Platelet-derived vesicles restore fertility in endometrial injury by modulating the endometrial immune niche in mice.

To investigate the inflammation and fibrosis of injured endometrial tissue in IUA patients, immunohistochemistry for key immune cell markers and Masson’s trichrome staining were performed. B-ultrasound imaging revealed that endometrial thickness was significantly reduced in these patients (~5.08 mm) compared to healthy controls (~8.18 mm) ( Fig. 1, A and B ), accompanied by increased fibrosis ( Fig. 1C ). Immunohistochemical analysis further demonstrated elevated infiltration of CD45 + leukocytes, CD68 + macrophages, CD3 + T cells, and CD56 + natural killer (NK) cells in the endometria of patients with endometrial injury compared to normal controls ( Fig. 1D ). These findings suggest that chronic inflammation is a characteristic feature of the endometrial microenvironment in patients with IUA. This inflammatory state is associated with reduced endometrial thickness and adverse reproductive outcomes, including infertility, recurrent pregnancy loss, and pregnancy-related complications. ( A ) Representative B-ultrasound images of healthy controls and endometrium-injured patients. ( B ) Base clinical information of enrolled patients ( n  = 38). ( C ) Masson’s trichrome staining (scale bars, 200 μm) in the endometrium of healthy controls ( n  = 9) and endometrium-injured patients ( n  = 9). ( D ) Immunohistochemical staining of CD45, CD68, CD3, CD56, and MPO (scale bar, 100 μm). Statistical significance was determined by Student’s t test (two-tailed). *** P  < 0.001; ** P  < 0.01. The data are presented as mean ± SD. We successfully isolated PEVs using a freeze-thaw-based method for subsequent characterization ( 18 , 19 ) ( Fig. 2A and fig. S1). Analysis by transmission electron microscopy (TEM) and dynamic light scattering (DLS) showed that the purified PEVs had a spherical morphology ( Fig. 2B ) and a homogeneous size distribution of around 100 nm, contrasting with the larger size of PLTs (fig. S2). The particles exhibited a negative zeta potential ( Fig. 2C ). Furthermore, the size of the PEVs remained stable when stored at low temperatures, indicating their high stability ex vivo ( Fig. 2D ). ( A ) Photograph of whole blood, PRP, PLTs, and PEVs in DPBS. ( B ) TEM image of PEVs (scale bar, 200 nm). ( C ) PEVs size distribution and zeta potential measured by dynamic light scattering (DLS). ( D ) Particle size of PEVs stored for 7 days at different temperatures (4°, −20°, and −80°C). ( E ) SDS-PAGE of PLT lysate and PEVs with Coomassie brilliant blue staining. ( F ) Western blot analysis of CD9, CD63, CD41, β-actin, GAPDH from the PLT lysate and PEVs. ( G ) Protein mass spectrometry of PEVs ( H ) LC-MS/MS and GO term enrichment analysis of the cellular component, biological process, and molecular function. ( I ) KEGG pathway analysis and fold changes of genes expressed in the chemokine signaling pathway, PLT activation, and VEGF signaling pathway. The data are presented as mean ± SD. To confirm the origin of the vesicles, we used Coomassie Brilliant Blue staining ( Fig. 2E ) and Western blot analysis. These results demonstrated significant enrichment of PLT vesicle marker proteins (CD9, CD41, and CD63) in the PEVs, while the cytoplasmic proteins of PLTs [glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin] were present at lower levels ( Fig. 2F and fig. S3). Furthermore, PEVs were found to carry a range of bioactive molecules, including growth factors, which are crucial for tissue repair, regeneration, and modulation of inflammatory responses. Notably, several key growth factors, including vascular endothelial growth factor A (VEGFA) and transforming growth factor–β (TGF-β) (fig. S4), were encapsulated within the vesicles, where their lipid bilayer membrane provided protection from extracellular degradation ( 20 ), thereby maintaining the stability of the cytokine cargo during transport to target cells or tissues. We further evaluated PEVs using liquid chromatography–tandem mass spectrometry (LC-MS/MS) to further characterize the therapeutic use of PEVs. Through gene ontology (GO) enrichment analysis, we classified these parental proteins into cellular components, biological processes, and molecular functions ( Fig. 2G ). The enrichment analysis elucidated that the proteins in PEVs were mainly derived from membrane-bounded vesicles (count = 475), extracellular organelles (count = 433), vesicles (count = 504), and extracellular vesicles (count = 432). The GO biological process analysis revealed that PEVs proteins were involved in the regulation of biological quality, cellular localization, vesicle-mediated transport, wound healing, regulation of signaling, immune system process, and cellular response to growth factor stimulus. The GO molecular functional analysis revealed that these components were involved in the binding of proteins, cell adhesion molecules, receptors, and adenosine 5′-triphosphate (ATP), which may contribute to the targeting property of PEVs to inflammatory sites ( Fig. 2H ). The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis demonstrated the involved pathways of PEVs, including PLT activation, chemokine signaling pathway, and VEGF signaling pathway ( Fig. 2I ). These results suggest the proteins within the PEVs could initiate signaling pathways that promote tissue repair and regeneration upon reaching their target cells. In our previous study, we have demonstrated that PEVs exhibit a strong ability to target inflammatory sites, owing to their intrinsic affinity for inflammatory sites ( 21 ). To mimic a clinical scenario, we refined the murine model with endometrial injury by performing intrauterine perfusion of 1′-dioctadecyl-3,3,3′,3′-tetramethylindotricarbocyanine iodide (DiR)-labeled PEVs (PEV-DiR), followed by cervix ligation with sutures ( Fig. 3A ). This allowed the PEVs to remain in the uterine cavity for 2 hours before loosening the slipknot and untangling the cervix. A sham model was used as a control. After 22 hours, major organs were harvested and imaged using an ex vivo imaging system. Notably, we observed a significantly higher retention of PEV-DiR in the uterus cavity of the model mice (292.25 ± 11.18) compared to the sham controls (130.25 ± 8.655; P  < 0.001), with no observable signals in other organs. The mean difference between the groups was 162.00 (95% confidence interval, 144.7 to 179.3), representing a 2.24-fold increase ( Fig. 3B and fig. S5A), which demonstrates the targeting ability of PEVs to the inflammatory injured endometrium. We showed that cell adhesion–related proteins, such as P-selectin, were retained on PEVs (figs. S5B and S6), which facilitate their binding to receptors like PSGL-1 expressed on injured endometrial areas in mice and humans (fig. S5, C and D). This targeting ability is likely attributed to the both specific ligands-receptors binding and the nanosized effect that facilitating their interaction with target cells ( 22 ) (fig. S5, E and F). ( A ) Schematic diagram of the timeline of animal experiments to test the in vivo targeting properties of PEVs [created in BioRender. Li, D. (2026) https://BioRender.com/d20k571 ]. ( B ) Ex vivo fluorescence imaging of major organs and uterus in KM mice after various treatments as indicated at 24 hours and corresponding quantitative analysis ( n  = 4 per group). ( C ) Gross views of the uterus after treatment at different time (scale bar: 1 cm) ( n  = 5 per group, G power > 0.8). ( D ) HE staining of the cross section of uterus after treatment at different time points (scale bars, 100 and 500 μm) and ( E and F ) corresponding quantitative analysis ( n  = 5 per group, G power > 0.8). ( G ) Live birth outcomes and representative coronal-sections μCT reconstructions of E14.5 fetuses of different treatment groups ( n  = 5 per group, G power > 0.8; scale bar, 1 cm). ( H ) Litter size among different treatment groups ( n  = 5 per group, G power > 0.8). ( I ) Scatter diagram of E14.5 placental and fetal weight for Sham group and PEV-treated group. ( J ) The structure of murine placenta [created in BioRender. Li, D. (2026) https://BioRender.com/w7wotkx ]. ( K ) H&E staining, Immunohistochemistry staining of laminin and TUNEL staining of placentas from different groups on E14.5. Statistical significance was determined by Student’s t test (two-tailed) and one-way analysis of variance (ANOVA) with the Bonferroni post hoc test. *** P  < 0.001, ** P  < 0.01. The data are presented as mean ± SD. To systematically assess the therapeutic potential of PEVs in promoting uterine recovery, we established a murine model of endometrial injury by administering a direct injection of 95% ethanol into the uterine horns ( 23 , 24 ). Then, mice were randomly