Platelet-rich plasma cues program M2 macrophage–derived exosome mimetics to regenerate the tendon–bone interface

Abstract

Surgical repair of rotator cuff injuries is frequently complicated by high retear rates, driven by persistent inflammation and inadequate tissue regeneration. Exosomes derived from M2 macrophages represent a promising therapeutic avenue due to their innate immunomodulatory and regenerative properties. However, their clinical application is hindered by low yields and complex purification processes. In this study, we employed an extrusion technique to isolate exosome mimetics from platelet-rich plasma (PRP) pretreated M2 macrophages. These engineered vesicles, termed PRP-M2-EM, acquired additional bioactive factors from PRP, which significantly enhanced their pro-angiogenic and immunoregulatory functions compared to standard M2-EM. This biomimetic engineering strategy successfully transposes the therapeutic benefits of PRP into a stable, nanoscale delivery system, overcoming the limitations of PRP’s short half-life and high production costs. Mechanistically, we identified the enrichment of miR-21a-5p within PRP-M2-EM as a primary driver of their superior efficacy. Further mechanistic investigation revealed that the high expression of miR-21a-5p in PRP-M2-EM targets and inhibits the tissue inhibitor of metalloproteinase 3 (TIMP3) gene, facilitating a regenerative environment. In conclusion, our study introduces engineered PRP-M2-EM as a potential therapeutic strategy. This approach promotes rotator cuff regeneration through enhanced angiogenesis and immunomodulation, with the miR-21a-5p/TIMP3 axis potentially contributing to these effects.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12951-026-04626-5.

Introduction

Rotator cuff injuries refer to damage to the muscles of the shoulder and can occur in a variety of situations, including traumatic injuries, chronic overuse, aging, and sports injuries [1]. Rotator cuff tears are a common form of rotator cuff injury, with symptoms including shoulder pain, shoulder muscle weakness, limited range of shoulder motion, and pain that worsens with activity, which can seriously affect quality of life [2, 3]. Rotator cuff repair is a common surgery in sports medicine. The primary objective of rotator cuff repair surgery is to reattach the torn tendons to the bone and restore proper shoulder function and strength [4]. Tendon–bone interface healing refers to the complex and challenging process of reattaching and regenerating a tendon to bone after injury or surgical repair. Factors influencing tendon bone healing include severity of injury, inflammation and infection, blood supply, appropriate surgical intervention and rehabilitation program [5]. Notwithstanding significant advances in surgical methods over recent decades, postoperative re-tear rates remain notably high, varying from 20% to 94% [6, 7]. Multiple strategies have been explored to improve tendon–bone integration, among which growth factor applications and stem cell-based interface tissue engineering have progressed considerably in recent years [8, 9]. While these approaches show promise in supporting tendon–bone interface healing, several associated challenges must be acknowledged. Current stem cell treatments are constrained by issues such as limited expansion capabilities ex vivo, donor scarcity, immune rejection risks, potential tumor formation, and ethical considerations [8, 9]. Their application is also limited by the cost of presentation and the potential risk of growth factors [10]. Therefore, the “Cell-free therapy” has gradually become the focus of our attention in recent years [11].

Exosomes are nanoscale extracellular vesicles, approximately 30 to 150 nm in size, that transport a diverse array of cellular constituents such as DNA, RNA, lipids, metabolites, and proteins [12]. Due to this cargo, they represent a promising cell-free therapeutic approach. However, the clinical application of natural exosomes is constrained by several limitations, notably low production yields, insufficient biological activity, and a lack of precise targeting capabilities [13]. Consequently, various strategies to modify native exosomes have been developed, giving rise to the field of engineered exosomes designed to augment their therapeutic efficacy [14]. In line with these efforts, we utilized exosome mimetics (EM). These nanovesicles, derived from cells, share key characteristics with natural exosomes but, as previously documented, offer significantly higher production yields and enhanced biological performance [15].

Macrophages are key cells in chronic inflammation and related pathologies, and activated macrophages are usually categorized into two phenotypes, M1 macrophages and M2 macrophages [16, 17]. Briefly, M1 macrophages are primarily involved in pro-inflammatory responses and M2 macrophages are primarily involved in anti-inflammatory responses [18]. Previous experimental evidence that tendon–bone interface healing after rotator cuff repair can be affected by modulating macrophage polarization in a permissive manner [19, 20]. It’s proved that M1 macrophage-derived exosomes can pro-inflammatory, enhance cell senescence and induce apoptosis [21, 22]. Therefore, whether M2 macrophage-derived exosome mimetics can promote the recovery of rotator cuff injury has become the focus of our concern.

Platelet-rich plasma (PRP) is an autologous or allogenic blood product obtained through centrifugation or apheresis to concentrate platelets within plasma. It is characterized by a high concentration of growth factors released from these platelets [23]. The therapeutic potential of PRP is largely attributed to its diverse array of biomolecules, such as key growth factors (GFs), cytokines, and chemokines, which collectively facilitate processes essential for tissue regeneration [24]. Consequently, PRP therapy has found broad clinical application across numerous fields, including orthopedics, dermatology, and the management of various musculoskeletal disorders [25–27]. Our prior research demonstrated its capacity for articular cartilage repair, achieved using a macroporous hydrogel scaffold composed of platelet lysate to sustainably recruit M2 macrophages from endogenous sources [28]. Furthermore, PRP exhibits immunomodulatory properties by attenuating M1 macrophage polarization while concurrently enhancing the M2 phenotype, underscoring its significant potential for managing inflammatory conditions [29]. However, the troublesome for fresh preparation, short half-life and costly production limit PRP’s in-depth applying dedication to clinics [30]. Due to the great potential of PRP to promote tissue repair and regeneration, PRP-preconditioning was obtained to explore the endogenous modification of M2 macrophages in order to enhance the ability of tendon–bone interface healing after rotator cuff repair.

To address these limitations, we developed PRP-pretreated M2 macrophages (PRP-M2-EM) for the therapy of tendon-bone interface healing, a bioengineered vesicle platform generated by combining PRP-mediated macrophage programming with exosome mimetic engineering technology. Compared with naturally secreted exosomes, exosome mimetics offer higher production efficiency, improved scalability, and enhanced functional customization. Importantly, PRP stimulation further enhanced the regenerative phenotype of M2 macrophages, endowing PRP-M2-EMs with superior immunomodulatory and pro-angiogenic capabilities. Comparative small RNA sequencing identified microRNA-21a-5p (miR-21a-5p) as being highly enriched in PRP-M2-EM compared to the M2-EM. Previous studies have shown that miR-21a-5p play a vital role in modulate macrophage polarization and angiogenesis in target in tissue inhibitor of metalloproteinases 3 (TIMP3) [31–33]. Based on this result and corroborating evidence from published work, we next sought to define the specific regulatory mechanism through which miR-21a-5p, when delivered via PRP-M2-EM, facilitates tendon–bone interface healing after rotator cuff repair.

In this investigation, M2-EM and PRP-M2-EM were synthesized using an extrusion methodology (Fig. 1A and B). Our findings indicate that PRP-M2-EM displays superior capabilities in macrophage polarization and angiogenesis regulation compared to M2-EM. This leads to a mitigated progression of tendon-bone interface healing in a rat rotator cuff repair model. Furthermore, we have elucidated the regulatory mechanism of PRP-M2-EM, revealing that PRP-M2-EM facilitates the delivery of miR-21a-5p to silence TIMP3, a crucial factor in the regulation of macrophage polarization and angiogenesis (Fig. 1C and D). In summary, our study posits that PRP-enhanced bioinspired exosome mimetics present a viable strategy for promoting the transition of macrophages from M1 to M2 and stimulating angiogenesis. Our findings further suggest that the therapeutic effects of PRP-M2-EMs may be partially associated with the miR-21a-5p/TIMP3 signaling axis, providing potential mechanistic insight into their role in promoting tendon–bone interface healing after rotator cuff repair.

Fig. 1.

(A) Schematic diagram of M2 macrophages polarization and PRP pretreated. (B) Schematic diagram of obtaining of M2-EM and PRP-M2-EM by extrusion method. (C) PRP-M2-EM facilitates polarization by delivering miR-21a-5p to suppress TIMP3 expression in macrophages. (D) PRP-M2-EM promoting angiogenesis by delivering miR-21a-5p to suppress TIMP3 expression in HUVECs

Results

Exosome mimetics isolation from M2 and PRP-M2

Tendon stem cell-derived exosomes promote tendon-to-bone healing of aged chronic rotator cuff tears by modulation of macrophage polarization [34]. However, exosomes secreted naturally present several challenges that limit their clinical application. The isolation and purification process of natural exosomes is complex, with low yield, and relies on cell culture conditions, resulting in high costs. Exosome mimetics can be produced on a large scale through chemical synthesis (such as liposomes) or cell membrane bionic technologies (such as compression and ultrasound methods) and are more likely to meet clinical needs [35–37]. By mechanically reorganizing the cell membrane by extrusion, the vesicles formed retain the lipid bilayer structure and some membrane proteins of natural exosomes with similar characteristics to exosomes but obtainable in large quantities. Therefore, we used the extrusion method to obtain EM for subsequent studies.

