Microneedle delivery of 3D bioprinted tendon stem/progenitor cells microfiber-derived extracellular vesicles enhances chronic massive rotator cuff tears healing Authors: Weiping He, Xiaopeng Yin, Shihai Jiang, Yi Zhang, Boxun Liu, Zhifeng Zhang, Woquan Ye, Shurong He, Tao Xu, Chang Liu, Kun Wang Published: 2026 ABSTRACT AbstractHigh retear rates after surgical repair of chronic massive rotator cuff tears (CMRCT) are primarily due to the failure to regenerate native tendon tissue. Extracellular vesicles (EVs) derived from rat tendon stem/progenitor cells (rTSPCs) present substantial potential benefits for the treatment of CMRCT. However, the limited yield and efficacy of rTSPCs-EVs produced through traditional methodologies (2D-cultured) severely restrict their widespread application. In this study, we proposed an innovative strategy integrating 3D bioprinted rTSPCs microfiber-derived EVs production (3D-rTSPCs-EVs) introducing a novel therapeutic approach for CMRCT. In vitro comparison demonstrating that 3D-rTSPCs-EVs are significantly superior to rTSPCs-EVs in promoting tenogenic differentiation and exerting anti-inflammatory effects. Meanwhile, to overcome the issues of rapid clearance and low bioavailability associated with the direct injection of EVs, the Hyaluronic Acid Methacrylate-based microneedle (HAMA-MN) scaffold was designed for the local delivery of 3D-rTSPCs-EVs (3D-rTSPCs-EVs@MN) at the tendon-to-bone junction. Subsequently, in vivo 3D-rTSPCs-EVs@MN facilitated the regeneration of a native, collagen matrix-rich tendon tissue as evidenced by improved histological, radiologic manifestations and biomechanical function. Furthermore, 3D-rTSPCs-EVs suppressed the inflammatory response by regulating macrophage polarization from the pro-inflammatory M1 phenotype to the pro-healing M2 phenotype. Therefore, our findings suggest that 3D bioprinted rTSPCs-microfiber derived EVs significantly ameliorates scar tissue formation to enhance biomechanical properties of injured rotator cuff, opening new avenues for efficient EVs-based therapies in CMRCT treatment.Keywords: Chronic massive rotator cuff tears, Extracellular vesicles, 3D bioprinting, Tendon to bone healing, Microneedles Introduction Chronic massive rotator cuff tears (CMRCT) account for a substantial proportion of rotator cuff injuries and are characterized by tendon retraction, chronic inflammation, and progressive fatty infiltration, which severely impair tendon-to-bone healing [1,2]. Persistent inflammatory responses at the repair site promote excessive M1 macrophage polarization and fibroblast accumulation, leading to disorganized collagen deposition and scar tissue formation rather than regeneration of the native tendon-bone interface. These pathological changes compromise the biomechanical properties of the repaired tendon and contribute to the high postoperative re-tear rate, which ranges from 11% to 94% after arthroscopic repair [3]. Therefore, there is an urgent need to develop regenerative therapeutic strategies that modulate collagen remodeling and inhibit scar tissue formation, thereby preventing postoperative re-tears in patients with CMRCT. Recently, biological therapies have shown potential for improving tendon-bone healing. Mesenchymal stem cells (MSCs), characterized by self-renewal and multilineage differentiation capacity, have shown considerable potential in regenerative medicine for enhancing tendon repair [4,5]. Among them, tendon stem/progenitor cells (TSPCs), a tissue-specific MSC subtype, possess tenogenic, chondrogenic, and osteogenic differentiation abilities and play important roles in extracellular matrix (ECM) remodeling and tendon-bone regeneration [[6], [7], [8], [9]]. Currently, numerous evidence indicates that the regenerative properties of TSPCs are predominantly driven by their extracellular vesicles (EVs), which have emerged as promising cell-free therapeutic agents for tendon regeneration. Compared with conventional MSC-derived EVs, TSPCs-EVs exhibit tendon-specific regenerative characteristics and are more effective in regulating the tendon healing microenvironment [10,11]. Zhao et al. reported that healthy TSPC-derived EVs broke the senescence-inflammation feedback loop between aged tendon stem cells and macrophages, enhancing tendon-to-bone healing in chronic aged rotator cuff tears [12]. He et al. found that TSPC-derived EVs promote tendon-bone healing after rotator cuff repair by enhancing BMSC fibrogenesis and modulating macrophage polarization [13]. These findings suggests that TSPCs-EVs might activate tendon-repair signaling pathways, enhance regeneration and reduce re-tear risks. Despite the regenerative potential of TSPCs-EVs in rotator cuff repair, unresolved limitations hinder their clinical translation. The low abundance and limited proliferation capacity of TSPCs result in insufficient EV production. In addition, conventional two-dimensional (2D) culture systems fail to adequately mimic the native cellular microenvironment, thereby impairing cell bioactivity, stemness maintenance, and EV secretion efficiency [14]. Therefore, developing optimized culture strategies to enhance the therapeutic efficacy and yield of TSPCs-EVs is desirable. To overcome these limitations, three-dimensional (3D) culture systems have been developed to enhance EV production and biological activity. Compared with conventional 2D culture, 3D culture better mimics the native cellular microenvironment and promotes cell-cell as well as cell-extracellular matrix interactions, thereby improving stem cell proliferation and EV secretion efficiency [[15], [16], [17]]. Among various 3D culture strategies, coaxial extrusion bioprinting enables the fabrication of high-density cellular microenvironments and has shown considerable potential in tissue engineering applications [18,19]. Our group have established 3D coaxial bioprinting platform [20], and based on our previous findings, EVs derived from 3D-cultured ADSCs were found to stimulate both neovascularization and collagen synthesis, which collectively facilitated tissue regeneration in wound healing models [21]. In addition, 3D-EVs-functionalized HAMA-CPC hydrogels enhanced osteogenesis via vascularization which had good potential for bone defect repair [22]. Nevertheless, the role of 3D-coaxial bioprinting-TSPCs-derived EVs in rotator cuff tears healing remains unexplored. Meanwhile, directly injected EVs are rapidly cleared in vivo because of enzymatic degradation and immune elimination, resulting in short retention time and insufficient therapeutic efficacy at the tendon-bone interface. Although hydrogel-based delivery systems can partially prolong EV release, maintaining sustained local retention remains challenging. Recently, microneedle (MN)-based delivery systems have attracted increasing attention in tissue repair due to their minimally invasive properties and capability for localized sustained delivery. Hence, we integrated 3D-rTSPCs-EVs into a hyaluronic acid methacrylate (HAMA)-based microneedle platform to enhance local retention and therapeutic efficacy for rotator cuff healing. Here, we fabricated the high-density rTSPCs-laden 3D microfibers to generate EVs by using 3D-coaxial bioprinting. In vitro assays elucidated the tenogenic and anti-inflammatory capacities of 3D-rTSPCs-EVs. To improve local retention and delivery efficiency at the tendon-bone interface (TBI), 3D-rTSPCs-EVs were further incorporated into a hyaluronic acid methacrylate (HAMA)-based microneedle (MN) platform for the treatment of CMRCT. The therapeutic effects of 3D-rTSPCs-EVs@MN on tendon-bone healing, collagen remodeling, inflammatory regulation, and functional recovery were systematically investigated in a rat CMRCT model. This study provides a promising strategy for rotator cuff regeneration based on 3D-TSPCs-EVs-mediated regenerative therapy. Materials and methods Isolation and culture of rat tendon stem/progenitor cells (rTSPCs) rTSPCs were isolated from tendon tissues of rats (12-weeks, Achilles tendons), respectively, using identical enzymatic digestion protocols. Briefly, harvested tissues were minced into fragments and digested in 2 mg/mL type I collagenase (Sigma) at 37 °C for 2 h. The slurry was centrifuged (1000 rpm, 5 