assigned to one of three groups: healthy (Sham) mice, untreated (Untx) mice with endometrial injury, and PEV-treated mice, which received intrauterine administration of PEVs (50 μg per uterine horn daily for three consecutive days) (fig. S7). Gross morphology and histological analyses of the uterine tissue were performed to assess recovery over time. In Untx mice, significant damage to the endometrium and myometrium was observed, resulting in structural and functional impairments of uterine tissue compared to healthy (Sham) controls. To address potential confounding effects of the infusion procedure itself and to benchmark against conventional therapy, we included two additional control groups: mice receiving intrauterine infusion of phosphate-buffered saline (PBS) and mice receiving subcutaneous injection of estradiol. Notably, neither PBS nor estradiol treatment resulted in significant improvement in uterine morphology, endometrial thickness, fibrosis reduction, or subsequent pregnancy outcomes compared to the Untx group (fig. S8), effectively ruling out placebo effects and highlighting the promising efficacy of PEVs. Conversely, PEV-treated mice exhibited progressive uterine recovery with minimal endometrial cavity fluid, IUA, and deformation ( Fig. 3C ). Specifically, the endometrial thickness was reduced by approximately 53% in the Untx group (from 150 to 70 μm), reflecting significant injury. In contrast, PEVs remarkably increased endometrial thickness from approximately 58 μm to an average of 157 μm on day 14 postmodeling, achieving approximately 99.3% recovery, indicating the impressive regenerative efficacy of PEVs ( Fig. 3D ). In addition, the number of glands exhibited the same trends as endometrial thickness, reaffirming the promoting effect of PEVs on endometrial regeneration ( Fig. 3, E and F ). Masson’s trichrome staining was used to assess the extent of fibrosis, which demonstrated that PEVs treatment effectively attenuated collagen deposition [reducing from 11.00% ± 1.48% in the untreated model group to 3.57% ± 1.02% ( P  < 0.01)], suggesting a reduction in fibrotic tissue formation (fig. S9, A and B). Understanding that successful endometrial regeneration relies on angiogenesis to facilitate the delivery of oxygen and nutrients necessary for tissue repair ( 25 ), we subsequently investigated the angiogenic potential of PEVs in vivo. Immunostaining for CD31, a marker of endothelial cells, revealed that PEVs treatment effectively promoted angiogenesis, as evidenced by a significant increase in CD31-positive endothelial cells [from 2.82% ± 0.98% in the untreated model group to 6.91% ± 1.03% ( P  < 0.001)] (fig. S9, C and D). Having established the regenerative efficacy of PEVs, we next evaluated their potential thrombotic risk, a critical safety concern for PLT-derived therapeutics. Western blot analysis revealed that the expression of cyclooxygenase-1 (COX-1) and thromboxane A synthase (TXAS) ( 26 ) in PEVs was substantially diminished compared to their parent PLT (grayscale values: COX-1, 0.27 ± 0.04 in PEVs versus 1.77 ± 0.33 in PLT; TXAS, 0.32 ± 0.04 in PEVs versus 1.18 ± 0.07 in PLT; fig. S10, A and B). This minimal residual expression, combined with an intrauterine administration route that minimizes systemic exposure, was functionally insufficient to catalyze substantial TXA 2 synthesis. This conclusion was directly supported by the absence of a significant elevation in the urinary level of 11-dehydro-TXB 2 (a stable systemic metabolite of TXA 2 , normalized to creatinine) in PEV-treated mice compared to the Sham group ( P  > 0.05; fig. S10C). Furthermore, to assess the functional consequence in tissues most susceptible to microthrombosis, we performed a histopathological examination of the renal and pulmonary vasculature. Consistently, no detectable microthrombi were observed in the glomeruli of the kidneys or the vasculature of the lungs across all groups (fig. S10, D and E). These data collectively demonstrate that intrauterine delivery of PEVs presents a favorable safety profile with minimal thrombotic risk. To evaluate the functional recovery of uterine fertility, we analyzed embryo implantation and fetal development at embryonic day 14.5 (E14.5) ( 27 , 28 ). Western blot analysis revealed that the expression levels of key endometrial receptivity markers, including leukemia inhibitory factor (LIF), integrin subunit beta 3 (ITGB3), and estrogen receptor (ER), were reduced in the Untx group compared to the Sham group. PEVs treatment notably restored the expression of these proteins, particularly during the late recovery phase (fig. S11). Furthermore, we assessed the inflammatory levels and found that the elevated expression of pro-inflammatory cytokines [tumor necrosis factor–α (TNF-α), interleukin-6 (IL-6), and IL-1β] in the Untx group was significantly suppressed following PEVs treatment (fig. S12), suggesting a resolution of inflammation. Consistent with these molecular findings, functional fertility assessment demonstrated that the number of viable fetuses was significantly improved in PEV-treated mice compared to the untreated injury group ( P  < 0.001) ( Fig. 3, G and H ). Moreover, the average weights of embryos (0.3036 g ± 0.04114 g versus 0.3134 g ± 0.04402 g) and placentas (0.1586 g ± 0.04044 g versus 0.1534 g ± 0.04169 g) from PEV-treated mice were comparable to those of healthy (Sham) controls ( P > 0.05), indicating that the repaired uterine environment was capable of supporting normal fetal development ( Fig. 3I and fig. S13, A to C). Specifically, micro–computed tomography (μCT) confirmed that the skeleton of fetuses from the PEV-treated mice was in normal development, which was comparable to the healthy (Sham) mice. To assess whether PEVs treatment repaired the damaged endometrium, hematoxylin and eosin (H&E) staining, immunohistochemistry of laminin expression, and terminal deoxynucleotidyl transferase–mediated (TUNEL) staining were conducted in placental sections. The placenta from the PEV-treated mice exhibited normal structures in the three compartments (maternal decidua, junctional zone, and labyrinth zone) ( Fig. 3J ). The laminin (a marker of the fetal endothelium and its associated basement membrane) immunostaining revealed good fetal vascularization of the labyrinth in the PEV-treated mice. The apoptosis of placental cells of mice in the PEV-treated mice was not significantly different from that in the healthy (Sham) mice, further indicating the quality of placenta ( Fig. 3K and fig. S13, D and F). To further depict the sophisticated alterations of endometrial cells and the local immune microenvironment after treatment, 10X Genomics single-cell RNA sequencing (scRNA-seq) was applied to analyze the discrepancy in gene transcription between Untx and PEV-treated uterus isolated from mice (day 14 after modeling) at the same estrous cycle. Mice uterus is a hollow, tube-like tissue composed of luminal and glandular epithelium, stromal compartments, vascular networks, and complex immune cell populations. On the basis of the recognized marker genes ( Ptprc ), we divided all cells into two groups, namely the nonimmune cell population and the immune cell population (fig. S14). In the PEV-treated mice, immune cells infiltration declined obviously, while the proportion of non-immune cells recovered ( Fig. 4A ). ( A ) Uniform manifold approximation and projection (UMAP) map showing the map of immune cells and non-immune cells in mice of the Untx and PEVs groups, and the percentage of each cluster. ( B ) UMAP of nonimmune cells in the Untx and PEVs group. ( C ) Relative proportions of fibroblasts, epithelial cells, endothelial cells, mesothelial cells, myocytes, and pericytes. ( D ) Immunofluorescent images and quantitative statistics of CK-19 and vimentin in Untx and PEVs group on day14 postmodeling ( n  = 3 per group; scale bar, 20 μm). ( E ) Dot plot of relative expression of receptivity markers ( Pgr , Hoxa10 , Itgb3 , Igf1 , and Igf2 ), inflammatory markers ( Il1b , Il1a , and Tnfa ) and cellular senescence markers ( Cdkn1a , Cdkn2a , Trp53 , etc.) in the Untx and PEVs groups. ( F ) Up-regulation and down-regulation of GO enrichment analysis of endothelial cells. ( G ) Cell cycle analysis of endothelial cells in Untx and PEVs groups. ( H ) Expression of Mki67 , Top2a , and Pcna in G 1 , S, G 2 /M phase in endothelial cells between Untx and PEVs group. Statistical significance was determined by one-way analysis of variance (ANOVA) with the Bonferroni post hoc test. *** P  < 0.001, ** P  < 0.01. The data are presented as mean ± SD. We first characterized the nonimmune alterations in the uterus in PEV-treated mice versus untreated mice. Nonimmune cells can be further divided into six subgroups according to multiple