EM were harvested from M2 macrophages, including a group that had been pretreated with PRP. Red blood cell (RBC), platelet (PLT), and white blood cell (WBC) concentrations of the whole blood and PRP groups were summarized in Supplementary Table S1. PDGF, TGF-β1, VEGF, and IGF-1 were detected by ELISA (Figure S1). Subsequent purification was achieved by serial extrusion, first through a 5 μm and then a 1 μm polycarbonate membrane filter, culminating in a final step using a 100 kDa molecular weight cutoff centrifugal filter. (Fig. 2A). We characterized the physicochemical properties of PRP-M2-EM and compared them with M2-EM. Transmission electron microscopy (TEM) images of both EMs revealed an intact, round shape with a complete membrane structure (Fig. 2B).

Fig. 2.

Characterization of M2-EM and PRP-M2-EM. (A) Schematic diagram of obtaining of M2-EM and PRP-M2-EM by extrusion method. (B) Transmission electron microscopy images of M2-EM and PRP-M2-EM. Scale bar: 100 nm. (C) Assessment of exosome mimetics surface markers based on western blot analysis. (D) The size distribution of M2-EM and PRP-M2-EM analyzed by nanoparticle tracking analysis. (E and F) Analysis of the mean hydrodynamic diameter and surface charge (zeta potential) for both vesicle types, conducted using dynamic light scattering. (G and H) Quantification of the average protein content and particle yield prepared from 10⁶ cells. (I to L) Stability analysis of M2-EM and PRP-M2-EM in PBS and 10% FBS. (n = 3, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

Western blot analysis showed a marked increase in surface markers, including HSP70, TSG101 and CD63, on the isolated nanoparticles compared to RAW264.7 (Fig. 2C).

The size distribution of M2-EM and PRP-M2-EM, as determined by nanoparticle tracking analysis (NTA), showed mean diameters of 127.5 ± 12.1 nm and 124.7 ± 3.8 nm, respectively (Fig. 2D and E). These measurements align with the data obtained from TEM analysis. The zeta potential of both EMs are further demonstrated by dynamic light scattering (DLS) shown in Fig. 2F. The average zeta potential of M2-EMs was −21.6 mV, while that of PRP-M2-EMs was −20.5 mV.

The mean protein concentrations of EM from 10⁶ cells were 22.9 ± 2.7 µg/mL (M2-EM) and 25.7 ± 1.2 µg/mL (PRP-M2-EM) (Fig. 2G). The single cell yields of M2-EM and PRP-M2-EM were 8760 ± 960 particles (M2-EM) and 10,180 ± 896 particles (PRP-M2-EM) (Fig. 2H).

Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) detected the protein composition of M2, PRP-M2, M2-EM and PRP-M2-EM, showed most proteins were preserved in EM (Figure S2). The size of both M2-EM and PRP-M2-EM showed good stability after being immersed in PBS and 10% FBS for one week (Fig. 2I to L).

Furthermore, the CCK-8 assay results demonstrated that during the co-culture periods with M2-EM or PRP-M2-EM, the cell viability of RAW264.7 and HUVECs remained largely unchanged (Figure S3). This indicates that neither M2-EM nor PRP-M2-EM exerted any appreciable cytotoxic effects on these two cell types.

PRP-M2-EM promote M2 macrophages polarization

The overexpression of macrophages or their polarization to the M1 phenotype is capable of promoting apoptosis, inhibiting cellular proliferation, and triggering the excessive secretion of extracellular matrix by fibroblasts. The consequent outcome is peritendinous fibrosis alongside the development of scar tissue [38].

In addition, sustained inflammation initiates the release of multiple inflammatory factors, which subsequently obstruct the regeneration of the fibrocartilage enthesis and the remodeling process of intra-articular grafts [39]. Cui et al. discovered that Engineered exosomes derived from M2 macrophages facilitate immune modulation by converting macrophages from a pro-inflammatory state to an anti-inflammatory phenotype, thereby promoting macrophage polarization [40]. In a separate study utilizing a rabbit model, Jiang et al. found that a PRP-GelMA hydrogel facilitates osteochondral regeneration by inducing polarization toward the M2 macrophage phenotype [41]. Building upon the aforementioned research, we aim to delve into the potential of PRP-M2-EM in modulating macrophage polarization.

RAW264.7 were treated with DID-labeled M2-EM and PRP-M2-EM to investigate their endocytosis. Fluorescence microscopy confirmed the successful cellular uptake of both M2-EM and PRP-M2-EM, with each type of vesicle observed to accumulate primarily in the perinuclear region (Fig. 3A). A quantitative assessment of fluorescence intensity indicated that the internalization efficiency did not differ significantly between the two vesicle groups.

Fig. 3.

PRP-M2-EM promotes macrophage polarization in vitro. (A) DiD-labeled M2-EM and PRP-M2-EM uptake by RAW264.7. Scale bar: 100 μm. (B and C) Western blot detection and quantitative analysis of iNOS and Arginase-1. (D) qRT-PCR analysis of iNOS, CD86, CD206, or Arg-1 gene expression of RAW264.7 (E and F) Effect of M2-EM and PRP-M2-EM on the expression of CD163 and Arg-1 on RAW264.7 via immunofluorescence staining. Scale bar: 100 μm. (n = 3, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0. 01; ***p < 0.001)

To assess the capacity of these vesicles to induce a phenotypic shift in macrophages, RAW264.7 cells were subjected to one of four conditions: phosphate-buffered saline (PBS), lipopolysaccharide (LPS) at 100 ng/mL, LPS+M2-EM (50 µg/mL), or LPS + PRP-M2-EM (50 µg/mL). Analysis via western blot demonstrated that LPS stimulation elevated the expression of iNOS, a marker for the M1 phenotype, while concurrently reducing the expression of Arginase-1, a characteristic M2 marker (Fig. 3B and C). These findings were corroborated by real-time PCR, which indicated that LPS treatment upregulated the mRNA expression of CD86 and iNOS but downregulated the transcripts for CD206 and Arginase-1 (Fig. 3D).

M2-EM can reverse the protein expression increased in iNOS and the protein expression decreased in Arg1 caused by LPS, and PRP-M2-EM shows a more significant effect (Fig. 3D). Meanwhile, M2-EM and PRP-M2-EM downregulated the mRNA levels of CD86 and iNOS and upregulated the mRNA levels of CD206 and Arg1(Fig. 3D).

Immunofluorescence analyses further confirmed that LPS stimulation upregulated the expression of M1 macrophage surface marker CD86 in RAW264.7 cells (Fig. 3E). However, M2-EM and PRP-M2-EM treatment reduced the LPS-induced upregulation of CD86 (Fig. 3E). Moreover, M2-EM and PRP-M2-EM upregulated the expression of the M2 macrophage marker CD163, with PRP-M2-EM showing a more pronounced effect (Fig. 3F).

The polarization of macrophages in different treatment groups was evaluated using flow cytometry analysis (Figure S4). Compared with the control group, LPS stimulation significantly increased the expression of CD86, indicating successful induction of the inflammatory phenotype. Partial reversal of the LPS-induced changes was achieved using M2-EM treatment, while PRP-M2-EM treatment exerted a more significant regulatory effect on macrophage polarization and increased the expression of CD206. Specifically, the proportion of M2 macrophages significantly increased, and compared with the LPS group and the LPS + M2-EM group, the number of M1 macrophages in the LPS + PRP-M2-EM group was significantly reduced.

Based on the above results, M2-EM promoted the polarization of macrophages from a pro-inflammatory phenotype toward an anti-inflammatory phenotype. Furthermore, these findings suggest that PRP-M2-EM exhibits enhanced immunomodulatory activity and more effectively facilitates macrophage polarization toward a reparative phenotype.

PRP-M2-EM promote angiogenesis of endothelial cells in vitro

To further evaluate the pro-angiogenic effects of PRP-M2-EM, we examined their uptake by HUVECs. The exosome mimetics were labeled with DiD (violet) and nuclei were counterstained with DAPI (blue). A distinct violet fluorescent signal was detected within the cytoplasmic region of HUVECs, indicating successful cellular internalization of both M2-EM and PRP-M2-EM (Fig. 4A). Quantitative analysis of fluorescence intensity showed comparable uptake levels between the two formulations.