min), and the pellet was washed twice with phosphate-buffered saline (PBS) to terminate digestion. The digested tissue was resuspended in α-MEM supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin-neomycin, followed by incubation at 37 °C in 5% CO2. During primary culture, the medium was replaced every 3 days, and residual tissue fragments were removed on day 7. Cells at passage 2 were utilized for identification and subsequent experiments. For cell characterization, flow cytometry was performed to evaluate the expression of mesenchymal stem cell-associated surface markers CD29, CD44, and CD90, as well as the hematopoietic marker CD45. In addition, cell viability was assessed using a CCK-8 assay at days 1, 3, and 5. To further evaluate multilineage differentiation potential, osteogenic differentiation was induced using osteogenic medium followed by alkaline phosphatase (ALP) and Alizarin Red staining, while chondrogenic differentiation was assessed by Alcian Blue staining after culture in chondrogenic induction medium. Tendon-related gene expression, including COL-I, COL-III, and TNMD, was analyzed by RT-qPCR. Preparation and characterization of rTSPCs-EVs and 3D-rTSPCs-EVs The coaxial cell-laden microfiber fabrication protocol followed our previously established methodology. Briefly, a coaxial needle system (inner diameter: 0.38 mm; outer diameter: 1.1 mm) was employed to generate core-shell microfibers. The outer sheath consisted of sterile sodium alginate solution (1.5% w/v, medium viscosity, molecular weight: 120–190 kDa, analytical grade), while the core phase contained tendon stem/progenitor cells (rTSPCs) suspended in complete culture medium. Controlled extrusion was achieved using a dual-channel injection pump, with flow rates adjusted to 15–30 mL/h for the alginate sheath and 3–10 mL/h for the cellular core. The fabricated microfibers were immediately extruded into a sterile 3% (w/v) calcium chloride solution for ionic crosslinking at room temperature (22–25 °C) for 10 min to ensure sufficient structural stability and mechanical integrity. Subsequently, the crosslinked microfibers were gently washed three times with sterile phosphate-buffered saline (PBS) to remove residual calcium ions and then transferred into growth medium for subsequent culture. EVs from 2D-cultured (normal-cultured) rTSPCs and 3D-cultured (3D-microfibers-cultured) rTSPCs were isolated and purified using commercial kits (HQExoA002L3001, Huamei BioTech Co. Ltd., China) as previously described. Comprehensive characterization of EVs included: 1. Morphology: transmission electron microscopy (TEM) imaging. 2. Size distribution: nanoparticle tracking analysis (NTA). 3. Surface markers: western blot (WB) detection of EVs proteins. Moreover, we performed LC-MS proteomic analysis to detect the differences in rTSPCs-EVs and 3D-rTSPCs-EVs protein contents. KEGG pathway, Gene Ontology (GO) and heatmap analysis were conducted using R (version 4.3.1) and various bioconductor packages to identify biological processes implicated by proteins differentially expressed between rTSPCs-EVs and 3D-rTSPCs-EVs models. Our analysis initiated with limma package for preprocessing and identifying differentially expressed genes. ClusterProfiler enabled subsequent GO enrichment analysis, and ggplot2 facilitated the visualization, with color gradients denoting p-value significance. EVs labeling and cellular uptake analysis Commercially available rat tendon fibroblasts were obtained from Procell (Wu Han, China). Cells were cultured in DMEM supplemented with 10% FBS, 1% antibiotic in the incubator at 37 °C under a humidified 5% CO2 atmosphere. To evaluate EVs internalization by tendon fibroblast, rTSPCs-EVs and 3D-rTSPCs-EVs were fluorescently labeled with PKH67 (Sigma-Aldrich, USA). Briefly, EVs were incubated with 2 μM PKH67 in Diluent C for 5 min at room temperature. Unbound dye was quenched with 1% (w/v) bovine serum albumin (BSA) in PBS, followed by two rounds of ultracentrifugation (100,000×g, 4 °C, 70 min) to remove residual dye aggregates. For uptake assays, tendon fibroblasts were seeded onto glass-bottom confocal dishes (NEST, China) and allowed to adhere overnight. Labeled EVs (5 × 109 particles/mL in exosome-depleted medium) were administered to cells, followed by 6, 12, and 24 h incubation under standard culture conditions. Post-incubation, cells were washed thrice with PBS, fixed with 4% paraformaldehyde for 10 min, and permeabilized with 0.1% Triton X-100. Cytoskeletal F-actin and nuclei were counterstained with rhodamine-phalloidin (Solarbio, China) and DAPI (Solarbio, China), respectively. Immunofluorescence staining in vitro Immunofluorescence staining of tendon fibroblasts Immunofluorescence (IF) staining was conducted to evaluate protein expression. Cultured tendon fibroblasts under experimental conditions were fixed with 4% paraformaldehyde for 20 min at room temperature. After three washes with PBS, cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min. Non-specific binding sites were blocked with 5% (w/v) bovine serum albumin (BSA; Sigma-Aldrich) in PBS for 1h at 37 °C. Samples were then incubated with primary antibodies collagen type I (Col1a1), collagen type III (Col3a1), and tenomodulin (TNMD), Interleukin-6 (IL-1β), and Tumor Necrosis Factor-alpha (TNF-α), (Abcam, 1:500) overnight at 4 °C. Following PBS washes, Alexa Fluor-conjugated secondary antibodies (Servicebio; 1:500 dilution) and 4′,6-diamidino-2-phenylindole (DAPI; Servicebio) were applied for 30 min at 37 °C in darkness. Fluorescent images were acquired using a CLSM (Confocal Laser Scanning Microscope, LSM880, German). Semi-quantitative analysis of fluorescence intensity was performed using ImageJ software (NIH). Immunofluorescence staining of macrophage RAW 264.7 macrophages (5 × 105 cells/well) were plated in six-well plates. The control group was cultured in complete medium (DMEM supplemented with 10% FBS and 1% penicillin–streptomycin), whereas the LPS group was treated with 20 ng/mL LPS under identical culture conditions. The LPS+3D-rTSPCs-EVs group was co-treated with LPS and 3D-rTSPCs-EVs. After 48 h, cells were harvested for analysis. iNOS and CD163 (Abcam) were employed to evaluate macrophage polarization. Confocal microscopy and Semi-quantitative analysis were described as 2.4.1. Quantitative real-time polymerase chain reaction (qRT-PCR) Gene expression profiling was performed using qRT-PCR to evaluate tenogenic differentiation markers. Total RNA was isolated from cells with TRIzol reagent (DP424, Tiangen Biotech, China) and stored at −80 °C until analysis. RNA samples were reverse-transcribed into complementary DNA (cDNA) using the SuperScript First-Strand Synthesis System (SYT013, Seyotin, China), followed by PCR amplification with recombinant Taq DNA polymerase (SYT103, Seyotin, China). The expression levels of tendon-specific genes—Col1a1, Col3a1, and TNMD—were quantified at Day 3 post-treatment using the LightCycler 480 Real-Time Fluorescence PCR System (Roche Diagnostics). All reactions were normalized to the housekeeping gene β-actin, and relative expression was calculated via the 2−ΔΔCt method. The primer sequence was shown in Table 1. Table 1Primer sequences.Table 1GenesForward (5′-3′)Reverse (5′-3′)COL-ⅠCTCTCCTGGCTCTCCTGGTCAGGGAGACCGTTGAGTCCACOL-ⅢGACCTGGAGAGCGTGGACTGTCTCCCGGAAGACCGTTTTenomodulinTGGACTAGTGTTTGGTATCCTGGCTTCTTCTTCTCGCCATTGCTGTGAPDHAACGGGAAACCCATCACCATGGCGGAGATGATGACCCTTT Primer sequences. Western blot analysis Cells were lysed in RIPA buffer containing protease/phosphatase inhibitors. Protein concentrations were determined via BCA assay (Beyotime Biotechnology). Equal protein aliquots were separated by 10% SDS-PAGE, transferred to PVDF membranes, and blocked with 5% skim milk. Membranes were incubated overnight at 4 °C with primary antibodies against Col1a1 (1:1500), Col3a1 (1:1000), TNMD (1:1000), IL-1β (1:500), IL-6 (1:1000), and TNF-α (1:500) (all from Abcam). After HRP-conjugated secondary antibody incubation for 90 min, bands were visualized using a ChemiDoc system (Bio-Rad). Preparation method of hyaluronic acid methacrylate microneedle (HAMA-MN) The microneedles were prepared as previously described, with modifications according to the product instructions. Briefly, first, 