markers (fig. S15), including fibroblasts, epithelial cells, endothelial cells, mesothelial cells, myocytes, and pericytes ( Fig. 4, B and C ). Aligning with the scRNA-seq results, the uterus in PEV-treated mice possessed a comparable fluorescence intensity of CK19 (a marker of the epithelial cells) and vimentin (a marker of the fibroblasts stromal cells) to the healthy (Sham) mice, which was significantly diminished in Untx mice ( Fig. 4D and fig. S16). These data indicated that the damaged endometrium has largely recovered upon the PEVs treatment. Stromal cells, the most abundant cell type in the endometrium, marked by the high expression of Col6a1 , Col6a4 , and Dcn , are one of the key contributors to the repair process to achieve adequate endometrial thickness ( 29 ). The dot plot showed that the receptivity markers, like Pgr , Hoxa10 , Itgb3 , Igf1 , and Igf2 , were highly expressed in the repaired endometrium. While the inflammatory markers ( Il1b , Il1a , Tnfa , etc.) and cellular senescence markers ( Cdkn1a , Cdkn2a , Trp53 , etc.) were down-regulated in the PEV-treated mice ( Fig. 4E and fig. S17A). All these changes have created a good nutritional environment for the implantation and development of embryos. In addition, vascular networks formed by endothelial cells extend throughout the endometrium, delivering nutrients to different areas and playing a crucial role in tissue repair and embryo implantation. Using GO analysis of highly expressed genes specific to endothelial cells, we found that the terms extracellular matrix structural constituent, positive regulation of cell population proliferation, PLT-derived growth factor binding, angiogenesis, and the blood vessel development were highly enriched, and the terms related to Inflammatory response, immune response, and neutrophil chemotaxis were down-regulated in mice receiving PEVs, suggesting that PEVs can promote angiogenesis and vascular remodeling ( Fig. 4F and fig. S17B). By cell cycle assays, we observed that endothelial cells in mice receiving PEVs staged in the G 2 /M phase more obviously compared to the control ( Fig. 4G ). Expression of the proliferation markers Mki67 , Top2a , and Pcna was observed in the G 2 /M phases of endothelial cells in endometrium and was markedly increased in the PEV-treated endometrium ( Fig. 4H ). Pericytes, which contribute to the development, remodeling, structure, and permeability of the vascular system, proliferated more rapidly versus controls (fig. S17C). These results suggested that PEVs promoted vascular network reconstruction by reshaping non–immune cell phenotypes. We next questioned whether PEVs can regulate immune cells orientations in injured endometrium. KEGG pathway enrichment analysis of the whole uterus revealed a significant involvement of the immune system in the healing process (fig. S18). On the basis of the known conserved marker gene expression profiles ( Fig. 5A ), seven major cell types of immune cells were annotated ( Fig. 5B ). The immune cells generally evolved to favor tissue repair, with inflammatory-related immune cells markedly decreased, like macrophages, monocytes, T cells, dendritic cells (DC), and granulocytes ( Fig. 5C and figs. S19 and S20). Among all the changed immune cells, M1/M2-type macrophage balance has been reported to be critical in wound healing and tissue reconstruction. In line with the single cells clustering results, flow cytometry at different time points has verified a shift from M1-type macrophages to M2-type macrophages in the PEV-treated mice, especially at the late stage of treatment, which is conducive to shaping an anti-inflammatory and tissue-repair microenvironment ( Fig. 5, D and E , and fig. S21). Subsequently, to delve deeper into the fate of macrophages, we further reclustered the macrophages into M1, M2a, M2b, M2c, and M2d subtypes based on the exclusive marker genes (fig. S22). The macrophages are inclined to change forward to M2a and M2d subtypes ( Fig. 5F ), while the M1, M2b, and M2c type macrophages are decreased in mice receiving PEVs in contrast to the Untx mice ( Fig. 5G ), which was also in alignment with the pseudotime trajectory analysis ( Fig. 5H ). The inflammatory cytokines of macrophages are also remarkably declined in PEV-treated mice ( Fig. 5I ). The M2a subtype is associated with tissue repair and anti-inflammatory responses, while the M2d subtype is linked to the regulation of inflammation and promoting angiogenesis (the formation of new blood vessels). M2a macrophages are induced by IL-4 and IL-13, and exhibit elevated levels of mannose receptor ( Mrc1 ), arginase 1 ( Arg1 ), and transforming growth factor beta 1 ( TGF-β1 ). Arg1 can activate the arginine pathway, yielding ornithine, a precursor of polyamines and collagen, to contribute to extracellular matrix formation and facilitate tissue repair. Meanwhile, M2d macrophages are formed by synergistic induction of Toll-like receptor agonists and adenosine receptor agonists. They secrete IL-10 and VEGF, which are associated with angiogenesis. Accordingly, the up-regulation of the M2a and M2d subtypes was positively correlated with the regeneration of the injured endometrium. ( A ) A dot plot displaying conserved marker genes in each immune cells population. ( B ) t-SNE of immune cells from uterus in different treatment groups, revealing seven different cell populations. ( C ) The absolute percentage of different types of immune cells in Untx and PEVs group. ( D and E ) Flow cytometry and quantification demonstrated the percentage of M1 macrophages (F4/80 + CD80 + ), M2 macrophages (F4/80 + CD206 + ), and the ratio of M1/M2 in uterus on days 3, 7, and 14 in the Untx and PEVs groups ( n  = 3 per group). ( F ) UMAP of macrophages, revealing five clusters (M1, M2a, M2b, M2c, and M2d). ( G ) The relative percentage of macrophages in Untx and PEVs group. ( H ) Pseudotime ordering of Macrophages. ( I ) Relative expression of inflammatory markers and cytokines secreted by macrophages in the Untx and PEVs groups. ( J ) KEGG enrichment analysis of Macrophages. ( K ) GSEA of macrophages in the uterus (oxidative phosphorylation, mitochondria, and respiratory electron transport). ( L ) The dotplot displays the differentially expressed genes (DEGs) of mitochondrial function, mitophagy, detoxification of ROS, oxidative stress and OXPHOS in Untx and PEVs groups. Statistical significance was determined by Student’s t test (two-tailed). *** P  < 0.001, ** P  < 0.01. The data are presented as mean ± SD. We next sought to determine the mechanism by which PEVs influence macrophage polarization. We performed a subpopulation enrichment analysis of macrophages by excavating the scRNA-seq data. As shown in the volcano plot (fig. S23A), a total of 1858 differentially expressed genes (DEGs; P  < 0.05) were identified in the Untx group and PEV-treated group, with 1365 genes up-regulated and 493 genes down-regulated. Moreover, KEGG enrichment analysis was carried out to identify the enrichment pathways of DEGs. Notably, the KEGG results revealed that DEGs up-regulated in the PEV-treated group were significantly enriched in oxidative phosphorylation, metabolic pathways, mitophagy, autophagy, and citrate cycle ( Fig. 5J ). Consistently, GSEA enrichment analysis of these DEGs showed significant down-regulation of inflammatory response (fig. S23B) and up-regulation of Oxidative phosphorylation, mitochondria, and respiratory electron transport following PEVs treatment ( Fig. 5K ). According to the GSEA results, the relative expression levels of genes involved in mitochondrial function and homeostasis, including antioxidative stress, oxidative stress, and mitophagy, are presented as a dot plot ( Fig. 5L ). As a result, PEVs markedly enhanced antioxidative stress–related gene expression but exerted the opposite effect on oxidative stress–related genes. Genes involved in mitophagy were also significantly regulated by PEVs. These observations suggest that mitochondria, which participate in oxidative stress metabolic processes and mitophagy regulation, might play critical roles in the inhibitory effect of PEVs on the inflammation of macrophages. To delineate how PEVs affect macrophage polarization, a macrophage cell line (RAW264.7) and primary cells [bone marrow–derived macrophages (BMDMs); fig. S28, A and B] were used to assess in vitro. The results showed that PEVs significantly inhibited the expression level of major histocompatibility complex II (MHCII) and CD80 in a dose-dependent manner ( Fig. 6A and fig. S28C). In line with these results, PEVs displayed a blockade on IL-6 and TNF-α ( Fig. 6B and figs. S24, A and B, and S28D), which resulted in their lower secretion by macrophages. Consistently, the mitophagy and antioxidation levels indicated by