Repair of female reproductive function by GDF-9–overexpressing extracellular vesicles via ACVR1B/SMAD2 regulation in ovarian granulosa Authors: Sichen Zhang, Huidong Liu, Tianyue Zhang, Yansong Wang, Mengtong Zhang, Xiao Zhang, Hansong Li, Shuai Huang, Lin Liang, Ke Cheng, Shaowei Wang Published: 2025 ABSTRACT Aging in females is characterized by dysfunctional ovaries (DOs). While human umbilical cord mesenchymal stem cell (HucMSC)–derived extracellular vesicles (EVs) have shown promise in ameliorating DO, the mechanisms underlying their effects remain poorly understood. In this study, we investigated the therapeutic potential of growth differentiation factor 9 (GDF-9)–overexpressing EVs (GOEs) on granulosa cells. We identified the activin A receptor type 1B (ACVR1B) as a critical target for the restoration of ovarian function. Genetic modification of HucMSCs to overexpress GDF-9 resulted in the production of GOEs, which, through ACVR1B activation, induced SMAD2 phosphorylation in the nucleus, thereby rescuing ovarian function. These findings provide previously unknown insights into the mechanisms by which EVs mediate ovarian recovery and propose GOEs as a potential therapeutic strategy for the treatment of DO. INTRODUCTION Ovaries are essential gonads in the female reproductive system, responsible for housing oocytes and producing hormones crucial for reproductive function and sexual characters (1–3). Ovarian dysfunction encompasses a spectrum of conditions, from hormonal irregularities to structural anomalies, disrupting normal ovarian processes and leading to manifestations like irregular menstrual cycles, infertility, ovarian cysts, polycystic ovary syndrome, premature ovarian failure, and ovarian tumors (3, 4). Despite being a longstanding challenge in women’s health for over three decades, effective therapeutic interventions for dysfunctional ovary (DO) have remained elusive (5, 6). However, recent strides in regenerative medicine have paved the way for innovative approaches to address this issue (3, 7). Our research team has spearheaded the development of the ovarian regenerative patch (ORP) (8), a groundbreaking therapeutic strategy aimed at tackling DO. ORP integrates human umbilical cord mesenchymal stem cells (HucMSCs)–derived secretomes, enriched with vital bioactive molecules, into a clinically approved surgical pelvic scaffold (9–11). This pioneering approach enables the targeted delivery of therapeutic factors directly to the ovary, facilitating tissue repair and regeneration (12, 13). Further investigation has unveiled that the regenerative effects of ORP are driven by MSC-derived extracellular vesicles (EVs) (14, 15), which harbor bioactive factors like growth differentiation factor 9 (GDF-9), bone morphogenetic protein 15 (BMP-15), and other members of the transforming growth factor–β (TGF-β) superfamily (16–18). GDF-9, an oocyte-specific member of TGF-β superfamily, plays a critical regulatory role in early folliculogenesis. It is expressed in oocytes beginning at the primary follicle stage and is essential for promoting oocyte growth and the acquisition of meiotic competence. In addition, GDF-9 exerts indirect effects on oocyte maturation by modulating the function of surrounding cumulus cells. Specifically, it up-regulates the expression of cumulus expansion-related genes, including hyaluronan synthase 2 (HAS2), pentraxin 3 (Ptx3), and tumor necrosis factor–α–induced protein 6 (TNFAIP6), all of which are crucial for extracellular matrix remodeling and the establishment of ovulatory capacity. Moreover, GDF-9 is indispensable for granulosa cell (GC) proliferation, a process fundamental for the formation of the theca cell layer and the overall progression of follicular development. At the molecular level, GDF-9 signals through TGF-β type I and type II receptors, primarily ALK5 [activin A receptor type 1B (ACVR1B)] and BMPRII, to initiate SMAD2/3 phosphorylation. The resulting SMAD2/3–SMAD4 complex translocate into the nucleus to modulate the transcription of downstream target genes. Through this canonical SMAD-dependent pathway, GDF-9 not only promotes GC mitotic activity but also inhibits premature luteinization, thereby ensuring proper follicular maturation (19, 20). Despite its critical biological functions, GDF-9 exhibits limited stability and poor bioavailability in vivo, posing notable challenges for therapeutic translation. Our study addresses these limitations by delivering GDF-9 via MSC-derived EVs, which serve as protective and targeted delivery carriers. This EV-based strategy enhances the stability and cellular uptake of GDF-9, offering a novel therapeutic platform—ORP—for the restoration of ovarian function in cases of ovarian insufficiency. This study delved into the mechanistic underpinnings driving the regenerative effects of ORP. Initially, we characterized the structural composition of EVs derived from HucMSCs, shedding light on their functional properties (21–23). Subsequently, through in vitro models, we explored the efficacy of EVs in treating GCs, pivotal players in ovarian function (24, 25). Using RNA