Fig. 4.

PRP-M2-EM promotes angiogenesis in vitro. (A) Cellular uptake of DiD-labeled M2-EM and PRP-M2-EM by HUVECs, visualized via fluorescence microscopy, with accompanying quantitative assessment. Scale bar: 100 μm. (B and E) Western blot detection and quantitative analysis of VEGF. (C and D) Migratory potential was assessed using a transwell chamber assay, followed by quantitative measurement of migrated cells. Scale bar: 100 μm. (F to H) Angiogenic tube formation examined in HUVECs, with the corresponding quantitative analyses. Scale bar: 100 μm. (I and J) Cell scratch assay and cell migration rate of HUVECs cells at 0, 12 and 24 H. Scale bar is 100 μm. (K and L) qRT-PCR analysis of ANG1 and VEGF gene expression of HUVECs. (n = 3, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

Subsequently, to assess the ability of these vesicles to promote angiogenesis and the potential enhancing effect of PRP pre-treatment, HUVECs were exposed to LPS (100 ng/mL), M2-EM (50 µg/mL), or PRP-M2-EM (50 µg/mL) under various experimental conditions.

Western blot assay results suggested the expression of VEGF (important in vasculogenesis and angiogenesis) decreased after LPS incubated, and reversed by both EM, which PRP-M2-EM performed better (Fig. 4B and E).

The migratory capacity of HUVECs, as evaluated by transwell assay, was markedly reduced following LPS administration relative to the control. This inhibitory effect was counteracted by the introduction of EM. Notably, PRP-M2-EM exhibited a more potent capacity to enhance HUVECs migration than M2-EM (Fig. 4C and D).

In the tube formation assay, the LPS-treated group displayed a significant reduction in the complexity of the capillary network, evidenced by decreases in the tube formation (Fig. 4F), the total number of nodes (Fig. 4G), and total tube length (Fig. 4H) compared to the control group. The impairment of this angiogenic capability was effectively mitigated by treatment with M2-EM, with PRP-M2-EM producing an even more substantial restorative effect. A statistically significant difference was observed in both branch points and total length between LPS group and PRP-M2-EM group.

Consistent with these functional data, gene expression analysis revealed that LPS suppressed the mRNA expression of the pro-angiogenic factors VEGF and ANG1 (Fig. 4K and L). Both types of EM were able to counteract this LPS-induced downregulation. Importantly, PRP-M2-EM was significantly more effective at upregulating the expression of VEGF and ANG1 than M2-EM.

The impact on cellular migration was further confirmed via a scratch wound healing assay. Representative images captured at various time points post-treatment are shown in Fig. 4I. After 12 h, the wound closure in EM-treated groups was more advanced than in the LPS group. By 24 h, the scratch width in the PRP-M2-EM group was nearly identical to that of the control group, indicating a rescue of the migratory deficit (Fig. 4I and J).

Collectively, these in vitro findings indicate that PRP-M2-EM potently enhances angiogenic processes, an effect likely mediated through the pronounced upregulation of key angiogenic genes including VEGF and ANG1.

PRP-M2-EM improved tendon–bone interface healing

To investigate the therapeutic potential of M2-EM and PRP-M2-EM under physiological conditions, a rat model of rotator cuff injury was established. Surgically, the supraspinatus tendon (SST) was penetrated with a 5–0 PDS suture and subsequently transected. Following debridement of the fibrocartilaginous region and a portion of the subchondral bone at the insertion, a bone tunnel was drilled laterally across the proximal humerus using a 22-gauge needle. The PDS suture was then threaded through this tunnel to reattach the tendon to its anatomical insertion. Finally, the deltoid muscle and skin were closed in layers. 81 10-week-old Sprague-Dawley rats were randomly allocated into three experimental cohorts: a PBS control group, an M2-EM treatment group, and a PRP-M2-EM treatment group. Treatments consisted of weekly right shoulder joint cavity injections of either 500 µg of the respective vesicles suspended in 500 µL PBS, or PBS vehicle alone. Post-operatively, at 1 week, three animals from each group were euthanized, while at 4 and 8 weeks, twelve animals from each group were euthanized, and the supraspinatus tendon–humerus complex was harvested for analysis. (Fig. 5A).

Fig. 5.

Histological analysis of rat rotator cuff tendon and its insertion into the humerus after treated. (A) Schematic diagram of the experimental design for assessing the macrophage polarization and angiogenesis effects of M2-EM and PRP-M2-EM in rotator cuff repaired rat model. (B) Representative H&E staining images of humerus post-surgery at 4 and 8 weeks. (C) Representative Masson staining images of humerus post-surgery at 4 and 8 weeks. (D) Representative Safranin O/Fast green staining images of humerus post-surgery at 4 and 8 weeks. (E) Histological scores of repaired rotator cuffs in different treatment groups. (F) Tendon-bone interface maturing score of different treatment groups. (G) VEGF immunostaining images of humerus at tendon-bone interface at 4 weeks post-surgery. (H) Representative images of immunofluorescence staining (n = 3) of iNOS at tendon-bone interface at 1week post-surgery. (I) The macroscopic view of the biomechanical test. The statistic results of (I) Peak failure load, (J) Stiffness and (K) Energy to Yield load at 4 and 8 weeks. Scale bars, 250 μm (white) or 100 μm (black). (n = 6, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

H&E staining images at 4 weeks showed the degree of inflammatory infiltration from M2-EM and PRP-M2-EM was lower than that of the PBS groups, and at 8 weeks PRP-M2-EM group exhibited denser and more organized collagen fiber between tendon and bone (Fig. 5B).

The results of Masson’s staining showed the blank group exhibited higher cell density and a more disorganized cell arrangement than those of the two experimental groups at 4 weeks, and tendon collagen fibers in the blank group appeared more disorganized. And the collagen arrangement in treatment groups was more orderly and the structure more evacuated than NS group at 8 weeks (Fig. 5C). Based on H&E and Masson’s staining, histological evaluation revealed markedly superior scores in the PRP-M2-EM treatment group relative to all other groups (Fig. 5E). Enhanced formation of chondrocytes and fibrocartilage at the tendon–bone interface serves as a key indicator of improved healing [42, 43]. Safranin-O staining further demonstrated that both M2-EM and PRP-M2-EM promoted the deposition of chondrocytes and fibrocartilaginous tissues at this interface. Notably, the PRP-M2-EM group exhibited a stronger pro-chondrogenic effect, with significantly greater tissue formation observed at both 4 and 8 weeks (Fig. 5C). When assessed using established maturation criteria, this group also displayed significantly higher tendon maturation scores at these time points compared to the other groups (Fig. 5F).

Immunohistochemical analysis for VEGF revealed a higher number and density of newly formed blood vessels in the PRP-M2-EM group than in the control group (Fig. 5G).

To evaluate local inflammation, immunofluorescence staining for iNOS was performed. The results indicated substantially lower iNOS expression in the PRP-M2-EM group one week after surgery compared to the PBS group (Fig. 5H), suggesting that PRP-M2-EM enhances bone regeneration at the healing interface by attenuating inflammatory responses.

In in vitro experiments, rotator cuff tissue in the PRP-M2-EM group demonstrated less inflammation and more neovascularization at 4 weeks, and superior tendon regeneration and more orderly collagen fiber alignment at 8 weeks. This may be related to the fact that the PRP-M2-EM group had more bioactive factors.

To evaluate the biomechanical properties of tendon–bone healing after rotator cuff repair, the humerus–supraspinatus tendon complex was harvested at 4 and 8 weeks postoperatively and subjected to biomechanical testing using a universal material testing system (Fig. 5I). Peak failure load (Fig. 5J), stiffness (Fig. 5K), energy to failure (Fig. 5L), and failure mode were analyzed to assess the quality of tendon–bone interface healing at different time points. In the PBS group, failure at the tendon–bone interface was more frequently observed at both 4 and 8 weeks postoperatively. In contrast, tendon rupture was predominantly observed in all treatment groups at 8 weeks, indicating improved mechanical integrity and enhanced tendon–bone interface healing following treatment. Biomechanical evaluation demonstrated that the PRP-M2-EM treatment group exhibited significantly improved peak failure load (Fig. 5J), stiffness (Fig. 5K), and energy to failure (Fig. 5L) compared with the other groups at 8 weeks, indicating enhanced tendon–bone interface healing and superior mechanical integrity.

miR-21a-5p is enriched in PRP-M2-EM

The therapeutic potential of extrusion-prepared EM originates from its capacity to carry specific microRNAs (miRNAs), functioning in a manner similar to exosomes [44]. Our investigation into the mechanism through which PRP-M2-EM suppresses inflammatory responses and enhances vascularization in rotator cuff damage involved a comparative analysis of miRNA expression. This was accomplished by sequencing and computationally analyzing the miRNAs present in both M2-EM and PRP-M2-EM.