Nrf2, Beclin-1, P62, and LC3BII/I tended to increase as the dosage increased ( Fig. 6, C and D , and figs. S25, S26A, and S28, E and J). To elucidate the upstream signaling mechanism by which PEVs trigger mitophagy, we investigated the adenosine 5′-monophosphate–activated protein kinase (AMPK)–Unc-51-like kinase 1 (ULK1) pathway, a central regulator of energy homeostasis and mitophagy initiation. Western blot analysis revealed that PEVs treatment robustly promoted the phosphorylation of both AMPK and ULK1 in lipopolysaccharide (LPS)–stimulated macrophages cells in a dose-dependent manner ( Fig. 6E and figs. S26B and S28, F and K), indicating activation of this key pathway. ( A ) Flow cytometry of MHCII and CD80 among RAW264.7 cells in different treatment group and corresponding quantitative analysis ( n  = 4 per group). ( B ) Cytokines secreted by RAW264.7 after incubation with PEVs of different treatment groups ( n  = 4 per group). ( C ) Western blot analysis of autophagy protein expression (Nrf2, Beclin-1, P62, and LC3BI/II) in RAW264.7 cells treated with different groups. ( D ) GSEA enrichment analysis of RAW264.7 cells incubated with PEVs with or without LPS activation (Mitophagy_animal). ( E ) PEVs modulate the AMPK/ULK1 signaling pathway in RAW 264.7 cells. ( F ) Representative TEM images of autophagsome in RAW24.7 cells with or without PEVs and LPS treatment ( n  = 5 per group; scale bar, 2 μm). ( G ) JC-1 aggregates (Red) and monomers (green) staining in RAW264.7 for detecting mitochondria membrane potential ( n  = 4 per group; scale bars, 10 μm). ( H ) Representative images stained with MitoTracker of RAW264.7 incubated with PEVs with or without LPS activation ( n  = 4 per group; scale bar, 10 μm). ( I ) Representative images stained with MitoSOX of RAW264.7 incubated with PEVs with or without LPS activation ( n  = 4 per group; scale bar, 10 μm) and ( J to M ) fluorescence quantitative analysis. ( N and O ) Flow cytometry and ELISA results for MHCII and CD80 expression in RAW264.7 cells of different treatment groups pretreated with 3-methyladenine (3-MA) and quantitative analysis ( n  = 4 per group). **** P  < 0.0001, *** P  < 0.001, ** P  < 0.01. The data are presented as mean ± SD. Moreover, TEM imaging showed that LPS-treated cells exhibited less characteristic bilayer-membraned autophagosomes, while PEVs significantly reversed the level of mitophagy and increased the quality and quantity of mitochondria ( Fig. 6, F and J ). To prevent cellular damage, reactive oxygen species (ROS)–producing mitochondria are constantly eliminated through mitophagy, a specialized form of autophagy. As the results shown, LPS treatment resulted in low mitochondrial membrane potential and impaired mitochondria, while PEVs treatment significantly recovered the function of mitochondria ( Fig. 6, G and K , and fig. S28G). Similar results were obtained from the other two types of mitochondria-specific makers that distinguish respiring mitochondria (MitoTracker Red CMXRos) ( Fig. 6, H and L , and fig. S28H) and ROS-damaged mitochondria ( Fig. 6, I and M , and fig. S28I), respectively. Collectively, PEVs may alleviate inflammation of macrophages stimulated by LPS via promoting mitophagy and mitochondria rejuvenation (figs. S27 and S24C). To further verify this result, 3-methyladenine (3-MA), an inhibitor of mitophagy/autophagy, was added with macrophages. As expected, 3-MA pretreatment resulted in exacerbated inflammation and masked the anti-inflammatory effect of PEVs to LPS-induced macrophages ( Fig. 6N and fig. S28, L and M). Moreover, the inhibitory effects of PEVs on inflammatory cytokines secretion in macrophages were also significantly attenuated ( Fig. 6O ). These findings collectively suggest that mitophagy induced by PEVs serves as a scavenger of mtROS by removing damaged mitochondria and promoting mitochondria rejuvenation, which in turn suppresses inflammation. In accordance with the inclusion and exclusion criteria outlined in fig. S29, eight patients with documented endometrial injury–related infertility were enrolled in this pilot study to evaluate the therapeutic potential of autologous PEVs (registration number: ChiCTR2500102829, ethics approval number: JD-LC2024023-I01). Autologous PEVs were prepared from a single blood draw, which yielded sufficient material for three intrauterine infusions. Thereby, this single-draw protocol is notably more convenient than the PRP procedure that requires a separate blood collection for each infusion ( Fig. 7, A to E and fig. S30). Before infusion, all PEVs preparations were subjected to a strict quality control protocol to ensure their safety and compliance with predefined release specifications (table S1 for detailed quality control criteria and results). Intrauterine PEVs infusion was performed by experienced reproductive medicine specialists under direct ultrasound guidance, as depicted in fig. S31. Following infusion, patients were positioned in the lithotomy position for at least 30 to 40 min to allow the PEVs to remain in the uterine cavity for an extended period, thereby enhancing their contact and interaction with the endometrial lining. Serial ultrasound assessments were performed pre- and postintervention for comparative analysis. No obvious adverse effects were observed, and no complications relevant to PEVs were reported in these patients (table S2). Treatment with PEVs, the majority of enrolled patients achieved clinical pregnancy, despite previous multiple failed embryo transfer cycles or cycle cancellations due to thin endometrium (table S4) ( A ) TEM image of human PEVs (scale bar, 100 nm). ( B ) Human PEVs size distribution and ( C ) zeta potential measured by dynamic light scattering (DLS). ( D ) SDS-PAGE of human PLT lysate and PEVs with Coomassie brilliant blue staining. ( E ) Western blot analysis of CD9, CD63, CD41, β-actin, GAPDH, and VEGFA from the human PLT lysate and PEVs. ( F ) B-ultrasound images in different cycles of enrolled patients. Patient no. 1, a 33-year-old woman with 2.5 years of secondary infertility, had previously failed implantation despite multiple endometrial proliferation therapies, including intrauterine PRP perfusion. Following the first cycle of PEVs treatment, embryo implantation occurred without measurable endometrial thickening. Although this pregnancy ended as a biochemical pregnancy, the implantation itself indicates a potential immune-mediated mechanism. A subsequent course of PEVs perfusion resulted in a successful clinical pregnancy, with confirmed fetal cardiac activity. Patient no. 2, a 35-year-old woman with 5 years of secondary infertility, had undergone three egg retrievals and 12 embryo transfers at other clinics between 2019 and 2023, experiencing multiple biochemical pregnancies, spontaneous miscarriages, and implantation failures due to recurrent thin endometrium. As an alternative approach, pre-embryo transfer intrauterine PEVs infusion was performed. Although endometrial thickness did not significantly change, the endometrial pattern improved, and the patient achieved a clinical pregnancy. During the most recent follow-up, however, she underwent induced labor at 24 weeks due to fetal spina bifida. Patient no. 3, a 35-year-old woman with a 2-year history of infertility, had previously undergone multiple treatments, including PRP therapy, but experienced two failed embryo implantation attempts. She then received three consecutive intrauterine infusions of PEVs. Following this intervention, the patient achieved a clinical pregnancy and subsequently delivered at term via natural delivery, with no placental complications reported. Patient nos. 4 to 8, all with a long history of infertility, had previously experienced repeated embryo transfer cancellations due to thin endometrium, causing substantial physiological and psychological burdens. Following PEVs intervention, all five patients achieved clinical pregnancy, with patient nos. 6 and 7 conceiving naturally 2 to 3 months later. These findings suggest that PEVs may provide sustained improvement in endometrial receptivity. However, because of the small cohort and potential confounding factors, these preliminary results should be interpreted cautiously, and further long-term, large-scale studies are needed to confirm optimal dose, efficacy, and clarify mechanisms. Collectively, this preliminary investigation establishes that intrauterine infusion of autologous PEVs constitutes a safe, well-tolerated, and efficient cell-free therapeutic strategy for endometrial injury–related infertility, demonstrating considerable translational potential. These promising findings warrant robust validation through large-scale randomized controlled trials to further corroborate the efficacy and mechanisms of PEV-based therapy.