0.025g of the photoinitiator 2,4,6-Trimethylbenzoyl Lithium Phenyl Phosphinate (LAP) was dissolved and diluted with PBS to 10 mL to prepare a 0.25% (w/v) working solution. Then, 50 mg of Hyaluronic Acid Methacrylate (HAMA) was completely dissolved at 30 °C in the prepared working solution to prepare a 5% (w/v) HAMA-150k hydrogel precursor solution. This precursor solution was dripped into a microneedle mold, and the bubble removal process was repeated three times using a negative pressure defoaming device at room temperature. Subsequently, it was heated and concentrated in a 25 °C oven for 3 h; this process was repeated 2-3 times until the microneedle tips were fully formed, followed by UV light curing at 450 nm for 10 s. Finally, more precursor solution was added to form the base film, which was dried in a 25 °C oven for 4 h before the microneedles were demolded. Preparation method of 3D-rTSPCs-EVs@MN The method for fabricating microneedles loaded with 3D-rTSPCs-EVs were similar to that described above. First, according to the aforementioned scale-up process, a 5% (w/v) HAMA-150k hydrogel precursor solution was prepared. Then, 10 mL of this precursor solution was mixed with approximately 8 × 1010 particles of 3D-rTSPCs-EVs to form a uniform 3D-rTSPCs-EVs hydrogel solution. This 3D-rTSPCs-EVs hydrogel solution was then added to the microneedle mold. Bubble removal was performed three times using a negative pressure defoaming device at room temperature. Subsequent heating and concentration in a 25 °C oven for 3 h was carried out, repeated 2-3 times until the microneedle tips were completely formed, followed by UV light curing at 450 nm for 10 s. Finally, more 3D-rTSPCs-EVs hydrogel solution was added to form the base film, which was dried in a 25 °C oven for 4 h before the microneedles were demolded. To evaluate the biodegradation behavior of the microneedle system in vivo, the weight loss ratio of the 3D-rTSPCs-EVs@MN patch was measured over 28 days after implantation. The residual mass of the microneedle patches was collected at predetermined time points (7, 14, 21, and 28 days), and the degradation ratio was calculated based on the initial weight of the patches. The prepared microneedles were used in animal experiments as soon as possible. Animal experiments Ethical approval All animal experiments were approved by the Institutional Animal Care and Use Committee of the Third Affiliated Hospital of Sun Yat-Sen University (Approval No.2023-02-300). Animal handling and procedures strictly adhered to the ARRIVE guidelines for preclinical studies. Experimental animals and grouping Forty-eight male Sprague-Dawley rats (18-week-old, 353 ± 15.26 g) were randomly assigned into five groups: Sham (n = 8), CMRCT (n = 10), MN (n = 10), rTSPCs-EVs@MN (n = 10) and 3D-rTSPCs-EVs@MN group (n = 10). 1. Sham group: according to previously published rotator cuff injury models [23], only deltoid muscle separation was performed without tendon detachment or repair. 2. CMRCT group: under aseptic conditions, rats were positioned laterally with scapulohumeral joint external rotation. A 3 cm transverse incision exposed the humeral head, supraspinatus (ST) and infraspinatus tendon (IT); subsequently full-thickness ST, IT were detached and delayed surgical repair after 4 weeks. The delayed repair involving scar tissue excision, tendon-bone footprint decortication, and crossed bone tunnel (0.5 mm) fixation via modified Mason-Allen technique. 3. MN group: the CMRCT model was established, then the tendon bone insertion (TBI) was overlapped with MN after the delayed repair, creating a layered part consisting of MN and TBI. 4. rTSPCs-EVs@MN and 3D-rTSPCs-EVs@MN group:the CMRCT model was established, then the tendon bone insertion (TBI) was overlapped with rTSPCs-EVs@MN and 3D-rTSPCs-EVs@MN after the delayed repair. Wound healing was assessed daily. No signs of infection or systemic toxicity were observed. All rats were sacrificed at 4 weeks post-repair (8 weeks after initial modeling) via CO2 inhalation. Shoulder joints were harvested for biomechanical testing, histology, and imaging analysis. For biomechanical testing, histological evaluation, and Micro-CT analysis, 5 samples from each group were randomly selected. For MRI analysis, 3 samples from each group were randomly selected for representative imaging evaluation. Histology evaluation The rats were euthanized, and the musculotendinous-tendon-humerus complexes were harvested for histological analysis. Coronal sections (3 mm thickness) were fixed, paraffin-embedded, and stained with hematoxylin&eosin (H&E), Masson's trichrome, safranin-fast green and picrosirius red staining. H&E-stained sections were semi-quantitatively scored using the modified Watkins system to assess inflammatory infiltration, vascularity, and tissue architecture. Masson's trichrome staining was performed to visualize collagen fibers (blue), muscle fibers (red), and nuclei (dark blue), enabling quantitative assessment of fibrotic area ratios. Safranin O-Fast Green staining was utilized to evaluate proteoglycan content in cartilage matrices, with glycosaminoglycans (GAGs) visualized as red deposits against a green-stained collagen background. Picrosirius red staining was employed to differentiate collagen I (orange birefringence) and collagen III (green birefringence) under polarized light microscopy (Leica DM2700P, Germany), with semi-quantitative analysis of collagen deposition. Collagen maturity and fiber alignment were further evaluated by measuring birefringence intensity and orientation patterns. Immunofluorescence analysis of tendon-bone interface tissue sections Immunofluorescence quantification of the collagen type I/III (Col I/Col III) ratio The Col I/Col III ratio indicates tendon regeneration maturity and scar fibrosis status. Tissue specimens encompassing the coronal plane of the tendon-bone interface were decalcified, dehydrated, and embedded in paraffin. The 5 μm-thick sections were permeabilized with 0.1% Triton X-100, and blocked with 5% BSA. Primary antibodies targeting Collagen I (1:100; Abcam) and Collagen III (1:100; Abcam) were applied overnight at 4 °C. After PBS washes, Alexa Fluor and FITC-conjugated secondary antibodies (1:500; servicebio) were incubated, followed by DAPI staining. Relative fluorescence intensity of Collagen I and Collagen III were quantified using ImageJ software. Immunofluorescence analysis of M1 macrophage polarization M1 and M2 macrophage polarization were analyzed at 4 weeks post-surgery, aligning with the recruitment window of macrophages after rotator cuff tear. Macrophage polarization was assessed by Immunofluorescence co-staining of CD163 (M2) and iNOS (M1), and the staining procedure were described as above. CD163+, iNOS+ cells were quantified using ImageJ (NIH) for histological and immunofluorescence assessment. Immunohistochemical analysis Paraffin-embedded tissue sections (5 μm) were deparaffinized, rehydrated, and subjected to antigen retrieval using citrate buffer (pH 6.0). After blocking with 3% hydrogen peroxide and 5% BSA, the sections were incubated overnight at 4 °C with primary antibodies against COL-I, COL-III, CD163, and iNOS. Subsequently, sections were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature, followed by DAB staining and hematoxylin counterstaining. Images were captured under a light microscope, and positive staining areas were quantified using ImageJ software. Biomechanical testing Biomechanical testing was performed on musculotendinous-tendon-humerus complexes harvested from euthanized rats. The cross-sectional area of the tendon insertion site was measured using a digital caliper. Following established protocols, a 0.9 mm steel wire was passed through a drilled hole in the humeral diaphysis and looped around the humeral head to prevent growth plate displacement. The musculotendinous portion was clamped with a vice grip lined with dry gauze for friction enhancement, while the distal humerus was stabilized using a silicone pad to avoid fracture and secured in the opposing grip. A material testing system (MTS Insight®, MTS Systems Corp, USA) applied sequential loading protocols: preloading (0.25 N, 5 s), preconditioning (3% strain, 1Hz, 5cycles), and