken up by native BMSCs invitro, Exos labeled with the PKH26 dye (red uorescence) were incubated with native BMSCs for 12 h and 24h. As shown in Fig. 12C, the PKH26-labeled Exos were found in the cytoplasm of native BMSCs. We next determined the transfer of miR-140 into recipient cells. After treatment with LIUPUS-BMSC-Exos or BMSC-Exos for 12h, BMSCs cells were harvested. qRT-PCR analysis demonstrated that miR-140 levels in native BMSCs were remarkably increased after incubation with LIUPUS-BMSC-Exos compared with those incubated with BMSC-Exos (Fig. 12D, P < 0.001). The results suggest that miR-140 can be shuttled into target cells by LIUPUS-BMSC-Exos.

miR-140 in LIPUS-BMSC-Exos Promotes Chondrogenesis and Inhibit Adipogenesis of MSCs
To explore the functional mechanism of the enriched miRNAs during chondrogenesis and adipogenesis of MSCs, the most abundant miRNA in LIPUS-BMSC-Exos, miR-140, was chosen as a potential target for further investigation. To fully unveil the underlying function of miR-140 in mediating chondrogenesis, a speci c miR-140 mimic/inhibitor was transfected into native BMSCs. When miR-140 was overexpressed by transfecting the mimic, the Bern scores and expression levels of chondrogenic genes (SOX9, aggrecan, and collagen II) were evidently upregulated (P < 0.001 for all), nearly replicating the chondrogenic effect of LIPUS-BMSC-Exos (Fig. 6B-E). Whereas, the intensity of ORO and expression of adipogenesis-related genes (Adipo, Retn and Pparg) were signi cantly down-regulated (P < 0.001 for all, Fig. 7B-E). On the other hand, when the elevated level of miR-140 conferred by LIPUS-BMSC-Exos was reversed by transfecting the inhibitor, the expression of chondrogenic genes decreased signi cantly and expression of adipogenesis-related genes increased signi cantly compared with that of cells incubated with LIPUS-BMSC-Exos only (P < 0.001, Fig. 6B-E, Fig. 7B-E ). However, miR-140 inhibitor group still exhibited signi cantly higher the Bern scores and expression levels of chondrogenic genes and signi cantly lower ORO intensity and expression of adipogenesis-related genes. These results together suggest that miR-140 was the major contributor to LIPUS-BMSC-Exos-mediated chondrogenesis of MSCs.

Discussion
In the present study, the feasibility of applying LIPUS-BMSC-Exos to promote brocartilage regeneration and alleviate fatty in ltration to induce the chondrogenesis of MSCs and was evaluated. We found that LIPUS-BMSC-Exos, the nanocarriers secreted by BMSCs which were preconditioned by LIPUS, could effectively promote BTI healing and ameliorate rotator cuff fatty in ltration through enhancing brocartilage regeneration, reducing supraspinatus muscle fat accumulation. We also showed that LIPUS-BMSC-Exos promoted chondrogenetic differentiation and inhibit adipoogenetic differentiation of native BMSCs in vitro. Moreover, we demonstrated that miR-140, a potently pro-chondrogenic miRNA, was extremely highly enriched in LIPUS-BMSC-Exos. To sum up, we showed that LIPUS-BMSC-Exos could promote the shift from adipogenic to chondrogenic differentiation of BMSCs via delivering miR-140, thus enhancing BTI brocartilage regeneration and ameliorate rotator cuff fatty in ltration.
Over the past years, the application of MSCs for cartilage tissue repair and regeneration has received considerable attention. 23,66,74 In recent years, an increasing number of studies have reported that the paracrine actions are the main mechanism by which MSCs exert their therapeutic effects. 35,61 Exos, recently identi ed as an important paracrine factors that mediate the intercellular communication, are critical effectors of MSCs. 33,52,57,61,70 It has been demonstrated that Exos can transfer proteins, RNA and other bioactive compounds into target cells and create an optimal microenvironment for maintaining cellular dynamic homeostasis, thus completing the transmission of biological information. 29,59 Due to speci c advantages of Exos therapy such as high stability, low immunogenicity, non-tumorigenicity and non-vascular thrombosis, 13,59 MSCs-Exos therapy has been receiving increasing attention for tissuse regeneration.