A receptive endometrium is vital for successful embryo implantation, providing an essential microenvironment and the biological support for the process ( 28 , 30 ). However, infertility resulting from endometrial injuries, particularly common conditions like IUAs and thin endometrium, impairs implantation success and poses a major challenge in global reproductive health ( 31 , 32 ). The successful repair of such injuries relies heavily on immune homeostasis, which is crucial for restoring tissue architecture ( 33 , 34 ), reestablishing receptivity ( 35 ), minimizing fibrosis ( 36 ), and promoting vascular regeneration ( 37 ). Notably, iatrogenic injury from improper curettage can trigger a chronic inflammatory response that disrupts this delicate balance, ultimately impairing endometrial regeneration and leading to recurrent implantation failure. An imbalance in M1/M2 macrophage polarization, dysregulation of T helper/regulatory T cells, and dysfunctional uterine NK cells following intrauterine injury drive excessive inflammatory cytokine release, impairing endometrial receptivity and leading to implantation failure. Currently, researches focus on reprogramming the immune microenvironment through targeted drug therapies and bioactive materials to enhance regeneration and provide novel therapeutic avenues for patients suffering from recurrent implantation failure due to endometrial injury ( 38 – 41 ). Current clinical treatments for endometrial injury include hysteroscopic surgery, estrogen pharmacotherapy, physical barriers, and stem cell–based therapies ( 8 , 9 ). However, these methods show limited and variable effectiveness, poorly understood mechanisms, and unsatisfactory restoration outcomes. Therefore, there is an urgent need to develop novel therapeutic strategies to improve clinical prognosis. PRP has recently emerged as a promising biological therapy for endometrial regeneration, promoting tissue repair via mechanisms such as angiogenesis stimulation, stem cell proliferation, and trophic factor delivery ( 42 , 43 ). Clinical reports indicate that PRP can increase endometrial thickness, improve microcirculation, and enhance embryo implantation outcomes ( 44 , 45 ). However, its clinical translation is limited by unclear mechanisms, batch variability, and potential immunogenicity ( 46 ). Inspired by our previous work, we applied PEVs, key functional components of PRP that overcome its heterogeneity and stability limitations. In a mouse model of severe endometrial injury, PEVs promoted regeneration and live births by driving macrophage polarization toward M2a/M2d phenotypes, which enhanced angiogenesis, reduced inflammation, and facilitated repair. Further sequencing analysis revealed that macrophages in injured endometrium exhibit mitochondrial dysfunction with enhanced glycolysis and disrupted TCA cycle, leading to accumulation of succinate and itaconate. Succinate stabilizes HIF-1α to drive inflammation, while itaconate, produced by Irg1 , modulates SDH activity and redox balance ( 47 – 49 ). Together, these metabolites impair OXPHOS , promote ROS generation, and reinforce an M1 pro-inflammatory metabolic state. Notably, our study revealed that within a defined concentration range, PEVs activated the AMPK-ULK1 signaling pathway in a dose-dependent manner. Acting as a cellular energy sensor, AMPK directly phosphorylated ULK1 at sites such as Ser 555 , while simultaneously alleviating mTORC1-mediated inhibition. This dual regulation ensured effective activation of the ULK1 complex, which initiated downstream mitophagy cascades, including recruitment of the class III PI3K complex and facilitation of LC3 lipidation ( 50 – 52 ). Consequently, mitophagy was enhanced, allowing selective clearance of damaged mitochondria and thereby mitigating inflammation-induced mitochondrial dysfunction. These findings suggest that PEVs restore mitochondrial homeostasis and break the cycle of inflammation and metabolic dysregulation by modulating the AMPK-ULK1-mitophagy axis, providing potential therapeutic strategies for endometrial injury–related disease. While our findings establish the AMPK-ULK1-mitophagy axis as a key mechanism through which PEVs modulate macrophages and promote endometrial repair, we recognize that the therapeutic actions of PEVs are likely multifaceted ( 53 ). Beyond mitochondrial regulation, other pathways, particularly those involving lipid metabolism, may contribute substantially. A more complete mechanistic map will not only advance our fundamental understanding but also enhance the rational application and optimization of PEV-based therapies for endometrial regeneration. Beyond these notable biological effects, PEVs also demonstrate unique advantages over PRP in terms of clinical translation. The surplus of clinical-grade allogeneic PLTs presents a valuable and readily available national resource for large-scale production of PEVs. Unlike PRP, which is dependent on an individual’s specific physiological condition, PEVs offer a universally applicable solution, making them especially beneficial for patients who may not be suitable candidates for PRP due to compromised health. In addition, PEVs are not only safe, accessible, and cost-effective, but also anucleated, which alleviates concerns regarding oncogenic risks. They can be produced in large quantities from either autologous PLTs or discarded allogeneic PLTs, with minimal batch-to-batch variability. Moreover, freeze-dried PEVs maintain stability and retain their therapeutic properties even when stored at room temperature, making them particularly suitable for clinical applications and furthering their potential for clinical translation. These findings highlight the potential of further clinical exploration of PEVs in endometrial injury. To our knowledge, no prior studies have investigated the underlying efficacy of PEVs in human models of endometrial injury. Although the sample size in this study was relatively small, the results indicate notable improvement in endometrial health following PEVs treatment. Specifically, patients showed increased endometrial thickness and a more pronounced triple-line pattern after treatment. These findings also suggest that when evaluating PEVs therapy for thin endometrium or IUAs, attention should be paid not only to changes in endometrial thickness but also to the potential of PEVs in enhancing immune tolerance and endometrial receptivity, which may contribute to improved embryo implantation outcomes. While our preclinical and early clinical data suggest that PEVs hold promise for endometrial regeneration and fertility restoration, several limitations must be acknowledged. Most notably, the human clinical data are derived from a small, single-center pilot trial with a limited sample size. Although the results are encouraging, showing improved endometrial parameters and pregnancy outcomes without serious adverse events, they remain preliminary and insufficient for definitive efficacy conclusions. The absence of a control group in this pilot study further limits the strength of clinical inference. We have registered a larger-scale clinical trial (ChiCTR2500102829) to address these limitations. This follow-up study is designed as a randomized, controlled trial with adequate statistical power to rigorously evaluate the safety and efficacy of PEV therapy in a well-defined patient population. Until those results are available, the clinical findings presented here should be interpreted with caution and viewed as exploratory. Future research should also focus on standardizing PEV isolation and characterization protocols, optimizing dosing regimens, and elucidating the long-term safety profile. In addition, comparative studies with existing therapies (e.g., PRP and estrogen supplementation) will be essential to establish the relative therapeutic advantage of PEVs. Despite these limitations, our integrated preclinical and early clinical data provide a foundational rationale for continued investigation into PEVs as a cell-free therapeutic strategy for endometrial injury and infertility. In summary, through murine models and a small-scale clinical study in humans, we comprehensively investigated the therapeutic potential of PEVs in repairing injured endometrial tissue. The results indicate that due to their low immunogenicity, stability, and nanoscale properties, PEVs represent a promising alternative to PRP. Furthermore, PEVs modulate the dysregulated immune microenvironment in injured endometrium by promoting macrophage polarization toward an anti-inflammatory phenotype, achieved through the induction of mitophagy and mitochondrial functional regeneration ( Fig. 8 ). This mechanism contributes to the restoration of endometrial thickness and improves embryo implantation rates. Importantly, preliminary evidence from a small clinical trial ( n  = 8) supports the potential of PEVs in treating endometrial injury–related infertility. These encouraging results enhance the feasibility of this therapeutic strategy and warrant further large-scale investigations. Given their simple preparation, low cost, and well-defined mechanism of action, PEVs offer a promising treatment option for patients with thin endometrium in clinical practice. PEVs can promote mitophagy of macrophages, alleviate inflammation, and induce anti-inflammatory polarization in vivo and in vitro. Using the PEVs to modulate the local immune homeostasis is shown to be effective for endometrium regeneration. LPS, lipopolysaccharide; PEVs, PLT-derived extracellular vesicles; IL-6, interleukin-6; IL-1β, interleukin-1β; TNF-α, tumor necrosis factor–α; ROS, reactive oxygen species; NRF2, nuclear factor erythroid 2–related factor 2 [created in BioRender. Li, D. (2026) https://BioRender.com/ym8jwi5 ].