tensile failure testing at 20 mm/s. Ultimate load (N), tensile stress (N/mm2), and stiffness (N/mm) were recorded during failure testing, and tensile stress (N/mm2) was calculated by normalizing the peak force to the pre-measured cross-sectional area. During biomechanical testing, most specimens failed at the tendon-bone junction area. Micro-CT analysis Following 24-h fixation in 4% paraformaldehyde, specimens were mounted on the scanning stage of a high-resolution micro-CT system (SkyScan 1176, Bruker, Germany). Scanning parameters included an X-ray source operating at 80 kV/300 μA, achieving an isotropic voxel resolution of 18 μm. Beam hardening artifacts were minimized using a 1.0-mm aluminum filter. Quantitative morphometric analysis was performed within a cylindrical region of interest (ROI, 5 × 5 mm2) to calculate bone volume fraction (BV/TV%) and bone mineral density (BMD, mg HA/cm3). Micro–magnetic resonance imaging analysis Four shoulders per group were randomly selected for magnetic resonance imaging (MRI) using a 3.0-T/160-mm preclinical MRI system (Agilent Technologies) under respiratory anesthesia (2% isoflurane in O2: N2O = 1:2) prior to euthanasia. Coronal T2-weighted sequences were acquired in the scapular plane, mimicking clinical shoulder MRI protocols. Tendon-to-bone junction integrity was assessed through three morphological parameters: thickness, continuity, and signal intensity heterogeneity. All micro-MRI images were independently evaluated by a board-certified musculoskeletal radiologist, with consensus reached among investigators regarding qualitative findings. Due to limited sample size (n = 3/group) and absence of validated scoring criteria, quantitative statistical comparisons were not performed. Statistical analysis In this study, all experimental data are presented as the mean ± standard deviation. To evaluate significant differences, one-way analysis of variance (ANOVA) and Student's t-tests were performed, with statistical significance defined as a P-value < 0.05. Statistical analyses and graphical visualization were conducted using GraphPad Prism 7 software. Results Characterization of the 3D-rTSPCs-EVs rTSPCs were first identified by flow cytometry and multilineage differentiation assays (Fig. S1). Flow cytometric analysis demonstrated that rTSPCs highly expressed the MSC-associated surface markers CD29, CD44, and CD90, while showing minimal expression of the hematopoietic marker CD45 (Fig. S1A–B), confirming the typical phenotypic characteristics of rTSPCs. In addition, both 2D-rTSPCs and 3D-rTSPCs maintained favorable cell viability during culture, although 3D-rTSPCs exhibited significantly enhanced proliferation at later time points compared with 2D-rTSPCs (Fig. S1C). Multilineage differentiation assays further demonstrated the stemness properties of rTSPCs. ALP and Alizarin Red staining confirmed osteogenic differentiation potential (Fig. S1D–E), whereas Alcian Blue staining demonstrated chondrogenic differentiation capability (Fig. S1F). Moreover, RT-qPCR analysis revealed that 3D-rTSPCs exhibited significantly elevated expression of tendon-related genes, including COL-I, COL-III, and TNMD, compared with 2D-rTSPCs (Fig. S1G), suggesting that the 3D culture microenvironment better preserved the tendon-specific phenotype of rTSPCs. Hollow and aligned microfiber fabrication via coaxial bioprinting supported high-density rTSPCs culture and subsequent EV production (Fig. 1A). Compared with conventional 2D culture, the 3D bioprinting system markedly enhanced EV production, resulting in approximately 20-fold higher EV protein yield and 98-fold greater EV particle yield per culture dish (Fig. S2). Western blot analysis confirmed the identity of EVs derived from 3D-rTSPCs, showing positive expression of the EV marker CD81, while the endoplasmic reticulum marker calnexin was undetectable (Fig. 1B). Transmission electron microscopy (TEM) further demonstrated that both rTSPCs-EVs and 3D-rTSPCs-EVs exhibited a characteristic cup-shaped morphology, consistent with the typical ultrastructural features of EVs (Fig. 1C). NTA analysis revealed that possessed an average particle size of approximately 126 nm, with no significant size difference compared with rTSPCs-EVs (Fig. 1D), suggesting that the 3D culture system did not markedly alter the basic physicochemical properties of EVs. These findings confirmed the successful isolation and characterization of 3D-rTSPCs-EVs suitable for subsequent functional studies. Culture of rTSPCs and the Characterization of rTSPCs-EVs and 3D-rTSPCs-EVs. (A) The Bright filed of rTSPCs in 3D microfibers. (B) Western blot analysis revealed the presence of EVs markers CD81 in both rTSPCs-EVs and 3D-rTSPCs-EVs samples, however, Calnexin is to confirm the absence of EVs content. (C) TEM of rTSPCs-EVs and 3D-rTSPCs-EVs. (D) Representative graph of EVs size distribution by Nanoparticle tracking system. (E-F) The KEGG enrichment and heatmap analysis of the rTSPCs-EVs and 3D-rTSPCs-EVs were screened by proteomics. (G) Endocytosis behavior of EVs by tendon fibroblast at different time points (6, 12, and 24 h). n = 3, *, P < 0.05; **, P < 0.01; ***, P < 0.001. Proteomic analysis was utilized to evaluate the rTSPCs-EVs and 3D-rTSPCs-EVs. The volcano plot revealed that we identified 315 differentially expressed genes (DEGs), including 99 upregulated and 216 downregulated proteins in 3D-rTSPCs-EVs relative to rTSPCs-EVs (Fig. S3). KEGG enrichment analysis demonstrated that the differentially expressed proteins were mainly associated with ECM-receptor interaction, TGF-β signaling, and NF-κB signaling pathways (Fig. 1E), which are closely involved in tendon regeneration and inflammatory regulation. Furthermore, heatmap analysis revealed that 3D-rTSPCs-EVs were enriched in multiple proteins associated with tenogenesis (CCN1, ITGB5, TXN, RHOA, and LATS1) and anti-inflammatory regulation (RGS10, MAP3K7, and HSP90B2P) (Fig. 1F), indicating that the biomimetic 3D microenvironment altered EV cargo composition toward a more regenerative phenotype. To further evaluate EV bioactivity, PKH67-labeled EVs were incubated with rat tendon cells, and cellular uptake was assessed by confocal microscopy. Both rTSPCs-EVs and 3D-rTSPCs-EVs were effectively internalized by tendon cells in a time-dependent manner (Fig. 1G). Quantitative fluorescence analysis further demonstrated that the uptake efficiency of 3D-rTSPCs-EVs was significantly higher than that of rTSPCs-EVs at 6, 12, and 24 h, suggesting enhanced cellular interaction and internalization capacity of EVs generated under the 3D culture condition. 3D-rTSPCs-EVs enhanced collagen deposition, tenoinductive ability and increased the collagen I/collagen III ratios in-vitro To evaluate the tenogenic potential of 3D-rTSPCs-EVs, the expression of tendon-related ECM proteins and tenogenic markers was assessed in tendon fibroblasts. RT-qPCR analysis demonstrated that treatment with 3D-rTSPCs-EVs markedly upregulated the mRNA expression levels of collagen I (Col-I), collagen III (Col-III), and tenomodulin (TNMD) compared with both the control and rTSPCs-EVs groups (Fig. 2D), indicating enhanced tenogenic induction capacity. Immunofluorescence analysis further revealed substantially increased deposition of Col-I and TNMD in the 3D-rTSPCs-EVs-treated fibroblasts (Fig. 2A–C). Although Col-III expression was also elevated, quantitative analysis demonstrated a significantly higher Col-I/Col-III ratio in the 3D-rTSPCs-EVs group (Fig. 2C), suggesting improved collagen matrix remodeling and a more mature tendon-like ECM phenotype. Western blot analysis confirmed the enhanced protein expression of Col-I, Col-III, and TNMD following 3D-rTSPCs-EVs treatment (Fig. 2E–F). Notably, the increase in Col-I expression was more pronounced than that of Col-III, further supporting the elevated Col-I/Col-III ratio observed in immunofluorescence analysis. These results collectively indicated the tenoconductive, tenoinductive, and tenogenic potential of 3D-rTSPCs-EVs. 3D-rTSPCs-EVs markedly accelerated tendon regeneration by enhancing the expression of tendon-related proteins. (A-C) Representative immunofluorescence images of Col-I, Col-III and TNMD in rat tendon fibroblasts, with quantitative analysis of immunofluorescence signal intensities normalized to the control group. (D) mRNA levels of Col-I, Col-III, and TNMD were analyzed by RT-qPCR. (E, F) Western blot analysis of Col-I, Col-III, and TNMD protein expression, with quantitative analysis of band gray intensities demonstrating a significantly elevated Col-I/Col-III ratio. n = 3, *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. 