PCA revealed that PC1 and PC2 accounted for 37.73% and 23.38% of the total variance, respectively (Fig. 6A). Notably, the M2-EM and PRP-M2-EM groups exhibited significant separation along the PC1 axis, indicating that PRP pretreated had a substantial impact on gene expression. The PC2 axis may reflect minor differences within or between groups, and no significant outliers are observed, confirming the high quality of the data.

Fig. 6.

miR-21a-5p is enriched in PRP-M2-EM. (A) Principal Component Analysis (PCA) of M2-EM and PRP-M2-EM. (B) Volcano plot illustrating the distribution of downregulated and upregulated miRNAs when comparing M2-EM and PRP-M2-EM. (C) Differential miRNA expression profiles between M2-EM and PRP-M2-EM visualized using heat map. (D) Diagram summarizing the anticipated target genes associated with the five most prevalent miRNAs showing increased abundance. (E) Enrichment results of biological processes linked to these genes, based on Gene Ontology (GO) analysis. (F) Pathway analysis of the above genes using KEGG annotation. (G) Quantitative measurement of miRNA expression levels in RAW264.7 cells following treatment with M2-EM and PRP-M2-EM, as determined by qRT-PCR. (n = 3, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

As illustrated by a volcano plot, a comparison with the M2-EM group identified 10 miRNAs that were significantly upregulated and 26 that were significantly downregulated in the PRP-M2-EM group (Fig. 6B).

The analysis of miRNA sequencing data yielded a heatmap visualizing differential expression, showcasing miRNAs that were upregulated, downregulated, or remained statistically unchanged when comparing PRP-M2-EM to M2-EM (Fig. 6C). Our subsequent research focused on the markedly upregulated miRNAs within PRP-M2-EM, leading to the selection of the top five most abundant candidates—mmu-let-7d-5p, mmu-miR-340-5p, mmu-let-7f-5p, mmu-miR-21a-5p, and mmu-miR-1b-5p—for further study. The target genes associated with these five miRNAs are displayed in Fig. 6D and provided the basis for ensuing Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses.Results from the biological process (BP) segment of the GO analysis indicated that the miRNAs’ associated target genes were chiefly involved in regulating transcription by RNA polymerase II, regulation of DNA-templated transcription, cell differentiation, regulation of gene expression, regulation of cell migration, protein polyubiquitination and roof of mouth development, among others (Fig. 6E). Molecular function (MF) of GO analysis exhibited that these five up-regulated miRNAs related target genes were closely related to protein binding, DNA binding, DNA-binding transcription factor activity, RNA polymerase II cis-regulatory region sequence-specific DNA binding, DNA-binding transcription factor activity, and chromatin binding, among others (Figure S5A). Cell component (CC) terms of GO analysis revealed an enrichment of the predicted miRNA target genes within several key locations, including the nucleus, cytoplasm, nucleoplasm, cytosol, and chromatin (Figure S5B). The KEGG pathway enrichment results identified the MAPK signaling pathway among the top ten most relevant pathways, suggesting its potential significant role in both macrophage polarization and angiogenesis (Fig. 6F).

To evaluate the transfer efficiency of miRNAs from the vesicles, RAW 264.7 cells were co-cultured with M2-EM and PRP-M2-EM, followed by quantification of the levels of five specific miRNAs (let-7d-5p, miR-340-5p, let-7f-5p, miR-21a-5p, and miR-1b-5p). Notably, miR-21a-5p expression was the most markedly elevated in cells treated with PRP-M2-EM, making it a primary candidate for further study (Fig. 6G).

This finding is particularly significant in light of prior research which has established that miR-21a-5p is implicated in modulating the MAPK signaling pathway and plays a critical part in polarizing macrophages and promoting blood vessel formation. Consequently, the transfer of miR-21a-5p by PRP-M2-EM is proposed as a likely mechanism for its regulation of these cellular processes.

miR-21a-5p in PRP-M2-EM regulates macrophages polarization and angiogenesis in vitro

While miR-21a-5p has been previously implicated in macrophage polarization and the regulation of angiogenesis in endothelial cells [45, 46], its specific function within the context of rotator cuff injury remains unexplored.

To elucidate the pivotal role of miR-21a-5p in modulating these processes, we employed both mimic and inhibitor oligonucleotides. RAW264.7 and HUVECs were transfected with NC-mimic (50 nM), miR-21a-5p-mimic (50 nM), NC-inhibitor (100 nM) and miR-21a-5p-inhibitor (100 nM) in different groups. Following a 48-h incubation period, the cells were collected for further analysis. Subsequently, the transfected cells in the different experimental groups received additional treatment with either LPS (100 ng/mL), PBS, M2-EM (1000 µg/mL), or PRP-M2-EM (1000 µg/mL).

Immunofluorescence analysis revealed that LPS stimulation significantly upregulated the M1 macrophage marker CD86 while downregulating the M2 marker CD206 in RAW264.7 cells, indicating a shift toward pro-inflammatory polarization (Fig. 7A and B). However, treatment with PRP-M2-EM effectively reversed this trend, attenuating the LPS-induced increase in CD86 expression and promoted the expression of CD206 (Fig. 7A and B). Strikingly, the anti-inflammatory impact of PRP-M2-EM was partially abolished by miR-21a-5p inhibitor, which reinstated elevated CD86 levels, suggesting miR-21a-5p is critical for PRP-M2-EM’s M1-suppressive action (Fig. 7A and B). Conversely, miR-21a-5p mimic synergized with PRP-M2-EM to further enhance the M2 marker CD206, amplifying the shift toward M2 polarization and reinforcing the attenuation of inflammation (Fig. 7A and B). Together, these data implicate miR-21a-5p as a key mediator of PRP-M2-EM’s ability to reprogram macrophage polarization from LPS-induced M1 to M2 phenotypes.

Fig. 7.

miR-21a-5p in PRP-M2-EM regulates macrophage polarization and angiogenesis in vitro. (A) The expression of CD86 (M1 macrophage marker) on RAW264.7 treated with NC-inhibitor, miR-21a-inhibitor, LPS and PRP-M2-EM respectively was examined via immunofluorescence staining. Scale bar: 100 μm. (B) The expression of CD206 (M2 macrophage marker) on RAW264.7 under different treatments examined via immunofluorescence staining. Scale bar: 100 μm. (C–F) Transwell assays were performed to evaluate HUVEC migration under different treatment conditions, along with quantitative analysis of migrated cells. Scale bar = 100 μm. (n = 3, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

The migration ability of HUVECs was evaluated by using the transwell experiment. Compared with the control group, the migration ability of HUVECs was weakened by LPS treatment (Fig. 7C to F). The incubation of PRP-M2-EM reversed the migration ability of HUVECs blocked by LPS, while miR-21a-5p-inhibitor partially eliminated the migration promoting effect of PRP-M2-EM (Fig. 7C to F). Furthermore, miR-21a-5p-mimic enhanced the effect of PRP-M2-EM in blocking the inhibition of migration ability induced by LPS. Compared with treatment with PRP-M2-EM alone, the number of migrating endothelial cells was greater (Fig. 7C to F). To further investigate whether VEGF signaling is involved in the therapeutic mechanism of PRP-M2-EM, VEGF protein expression was evaluated following modulation of miR-21a-5p expression (Figure S6). Western blot analysis showed that LPS stimulation significantly reduced VEGF expression compared with the control group, whereas PRP-M2-EM treatment partially restored VEGF protein levels. Furthermore, overexpression of miR-21a-5p further enhanced VEGF expression in PRP-M2-EM-treated cells, while inhibition of miR-21a-5p attenuated the PRP-M2-EM-mediated upregulation of VEGF. These findings suggest that PRP-M2-EM may regulate VEGF signaling through the miR-21a-5p, thereby contributing to its pro-regenerative effects. To sum up, these results indicate that miR-21a-5p in PRP-M2-EM may promote angiogenesis in HUVECs.

The findings indicate that a high concentration of miR-21a-5p within PRP-M2-EM contributes to the modulation of macrophage polarization and the promotion of angiogenic processes. This research validates the critical function of miR-21a-5p in coordinating these dual mechanisms during tendon–bone interface healing after rotator cuff repair, thereby opening novel therapeutic avenues for treating this and other disorders through miRNA-mediated strategies.