sequencing (RNA-seq) analysis, we identified ACVR1B as a crucial gene implicated in both normal ovarian physiology and dysfunction, serving as a significant target for therapeutic intervention (26). Further mechanistic elucidation revealed the regulatory role of GDF-9 as the upstream molecule of ACVR1B in GCs (27, 28). Upon binding, GDF-9 and ACVR1B synergistically activate the SMAD2 signaling pathway, promoting GC proliferation, follicle maturation, and ovulation. This comprehensive investigation provides a mechanistic framework elucidating the interplay between GDF-9, ACVR1B, and the SMAD2 signaling pathway in governing ovarian function (29, 30). Moreover, the elucidation of the therapeutic mechanism of EVs derived from HucMSCs lays a solid foundation for the development of EV-based therapies for DO (31, 32). These EVs can serve as a therapeutic delivery system to precisely deliver bioactive factors like GDF-9 to the ovary, promoting tissue repair and regeneration (Fig. 1). With further refinement and optimization, EV-based therapies hold significant promise for clinical translation, offering a minimally invasive and targeted approach for restoring ovarian function and improving outcomes for women affected by DO. EVs as a therapeutic delivery system for targeted GDF-9 delivery to the ovary. GDF-9 activates ACVR1B, resulting in SMAD2 phosphorylation in the nucleus, ultimately rescuing ovarian function. RESULTS EVs derived from umbilical cord MSCs can be isolated and purified using ultrafiltration techniques EVs are small, membrane-bound structures released by cells into their extracellular environment (33). Laden with an assortment of biomolecules including proteins, lipids, nucleic acids, and metabolites, EVs serve as mediators of intercellular communication and regulatory processes (34). In this investigation, our focus centered on the isolation and refinement of EVs derived from HucMSCs (fig. S1A). Using transmission electron microscopy (TEM), we scrutinized the morphological characteristics of the isolated EVs, affirming their integrity and structure (fig. S1B). To quantitatively evaluate the purification efficiency of EVs and GDF-9–overexpressing EVs (GOEs), we conducted a series of assessments using nanoparticle tracking analysis (NTA), bicinchoninic acid (BCA) assay, and enzyme-linked immunosorbent assay (ELISA). NTA revealed that the mean particle diameters of EVs and GOEs were 146 and 167.5 nm, respectively (fig. S1, C and E). Furthermore, the identification of EV-specific markers, such as CD63 and CD9, was achieved through Western blot analysis, substantiating the purity and authenticity of the isolated EV population (fig. S1D). In addition, to validate GDF-9 enrichment and compare protein profiles, we performed liquid chromatography–tandem mass spectrometry (LC-MS/MS) proteomic analysis of GOEs versus control EVs. Semiquantitative analysis using label-free quantification (LFQ) revealed a significant up-regulation of GDF-9 in GOEs compared to control EVs (fig. S1F). The relative intensity values showed a 2.9-fold increase in GOEs versus control EVs. To quantitatively validate the successful loading of GDF-9 into EVs, we performed ELISA-based detection to assess the absolute concentration of GDF-9 in both intracellular and extracellular compartments following transfection. Two different delivery systems—Lipofectamine MessengerMAX and CALNP mRNA in vitro transfection reagents—were used to introduce GDF-9 mRNA into MSCs, with three mRNA dosage levels: 1, 2, and 5 μg/ml. After 24 hours of transfection, cells and supernatants were collected separately for protein extraction and subsequent GDF-9 quantification. As illustrated in fig. S1G, both transfection methods showed dose-dependent increases of intracellular GDF-9 protein with increasing GDF-9 mRNA concentrations. The intracellular GDF-9 protein concentrations using Lipofectamine MessengerMAX transfection reached 19.99, 16.97, and 20.20 ng/ml at mRNA doses of 1, 2, and 5 μg/ml respectively. In comparison, CALNP transfection resulted in intracellular GDF-9 levels of 10.86, 20.95, and 29.05 ng/ml at the same respective doses. The extracellular (EV-associated) GDF-9 protein concentrations for Lipofectamine MessengerMAX–transfected cells were 33.65, 37.41, and 38.21 ng/ml, while those for CALNP were 10.71, 17.44, and 31.16 ng/ml, respectively (fig. S1H). These results confirm the effective expression and secretion of GDF-9 into the EV fraction, with a dose-dependent increase particularly evident in the CALNP-transfected groups. On the basis of CALNP’s superior dose-responsive GDF-9 production, we transfected MSCs with GDF-9 mRNA (untreated cells as controls) and isolated EVs via ultracentrifugation, yielding GOEs and control EVs. As shown in table S1, the particle concentrations were 2.56 × 1011 ± 8.14 × 1010 particles/ml for EVs and 1.28 × 1011 ± 2.14 × 1010 particles/ml for GOEs. Protein concentrations were 1353 μg/ml for EVs and 566 μg/ml for GOEs (table S1 and fig. S1I). Purity was calculated using the standard formulaPurity=Particles/mlProtein (μg/ml) On this basis, the calculated