3D-rTSPCs-EVs significantly suppressed the inflammatory response in tendon fibroblasts To investigate the anti-inflammatory effects of 3D-rTSPCs-EVs, tendon fibroblasts were stimulated with LPS to establish an inflammatory microenvironment in vitro. Immunofluorescence staining demonstrated that LPS stimulation significantly upregulated pro-inflammatory genes, including tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and interleukin-1β (IL-1β). Notably, 3D-rTSPCs-EVs treatment resulted in a more pronounced suppression of IL-6, TNF-α and IL-1β fluorescence intensity than rTSPCs-EVs treatment alone (Fig.S4A and 3A-D), indicating superior anti-inflammatory activity. Western blot analysis demonstrated that LPS stimulation significantly elevated the protein expression levels of IL-1β, IL-6, and TNF-α, whereas pretreatment with 3D-rTSPCs-EVs markedly suppressed the expression of these pro-inflammatory cytokines (Fig. 3E–F). In addition, immunofluorescence staining revealed that matrix metalloproteinase-9 (MMP9), a key matrix-degrading enzyme associated with inflammatory tendon degeneration, was significantly upregulated following LPS stimulation, while 3D-rTSPCs-EVs treatment effectively reduced MMP9 expression (Fig. S4B). 3D-rTSPCs-EVs attenuated inflammatory responses by suppressing pro-inflammatory cytokines. (A-B) Representative immunofluorescence images and quantitative analysis of IL-1β expression in rat tendon fibroblasts. (C-D) Representative immunofluorescence images and quantitative analysis of TNF-α expression in rat tendon fibroblasts. (E-F) Western blot analysis and quantitative analysis of IL-1β, IL-6, and TNF-α protein expression. n = 3, *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. Synthesis and characterization of 3D-rTSPCs-EVs@MN The fabricated 3D-rTSPCs-EVs@MN patch exhibited a translucent and uniformly distributed microneedle architecture with good structural integrity (Fig. 4A). Scanning electron microscopy (SEM) further demonstrated that the microneedles possessed a highly ordered array structure with sharp conical tips and homogeneous morphology (Fig. 4B–C). Cross-sectional SEM images revealed that the microneedles exhibited an approximate height of 800 μm, a basal diameter of approximately 200 μm, and an inter-needle spacing of 200 μm (Fig. 4D–E), indicating that the microneedle geometry was suitable for tissue penetration and localized drug delivery. Moreover, high-magnification SEM imaging revealed the presence of uniformly distributed nanoscale particle-like structures on the microneedle surface (Fig. 4F). These structures exhibited a spherical morphology with diameters of approximately 100–150 nm, which was consistent with the size distribution of 3D-rTSPCs-EVs determined by TEM and NTA analyses (as shown in Fig. 1C–D). To further verify the incorporation of EVs into the microneedles, 3D-rTSPCs-EVs were labeled with PKH67 prior to fabrication. Fig. S5A demonstrated a homogeneous distribution of PKH67-positive signals throughout the microneedle array, providing additional evidence that the labeled EVs were successfully incorporated into the MN platform. To evaluate the skin penetration capability of the microneedles, the MN patch was inserted into rat skin tissue. Representative photographs and H&E staining demonstrated distinct microchannels at the insertion sites, confirming the effective puncture ability of the microneedles and successful penetration into skin tissue (Fig. 4G–H). Mechanical testing further demonstrated that there were no significant differences in compression strength among MN, rTSPCs-EVs@MN, and 3D-rTSPCs-EVs@MN groups (Fig. 4I), indicating that EV incorporation did not compromise the mechanical stability of the microneedle system. The release behavior of EVs from the microneedle platform was subsequently evaluated in vitro. Compared with free EVs, both rTSPCs-EVs@MN and 3D-rTSPCs-EVs@MN exhibited sustained release profiles over 21 days (Fig. 4J), suggesting that the HAMA-MN system effectively prolonged EV retention and release duration. In vivo living imaging demonstrated prolonged local retention of Cy5.5-labeled 3D-rTSPCs-EVs delivered via the microneedle platform over a 28-day period (Fig. 4K–L). The fluorescence signal gradually decreased over time, in agreement with the sustained release behavior observed in vitro. In addition, the in vivo degradation behavior of the microneedle patch was assessed over 28 days (Fig. S5B). The MN patch exhibited gradual degradation in vivo, with approximately 90% weight loss observed at day 28, indicating favorable biodegradability and supporting its suitability as a sustained EV delivery system for tendon repair applications. Characterization and in vivo retention of rTSPCs-EVs@MN. (A) Macroscopic appearance of the 3D-rTSPCs-EVs@MN patch. (B-F) SEM analysis of 3D-rTSPCs-EVs@MN images of the microneedle array. (B, C) Top-view SEM images showing the uniform distribution and morphology of the microneedles. (D) Cross-sectional SEM image demonstrating the sharp conical structure of the microneedles. (E, F) Enlarged SEM images showing the surface morphology of microneedles and the distribution of 3D-rTSPCs-EVs on the microneedle surface, as indicated by the dashed circles and arrows. (G) Photograph of the microneedle array into the skin tissue. (H) H&E staining of skin tissue after microneedle puncture. Red arrow indicates the puncture site. (I) Mechanical strength analysis of different MN. (J) Cumulative release profiles of rTSPCs-EVs, 3D-rTSPCs-EVs, rTSPCs-EVs@MN and 3D-rTSPCs-EVs@MN. (K-L) In vivo fluorescence imaging and quantitation of Cy5.5-labeled 3D-rTSPCs-EVs delivered by microneedle at indicated time points. n = 3, *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. 3D-rTSPCs-EVs@MN enhanced in vivo rotator cuff tendon-bone regeneration A rat model of CMRCT was established to evaluate the therapeutic efficacy of 3D-rTSPCs-EVs@MN in vivo (Fig. 5A–B). Representative photographs illustrated the procedures for rotator cuff exposure, tendon localization, surgical repair, and microneedle-mediated treatment administration (Fig. 5B), further supporting the reliability and reproducibility of the in vivo experimental process. Histological evaluation demonstrated superior tendon-to-bone healing in the 3D-rTSPCs-EVs@MN group compared with the CMRCT, MN, and rTSPCs-EVs@MN groups. H&E staining revealed that the 3D-rTSPCs-EVs@MN group exhibited markedly reduced inflammatory cell infiltration and improved tissue continuity at the tendon-bone interface, accompanied by more regularly aligned collagen fibers and a more organized tendon structure (Fig. 5C). In contrast, the CMRCT and MN groups displayed obvious fibrotic scar tissue formation, disordered collagen arrangement, and poorly regenerated tendon-bone interfaces. Histological scoring further confirmed that the 3D-rTSPCs-EVs@MN group achieved significantly higher tendon maturity scores than the CMRCT, MN, and rTSPCs-EVs@MN groups (Fig. S6A). Safranin O/Fast Green staining further demonstrated enhanced fibrocartilage regeneration at the tendon-bone interface following 3D-rTSPCs-EVs@MN treatment (Fig. 5D). Compared with the other treatment groups, the 3D-rTSPCs-EVs@MN group exhibited a substantially enlarged metachromatic region and increased proteoglycan deposition, indicating improved fibrocartilage formation and interface remodeling. Quantitative analysis of metachromasia area further supported these observations, showing significantly greater fibrocartilage regeneration in the 3D-rTSPCs-EVs@MN group relative to the CMRCT, MN, and rTSPCs-EVs@MN groups (Fig. S6B). 