PRP-M2-EM delivers miR-21a-5p to silence the target gene TIMP3 to function in RAW264.7 and HUVECs

The primary mechanism of action for miRNAs involves the suppression of their downstream target mRNAs. Existing literature indicates that miR-21a-5p plays a critical part in polarizing macrophages and promoting angiogenesis, primarily through its regulation of Tissue Inhibitor of Metalloproteinase 3 (TIMP3) expression. TIMP3 functions by obstructing the interaction between VEGF and its receptor, VEGFR-2, thereby inhibiting subsequent signaling cascades and the formation of new blood vessels [47]. Furthermore, prior investigations have demonstrated that the miR-21a-5p/TIMP3 axis is indispensable for neuroprotection, particularly in the context of modulating microglia and macrophage polarization [31].

Therefore, subsequent experiments were designed to determine if PRP-M2-EM mediates its effects by transferring miR-21a-5p to suppress TIMP3 expression and modulate function in RAW264.7 macrophages and HUVECs. Supporting this hypothesis, TargetScan predictions identified an evolutionarily conserved specific binding sequence for miR-21a-5p on the TIMP3 3′ UTR in both mouse and human genomes (Figure S7 and S8).

The functional role of miR-21a-5p was evaluated by transfecting these cell lines with either negative control or specific miR-21a-5p mimic and inhibitor oligonucleotides to analyze subsequent changes in TIMP3 expression.

Western blot analysis demonstrated that transfection with a miR-21a-5p mimic significantly suppressed TIMP3 protein levels in both RAW264.7 and HUVECs. Conversely, introduction of the miR-21a-5p inhibitor markedly elevated TIMP3 expression (Fig. 8A to D), confirming TIMP3 as a direct target of this miRNA. Subsequent investigation assessed the influence of M2-EM and PRP-M2-EM on TIMP3 regulation.

Fig. 8.

PRP-M2-EM mediates the suppression of the target gene TIMP3 in RAW264.7 and HUVECs through the delivery of miR-21a-5p. (A and B) Protein expression levels of TIMP3 in RAW264.7 transfected with NC-mimic, NC-inhibitor, miR-21a-5p-mimic, or miR-21a-5p-inhibitor were evaluated by western blot and quantified by densitometric analysis. (C and D) Protein expression levels of TIMP3 in HUVECs transfected with different treatments were evaluated by western blot and quantified by densitometric analysis. (E and F) Western blot and corresponding quantitative evaluation of TIPM3 levels in RAW264.7 treated with LPS, M2-EM, or PRP-M2-EM. (G and H) Western blot and corresponding quantitative evaluation of TIPM3 levels in HUVECs treated with different treatments. (I) qRT-PCR analysis of TIMP3 gene expression of RAW264.7 in different treatment. (J) qRT-PCR analysis of TIMP3 gene expression of HUVECs in different treatment. (K) The

The data further revealed that LPS stimulation notably increased TIMP3 expression. This induced expression was attenuated by treatment with either M2-EM or PRP-M2-EM, with the latter demonstrating a significantly more potent suppressive effect on TIMP3 protein levels (Fig. 8E to H).

Further, real-time PCR results in both HUVECs and RAW264.7 showed same results as Western blot (Fig. 8I and J).

Bioinformatic analysis using TargetScan predicted a conserved binding site for miR-21a-5p within the 3′-UTR of TIMP3 (Fig. 8K). To validate the direct interaction between miR-21a-5p and TIMP3, a dual-luciferase reporter assay was performed using wild-type (WT) and mutant (MUT) TIMP3 3′-UTR constructs. As shown in Fig. 8L, co-transfection of the miR-21a-5p mimic significantly reduced the luciferase activity of the WT TIMP3 reporter compared with the negative control group. In contrast, mutation of the predicted miR-21a-5p binding site abolished this inhibitory effect, as no significant difference in luciferase activity was observed between the MUT groups. These findings demonstrate that miR-21a-5p directly binds to the TIMP3 3′-UTR and negatively regulates TIMP3 expression.

Collectively, these findings suggest that PRP-M2-EM plays a critical role in modulating macrophage polarization and stimulating angiogenesis, potentially through the delivery of miR-21a-5p and subsequent suppression of TIMP3 expression.

binding site of TIMP3 and miR-21a-5p predicted by TargetScan. (L) Double luciferase gene reporting assay verified the binding sites of TIMP3 and miRNA-21a-5p. (n = 3, mean ± SD; *p < 0.05; **p < 0.01; ***p < 0.001)

Antagomir-21a reverses the effects of PRP-M2-EM in regulation of polarization and angiogenesis in rotator cuff injury rats

To evaluate the involvement of miR-21a-5p in the therapeutic effects of PRP-M2-EM on rotator cuff injury in vivo, we administered antagomir-21a in a rat model. Specifically, a miR-21a-5p antagomir (5 nmol) or a negative control antagomir (5 nmol) was injected into the right shoulder joint cavity of anesthetized rats using a microinjection system to suppress miR-21a-5p expression. This intervention was performed on the first day of each week for the initial three weeks following surgical reconstruction. Concurrently, PRP-M2-EM (500 µg in 500 µL PBS) was administered into the right shoulder joint cavity for the first three days of each week during this initial three-week period. Subsequently, PRP-M2-EM injections were continued once a week for an additional three weeks. Control animals received joint cavity injections of PBS, the vehicle solution for PRP-M2-EM, following the same schedule. (Fig. 9A).

Fig. 9.

Antagomir-21a-5p counteracts the therapeutic benefits of PRP-M2-EM by impairing macrophage polarization and angiogenesis in rotator cuff injury model. (A) Schematic representation of the experimental strategy. (B) Representative H&E staining images of humerus post-surgery. (C) Representative Masson staining images of humerus post-surgery. (D) Representative Safranin O/Fast green staining images of f humerus post-surgery. (E) Histological scores of repaired rotator cuffs in different treatment groups at 8 weeks. (F) Bone-tendon interface maturing score of different treatment groups at 8 weeks. (G) VEGF immunostaining images of humerus at tendon-bone interface at 4 weeks post-surgery. (H) Representative images of immunofluorescence staining (n = 3) of Arg-1 at tendon-bone interface at 1week post-surgery. The statistic results of (I) Peak failure load, (J) Stiffness and (K) Energy to Yield load at 8weeks. Scale bars, 250 μm (white) or 100 μm (black). (n = 6, mean ± SD; ns, no significant difference; *p < 0.05; **p < 0.01; ***p < 0.001)

H&E staining images at 8 weeks showed PRP-M2-EM effectively reduced the inflammatory cell infiltration after rotator cuff reconstruction, while the improvement was counteracted by antagomir-21a (Fig. 9B).

Masson staining results at 8 weeks indicated that, compared with the control group, the PRP-M2-EM group exhibited a more regular arrangement of collagen fibers, along with a looser structure. At the tendon-bone junction, there was an increase in chondrocyte production and fibrocartilage formation, suggesting enhanced tendon–bone interface healing after rotator cuff repair (Fig. 9C).

According to H&E staining and Masson’s staining, the histological scores of the Antagomir-21a-5p group were significantly lower than PRP-M2-EM group at 8 weeks (Fig. 9E).

Safranin-O staining results of the 8th week showed that PRP-M2-EM promoted the growth of chondrocytes and fibrocartilage tissue at the tendon-bone junction. However, antagomir-21a weakened the effect mediated by PRP-M2-EM (Fig. 9D).

According to the maturation score criteria, the Antagomir-21a-5p group showed significantly lower tendon maturation at 8 weeks compared to the PRP-M2-EM group (Fig. 9F).

The number and density of newly formed blood vessels in the PRP-M2-EM groups were higher than other groups, as shown by the VEGF immunohistochemical images (Fig. 9G).

To assess the inflammatory status at the tendon–bone interface one week after surgery, immunofluorescence staining was performed for Arg-1. The results, presented in Fig. 9H, revealed a notable reduction in Arg-1 expression in the Antagomir-21a-5p group compared to the PRP-M2-EM group.

Biomechanical analysis demonstrated that the PRP-M2-EM + antagomir-NC group exhibited significantly enhanced mechanical properties compared with the PBS + antagomir-NC group, including increased Peak failure load (Fig. 9I), stiffness (Fig. 9J), and energy to yield load (Fig. 9K). However, inhibition of miR-21a-5p by antagomir-21a-5p partially abolished these beneficial effects. These results suggest that miR-21a-5p plays an important role in PRP-M2-EM-mediated enhancement of tendon–bone healing.

These results suggest that antagomir-21a can partially counteract the impacts of PRP-M2-EM on macrophage polarization and angiogenesis regulation following rotator cuff reconstruction. This finding further validates that PRP-M2-EM is capable of transferring miR-21a-5p to the rotator cuff of rats, promoting M2 polarization and angiogenesis in vivo.