Infertility, defined as the inability to achieve pregnancy after 12 months of regular unprotected intercourse, affects more than 50 million individuals globally, with profound emotional and societal implications ( 1 ). This highlights the urgent need for accessible and effective reproductive healthcare solutions. Among the diverse causes of infertility, endometrial injury has emerged as a notable barrier to the success of assisted reproductive technologies. Intrauterine adhesion (IUA), a common cause of refractory uterine infertility, is often associated with clinical complications such as hypomenorrhea, amenorrhea, and infertility ( 2 , 3 ). Adequate endometrial thickness is critical for both implantation and sustaining pregnancy ( 4 , 5 ). However, various factors, including repeated curettage, endocrine dysfunction, medical procedures, and infections, can compromise the integrity of the basal endometrium, resulting in thinning and impaired functionality ( 6 , 7 ). Current therapeutic approaches for endometrial injury include hormonal treatments (e.g., estrogen), cytokine administration, sildenafil citrate, low-dose aspirin, surgical interventions, and stem cell therapy ( 8 , 9 ). Among these, platelet-rich plasma (PRP), an autologous product with platelet concentrations above physiological levels, has shown promise in promoting endometrial healing and improving receptivity ( 10 ). Despite its potential, the precise molecular mechanisms behind PRP’s regenerative effects remain unclear, which limits its optimization for complex clinical cases. In addition, PRP therapy faces practical challenges, such as the need for repeated blood collection, variability in platelet concentration, and the absence of standardized preparation protocols. The instability of platelets during storage and transportation also presents considerable hurdles, as platelet activation can trigger inflammatory responses, ultimately compromising the therapeutic outcome ( 11 ). Platelet (PLT)-derived extracellular vesicles (PEVs), nanoscale bilayer vesicles derived from PLTs, retain key biological properties of their parent cells while exhibiting unique advantages attributable to their nanoscale dimensions ( 12 ). These vesicles are highly effective in delivering bioactive molecules, enhancing cellular communication, and modulating immune responses ( 13 – 15 ). PEVs demonstrate superior stability, storage flexibility, and reduced immunogenicity compared to intact PLTs ( 16 ). In this study, we identify PEVs as a key factor underlying the regenerative potential of PRP in treating endometrial injury ( 17 ). Following intrauterine administration, PEVs exhibited prolonged retention within the uterine environment and demonstrated robust regenerative effects, significantly improving live birth rates compared to controls in mice model. Mechanistic investigations revealed that PEVs drive macrophage polarization at inflammation sites through mitophagy pathways, thereby enhancing tissue repair and restoring homeostasis. Furthermore, preliminary clinical evidence from a small cohort of patients with endometrial injury suggests that PEVs may support endometrial repair and create conditions conducive to pregnancy. These findings establish a mechanistic basis for the therapeutic application of PEVs and pave the way for their clinical use in endometrial regeneration, marking a meaningful advancement in reproductive medicine.

Female KM mice (6 to 8 weeks old) weighing approximately 25 to 28 g were purchased from Beijing Vital River Laboratory Animal Technology Co. Ltd. The mice were housed in a specific pathogen–free barrier facility under controlled conditions, with a constant temperature of 25° ± 2°C, humidity maintained between 45 and 55%, and a 12-hour light/dark cycle. They were given unrestricted access to food and water. All animal experiments were conducted in accordance with the ethical guidelines approved by Ethics Committee of Soochow University (approval no. SUDA20231012A02). The sample sizes for the experimental groups were determined and approved by the relevant regulatory authorities to ensure adequate statistical power while considering feasibility and ethical concerns. Before the experiments, the mice were acclimatized for 1 week. Vaginal smears were performed daily between 08:00 and 10:00 a.m. to monitor the estrous cycle stages. Only female mice with regular estrous cycles were selected for inclusion in the subsequent experiments. The mice were randomly assigned to three experimental groups: the Sham group, which underwent laparotomy only; the Untx group, which underwent uterine injury surgery without any treatment; and the PEV-treated group, which received intrauterine PEVs [50 μg per uterine horn per day ( 54 – 56 )] following uterine injury surgery. Endometrial injury was induced in the mice using a previously established protocol. Mice were anesthetized via intraperitoneal injection of 1% pentobarbital sodium at a dose of 10 μl/g. A small vertical incision was made in the lower abdomen to expose the uterus. Ethanol (95%) was then injected into the uterine horn using a 31-gauge insulin syringe, and the uterus was clamped with microhemostatic clamps for 1 min. The uterus was subsequently rinsed repeatedly with 0.9% saline to remove the ethanol. After the abdominal cavity was washed with 0.9% saline, the incision was closed with 4-0 absorbable sutures. All animal experiments were independently repeated three times. This study was approved by the Ethics Committee of The Second Affiliated Hospital of Soochow University (no. LC2024023), and informed consent was obtained from each participant. Endometrial biopsies were collected from nine patients with moderate/severe endometrial injury and nine normal controls (nine Asian females; for details, see Fig. 1B ) for historical analysis during hysteroscopy for infertility screening in the late proliferative phase of the menstrual cycle. Patients with IUA were diagnosed by a skilled gynecologist based on the criteria recommended by the American Fertility Society. Control subjects were women with an endometrial thickness ≥7 mm (measured by B-ultrasound) and no endometrial lesions were found under hysteroscopy. CD138-positive endometrial tissue, which indicated endometritis, was excluded from the enrollment. PLTs were isolated from allogenic KM mice following a previously described protocol. In brief, whole blood was drawn from the mouse’s sinus into a tube containing 200 μl of ACD anticoagulant (KWS, China) and 1 mM PGE1 (HY-B0131, MCE). The blood was then centrifuged at 100 g for 15 min to remove red blood cells. The supernatant was collected and further centrifuged at 800 g for 20 min to pellet the PLTs. To purify the PLTs, the pellet was resuspended in 500 μl of sterile Dulbecco’s PBS (DPBS; P1006, Solarbio) and centrifuged again. PEVs were prepared using the repeated freezing and thawing method. This approach is cost-effective, avoids introducing exogenous activators, minimizes contamination risk, and is more amenable to future large-scale translational applications. A high yield of ~200 μg of PEVs was obtained from one mouse blood, which is two to three times higher than that achieved with the thrombin activation method. To concentrate the PEVs, the samples were frozen at −80°C for 24 hours and then thawed at 37°C for 40 min. Following each freeze-thaw cycle, the samples were subjected to sequential centrifugations: 300 g for 10 min, 1500 g for 15 min, and 3200 g for 20 min, followed by ultracentrifugation at 100,000 g for 2 hours to concentrate the PEVs into granules. Using this standardized procedure, we typically obtained approximately 200 μg of PEVs from one mouse blood. Protein concentration was quantified by bicinchoninic acid (BCA) assay to ensure reproducibility across batches. To ensure consistency in PEVs quality across different donors and batches, we implemented several quality control steps. The morphology of the PEVs was examined using TEM. Protein expression was analyzed by SDS–polyacrylamide gel electrophoresis (PAGE), Western blotting, and mass spectrometry. The size distribution and temporal changes in the PEVs were assessed using DLS. A 20-μl aliquot of fresh PEVs solution was placed onto a 300-mesh copper grid (BZ11023a, EMCN) supported on filter paper and incubated for 5 min. After incubation, the grid was stained with 1% phosphotungstic acid solution for 5 min for negative staining, followed by washing with distilled water and air-drying. The prepared sample was then examined using a FEI TF20 transmission electron microscope at 200 kV, and images were captured for analysis. The particle size and zeta potential of PEVs