3D-rTSPCs-EVs@MN enhances rotator cuff healing in a CMRCT. (A) Schematic timeline of the animal experiment process. (B) Surgical procedure for CMRCT modeling and treatment: (i) Surgical exposure of the rotator cuff; (ii) Identification of the rotator cuff tendon; (iii) Tendon localization and positioning; (iiii) Surgical repair of the rotator cuff tear; (iiiii) Microneedle-based therapeutic treatment with different MN. (C) H&E staining of the tendon-bone interface at 4 weeks postoperatively. (D) Safranin O/Fast Green staining demonstrating enhanced fibrocartilage regeneration at the tendon-bone interface in the 3D-rTSPCs-EVs@MN group. n = 5, Abbreviation: T, Tendon; I, insertion; B, bone. Micro-CT and MRI analysis Micro-CT evaluation demonstrated substantial differences in tendon–bone healing among the experimental groups at 4 weeks postoperatively (Fig. 6A). Compared with the CMRCT and MN groups, the 3D-rTSPCs-EVs@MN-treated group exhibited markedly enhanced bone regeneration and improved structural integration at the tendon–bone insertion site, as evidenced by a more continuous and compact trabecular architecture around the bone tunnel region. In contrast, obvious bone defects and reduced trabecular formation were observed in the untreated repair groups. Quantitative micro-CT analysis further confirmed these findings (Fig. 6B). The 3D-rTSPCs-EVs@MN group showed significantly increased bone volume fraction (BV/TV), bone mineral density (BMD), trabecular number (Tb.N), and trabecular thickness (Tb.Th) compared with the CMRCT, MN, and rTSPCs-EVs@MN groups. Although the sham group maintained the highest overall bone quality parameters, the 3D-rTSPCs-EVs@MN treatment markedly restored peri-insertion bone microarchitecture toward normal levels, indicating superior osteogenic remodeling and mineralized tissue formation at the tendon–bone interface. Radiological and biomechanical evaluations of tendon-bone healing after different MN treatment. (A) Representative micro-computed tomography (micro-CT) images of the rat rotator cuff tendon-bone complex in different groups at 4 weeks post-treatment (n = 5). Red arrows indicate the bone tunnel/repair site. (B) Quantitative analysis of bone parameters at the tendon-bone interface, including bone volume/tissue volume (BV/TV), bone mineral density (BMD), trabecular number (Tb.N), and trabecular thickness (Tb.Th) (n = 5). (C) Representative magnetic resonance imaging (MRI) scans of the rotator cuff in each group (n = 3). Red arrows highlight the tendon-bone enthesis, showing improved tendon continuity and reduced signal intensity in the 3D-rTSPCs-EVs@MN group compared to the CMRCT and control groups. (D) Biomechanical analysis of the repaired tendon-bone complex, including ultimate failure load, cross-sectional area, stiffness, and modulus (n = 5). *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, not significant. Abbreviation: T, Tendon; I, insertion; B, bone. Representative coronal T2-weighted MRI images are presented in Fig. 6C. The sham group displayed an intact tendon continuity and a normal tendon–bone junction. Obviously, the CMRCT group exhibited pronounced hyperintense signals, discontinuous tendon structures, cystic-like changes in the humeral head, and obvious defects at the repaired insertion site. The MN-treated group showed partial tissue filling at the repair site but still demonstrated irregular tendon morphology and incomplete continuity. Similarly, the rTSPCs-EVs@MN group exhibited partial improvement with reduced signal intensity and scar tissue formation around the insertion area; however, structural continuity of the rotator cuff remained incomplete. Notably, the 3D-rTSPCs-EVs@MN group demonstrated significantly lower signal intensity, improved tendon continuity, a more organized tendon–bone interface, and increased tendon thickness compared with all other repair groups, indicating enhanced maturation and structural remodeling of the repaired rotator cuff tissue. Repaired tendons Augmented with 3D-rTSPCs-EVs showing better biomechanical properties Biomechanical assessments revealed distinct structural properties among the experimental groups (Fig. 6D). The sham group exhibited the lowest tendon cross-sectional areas (CSAs) (4.08 ± 0.53 mm2), reflecting the native tendon morphology, whereas the CMRCT group showed a markedly increased CSA (8.27 ± 1.33 mm2), which may be attributed to excessive scar tissue formation and disorganized ECM deposition after injury repair. In contrast, tendons treated with 3D-rTSPCs-EVs@MN demonstrated a marked reduction in CSA (5.76 ± 0.48 mm2) relative to the MN (7.34 ± 0.71 mm2) and rTSPCs-EVs@MN group (8.30 ± 0.85 mm2), showing that 3D-rTSPCs-EVs-mediated tendon neogenesis might mitigate scar-related tissue hypertrophy. Functional mechanical assessments further confirmed the superior repair quality induced by 3D-rTSPCs-EVs@MN treatment. The ultimate load in the 3D-rTSPCs-EVs@MN group (45.19 ± 1.44 N) was significantly higher than that in the rTSPCs-EVs@MN group (30.02 ± 1.94 N), indicating enhanced load-bearing capacity at the tendon–bone interface. Similarly, both stiffness and stress values were markedly improved in the 3D-rTSPCs-EVs@MN group compared with the other repair groups, further demonstrating superior biomechanical integrity and functional restoration. Ultimate load, stiffness, and stress are widely recognized indicators for evaluating tendon-to-bone healing quality in rotator cuff repair models [24]. The improved biomechanical performance observed in the 3D-rTSPCs-EVs@MN group may be associated with enhanced collagen fiber organization and ECM remodeling, which was consistent with the previous histological observations from picrosirius red staining. Furthermore, all specimens failed at the tendon–bone interface during biomechanical testing, indicating that the measured mechanical properties primarily reflected the healing quality of the tendon–bone junction rather than the mid-substance tendon tissue. These findings demonstrated that 3D-rTSPCs-EVs@MN treatment substantially enhanced both structural and functional tendon–bone healing after rotator cuff repair. 3D-rTSPCs-EVs@MN promoted native and well-aligned collagen fibers formation To further evaluate ECM remodeling and tendon maturation during tendon–bone healing, histological, immunofluorescence, and immunohistochemical analyses were performed at 4 weeks postoperatively. Masson's trichrome staining demonstrated substantial differences in collagen deposition and fiber organization among the experimental groups (Fig. 7A). The sham group displayed a well-organized native tendon architecture with densely aligned collagen bundles at the tendon–bone interface. In contrast, the CMRCT group exhibited sparse and disorganized collagen deposition accompanied by irregular fibrous scar tissue formation. Although partial collagen regeneration was observed in the MN and rTSPCs-EVs@MN groups, the collagen fibers remained loosely arranged and lacked the highly ordered parallel orientation characteristic of native tendon tissue. Notably, the 3D-rTSPCs-EVs@MN group showed markedly enhanced collagen deposition with densely packed and longitudinally aligned collagen fibers bridging the tendon–bone interface, indicating superior ECM remodeling and more mature tendon-like tissue regeneration. Quantitative analysis further confirmed significantly increased collagen content in the 3D-rTSPCs-EVs@MN group compared with the other repair groups (Fig. S7A). 