As for cartilage regeneration, the role of native MSCs-derived Exos for chondrogenesis is inde nable. Liu  77 demonstrated the e cacy of human embryonic MSC exosomes in promoting cartilage regeneration, and the utility of MSC exosomes as a ready-to-use and "cell-free" therapeutic alternative to cell-based MSC therapy in vivo. As for BTI healing, Wang et al 71 showed that adipose stem cell-derived Exos can decrease fatty in ltration and enhance BTI healing in vivo. In accordance with Wang et al 71 research, the present study also revealed improved BTI healing which can be judged by the decrease of brous scar tissue formation, augment of brocartilaginous enthesis regeneration and better biomechanical properties. Meanwhile, although signi cant suppression of adipogenesis of native BMSCs were determined by BMSC-Exos, we did not found obvious prochongdrogenesis effect of native BMSCs by BMSC-Exos in vitro, which was in consistent with Jing et al 28 results. Actually, the results of in vivio and in vitro are not contradictory in the present study. The enhanced BTI healing in the BMSC-Exos group may be ascribed to 2 aspects: First, the anti-in ammatory effects of MSC-Exos such as reduce the in ltration of in ammatory cells into the BTI interface, which can decrease the formation of brous scar tissue and enhance the regeneration of the normal tendon-bone insertion site. 31,39 Second, the angiogenesis effects of MSC-Exos can provide su cient vascular invasion around the BTI junction, which is essential for promoting BTI healing and possibly contributes to aggregation and chondrogenesis of local stem cells after rotator cuff repair. 26 Combined with all of these results, we can draw conclusions carefully that MSC-Exos may not possess obvious chondrogenetic effect alone,but they can still exhibit a degree of facilitating effect on brocartilage regeneration and BTI healing enhancement in vivo by promoting angiogenesis, decreasing in ltration of in ammatory cells and alleviating adipogenesis of MSCs.
Over the last two decades, LIPUS therapy has been proved to be an effective strategy to stimulate tendonbone junction healing. 8, 43-48, 50, 60 In these studies, an standard 30 mW/cm 2 intensity of LIPUS transcutaneous therapy in vivo turned out to enhance BTI healing through promoting angiogenesis, cartilage formation and maturation, endchondral bone formation and increaing the mechanical properties of the healing tissues of the tendon-bone junction. 50,75 LIUS is also reported to enhance repair of articular cartilage in animal model of cartilage defect. 11,58 In a rabbit full-thickness osteochondral defects model, Cook et al 11 demonstrate that daily low-intensity pulsed ultrasound had a signi cant positive effect on the healing of osteochondral defects. Meanwhile, LIPUS can induce chondrogenesis of MSCs in vitro by enhancing the synthesis of matrix proteins such as collagen type II and proteoglycans and expression of chondrogenic markers such as Sox-9 and TIMP-2. 37 Moreover, LIPUS preconditioning of BMSCs in vitro proved to be an effective cue to upregulate chondrogenic differentiation of MSCs in vivo. 12 As valid effect of LIPUS therpy for BTI healing has been com rmed and the fact that biological functions of the generated Exos were determined by the state of the Exos donor cells, direct use of Exos derived from LIPUS preconditioned MSCs may represent a preferable and promising biotherapy agent for BTI regeneration. Thus, the present study construct Exos derived from BMSCs induced by LIPUS stimulation for brocartilage regeneration of BTI healing. What' more, as fatty in ltration of rotator cuff is the other factor direct associated with unsatisfactory prognosis besides poor BTI healing, we also concern on the effect of LIPUS-BMSC-Exos on fatty in ltration and adipogenetic differentiation of BMSCs. Accordingly, we hypothesized that LIPUS preconditioning might impart Exos derived from MSCs with chondrogenic potential to generate a valid chondrogenic microenvironment. As anticipated, the present results demonstrate that LIPUS-preconditioned BMSCs release Exos with potent chondrogenesis-inducing functions compared to BMSC-Exos both in vivo and in vitro. Meanwhile, LIPUS-BMSC-Exos also exhibit inhibit effect on adipogenesis of BMSCs and fatty in ltration of supraspinatus muscle. These data threw light on the possible mechanism of shift from adipogenetic differentiation to chondrogenetic inconclusive. This is also the probable cause why miR-140 inhibitor could not totally reverse the chondrogenetic effect of LIPUS-BMSC-Exos.