purities were 1.89 × 108 particles/μg for EVs and 2.13 × 108 particles/μg for GOEs (table S1). When normalized to 2 × 109 particles, ELISA detected 3.42 ng GDF-9 in EVs versus 12.03 ng in GOEs (fig. S1J, P = 0.0065), demonstrating a 3.5-fold enrichment and confirming CALNP’s efficacy in generating functional GDF-9–loaded EVs. This ratio closely matched the GDF-9 protein expression levels measured by LC-MS/MS. This differential expression was statistically significant (P < 0.05), suggesting that GDF-9 is enriched in GOEs under the experimental conditions. Through this systematic approach, we meticulously delineated the properties and composition of HucMSC-derived EVs, laying the groundwork for further investigations into their functional roles and therapeutic potential. EVs exhibit potential in enhancing the viability of GCs following cisplatin-induced injury in vitro To explore the functional implications of EVs in cellular dynamics associated with survival and apoptosis, we used human GCs as the experimental model in vitro (35). Furthermore, we established a model simulating DOs using cisplatin at a concentration of 2.7 μg/ml, which was confirmed using the Cell Counting Kit-8 assay (Fig. 2A) (31).We divided the cells into four groups: control, DO, DO treatment with EVs, and DO treatment with GOEs. In addition, flow cytometry was used to evaluate cell viability, apoptosis, and death within each group (Fig. 2B). In the group of DO treated with GOEs, the percentage of viable cells is significantly higher compared to both the DO group and approaches that of the treated with EVs group (Fig. 2C). In addition, in the DO group treated with GOEs, there is a significant decrease in the percentage of apoptotic cells compared to the DO group and DO treat with EVs (Fig. 2D). Meanwhile, no significant differences were found in the proportion of mechanically damaged and dead cells among the four groups (Fig. 2E). These findings suggest that GOEs confer stronger antiapoptotic effects than EVs, further supporting their therapeutic potential in preserving GC integrity in the context of ovarian dysfunction. Briefly, GDF-9–overexpressing MSCs were cultured in the upper chamber, while cisplatin-treated GCs (DO) were cultured in the lower chamber. After 24 hours, apoptosis levels in GCs were assessed. However, there was no statistically significant difference in apoptosis rates between the GOE-treated group and the MSCs group (Fig. 2, B and F). These findings suggest that the antiapoptotic effects observed are primarily mediated by GDF-9–rich EVs rather than other factors secreted by MSCs. This supports the hypothesis that EVs themselves are sufficient to confer the therapeutic benefits observed in GC survival and ovarian function restoration. The potential of EVs in enhancing GC viability. (A) Schematic illustration of the in vitro DO model; concentration of cisplatin, 2.7 μg/ml; EV/GOE, 1 × 106 particles/ml; time of treatment, 24 hours (h). (B) Flow cytometry analysis of GC viability, apoptosis, and cell death. (C to E) Quantification of flow cytometry data, showing the efficacy of EVs in treating DO in vitro. (F) No statistically significant difference in apoptosis rates between GOEs and GDF-9–overexpressing MSCs. n = 3, data are presented as means ± SD. This figure was created in part using BioRender.com. Created in BioRender. Zhang, S. (2025) https://BioRender.com/zv1tbzt. IC50, median inhibitory concentration. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant. GDF-9 playing a pivotal role in mediating the improvement of GCs viability and proliferation in response to EVs EVs demonstrated a significant capacity to increase the quantity of GCs compared to the control group. EVs encapsulate multiple molecules (36). To investigate the target molecules within EVs for treating DO (37), we transfected GCs with either GDF-9 mRNA to induce GDF-9 overexpression or si-GDF-9 mRNA to inhibit GDF-9 expression (Fig. 3A). Following treatment with EVs, the cell counts of GDF-9–overexpressed GCs was higher than that of the control group. However, in the absence of GDF-9 gene expression, EV treatment failed to enhance the number of GCs (Fig. 3B). In addition, apoptosis in GDF-9 GCs treated with EVs was reduced compared to si-GDF-9 GCs treated with EVs, as confirmed by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) assay (Fig. 3, C and D). Moreover, the proliferation marker Ki-67 exhibited higher expression levels in GDF-9 GCs compared to si-GDF-9 GCs (Fig. 3, E and F). GDF-9 is critical for GC viability and proliferation in EV treatment. (A) Schematic illustration of in vitro GC induction. (B) Cell count analysis showing higher viability in GDF-9–overexpressing GCs and lower viability in si-GDF-9 GCs compared to the control group. (C and D) TUNEL assay showing reduced apoptosis in GDF-9–overexpressing GCs versus si-GDF-9 GCs treated with EVs. (E and F) Ki-67 staining indicating higher proliferation in GDF-9–overexpressing GCs compared to si-GDF-9 GCs treated with EVs (scale bar, 50 μm). n = 3. Data are presented as means ± SD.