3D-rTSPCs-EVs@MN promoted collagen remodeling and mature tendon-like tissue regeneration at the tendon–bone interface. (A) Representative Masson's trichrome staining images of the tendon–bone healing interface at 4 weeks postoperatively (n = 5). (B) Immunofluorescence staining of collagen type I (COL-I, green) and collagen type III (COL-III, red) at the tendon–bone insertion (n = 5). (C) Representative Picrosirius Red staining images observed under polarized light microscopy (n = 5). Mature collagen type I fibers are displayed in yellow, whereas immature collagen type III fibers are shown in green. (D-E) Immunohistochemical staining of COL-I and COL-III at the tendon–bone healing interface (n = 5). Abbreviation: T, Tendon; I, insertion; B, bone. To further investigate collagen subtype remodeling at the healing interface, immunofluorescence and immunohistochemical staining of collagen type I (COL-I) and collagen type III (COL-III) were performed (Fig. 7B–D and E). Collagen type I (COL-I), which represents mature tendon extracellular matrix (ECM), was strongly expressed in both the sham and 3D-rTSPCs-EVs@MN groups, whereas markedly weaker COL-I expression and fluorescence intensity were observed in the CMRCT and MN groups. In contrast, collagen type III (COL-III), a marker associated with immature scar tissue formation and fibrotic remodeling, was highly expressed in the CMRCT group and remained relatively elevated in the MN and rTSPCs-EVs@MN groups. Importantly, the 3D-rTSPCs-EVs@MN group exhibited significantly reduced COL-III expression together with enhanced and more continuously distributed COL-I deposition at the tendon–bone interface compared with the untreated repair groups. Consequently, the COL-I/COL-III ratio was substantially increased in the 3D-rTSPCs-EVs@MN group, indicating accelerated transition from scar-associated matrix deposition toward mature tendon-like ECM remodeling and improved collagen matrix organization (Fig. S7B–D). Picrosirius red staining under polarized light microscopy further demonstrated the enhanced collagen maturation induced by 3D-rTSPCs-EVs@MN treatment (Fig. 7C). The sham and 3D-rTSPCs-EVs@MN groups predominantly exhibited strong yellow-to-orange birefringence, representing densely packed and highly organized mature collagen type I fibers. In contrast, the CMRCT and MN groups displayed predominantly green birefringence, indicative of immature and loosely organized collagen type III fibers. Although the rTSPCs-EVs@MN group showed partial improvement in collagen organization, the collagen alignment and birefringence intensity remained inferior to those observed in the 3D-rTSPCs-EVs@MN group. Quantitative birefringence analysis further confirmed significantly enhanced collagen maturation in the 3D-rTSPCs-EVs@MN-treated tissues (Fig. S7E). 3D-rTSPCs-EVs regulated macrophage polarization To investigate the immunomodulatory effects of 3D-rTSPCs-EVs on macrophage polarization, immunofluorescence staining was first performed in LPS-stimulated RAW264.7 macrophages (Fig. 8A–B). iNOS and CD163 were used as representative markers for pro-inflammatory M1 and anti-inflammatory M2 macrophages, respectively. Compared with the LPS group, treatment with rTSPCs-EVs partially reduced iNOS-positive staining and increased CD163 expression. Notably, the LPS+3D-rTSPCs-EVs group exhibited a markedly lower proportion of iNOS-positive macrophages together with significantly enhanced CD163-positive staining compared with both the LPS and LPS + rTSPCs-EVs groups, indicating that 3D-rTSPCs-EVs more effectively suppressed pro-inflammatory macrophage activation and promoted polarization toward the anti-inflammatory M2 phenotype. 3D-rTSPCs-EVs promoted macrophage polarization toward the anti-inflammatory M2 phenotype. (A-B) Representative immunofluorescence staining images and quantitative analysis of RAW264.7 macrophages under different treatments. iNOS (green) was used as a marker of M1 macrophages, while CD163 (red) represented M2 macrophages (n = 3). Nuclei were counterstained with DAPI (blue). (C) Representative immunofluorescence staining images of iNOS and CD163 at the tendon–bone healing interface in vivo (n = 5). (D-E) Representative immunohistochemical staining images of iNOS and CD163 at the tendon–bone insertion site (n = 5). Abbreviation: T, Tendon; I, insertion; B, bone. To further evaluate macrophage polarization during tendon–bone healing in vivo, immunofluorescence and immunohistochemical staining of iNOS and CD163 were performed at the tendon–bone interface at 4 weeks postoperatively (Fig. 8C–E). Compared with the sham group, the CMRCT group exhibited markedly increased iNOS-positive staining, indicating persistent M1 type macrophage during tendon–bone healing. Although the MN and rTSPCs-EVs@MN groups partially alleviated inflammatory responses, obvious iNOS-positive areas were still observed. Notably, the 3D-rTSPCs-EVs@MN group showed significantly reduced iNOS expression together with enhanced CD163-positive staining at the tendon–bone interface, suggesting enhanced polarization toward the anti-inflammatory M2 phenotype. Quantitative analysis further confirmed that the 3D-rTSPCs-EVs@MN group exhibited the lowest iNOS-positive area and the highest CD163-positive area among all repair groups (Fig. S8). These findings demonstrated that 3D-rTSPCs-EVs@MN effectively modulated macrophage polarization toward an anti-inflammatory microenvironment favorable for tendon–bone healing and tissue regeneration. Discussion Our results demonstrated that 3D coaxial bioprinting enables the efficient generation of bioactive EVs from scaffold-cultured rTSPCs. Compared with conventional rTSPCs-EVs, 3D-rTSPCs-EVs exhibited enhanced tenogenic and anti-inflammatory properties both in vitro and in vivo. In a rat model of CMRCT, 3D-rTSPCs-EVs promoted mature collagen deposition, fibrocartilage regeneration, and tendon-to-bone integration, ultimately improving biomechanical strength. Moreover, 3D-rTSPCs-EVs enhanced immunomodulatory activity by promoting M2 macrophage polarization while suppressing M1 activation. Mechanistically, proteomic analysis further suggested that these effects were associated with enrichment of tenogenesis-related and anti-inflammatory proteins. These findings indicate that 3D-rTSPCs-EVs represent a promising cell-free therapeutic strategy for rotator cuff repair. Recent studies have highlighted the therapeutic potential of stem cell-derived EVs for tendon regeneration. Wang et al. demonstrated that BMSC-derived exosomes promote tendon–bone healing by regulating macrophage M1/M2 polarization through circRNA1052 [25], while Lin et al. reported that BMSC-derived exosomes enhance tendon regeneration by promoting TDSC migration and differentiation [26]. TSPCs-derived EVs have emerged as attractive therapeutic candidates for tendon regeneration because they preserve tendon-specific biological characteristics from parental TSPCs and actively regulate the tendon healing microenvironment [10,11,13]. However, conventional 2D culture systems fail to recapitulate the anisotropic architecture and biomechanical cues of native tendon tissue, thereby limiting EV production and regenerative bioactivity [14]. Consistent with previous studies demonstrating the advantages of biomimetic 3D culture [15,17], our tendon-mimetic 3D bioprinting platform supported high-density rTSPCs culture and generated EVs enriched in proteins involved in ECM remodeling, TGF-β signaling, tenogenesis, and inflammatory regulation. Notably, proteomic analysis identified increased expression of tenogenesis-related proteins (CCN1, ITGB5, TXN, RHOA, and LATS1) and anti-inflammatory regulators (RGS10, MAP3K7, and HSP90B2P), which may underlie the enhanced tenogenic induction and immunomodulatory capacity of 3D-rTSPCs-EVs. Moreover, the higher cellular uptake efficiency of 3D-rTSPCs-EVs further suggests that the biomimetic microenvironment improves not only EV production but also their regenerative potential. The therapeutic application of rTSPCs-EVs in CMRCT remains limited by rapid clearance and insufficient retention at the injury site. To address these limitations, we employed a HAMA-based microneedle platform for localized and sustained EV delivery. The MN exhibited adequate mechanical strength for tissue penetration and enabled sustained EV release for up to 21 days with prolonged in vivo retention for 28 days, corresponding to the sequential inflammatory, proliferative, and remodeling phases of tendon healing. Previous studies demonstrated that hydrogel encapsulation effectively improves EV retention and therapeutic efficacy in tissue regeneration applications [[27], [28], [29]], the microneedle-assisted delivery strategy further facilitated localized EV accumulation within dense tendon tissue through minimally invasive administration. In addition, the gradual degradation of the HAMA-MN patch supported continuous EV release without requiring secondary removal, highlighting its potential for clinical translation in rotator cuff repair. Successful tendon regeneration requires restoration of a highly organized tendon–bone interface that closely resembles native tissue [30,31]. As collagen I provides mechanical strength whereas collagen III is associated with fibrotic