Our ndings broaden the understanding of the regenerative potential of LIPUS-treated-Exos on chondrogenesis of MSCs, and suggest the application of LIPUS-BMSC-Exos as an off-the-shelf bio-tool for inducing the stable differentiation of MSCs, enhancing BTI healing and ameliorate fatty in ltration of the rotator cuff. In contrast to cumbersome postoperative LIPUS therapy or expensive inducing hormonal cocktails preconditioning which containing transforming growth factor β (TGF-β), insulin-like growth factor (IGF), dexamethasone, insulin, transferrin, selenium, and ascorbic acid, 40 an local injection of LIPUS preconditioned MSC-derived Exos during operation is absolutely economical, simple, and convenient. Moreover, contrast to classical MSC-derived Exos therapy, the LIPUS preconditioned MSCderived Exos demonstrate substantially superior pro-chondrogenesis e cacy on BTI brocartilage regeneration and also better anti-adipogenesis effect on prevent supraspinatus fatty in ltration.
This study has some limitations. First, although we continued to observe the generation of brocartilage tissue at the BTI junction which were proved to be most di cult part to be regenerated, the biological and biomechanical properties of BTI junction is also determined by the other 2 zones of BTI, namely the subchondral bone and the tendon. It is still unclear whether LIPUS-BMSC-Exos can make a signi cant difference on the endochondral calci cation, ossi cation, tendon collagen ber formation, peritendinous brosis and tendon adhesion processes. Therefore, regeneration of the 3 zones of the native enthesis (tendon, brocartilage, bone) likely requires more than just simple application of LIPUS-BMSC-Exos at the surgical repair site. Second, in many cases, LIPUS treatment entails 20 min of daily stimulation at intensity of 30 mW/cm 2 , with a frequency of 1.5 MHz in pulsed-wave mode (0.2-s burst sine waves repeated at 1.0 kHz). 8, 24,39,67 Most studies, including this one, have only studied a single LIPUS dose. Different dosages or use of different LIPUS stimulation parameters may reveal different effects of LIPUS on MSC-Exos components and function, thus leading to different outcomes of rotator cuff healing in this speci c model. Third, it is noteworthy that inhibition of miR-140 did not thoroughly block the bene cial effects of LIPUS-BMSC-Exos on chondrogenetic differentiation in vitro, indicating the implication of other miRNA or proteins that might in uence the regulatory effects of the LIPUS-BMSC-Exos through some unknown mechanisms. Finally, although there are many advantages of mice as a model for acute supraspinatus tendon injury repair, 36 mice are quadrupedal animals and joint motion and loading on the healing site are di cult to control owing to the inability to reliably immobilize the extremity or restrict weight bearing. 72 As mechanical loading is critical to the processes of brocartilage formation, mineral accumulation and collagen metabolism that are closely involved in a healing enthesis, it is not suitable to simply translate our ndings into rotator cuff tear repair in humans. Meanwhile, The pathophysiological process of the acute rotator cuff injury model is different from that of chronic rotator cuff injury, and the role of LIPUS-BMSC-Exos in chronic rotator cuff injury is unclear and needs further exploration.
In conclusion, our results demonstrate that Exos derived from LIPUS-preconditioned BMSCs are able to promote BTI brocartilage regeneration and ameliorate supraspinatus fatty in ltration in a murine rotator cuff repair model. The potential mechanism may be the positive regulation of pro-chondrogenetic and anti-adipogenetic of MSCs differentiation primarily through delivering miR-140. LIPUS-BMSC-Exos may become an innovative "Exos based strategy" for inducing chondrogenic differentiation of native MSCs, thereby proposing a new therapeutic method to BTI regeneration and rotator cuff healing in the near future.