*P < 0.0.5, **P < 0.01, ***P < 0.001; ns, not significant. This figure was created in part using BioRender.com. Created in BioRender. Zhang, S. (2025) https://BioRender.com/iz9gv36. DAPI, 4′,6-diamidino-2-phenylindol. Total RNA-seq analysis unveiled ACVR1B as the principal regulator in GCs To figure out the target gene on ovaries associated with ovarian function, we used ovarian dysfunction models with cisplatin following our previous research, and then the ovaries from normal and DO rats are sampled and analyzed by total-RNA sequence (Fig. 4A). RNA-seq analysis identified 521 differentially expressed genes (DEGs) between the control and DO groups (fig. S2A), among which GDF, a key oocyte-derived member of the TGF-β superfamily located on chromosome 5, was significantly down-regulated (fig. S2B). KEGG pathway enrichment revealed notable dysregulation in cytokine-cytokine receptor interaction pathways, indicating impaired signaling between oocytes and surrounding GCs (Fig. 4B). ACVR1B is a key regulator in GCs. (A) Schematic illustration of rat DO models, sampled and analyzed by total RNA-seq. (B) KEGG pathway enrichment analysis revealing the cytokine-cytokine receptor interaction signaling pathway. (C) Volcano plot of RNA-seq analysis. (D and E) Western blot detection of GDF-9, ACVR1B, SMAD2, and the phosphorylation status of SMAD2. (F and G) Western blot analysis showing that inhibition of GDF-9 gene expression prevents EVs from stimulating GDF-9 signaling, leading to decreased protein levels of ACVR1B and SMAD2/p-SMAD2. n = 3. Data are presented as means ± SD. *P < 0.0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant. This figure was created in part using BioRender.com. Created in BioRender. Zhang, S. (2025) https://BioRender.com/k36cu40. To further explore the mechanism of EV-mediated repair, we conducted RNA-seq on ovaries from DO rats treated with EVs. Comparative analysis between DO and EV-treated ovaries revealed 1622 DEGs (fig. S3). Notably, components of this pathway, including GDF-9 and its type I receptor ACVR1B, were among the enriched signals (27, 38). According to the canonical pathway, GDF-9 binding to ACVR1B is known to activate SMAD2 phosphorylation, which in turn forms a complex with SMAD4 and translocates to the nucleus to regulate gene expression associated with GC proliferation and follicular development (Fig. 4C). Furthermore, to elucidate gene expression patterns under varied conditions, Western blot analysis was conducted to assess the levels of GDF-9, ACVR1B, SMAD2, and the phosphorylation status of SMAD2. The GDF-9 protein level exhibited a significant increase upon EV treatment in GCs. Conversely, when the GDF-9 gene was suppressed by follistatin (39), an antagonist of GDF-9, EVs failed to stimulate GDF-9 signaling. Similarly, the protein levels of ACVR1B expression were decreased when the GDF-9 gene was inhibited. This observation supports the notion that EVs enhance GC function via the GDF-9/ACVR1B signaling pathway (Fig. 4, D and E). Subsequently, the reverse experiment for the GDF-9/ACVR1B/SMAD2 pathway was formulated. GCs were subjected to either stimulation or inhibition of the GDF-9 or ACVR1B gene, followed by treatment with EVs. The protein levels of these cells were then evaluated via Western blot analysis. GCs were subjected to genetic manipulations targeting GDF-9 and ACVR1B, followed by the reintroduction of functional genes via rescue constructs. Upon restoration of GDF-9 expression in GCs, a consequential reactivation of ACVR1B-mediated signaling pathways was observed, evidenced by the phosphorylation of downstream effector SMAD2. When the ACVR1B gene was stimulated by ALK-4 in GCs (27), but the GDF-9 gene was knocked out, a notable absence of SMAD2 protein expression in GCs was observed. The knockout of ACVR1B using the inhibitor SB431542 in GCs results in the inability of GCs to express the specific protein GDF-9 (Fig. 4, F and G) (40). This observation underscores the critical role of ACVR1B in regulating GDF-9 expression in GCs, highlighting its significance in mediating GC function and reproductive processes. This compelling evidence underscores the hierarchical relationship within the GDF-9/ACVR1B/SMAD2 pathway, affirming GDF-9 as the upstream regulator crucial for initiating ACVR1B-mediated signaling cascades in GCs. GOEs can make ovarian function in rats GOEs present a promising avenue for therapeutic intervention in DO rat models induced by cisplatin. To comprehensively investigate the impact of GOEs on ovarian dysfunction, we designed a controlled animal experimental study. To investigate the administration and biodistribution of EVs in the body, live imaging revealed that the injected EVs were distributed in both ovaries and were cleared within 48 hours (fig. S4). Furthermore, EVs were observed to accumulate in the follicles, most likely being taken up by GCs within the follicles (fig. S5). Four distinct groups were established, including a control group, a DO group, a DO + EVs group, and a DO group treated with GOEs (Fig. 5A). Upon administering purified GOEs to DO rats, a significant increase in body weight, ovarian size, and