Discussion

In our study, exosome mimetics extracted by extrusion of M2 macrophages were shown to have an effect on promoting the proliferation, migration, and angiogenesis of HUVECs. And it has the effect of promoting macrophage polarization to M2 type. And, by pretreating M2 macrophages with PRP, we found that PRP-M2-EM gained a stronger capacity and accelerated tendon–bone interface healing after rotator cuff repair in rat model. By miRNA-seq analysis, we found that the therapeutic effect of PRP-M2-EM might be associated with miR-21a-5p delivered by EM. It was shown by subsequent in vivo and in vitro experiments that miR-21a-5p may exert its biological effects by targeting TIMP3.

In recent years, exosome mimetics, as nanoscale vesicles secreted by cells, have shown great potential in disease treatment. Mesenchymal stem cell exosomes promote cardiac repair after myocardial infarction [48]. With engineered modifications, exosomes mimetics can precisely deliver drugs to diseased tissues [49]. However, conventional exosome extraction yields are low and new extraction methods are needed to improve efficiency. Extrusion, a technique that reconfigures cell membranes to form exosome-like vesicles through physico-mechanical forces, has shown unique advantages in exosome mimetics extraction and engineering modification in recent years [37]. Platelet-Rich Plasma (PRP) is derived from autologous blood through centrifugation, producing a highly concentrated platelet solution containing a high concentration of growth factors (GFs) and cytokines. This bioactive formulation has the potential to enhance tissue repair, regulate inflammatory responses, and support regenerative processes during tendon–bone interface healing after rotator cuff repair [50–52]. However, the GFs (e.g., PDGF, TGF-β) released by PRP are degraded within hours in vivo, which makes it difficult to maintain a sustained effect. PRP still faces many problems in its clinical application, and optimization of the preparation process is needed in the future to realize its clinical potential. The ability to inhibit excessive inflammation and promote tissue regeneration contained in M2 macrophage exosomes gives them a unique potential in the treatment of disease and has attracted our attention to their use in the healing of rotator cuff injuries.

The healing process after rotator cuff injury is highly dependent on the precise modulation of the inflammatory response [53], but excessive or persistent inflammation can lead to repair failure and the formation of fragile scar tissue rather than a functional tendon-bone interface [54]. Moderate inflammation activates tendon stem cells and angiogenesis, but excessive inflammation exacerbates oxidative stress, leading to apoptosis and early repair failure [55]. Early in the healing phase of rotator cuff injuries, M1 macrophages infiltrate and act as inflammatory mediators, and a reduction in M1 macrophage accumulation helps promote tendon–bone interface healing after rotator cuff repair [56]. A regulated inflammatory milieu facilitates the recruitment of functionally active cells and establishes a critical foundation for ensuing tissue regeneration processes [57]. Following this, polarization of M2 macrophages leads to the secretion of TGF-β3 and IGF-1, factors known to suppress fibrotic processes while enhancing collagen production. Our findings indicate that PRP-M2-EM suppressed LPS-induced M1 polarization in RAW264.7 cells and concurrently enhanced their differentiation toward an M2 phenotype. Consistent with these in vitro results, animal studies showed that PRP-M2-EM administration effectively reduced early postoperative infiltration of M1 macrophages.

The rotator cuff tendon, as a low-vascularized tissue, is highly dependent on the process of angiogenesis for repair after injury, and moderate neovascularization is essential for healing. Prolonged ischemia leads to disorganized tendon collagen, fatty infiltration, poor healing ability in sparsely vascularized areas, and a high rate of postoperative re-tear [58]. Neovascularization promotes healing by secreting factors such as PDGF and SDF-1 and recruiting tendon stem cells and mesenchymal stem cells to the injury site [59]. Our study investigated the pro-angiogenic effect of M2-EM on HUVECs and found that M2-EM has the ability to promote angiogenesis, cell migration, and neovascularization of tendon-bone surfaces, which promotes tendon healing.PRP-M2-EM obtained from PRP-pretreated M2 macrophages exerted better effects.

Acting as integral components of exosome mimetics, miRNAs post-transcriptionally modulate gene expression through binding to the 3′-untranslated regions of target mRNAs, participating in numerous physiological and pathological processes including cellular differentiation, metabolic regulation, and immune modulation [60–62]. MicroRNAs (miRNAs), which function as essential constituents of exosome mimetics, regulate gene expression post-transcriptionally by binding to the 3′-untranslated regions of target mRNAs. They participate in a wide range of physiological and pathological processes, such as cell differentiation, metabolic homeostasis, and immune response modulation [63, 64]. Our investigation revealed that miR-21a-5p, which was highly abundant in PRP-M2-EM, enhanced the polarization of RAW264.7 cells toward an M2 macrophage phenotype while attenuating LPS-induced M1 polarization. It also stimulated angiogenic activity in HUVECs. Due to the known role of miRNAs in post-transcriptional regulation, we further explored the underlying mechanisms and identified TIMP3 as a putative target gene of miR-21a-5p.

Tissue inhibitor of metalloproteinases 3 (TIMP3) is an endogenous inhibitor of matrix metalloproteinases (MMPs) and performs vital functions in various physiopathological contexts by controlling extracellular matrix (ECM) turnover, angiogenesis, and inflammatory processes [65–67]. These mechanisms are also critically involved in tendon repair. In the current study, we found that miR-21a-5p may negatively regulate TIMP3 expression. Overexpression of miR-21a-5p was associated with enhanced M2 macrophage polarization and angiogenesis, potentially through suppression of TIMP3. KEGG pathway analysis further suggested that TIMP3 may be involved in the MAPK signaling pathway; however, the precise mechanisms require further investigation.

Our results demonstrate that the miR-21a-5p carried by PRP-M2-EM confer enhanced biological functionality, promoting macrophage polarization and angiogenesis to further promote tendon-bone interface healing. Unlike living stem cells, exosome mimetics lack an intact nucleus and do not possess proliferative capacity, which may reduce the potential risk of tumor formation and contribute to a favorable safety profile. Exosome mimetics obtained by the push-push method improved production yield while retaining the key cargoes of exosomes, thereby providing a cost-effective alternative to conventional exosome-based therapies. PRP pretreatment provided M2 macrophages with richer bioactive substances, solved the problem of rapid PRP degradation and high preparation cost, and improved the efficiency of PRP utilization. These attributes render PRP-M2-EM a more favorable therapeutic approach, offering a novel perspective for the treatment of tendon–bone interface healing after rotator cuff repair.

Nevertheless, several limitations still remain. First, this study was conducted in a rat rotator cuff repair model, which may not fully recapitulate the complex biomechanical environment and healing characteristics observed in humans. Second, although PRP-derived products exhibit promising therapeutic potential, variability in PRP composition between different preparations may affect treatment consistency and reproducibility. In addition, the long-term biodistribution, immunogenicity, and optimal dosing regimen of PRP-M2-EM were not systematically evaluated in the current study. Furthermore, although our results suggest that the miR-21a-5p/TIMP3 axis may participate in regulating macrophage polarization and angiogenesis, the mechanistic interpretation should remain cautious given the complexity of macrophage signaling networks and the possible involvement of multiple interconnected pathways. Finally, the current follow-up period was relatively limited, and longer-term studies, particularly in large-animal models, are still required to further evaluate the durability, biosafety, and translational potential of this therapeutic strategy.

Conclusion

Employing an extrusion-based methodology, we successfully generated M2 macrophage-derived exosome mimetics (M2-EM) and platelet-rich plasma-modified exosome mimetics (PRP-M2-EM). This biomimetic engineering approach effectively encapsulates the regenerative properties of PRP within a stable, nanosized vesicular system. Notably, our results indicate that pre-treatment of M2 macrophages with PRP yields exosome mimetics with an enhanced capacity to induce M2 macrophage polarization and stimulate angiogenesis during rotator cuff repair. Further investigation identified significant enrichment of miR-21a-5p in PRP-M2-EM, and we found that these vesicles exerted significant therapeutic effects by delivering miR-21a-5p, which might function by inhibiting the expression of TIMP3. This study introduces an endogenous modification strategy for engineering exosome mimetics using PRP and elucidates the specific regulatory mechanisms involved. Our findings offer an innovative treatment paradigm for clinical management of tendon–bone interface healing after rotator cuff repair.