were measured using DLS. To prepare the sample, 500 μl of DPBS was added to the PEVs, and the resulting solution was placed into a quartz cuvette. The size distribution and zeta potential were measured at 25°C, and the data were analyzed using FLEX software. To assess the stability of the PEVs, particle size measurements were taken over a 7-day period, with DLS measurements conducted daily to monitor any changes. The PEVs were resuspended in a loading buffer (WB2001, NCM Biotech) and loaded onto a 10% polyacrylamide gel (PG122, Epizyme). Protein separation was performed using a running buffer (G2018-15, Servicebio) in an electrophoresis system (Bio-Rad). After electrophoresis, the protein bands were stained with Coomassie Brilliant Blue (PS111, Epizyme) for visualization. The samples were lysed on ice using RIPA buffer (PC101, Epizyme) with a cocktail of phosphatase and protease inhibitors (P1045, Beyotime). Protein concentrations were determined by BCA assay, and equal amounts of protein were loaded onto SDS-PAGE gels (PG122, PG123, Epizyme). The separated proteins were transferred to polyvinylidene difluoride membranes (0.22 μm, ISEQ00010, Sigma-Aldrich). Membranes were blocked with nonfat milk for 2 hours at room temperature, followed by overnight incubation at 4°C with primary antibodies diluted 1:1000. After washing with tris-buffered saline with Tween 20 (G0001, Servicebio), membranes were incubated for 1 hour with secondary antibodies at a 1:5000 dilution. Band detection was performed using ECL reagent ( P10060 , NCM Biotech) and protein levels were quantified with ImageJ software. The primary antibodies used were as follows: CD9 rabbit polyAb (20597-1-AP, Proteintech), CD63 rabbit polyAb (Abs132700, Absin), CD41 rabbit polyAb (TD7456S, Abmart), β-actin rabbit mAb (4970T, CST), GAPDH rabbit polyAb (10494-A-AP, Proteintech), VEGFA rabbit mAb ( Ab214424 , Abcam), TGF-β mouse mAb (GTX21279, GeneTex), P-selectin/CD62P mouse mAb (60322-1-Ig, Proteintech), sodium potassium ATPase rabbit pAb (ab58475, Abcam), LIF rat mAb ( Ab138002 , Abcam), ITGB3 rabbit mAb (13166T, CST), estrogen receptor alpha rabbit mAb (Ab32063, Abcam), Nrf2 rabbit mAb (12721T, CST), Beclin-1 rabbit mAb (3495S, CST), SQSTM1/P62 rabbit pAb (5114T, CST), LC3B mouse mAb (83506S, CST), IL6 rabbit pAb (A0286, Abclonal), IL1β rabbit pAb (A16288, Abclonal), TNF alpha rabbit pAb (P010076, Epizyme), COX1 rabbit mAb ( Ab109025 , Abcam), thromboxane synthase Rabbit mAb (A5173, Abclonal), phospho-AMPK alpha (Thr 172 ), rabbit pAb (2535T, CST), AMPK alpha 1 rabbit mAb (R013729, Epizyme), phospho-ULK1 (Ser 556 ) Rabbit mAb (R014139, Epizyme), and ULK1 rabbit mAb (R014176, Epizyme). Secondary antibodies used were anti-mouse immunoglobulin G (IgG) horseradish peroxidase (HRP)–linked (SA00001-1), anti-rabbit IgG HRP-linked (SA00001-2), and anti-rat IgG HRP-linked (SA00001-15), all from Proteintech. LC-MS/MS was conducted on PEVs to identify the proteins. The samples were added with 800 μl of SDT cracking solution, mixed and boiled at 100°C for 5 min, ultrasound, centrifuged, and the supernatant was taken. Then, the liquid phase separation was carried out by liquid chromatography column and then analyzed by mass spectrometry with QExactive HF-X mass spectrometer (Thermo Fisher Scientific) for 60 min. The raw data of mass spectrometry were searched by MaxQuant 1.5.5.1 software, and the results of protein identification and quantitative analysis were obtained. RAW264.7 cells (catalog number TCM13) were purchased from the Cell Bank, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. These cells were cultured in low-glucose Dulbecco’s modified Eagle’s medium (L-DMEM, Gibco), supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin (C100C5, NCM Biotech) to provide essential nutrients and prevent bacterial contamination. The cells were incubated at 37°C in a humidified atmosphere with 5% CO 2 . Bone marrow–derived macrophages (BMDMs) were isolated from the femurs and tibias of C57BL/6J mice. After euthanasia, bones were dissected and marrow was flushed with cold PBS. The cell suspension was passed through a 70-μm strainer, and erythrocytes were lysed with ACK buffer. Cells were plated in complete DMEM supplemented with M-CSF (20 ng/ml; RP01216, Abclonal) and cultured for 7 days. Fresh medium with M-CSF was added on day 3 and fully replaced on day 6. Mature, adherent BMDMs were harvested for experiments. Mice were initially anesthetized using isoflurane mixed with oxygen, and then DiR-labeled PEVs (DiR, HY-D1048, MCE) were administered into the uterus. Four mice were included in each group. To simulate the clinical scenario where patients remain in the lithotomy position after intrauterine perfusion, the cervix was ligated for 2 hours to retain the PEVs in the uterine cavity. After a 24-hour period, the retention of PEVs in the uterus was monitored using the IVIS Spectral Imaging System (PerkinElmer Ltd.). In addition, major organs were harvested for further in vitro imaging. Fluorescence intensity was quantified as total radiance ([p/s]/[μW/cm 2 ]) using the IVIS Living Image 4.2 software. Confocal microscopy was employed to visualize the expression of DiR in the uterine tissue. For immunofluorescence, cells or tissue sections were initially fixed in 4% paraformaldehyde for 15 min. Following fixation, samples were blocked with 10% FBS for 2 hours at room temperature. After blocking, the samples were incubated overnight at 4°C with primary antibodies, and subsequently with secondary antibodies for 1 hour at room temperature. Nuclear staining was performed using 4′,6-diamidino-2-phenylindole (DAPI; G1012, Servicebio). The following primary antibodies were used: phalloidin-488 (YP0059S, UElandy), PSGL-1 rabbit monoclonal antibody (1:1000, A23373, Abclonal), CK-19 rabbit polyclonal antibody (1:1000, 10712-1-AP, Proteintech), and vimentin rabbit monoclonal antibody (1:1000, Ab92547, Abcam). The secondary antibody, goat anti-rabbit Alexa Fluor 594 (A11012), was applied at a dilution of 1:200, and was purchased from Invitrogen. After treatment, the uteri from various time points in each group were harvested, rinsed with saline to eliminate excess blood, and fixed in 4% paraformaldehyde solution. The tissues were subsequently embedded in paraffin. Sections of 4 μm thickness were prepared from the paraffin blocks and subjected to H&E staining, Masson’s trichrome staining, and immunohistochemical (IHC) analysis. The stained sections were then mounted with neutral resin, and images were captured for further examination. The proportion of positively stained areas was quantified using ImageJ software, while measurements of endometrial thickness and the labyrinth zone area were performed using ImagePro Plus 6.0. Whole embryos were serially dehydrated in ethanol and incubated in 4% (w/v) phosphotungstic acid diluted in 70% EtOH for 4 days at 4°C. Embryos were scanned at 50 kV/800 μA with a 0.5-mm Al filter at a resolution of 17.1 μm using a benchtop μCT scanner (SkyScan 1174, Bruker). two-dimensional images were reconstructed using NRecon and DataViewer software. Mice were anesthetized by intraperitoneal injection of 1% pentobarbital sodium, and sterile PBS was infused from the left ventricle to remove circulating blood cells in the tissue. Single-cell RNA sequences were obtained from the whole uterus of untreated and PEV-treated mice. The tissue was digested with enzymes and processed into a single-cell suspension. The oligo (dT)–based complementary DNA database is a droplet-partitioning barcode using the Chromium Single Cell Controller (10-fold genomics) system in the National Cancer Institute–Center for Cancer Research (NCI-CCR) single-cell analysis tool. We carried out the removal of dead cells and adjusted the cells to the required concentration. Sequencing was performed on a NovaSeq (Illumina) at the NCI-CCR sequencing facility, Bioinformatic analysis was performed using the OmicStudio tools at www.omicstudio.cn/tool . As for tissue, mice uterine tissues were minced with scissors and digested at 37°C for 40 min with type IV collagenase (1.0 mg/ml; Sigma-Aldrich, C5138) prepared in RPMI 1640. The digested tissues were then ground and passed through a 70-μm strainer to obtain a single-cell suspension, and for cells, they can be stained after washing with 3% bovine serum albumin at 300 g , 4 min, 4°C. The cell surface was then stained using the following flow cytometry antibodies: anti-mouse CD45-PE/Cy7 (BioLegend, 157206), anti-mouse F4/80-FITC (BioLegend, 123108), anti-mouse F4/80-APC (BioLegend, 123116), anti-mouse CD80-PE (BioLegend, 104708), anti-mouse CD206-APC (BioLegend, 141708), anti-mouse CD11b-Percp (BioLegend, 