weight were observed compared to untreated DO rats (Fig. 5, B to E, and fig. S6). The measured ovarian sizes in the control, DO, DO + EVs, and DO + GOEs groups were 6.867 ± 0.2200, 5.453 ± 0.2189, 6.858 ± 0.2065, and 7.617 ± 0.1435 mm, respectively. Ovarian size was significantly reduced in the DO group compared to the control (P = 0.042), DO + EVs (P = 0.0017), and DO + GOEs (P < 0.0001) groups. The ovarian index can provide a normalized parameter that accounts for interindividual variability in body size and offers a more accurate representation of ovarian physiological function. Our results demonstrated a significant increase in the ovarian index in the DO + GOEs group (0.6215 ± 0.01916) compared to both the DO group (0.3821 ± 0.01299, P = 0.0001) and the DO + EVs group (0.4927 ± 0.03629, P = 0.037) (Fig. 5F). GOEs improve ovarian function in DO rats. (A) Schematic illustration of GOE design and animal experimental setup. (B and C) Body weight of rats before and after DO modeling or GOE treatment. (D and E) Ovarian size and weight of rats after DO modeling or GOE treatment. Scale bar, 2 cm. (F) Analysis of ovarian index. (G) Ovarian histology posttreatment. Representative sections are shown in the top row; black boxes indicate magnified regions. Arrowheads indicate PMF, arrows indicate SCF, and stars indicate ATF. Scale bar, 100 μm. (H to J) Quantification of follicles in control, DO, and GOE treatment groups. n = 6. Data are presented as means ± SD. *P < 0.0.5, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant. This figure was created in part using BioRender.com. Created in BioRender. Zhang, S. (2025) https://BioRender.com/c7lj7bl. H&E, hematoxylin and eosin. Moreover, estrous cycle monitoring (fig. S7, A to E) revealed that the ratio of proestrus and estrus phases in the DO + GOEs group (0.33 and 0.27, respectively) were significantly longer than in the DO group (0.12 and 0.07, respectively; P < 0.001 and P = 0.015). In contrast, the ratio of metestrus and diestrus phases in the DO + GOEs group (0.17) were significantly shorter compared to the DO group (0.44; P < 0.001) (fig. S7F). In addition, serum hormone analysis, including estrogen (E2), follicle-stimulating hormone (FSH), and anti-Müllerian hormone (AMH) demonstrated increased levels of E2 and AMH, while FSH levels were reduced in the DO + GOEs group compared to the DO group. These findings suggest that GOEs effectively restore ovarian function in the DO model (fig. S7, G to I). Fertility rescue experiments further supported these findings. Sperm embolus was observed 24 hours after mating (fig. S8A), and pregnancy was confirmed 15 to 20 days later (fig. S8B). The pregnancy rate in the DO + EVs and DO + GOEs groups was 26.7%, significantly higher than the DO group, which had a pregnancy rate of 0% (fig. S8F). Litter size and growth were also recorded. The DO + GOEs group showed a significantly larger litter size (17 ± 2.32) compared to the DO (0) (P = 0.001) and DO + EVs (7 ± 1.58) (P = 0.048) groups (fig. S8, C and G). The weight of pups in the DO + GOEs group was similar to the control group at 14 days of age (fig. S8I). Pup viability was assessed on the basis of movement (movie S1), sucking reflex (fig. S8E and movies S2 and S3), righting reflex (movies S4 and S5), and skin color (fig. S8D). The pup viability score (table S2) in the DO + GOEs group (6.16 ± 2.167) was significantly higher than the DO + EVs group (4.16 ± 2; P = 0.04) (fig. S8H). Furthermore, histological examination revealed a notable augmentation in the number of primordial follicles (PMFs) and secondary follicles (SCFs) within the GOE-treated group, accompanied by a marked reduction in atretic follicles (ATF) (Fig. 5G). The number of PMF was significantly reduced in the DO group (6.17 ± 0.98) compared to the control (12.33 ± 1.91, P = 0.03), DO + EVs (12.00 ± 1.24, P = 0.04), and DO + GOEs (12.00 ± 1.37, P = 0.04) groups (Fig. 5H). Similarly, SCFs were markedly decreased in the DO group (3.17 ± 0.31) relative to the control (12.33 ± 1.65, P = 0.0002), DO + EVs (9.67 ± 0.88, P = 0.0067), and DO + GOEs (15.17 ± 1.58, P < 0.0001) groups, with a notable increase in the DO + GOEs group compared to DO + EVs (P = 0.023) (Fig. 5I). The number of antral follicles increased in the DO group (12.33 ± 1.17) compared to the control (4.83 ± 0.40, P = 0.0001), DO + EVs (9.00 ± 0.77, P = 0.041), and DO + GOEs (5.33 ± 0.71, P = 0.0001), while DO + GOEs significantly reduced antral follicle numbers relative to DO + EVs (P = 0.022) (Fig. 5J). These compelling findings collectively suggest a multifaceted mechanism of action for GOEs, encompassing the promotion of follicular development, suppression of follicular atresia, and enhancement of overall ovarian function. Such robust therapeutic effects underscore the potential utility of GOEs as a novel and promising therapeutic modality for mitigating ovarian dysfunction and restoring ovarian health in conditions characterized by DO. This study provides valuable insights into the therapeutic potential of GOEs and highlights their promising role in addressing ovarian dysfunction, thereby