healing [32,33], promoting collagen remodeling is critical for functional repair. In our study, 3D-rTSPCs-EVs@MN enhanced collagen deposition, increased the collagen I/III ratio, improved collagen fiber alignment, and promoted peri-insertion bone regeneration, indicating coordinated restoration of both soft and hard tissues. The structural improvements translated into superior biomechanical performance, as evidenced by significantly enhanced ultimate load, stress, and stiffness of the repaired tendon. These findings are consistent with previous reports demonstrating that stem cell-derived EVs promote collagen remodeling and fibrocartilage regeneration during tendon healing [34,35]. Notably, most specimens failed at the tendon–bone junction during biomechanical testing, indicating that the measured mechanical properties primarily reflected the integrity of tendon–bone healing. Although the representative biomechanical parameters evaluated in this study sufficiently reflected tendon–bone healing quality, incorporating additional indices, such as elastic modulus and fracture energy, would provide a more comprehensive assessment of the functional restoration achieved by 3D-rTSPCs-EVs@MN. Furthermore, similar biomechanical improvements have been reported in recent EV-based tendon regeneration studies. Zhang et al. demonstrated that PRP-Exos significantly enhanced the maximum breaking load and stiffness of the Achilles tendon-bone insertion [36]. Likewise, Zhang et al. found that grapefruit exosome-loaded microneedles restored tendon Young's modulus and tensile strength [37]. Compared with conventional TSPCs-EVs, however, 3D-rTSPCs-EVs achieved more organized collagen remodeling and greater biomechanical recovery, which may be attributed to the enhanced regenerative bioactivity conferred by the tendon-mimetic 3D bioprinting microenvironment. These findings demonstrate that optimizing both EV bioactivity and localized delivery effectively promotes functional tendon–bone regeneration following chronic massive rotator cuff injury. Macrophage polarization plays a central role in regulating the inflammatory microenvironment during rotator cuff healing [38,39]. While M1 macrophages mediate persistent inflammation and collagen degradation, M2 macrophages facilitate inflammation resolution, ECM remodeling, and tissue regeneration [40]. Our results demonstrated that 3D-rTSPCs-EVs markedly promoted M2 polarization while suppressing M1 activation both in vitro and in vivo, suggesting that immunomodulation represents an important mechanism underlying their therapeutic efficacy. These findings are consistent with accumulating evidence indicating that stem cell-derived EVs regulate tendon healing through immunomodulatory mechanisms. He et al. reported that SHED-derived exosome-mimetics promotes rotator cuff tendon-bone healing via macrophage immunomodulation [41]. Moreover, Zhang et al. [12] demonstrated that h-TSC-Exos could skew macrophage polarization from the M1 phenotype to the M2 phenotype, attenuating TSCs senescence. Compared with conventional rTSPCs-EVs, our findings further indicate that the biomimetic 3D culture microenvironment enhances the immunoregulatory capacity of rTSPCs-EVs, likely through alterations in EV cargo composition. Beyond regulating inflammation, M2 macrophages have been shown to promote collagen remodeling, suppress oxidative stress and fibrosis, and improve tissue repair quality. Shao et al. reported that M2 macrophages reduced the production of ROS and inflammatory cytokines to improve the wound-healing quality with more collagen deposition and higher ratio of collagen I/III [42]. Sun et al. demonstrated that M2 macrophage membrane-mediated biomimetic nanoparticles effectively attenuated tendon adhesion formation by suppressing inflammation and fibrosis [43]. In agreement with these studies, the enhanced M2 polarization induced by 3D-rTSPCs-EVs was accompanied by increased collagen I deposition, a higher collagen I/III ratio, and reduced fibrotic remodeling, ultimately contributing to improved tendon–bone healing. These findings suggest that 3D-rTSPCs-EVs may facilitate ECM remodeling and attenuate shoulder fibrosis by driving macrophage polarization toward the M2 phenotype. Although 3D-rTSPCs-EVs exhibited significant regenerative potential for chronic rotator cuff repair, several limitations remained unresolved. First, although 3D-rTSPCs-EVs@MN significantly improved tendon-bone healing and biomechanical recovery, complete restoration of native glenohumeral function remains challenging in CMRCT. Future studies should further evaluate whether this strategy can restore shoulder force-couple balance and prevent long-term rotator cuff arthropathy under chronic pathological conditions [44]. Second, although the HAMA-MN platform improved local retention and sustained release of EVs, the long-term stability, degradation kinetics, and bioactivity of EVs within the hydrogel matrix were not directly characterized. In addition, the relationship between EV release kinetics and the stage-specific biological responses during tendon healing warrants further investigation. Third, 3D-rTSPCs-EVs lack active targeting capability, and incorporation of targeting ligands or stimuli-responsive delivery strategies may further improve local therapeutic specificity and delivery efficiency. Finally, only representative biomechanical parameters and short-term therapeutic outcomes were evaluated in the present study. Future studies incorporating more comprehensive biomechanical assessments, including elastic modulus and fracture energy, together with long-term evaluations in clinically relevant large-animal models, will provide stronger evidence for the durability and translational potential of this therapeutic strategy. Conclusion This study demonstrated that 3D-bioprinted rTSPCs microfibers-derived EVs (3D-rTSPCs-EVs) significantly enhanced tendon-bone healing in CMRCT by promoting tenogenic differentiation, collagen maturation, and M2 macrophage polarization. Delivered via HAMA-MN, 3D-rTSPCs-EVs@MN improved biomechanical strength, reduced inflammation, and suppressed scar formation, offering a novel strategy for rotator cuff healing. CRediT authorship contribution statement Weiping He: Conceptualization, Data curation, Writing – original draft. Xiaopeng Yin: Conceptualization, Formal analysis, Writing – original draft. Shihai Jiang: Conceptualization, Investigation, Methodology, Writing – original draft. Yi Zhang: Methodology, Software. Boxun Liu: Methodology, Software. Zhifeng Zhang: Data curation, Formal analysis, Investigation. Woquan Ye: Formal analysis, Investigation, Methodology. Shurong He: Formal analysis, Investigation, Methodology. Tao Xu: Conceptualization, Software, Supervision, Visualization, Writing – review & editing. Chang Liu: Conceptualization, Data curation, Formal analysis, Supervision, Writing – review & editing. Kun Wang: Conceptualization, Formal analysis, Funding acquisition, Writing – review & editing. Declaration of competing interests The authors report no conflict of interest. Acknowledgements The authors would like to express their sincere gratitude to Huaqing Zhimei Biotechnology and Medprin Biotech Company for providing the experimental equipment. Data can be accessed by contacting the corresponding author. This work was supported by the General Program of Guangdong Provincial Natural Science Foundation (Grant No.2025A1515012753), Dongguan Science and Technology of Social Development Program (20221800906152), National Natural Science Foundation of China (no.82372402), Natural Science Foundation of Guangdong (20221800906152), and National Natural Science Foundation Cultivation Project of The Third Affiliated Hospital of Sun Yat-sen University (2026GZRPYQN06). Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103553. Contributor Information Tao Xu, Email: xut@tsinghua-sz.org. Chang Liu, Email: liuch398@mail2.sysu.edu.cn. Kun Wang, Email: wangk@mail.sysu.edu.cn. Supplementary data The following are the Supplementary data to this article: Multimedia component 1 Multimedia component 2 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Multimedia component 2 Data Availability Statement Data will be made available on request.