Materials and methods

Cell culture, polarization and PRP pretreatment

Human umbilical vein endothelial cells (HUVECs) were purchased from the National Collection of Authenticated Cell Cultures (Shanghai, China) and cultured in Endothelial cell medium (Sciencell, USA) with 10% fetal bovine serum (FBS, Gibco, NY, USA), 1% penicillin/streptomycin. RAW264.7 monocytes were obtained from the cell bank of the Chinese Academy of Sciences. The RAW264.7 monocytes were cultured within the DMEM (Gibco BRL, Grand Island, NY) that contained the 10%FBS, 1% penicillin/streptomycin. RAW264.7 monocytes were cultured with 100 ng/mL lipopolysaccharide (LPS) and 20 ng/mL IFN-γ for 24 h to generate M1-polarized macrophages or 20 ng/mL interleukin-4 (IL-4) and 20 ng/mL interleukin-13 (IL-13) for 24 h to generate M2-polarized macrophages [68]. M2 macrophages were pretreated with 10% PRP solution for 48 h until collected for extraction of EM.

Preparation of Platelet-rich plasma

Blood samples were acquired from healthy donors at the Blood Transfusion Center of The Second Affiliated Hospital and Yuying Children’s Hospital of Wenzhou Medical University. All procedures involving human subjects were conducted in accordance with the ethical standards set by the institutional review board (2021-k-81-02), with informed consent obtained and participant privacy rigorously maintained. Platelet-rich plasma was generated using a previously established protocol [28]. Briefly, whole blood was collected and combined with a 10% (v/v) sodium citrate anticoagulant solution. Subsequent centrifugation at 300 × g for 20 min separated red blood cells, and the supernatant was further centrifuged at 900 × g for 10 min to remove leukocytes. The resulting supernatant was subjected to three freeze–thaw cycles using liquid nitrogen for freezing and a 37 °C water bath for thawing. Following this, the solution was centrifuged at 10,000 × g for 5 min to pellet platelet membrane debris. The final supernatant was collected, lyophilized, and reconstituted in PBS at 40 wt% to obtain the platelet-lysate rich plasma solution, which was stored at −20 °C until use. PRP activation was performed by incubation with 10% CaCl2 and 1000 U of thrombin (Beyotime, Beijing, China) [69]. To reduce donor-dependent variation, PRP collected from multiple donors was pooled and homogenized before downstream experiments. The concentrations of representative growth factors, including platelet-derived growth factor (PDGF), transforming growth factor-β1 (TGF-β1), vascular endothelial growth factor (VEGF), and insulin-like growth factor-1 (IGF-1) were quantified using ELISA kits (MultiSciences, Hangzhou) according to the manufacturer’s instructions. GFs in all kinds of PRP were assessed for each of the three donors. All PRP preparations were generated using the same protocol to improve batch-to-batch consistency and reproducibility.

M2-EM and PRP-M2-EM preparation and characterization

M2 macrophages were resuspended in PBS, and EM were isolated and purified using a sequential extrusion-based protocol according to an established method [15]. Cell suspensions were processed using a mini-extruder (Avanti Polar Lipids, AL) by successive passages through polycarbonate membrane filters with pore sizes of 5 μm and 1 μm (Whatman, UK). In order to remove the cell debris and large vesicles the sample was centrifugated at 10,000 g for 10 min at room temperature. The EM were concentrated and purified using a 100 kDa centrifugal filter unit (Millipore, USA). This process involved centrifugation at 1,000 × g for 15 min at room temperature with PBS, repeated three times with equal volumes of PBS. Following purification, the samples were stored at −80 °C.

The protein content of the exosome mimetic (EM) preparations was determined using a bicinchoninic acid (BCA) assay kit (Beyotime, China). The characterization process involved evaluating exosomal markers by western blot and quantifying particle concentration via Nanoparticle Tracking Analysis (NTA). Particle size distribution and zeta potential were assessed using dynamic light scattering (DLS) performed on a Zetasizer Nano ZS90 system (Malvern, UK). Furthermore, morphological examination was conducted using transmission electron microscopy (TEM; Hitachi HT-7700, Japan).

Cell viability assay

The cell vitality response to M2-EM and PRP-M2-EM was detected using a CCK-8 kit (meilunbio, China). After being exposed to PBS, M2-EM or PRP-M2-EM in 24 h, 48 h, and 72 h, HUVECs and RAW264.7 in 96-well plates were supplemented with 100 µL of working solution containing CCK-8 reagent 10 µL for 2 h at 37 °C. The absorbance at 450 nm was detected by a microplate reader.

Celluar uptake assay

M2-EM and PRP-M2-EM were labeled with the lipophilic carbocyanine dye DiD (Beyotime, China). DiD-labeled EMs were incubated with RAW264.7 and HUVECs in 24-well plates for 6 h. After washed with PBS, samples were stained with Actin-Tracker Green (Beyotime Biotechnology), DAPI (Beyotime Biotechnology) and imaged by fluorescence microscope.

Transwell migration assay

Transwell assay was applied to measure the ability of cell migration. HUVECs (2 × 10⁴) were incubated in the upper chamber of 24-well transwell plates (Corning). After incubated with corresponding treatmens for 18 h, the migrated cells were fixed with paraformaldehyde and stained with 0.2% crystal violet (Beyotime, China). The level of migration was observed using optical microscope (SZ61TR, Olympus, Japan).

Tube formation assay

Tube formation assay was applied to measure the ability of HUVECs to generate vessels. A total of 50 µl of Matrigel Matrix (Corning) was added to each well of 96-well plates and was incubated in a cell incubator for 1 h to gelation. HUVECs (4 × 10⁴) were incubated in 96-well plate for 8 h and observed under an optical microscope (SZ61TR, Olympus, Japan) after 8 h.

qRT-PCR analysis

Total RNA (tRNA) was extracted from cells using TRIzol reagent (Sangon, China). Following isolation, RNA concentration was quantified with a Nanodrop 2000 spectrophotometer, and complementary DNA (cDNA) was synthesized with the RevertAid First Strand cDNA Synthesis Kit (Takara). Quantitative polymerase chain reaction (qPCR) was subsequently performed using SYBR Green detection chemistry (Takara) on a LightCycler^®96 Real-Time PCR System (Roche, IN, USA). All primer sequences employed in HUVECs and RAW264.7 experiments are detailed in Table S2 (Sangon Biotech, Shanghai, China). Reverse transcription procedures adhered strictly to manufacturer protocols, while stem-loop primers were acquired from Servicebio (Table S3). Additional miRNA primers used in this study were likewise supplied by Servicebio (Table S4).

Western blotting (WB) assay

Protein separation was performed using 10% SDS-PAGE, after which the proteins were electrophoretically transferred onto a polyvinylidene fluoride (PVDF) membrane. The membranes were blocked in 5% non-fat milk for two hours and subsequently incubated with primary antibodies at 4 °C overnight. Following this, the membranes were treated with a goat anti-rabbit IgG secondary antibody (abcam) for a two-hour incubation period. Target protein bands were detected using an enhanced chemiluminescence method, visualized with a ChemiDoc XRS+ imaging system, and quantitatively analyzed with Image Lab V3.0 software (Bio-Rad, USA).The primary antibodies were obtained from Abcam (Cambridge, United Kingdom): HSP70 (ab2787), TSG101(ab133586), CD63(ab134045), Arginase-1(ab315110), iNOS (ab178945), VEGF (ab52917), β-actin (ab32572), GADPH (ab8245). The primary antibodies of TIMP3 was purchased from Boster Biological Technology (Wuhan, China).

Immunofluorescence

RAW264.7 cells were seeded onto cell slides in 12-well plates and maintained in culture for 24 h. Following this initial period, the cells received treatment with LPS, LPS combined with M2-EM, or LPS together with PRP-M2-EM at predetermined concentrations. Subsequently, the cells underwent fixation with 4% paraformaldehyde for 20 min and were then permeabilized with 0.5% Triton X-100 for 10 min. A blocking step was carried out using 5% goat serum (Solarbio, China) for one hour at 37 °C. The samples were then incubated overnight at 4 °C with primary antibodies against CD86, CD163, and CD206, all diluted to 1:200. Following primary antibody incubation, the cells were exposed for two hours at room temperature to Alexa Fluor^® 488 or 594 conjugated secondary antibodies at a 1:500 dilution. Cellular morphology and fluorescence were subsequently visualized and analyzed by fluorescence microscopy.