101230), anti-mouse Ly6G-PE (BioLegend, 164503), anti-mouse Ly6C-FITC (BioLegend, 128005), anti-mouse CD3-PE (BioLegend, 100206), anti-mouse CD4-AlexaFlour (BioLegend, 100425), and anti-mouse CD8a-APC (BioLegend, 100712). RAW264.7 cells were cultured in six-well plates in L-DMEM medium for 24 hours in the presence or absence of PEVs and LPS. Total RNA from the harvested cells was extracted with TRIZOL reagent (Invitrogen, Carlsbad, CA, USA). The RNA amount and purity of each sample were quantified using NanoDrop ND-1000 (NanoDrop, Wilmington, DE, USA). The RNA integrity was assessed by Bioanalyzer 2100 (Agilent, CA, USA) with RNA integrity number >7.0, and confirmed by electrophoresis with denaturing agarose gel. At last, we performed the 2 × 150 bp paired-end sequencing (PE150) on an Illumina Novaseq 6000 (LC-Bio Technology Co. Ltd., Hangzhou, China) following the vendor’s recommended protocol. Low-quality reads and adapter sequences were removed from raw reads. Then, reads were aligned against the mouse genome with v.GRCm38.91. On the basis of this matrix, differential genes expressed between two groups were identified (R version 3.5.1, DESeq2 version 1.20). The differentially expressed mRNAs were selected with fold change more than 1.5-fold with parametric F-test comparing nested linear models ( P value < 0.05) by R package edgeR ( https://bioconductor.org/packages/release/bioc/html/edgeR.html ). Cells were harvested, pelleted, and fixed with 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.0 to 7.5) at 4°C overnight. The fixed pellets were dehydrated using a gradient ethanol series (50, 70, 90, and 100%), and subsequently embedded in resin with silicone molds for 48 hours at 65°C. Trimmed blocks were then sectioned into ultrathin sections (70 to 90 nm thick) using Ultramicrotome (EM UC7, Leica). Then, sections were stained with lead citrate and uranyl acetate solutions respectively for 5 min and observed with a HITACHI HT 7800 electron microscope. The mitochondrial potential was analyzed via JC-1 staining. JC-1 probe aggregates in the matrix of the mitochondria to form the polymer at high mitochondrial membrane potential, producing red fluorescence (excitation/emission = 585/590 nm); and form the monomer at low potential, producing green fluorescence (excitation/emission = 510/527 nm). The JC-1 probe (2 μM, HY-K0601, MCE) was coincubated with cells at 37°C for 20 min, and photographed with a ZEISS confocal microscopy. To determine the change in mitochondrial membrane potential, the ratio of red/green fluorescence intensity was analyzed via ImageJ software. Cells were incubated with MitoTracker Red CMXRos (50 nM, M7512, Invitrogen) for 15 min and DAPI (G1012, Servicebio) for 15 min at 37°C, then washed with warm PBS, and subsequently observed and photographed using a ZEISS confocal microscope. To examine the mitochondrial superoxide, cells were incubated with MitoSOX Red mitochondrial superoxide indicator (2 μM, 40778ES50, YEASEN) for 20 min and Hoechst (P0133, Beyotime) for 20 min at 37°C away from light, and observed using a ZEISS confocal microscopy. Cytokines in cell culture supernatants and mouse uterine tissue homogenates were measured by enzyme-linked immunosorbent assay (ELISA) kits (88-7324-88, 88-7064-88, and 8-7013-88, Invitrogen) according to the manufacturer’s instructions. In brief, the culture supernatant and tissue homogenates were collected. Supernatant samples (100-μl well) were added to the plate with capture antibody coating and incubated at room temperature for 2 hours or overnight at 4°C and washed three times with 250 μl per well. Then, 200 μl per well blocking buffers were added for 2 hours. Then, 100-μl diluted detection antibodies were added to all wells and followed by incubation at room temperature for 1 hour, and washing three times with 250 μl per well washing buffer. One hundred microliters per well of diluted avidin-HRP was added, followed by incubation at room temperature for 30 min, and washing three times with 250 μl per well washing buffer. One hundred microliters per well of trimethylboron solution was added, followed by incubation at room temperature for 15 min. One hundred microliters per well of stop solution was added and the plate read at 450 nm and the data were analyzed. This study was approved by the Ethics Committee of The Second Affiliated Hospital of Soochow University, Suzhou, China (project no. LC2024023; approval no. JD-LC2024023-I01; and registration number: ChiCTR2500102829). Written informed consent was obtained from all participants before their inclusion in the study. The research was conducted at the Reproductive Medicine Center of The Second Affiliated Hospital of Soochow University. Three patients, who voluntarily consented to participate, were included in the study. These participants had previously experienced embryo transfer failures and were currently undergoing an artificial cycle. The study adhered to Good Clinical Practice guidelines and complied with relevant regulatory requirements, including the Declaration of Helsinki and the ICH Harmonised Tripartite Guidelines. Human PLTs were isolated according to a protocol optimized by our group. In brief, 20 ml of whole blood was collected from the autologous antecubital vein into an ACD anticoagulant tube. The blood was then centrifuged at 200 g for 12 min to remove red blood cells. If red blood cells were still visible in the supernatant, the centrifugation step was repeated at 200 g for 3 min. The resulting supernatant was collected and centrifuged at 1900 g for 5 min to pellet the PLTs. The PLT pellet was resuspended in 2 to 3 ml of sterile DPBS (P1006, Solarbio) and subjected to a second centrifugation to obtain purified PLTs. To concentrate PEVs, the PLTs underwent three cycles of freezing and thawing: frozen at −80°C for 24 hours and thawed at 37°C for 50 min. After each freeze-thaw cycle, the sample was centrifuged sequentially at 300 g for 10 min, 1500 g for 15 min, and 3200 g for 20 min. Final concentration of PEVs was achieved by ultracentrifugation at 100,000 g for 2 hours, yielding approximately 2500 μg of PEVs from 20 ml of human blood. The morphology of the PEVs was examined by TEM, and protein expression was analyzed by SDS-PAGE and Western blot. Before clinical use, the product underwent rigorous quality control, including sterility testing, endotoxin testing, and analysis of particle and protein concentrations. Human autologous PEVs were formulated in DPBS for perfusion. The final preparation was stored at −80°C. Before clinical use, the product underwent rigorous quality control, including sterility testing, endotoxin testing, and analysis of particle and protein concentrations. Human autologous PEVs (~400 to 500 μg) were formulated in DPBS for perfusion ( 57 ). The final preparation was stored at −80°C. Enrolled patients were women scheduled for frozen embryo transfer but with insufficient endometrial thickness due to thin endometrium or IUA. Inclusion criteria were as follows: women aged 25 to 45 years, with a body mass index between 18 and 30 kg/m 2 , normal baseline hormone levels, and a poor endometrium situation due to insufficient endometrial thickness, such as thin endometrium, IUA. Exclusion criteria included women aged over 45 years, those with other uterine conditions (e.g., uterine fibroids, adenomyosis, or uterine malformations), and those with serious medical disorders such as thrombophilia or cardiovascular and respiratory diseases. Oral antibiotics (Levofloxacin, Japan) were administered to all patients for 1 day following intrauterine perfusion to prevent infection. Surgical complications, including uterine perforation and postoperative lower abdominal pain, were recorded. In addition, body temperature and hematological parameters, such as leukocyte count and serum C-reactive protein, were monitored and documented. Data collection and analysis were not performed blind to the conditions of the experiments. All experiments included at least three biological replicates per group, and data are presented as mean ± SEM or SDs. The significance of differences between two groups was calculated by a two-tailed unpaired Student’s t test. In addition, analysis of variance (ANOVA) comparisons and Bonferroni’s post hoc tests were performed between more than two groups (multiple comparisons). All statistical analyses were performed using GraphPrism (version 9.5). Values of P  < 0.05 were considered significant. All intensities of fluorescence expression in the experiments were further calculated by ImageJ software. The standard symbols were presented as * P  < 0.05, ** P  < 0.01, *** P  < 0.001, and **** P  < 0.0001.