offering new avenues for the development of targeted interventions aimed at improving reproductive health outcomes in affected individuals. GOEs have the potential to enhance ovarian function by stimulating ACVR1B expression in GCs Furthermore, our study undertook a thorough exploration into the impact of GOEs on ACVR1B expression in GCs of DO, using a combination of immunohistochemistry (IHC) and immunofluorescence (IF) analyses. The utilization of both IHC and IF techniques provided a comprehensive assessment of ACVR1B expression within GCs, offering complementary perspectives on the observed effects (Fig. 6, A and C). The results from both methodologies consistently unveiled a significant up-regulation in ACVR1B expression within GCs of both the control and GOEs-treated DO groups in comparison to untreated DO ovaries (Fig. 6, B and D). This robust and convergent evidence strongly substantiates the hypothesis that GOE treatment augments ACVR1B expression in GCs of DO ovaries, thus delineating a promising therapeutic trajectory for mitigating ovarian dysfunction. By adopting this multifaceted approach, we not only bolstered the validity and reliability of our findings but also gained deeper insights into the intricate molecular mechanisms governing GOEs-mediated regulation of ACVR1B expression in the context of ovarian physiology and pathology. GOEs enhance ovarian function by stimulating ACVR1B expression in GCs in vivo. (A and B) IHC and IF detection of ACVR1B expression in GCs within ovaries. Representative sections are shown in the top row; black or white boxes indicate magnified regions. Scale bars, 100 μm (A) and 10 μm (B). (C and D) Quantification of ACVR1B expression by IHC and average fluorescence intensity of ACVR1B in control, DO, and GOE treatment groups. n = 3. Data are presented as means ± SD. ****P < 0.0001; ns, not significant. Moreover, in a bid to enhance the precision and specificity of our analyses, we integrated FSH receptor (FSHR) as a marker to demarcate GCs within the ovarian tissue microenvironment (Fig. 7, A and D) (41). FSHR, a pivotal receptor expressed on the surface of GCs, orchestrates critical processes implicated in follicular development and ovulation. Leveraging FSHR as a discerning marker facilitated the precise identification and visualization of GCs within the ovarian tissue sections, thereby enabling a meticulous characterization of ACVR1B expression dynamics within this cellular milieu (Fig. 7, B, C, and F). This strategic integration of FSHR not only optimized the accuracy and resolution of our investigations but also conferred invaluable insights into the intricate molecular cross-talk underpinning the effects of GOEs on ovarian function. Through this nuanced and comprehensive approach, our study offers compelling evidence delineating the pivotal role of GOEs in modulating ACVR1B expression in GCs, thereby paving the way for the development of targeted therapeutic interventions aimed at ameliorating ovarian dysfunction and advancing reproductive health outcomes. FSHR as a marker to detect GCs expressing ACVR1B after GOE treatment. (A) IF detection of FSHR expression in GCs within ovaries. Scale bar, 50 μm. (B) IF detection of ACVR1B expression in GCs within ovaries. Scale bar, 50 μm. (C) Codetection of FSHR and ACVR1B expression in GCs within ovaries. Scale bar, 5 μm. (D and E) Quantification of average fluorescence intensity for FSHR, ACVR1B, and coexpression of both markers in control, DO, and GOE treatment groups. n = 3. Data are presented as means ± SD. *P < 0.0.5, ***P < 0.001, ****P < 0.0001; ns, not significant. DISCUSSION The efficacy of EVs (42) in treating DO has been demonstrated in prior research, with nucleic acids such as mRNA, microRNA, and long noncoding RNA, along with proteins like ILK, implicated as potential key factors in mediating their therapeutic effects (33, 43, 44). However, despite these promising findings, a notable gap remains in the development of designed or modified EVs specifically tailored for DO treatment (1, 6, 7, 21). Building upon the foundation laid by our previous study investigating ORP, this current research aims to delve deeper into the underlying mechanisms of ORP while exploring novel therapeutic strategies for DO (8). The insights garnered from ORP served as the impetus for this investigation, prompting the exploration of GDF-9 as a potential therapeutic agent (16). Dysregulation of GDF-9 signaling has been implicated in various ovarian pathologies, including ovarian insufficiency and premature ovarian failure (45). In this study, we sought to bridge the gap in the development of tailored EV-based therapies for DO by engineering EVs to overexpress GDF-9 for targeted therapy (46). Our findings not only confirm the efficacy of GOE in vivo models but also shed light on the underlying molecular mechanisms driving their therapeutic effects (19). Mechanistically, the delivery of GDF-9 via EVs modulates key signaling pathways involved in ovarian function, particularly the BMP and TGF-β pathways (17). Specifically, GOEs were found to regulate the ACVR1B on