Flow cytometry

RAW264.7 macrophage polarization was analyzed by flow cytometry using CD86 and CD206 surface markers. After the indicated treatments, cells were collected, washed with PBS, and incubated with fluorochrome-conjugated anti-CD86 and anti-CD206 antibodies for 30 min at 4 °C in the dark. After washing, the cells were resuspended in PBS and analyzed using a flow cytometer (BD Biosciences, USA). At least 10,000 events were collected for each sample. CD86-positive and CD206-positive cells were quantified using FlowJo software.

miRNA sequencing

Total RNA isolation was performed with TRIzol reagent (Sangon, China) in accordance with the manufacturer’s instructions. RNA concentration was measured using a Nanodrop 2000 spectrophotometer, and RNA integrity was evaluated with an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). For small RNA library preparation, 1 µg of total RNA per sample was processed with the NEBNext Small RNA Library Prep Set for Illumina (Cat. No. NEB#E7330S, NEB, USA) as recommended. The procedure involved ligating adapters to both ends of the RNA molecules, followed by reverse transcription into cDNA and PCR amplification. Amplified products measuring 140–160 bp were selected and purified to construct the small RNA libraries. Library quality was verified using the Agilent Bioanalyzer 2100 system, and sequencing was carried out on the Illumina Novaseq 6000 platform, generating 150 bp paired-end reads. All small RNA sequencing and related analytical services were provided by OE Biotech Co., Ltd. (Shanghai, China).

Differentially expressed miRNAs were identified based on a q-value threshold of < 0.05 and an absolute fold change (FC) of > 2. The q-value was determined using the DEG algorithm within the R package for experiments with biological replicates, and by the Audic–Claverie statistic for those without replicates. Putative targets of differentially expressed miRNAs were predicted in animal systems using the software miRanda, applying the following parameters: S ≥ 150, ΔG ≤ − 30 kcal/mol, and requiring strict 5′ seed pairing. For plant miRNA target prediction, Targetfinder was employed. Enrichment analyses of Gene Ontology (GO) terms and KEGG pathways for the miRNA-targeted genes were conducted in R using the hypergeometric distribution test.

Dual-luciferase reporter gene assay

The wild-type (WT) sequence of the predicted miR-21a-5p binding site within the 3′-untranslated region (3′-UTR) of TIMP3, along with a corresponding mutant (MUT) sequence containing site-directed mutations in the seed region, were synthesized. These fragments were subsequently cloned into luciferase reporter vectors (Oulu Biotechnology Co., Ltd., Shanghai, China).

The constructed TIMP3-WT and TIMP3-MUT reporter plasmids were co-transfected with miR-21a-5p mimic or mimic negative control (NC) into HEK293T cells. After 48 h of transfection, cells were harvested and lysed. Luciferase activity was measured using a dual-luciferase reporter assay system (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Renilla luciferase was used as an internal control, and the relative luciferase activity was calculated as the ratio of Firefly luciferase activity to Renilla luciferase activity.

Animal model of rotator cuff injury

This animal study was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of Wenzhou Medical University. 81 male Sprague-Dawley rats, aged 10 weeks and weighing 400 ± 50 g, were obtained from the Animal Center of the Chinese Academy of Sciences. The rats were anesthetized and fixed in a supine position. Hair was removed from the shoulder area, and the skin was exposed and disinfected. A surgical incision was made at the humeral head to expose the rotator cuff tissue. Subsequently, a 5–0 PDS wire was inserted into the supraspinatus tendon (SST), which was then transected using a surgical blade. The fibrocartilage and part of the subchondral bone at the insertion site were carefully debrided, and a bone tunnel was created laterally through the proximal humerus using a 22-gauge needle. The PDS wire was passed through the bone tunnel to secure the SST to its original insertion site. The deltoid muscle and skin were sutured layer by layer. To prevent infection, penicillin G was administered once daily for three consecutive days. To investigate the effects of exosome mimetics (EM) on rotator cuff (RC) injury, the rats were randomly divided into three groups following RC injury modeling: the PBS group, the M2-EM group, and the PRP-M2-EM group.

Briefly, the rats were injected through joint cavity of the rat’s shoulder joint with vehicle only or 500 µg of M2-EM or PRP-M2-EM dissolved in 500 µL of PBS during the first three days of each week for four consecutive weeks following rotator cuff reconstruction. Subsequently, after the initial four-week treatment period, the injections of M2-EM or PRP-M2-EM were continued once a week for an additional four weeks. The PBS group were injected with PBS.

To investigate the role of miR-21a-5p in PRP-M2-EM-mediated regulation during rotator cuff injury modeling, 45 male 10-week-old Sprague-Dawley rats were randomly allocated into three experimental groups and administered either vehicle control or PRP-M2-EM treatments as previously outlined. Subsequently, antagomir injections were administered intra-articularly into the shoulder joint cavity. Specifically, on the first day of each week for four consecutive weeks following model establishment, anesthetized rats received injections of either miR-21a-5p antagomir (5 nmol) or negative control antagomir (5 nmol) into the right shoulder joint cavity using a microinjection system to suppress miR-21a-5p expression.

Histology

At designated intervals, the rats were euthanized at 1, 4 and 8 weeks. The supraspinatus tendon attachments to the humeral head were carefully dissected and isolated from surrounding soft tissues. All harvested specimens underwent fixation in 10% neutral buffered formalin for 72 h, followed by decalcification in EDTA solution for four weeks. Subsequently, the tissues were embedded in paraffin and sectioned at 5-µm thickness for histological processing, including hematoxylin and eosin (H&E) staining, Safranin O-Fast Green (S–O) staining, and Masson’s trichrome staining.

For immunohistochemical (IHC) analysis (n = 3), tissue sections were deparaffinized and treated with 3% hydrogen peroxide to block endogenous peroxidase activity. Following this, sections were incubated with primary antibodies overnight at 4 °C in a humidified environment, then with appropriate secondary antibodies for 60 min. Detection was performed using diaminobenzidine solution, with hematoxylin counterstaining applied to visualize cellular morphology.

For immunofluorescence staining (n = 3), paraffin sections were deparaffinized, rehydrated, and subjected to antigen retrieval using citrate buffer (pH 6.0). After permeabilization with 0.3% Triton X-100 and blocking with 5% bovine serum albumin for 1 h at room temperature, the sections were incubated overnight at 4 °C with primary antibodies against Arg-1 and iNOS. Subsequently, the sections were incubated with fluorophore-conjugated secondary antibodies for 1 h at room temperature in the dark. Nuclei were counterstained with DAPI, and the stained sections were observed and imaged using a fluorescence microscope. Quantitative analysis of fluorescence intensity was performed using ImageJ software.

Biomechanical test

Biomechanical testing was performed using a universal material testing system (Instron 5566, USA). Six rat supraspinatus tendon–humerus specimens from each group were subjected to biomechanical evaluation. For specimen preparation, the humerus and supraspinatus tendon were carefully preserved, while surrounding soft tissues were removed. To facilitate secure fixation and minimize slippage during testing, the humeral shaft was embedded in a custom mold using denture powder.

Prior to mechanical testing, each specimen was subjected to a preload of 1 N and preconditioned for 20 cycles at a displacement rate of 5 mm/min. Subsequently, the specimens were loaded to failure at a constant speed of 5 mm/min. The load–displacement curve was continuously recorded throughout the testing process. Ultimate failure load (N), stiffness (N/mm), energy to failure (J), and failure mode were analyzed. Stiffness was determined from the slope of the linear region of the load–displacement curve, while energy to failure was calculated as the area under the load–displacement curve until the point of failure. Failure modes, including tendon midsubstance rupture, tendon-to-bone interface failure, or bony avulsion, were documented for each specimen.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 10.0 (GraphPad Software, CA, USA). For comparisons between two groups, Student’s t-test was applied. For comparisons among multiple groups, one-way analysis of variance (ANOVA) was performed followed by appropriate post hoc multiple comparison tests (Tukey’s test). Normality of data distribution was assessed using the Shapiro–Wilk test prior to statistical analysis. All data were presented as mean ± standard deviation (SD). Biological replicates were used for all in vivo and in vitro experiments, and technical replicates were included where applicable to ensure measurement reliability. Exact p values were reported when available; otherwise, significance levels were indicated as *P < 0.05, **P < 0.01, ***P < 0.001.

Author contributions

C. L. and H. J. contributed equally to this work. C. L. and H. J. designed the study and wrote the main manuscript text. L. C., L. Z., Y. W., and H. S. performed the experiments and collected data. Z. L. and J. S. conducted data analysis and interpretation. J. W., J. D., and X. P. supervised the project, provided critical revisions, and secured funding. All authors reviewed and approved the final manuscript.

Funding

This work was funded by Zhejiang Provincial Medical and Health Science and Technology Plan Project (2023KY148), Zhejiang Provincial Medical and Health Science and Technology Plan Project (2020KY187), The Yuying Program of the Second Affiliated Hospital of Wenzhou Medical University (No. 813386), Zhejiang Provincial Medical Association Clinical Medicine Special Fund Project (2022ZYC-Z28), Wenzhou Medical Association Foundation (KT20220420085137419).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.