GCs, leading to the phosphorylation of SMAD2 and subsequent translocation into the nucleus (28–30). This ACVR1B/SMAD2 pathway serves as a critical regulator of GC proliferation, follicle maturation, and ovulation (47). Despite the promising findings of this study, several limitations must be acknowledged. One notable limitation is the challenge associated with purifying GDF-9–containing EVs and achieving a sufficiently high concentration of GDF-9 within these EVs (24). The purification process for EVs can be complex and may result in variable yields, affecting the overall efficiency of GDF-9 delivery. In addition, optimizing the concentration of GDF-9 within EVs is crucial to ensure adequate therapeutic efficacy while avoiding potential off-target effects. Moreover, the current results do not demonstrate statistically significant differences in ACVR1B expression between the EV and GOE groups. This may be attributed to the presence of endogenous GDF-9 within native HucMSC-derived EVs, which already contribute to ovarian function recovery. Although GDF-9 overexpression may further enhance this effect, the magnitude of improvement may not reach statistical significance within the current experimental conditions or sample size. In addition, biological constraints such as receptor saturation or feedback inhibition may limit the incremental benefits of GDF-9 enrichment. Future research efforts should focus on refining purification techniques and enhancing GDF-9–loading strategies to improve the therapeutic potential of GOEs for the treatment of DO. Overall, our study provides insights into the therapeutic potential of GOEs for the treatment of DO. By elucidating the molecular mechanisms underlying their therapeutic effects and demonstrating their efficacy in preclinical models, we lay the groundwork for the development of targeted EV-based therapies for DO. This work represents a significant step toward translating EV-based therapies into clinical practice and improving outcomes for individuals with ovarian dysfunction. MATERIALS AND METHODS Ethics statements All of the animal procedures and experiments protocols performed in this study were approved by the Animal Welfare Ethics Committee of Beijing Hospital, National Center of Gerontology (Ethical approval number is 2024BJYYEC-KY148-01). Animals care Female Sprague-Dawley rats or C57BL/6 mice, aged 4 to 8 weeks, were housed in a pathogen-free animal facility under standardized environmental conditions, including a controlled temperature of 21° to 25°C and relative humidity maintained between 40 and 70%. The animals were exposed to a 12-hour light/dark cycle, with lights on from 6:00 a.m. to 6:00 p.m., and were provided ad libitum access to standard rodent chow and water. Upon reaching 9 weeks of age, the rats were randomly assigned into four experimental groups: (i) Control group, (ii) DO group (DO model), (iii) DO + EVs group (treated with EVs), and (iv) DO + GOEs group (treated with GOEs). All animal experiments were conducted in strict accordance with the ethical guidelines and approved protocols of the Animal Care and Use Committee of Beijing Hospital, National Center for Gerontology. Cell culture GCs or MSCs were isolated from follicles in rat ovaries by enzymatic digestion using collagenase (Chemegen, CE12639) and hyaluronidase (CST, 55813S), followed by filtration to obtain a single-cell suspension (48). The cells are cultured in Dulbecco’s modified Eagle’s medium/F-12(FUSHENBIO, FSH073) or RPMI 1640 medium (Sigma-Aldrich BS, R7509) supplemented with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (Sigma-Aldrich, V900929), 1% l-glutamine (Sigma-Aldrich CS, W368401), and 1% insulin-transferrin-selenium (AbMole, M19988). They are incubated at 37°C in a 5% CO2 humidified incubator. Cells are passaged at 70 to 80% confluency using 0.25% trypsin/EDTA, with medium changes every 2 to 3 days (44). For cryopreservation, cells are suspended in 90% FBS and 10% dimethyl sulfoxide (BIO-ZL, ST038), frozen slowly, and stored in liquid nitrogen. Characterization can be done via morphology, flow cytometry, or immunostaining for GC-specific markers (49, 50). EVs collection, purification, and characterization Huc-MSCs (Simpon, C55101) were cultured in medium containing 10% FBS (LDBIO, LD8687) in a standard cell culture incubator set to 37°C with 5% CO2. When 80 to 90% confluency, the cells were rinsed by replacing the medium with phosphate-buffered saline (PBS, BIO-ZL, C0221A) every 30 min to remove residual FBS. For EV isolation, we replaced the culture medium with serum-free exosome-depleted medium during the final medium change before cell harvesting. After continuing culture for 24 hours, the conditioned medium was harvested. Subsequently, the collected medium underwent filtration through 0.22-μm filter to eliminate cellular debris and contaminants. The filtered supernatant was collected and performed sequential centrifugation steps: first at 800g for 5 min, then at 2000g for 5 min, and finally at 150,000g for 90 min, with all centrifugation steps conducted at 4°C (51, 52). The supernatant was carefully discarded, and the EV pellet was resuspended in an appropriate volume of PBS or another suitable buffer. EV size distribution and morphology were analyzed by TEM and NTA (8, 53). EV biomarkers included CD63 and CD9. GDF-9–overexpressed EVs To stimulate GDF-9 expression in HucMSCs, two different delivery systems—Lipofectamine MessengerMAX (LMRNA003) and CALNP mRNA in vitro (D-Nano, DN002) transfection reagents—were used to introduce GDF-9 mRNA into MSCs, with three mRNA dosage levels: 1, 2, and 5 μg/ml. GDF-9 expression of MSCs was quantified using ELISA kit (Raybiotech, ELH-GDF9-1). To isolate GOEs, the transfected HucMSCs were cultured in serum-free medium (FUSHENBIO, FS1183) for 24 hours, and then the conditioned medium was collected. The medium was centrifuged to remove cells and debris, followed by ultracentrifugation at 150,000g for 90 min to pellet the EVs. The protein concentration of GDF-9 in control EVs and GOE was quantitatively analyzed using the ELISA kit (Raybiotech, ELH-GDF9-1). Transmission electron microscopy TEM for observing the morphology and size of EVs or GOEs. Samples were prepared by fixing in a glutaraldehyde solution, followed by postfixation with osmium tetroxide. The samples were then dehydrated through a graded ethanol series, embedded in resin, and sectioned into ultrathin slices. These sections were stained with uranyl acetate and lead citrate to enhance contrast. Images were captured at various magnifications to assess the size, shape, and structural integrity of the EVs or GOEs (42). Nanoparticle tracking analysis Samples were first diluted in a particle-free liquid medium to achieve optimal concentration for measurement. The suspension was introduced into the NTA instrument, where particles were illuminated by a laser. Light scattering from individual particles was captured by a high-resolution camera, enabling real-time tracking of Brownian motion. The software calculated particle size based on diffusion coefficients using the Stokes-Einstein equation, providing detailed size distribution and particle concentration data. This method allows precise quantification and characterization of EV and GOE preparations. Western blots GCs were lysed using cell lysis buffer at 4°C for 30 min, followed by centrifugation at 15,000 rpm for 10 min. The resulting supernatant was collected and quantified using a BCA assay kit (ABCONE, B76382-500 T). Equal amounts of protein were loaded onto a polyacrylamide gel for SDS-PAGE separation. The separated proteins were then transferred onto polyvinylidene difluoride (PVDF, Acmec, AC17619) membranes. Subsequently, the PVDF membranes were incubated with primary antibodies overnight at 4°C and then with secondary antibodies at room temperature for 1 hour. The blots were visualized by an enhanced chemiluminescence kit (21). CCK-8 assay The CCK-8 assay (Solarbio, CA1210) measures cell viability and proliferation. The cells were seeded in plates and allowed to adhere. After induction or treatment, the reagent in the kit was added to each well and incubated for 1 to 4 hours. The reagent will be metabolized by viable cells to produce a colorimetric signal. The absorbance was measured at 450 nm under a microplate reader. The absorbance was proportional to the number of viable cells, and cell viability was calculated by comparing the absorbance of treated samples to control groups (54, 55). Enzyme-linked immunosorbent assay ELISA was used to quantify GDF-9 protein levels in EV preparations, serum level of E2, AMH, and FSH. Samples and standards were added to a 96-well plate precoated with specific antibodies. After incubation, unbound proteins were washed away, and a biotin-labeled detection antibody was added, followed by a streptavidin–horseradish peroxidase conjugate. The plate was then incubated with a chromogenic substrate (trimethylboron), resulting in a colorimetric change proportional to the amount of protein present. Absorbance was measured at 450 nm using a microplate reader, and concentrations were calculated on the basis of the standard curve. All measurements were performed in triplicate to ensure accuracy (56, 57). Liquid chromatography–MS/MS EVs or GOEs were separated by LC based on properties. Enriched EV and GOE samples were dried by vacuum centrifugation and redissolved in SDS/tris-HCl/dithiothreitol. Following lysis and reduction, proteins were alkylated using iodoacetamide. Next, trypsin was added to the samples and the resulting peptides were collected as a filtrate. The peptides of sample were desalted on C18 Cartridges (Empore SPE Cartridges MCX) and concentrated depending on vacuum centrifugation and reconstituted in 0.1% (v/v) formic acid. The peptides from each sample were analyzed by timsTOF HT (Bruker) connected to nanoElute (Bruker) in the DIA mode. MS1 were scanned at a mass range of 300 to 1500 mass/charge ratio, MS2 uses the dia-PASEF acquisition mode. Sixty-six windows were set for DIA mode in MS2 scanning. Each window accumulates a time of 50 ms. Under the dia-PASEF mode, the collision energy varies linearly with the collision energy of ion mobility 1/K0 (fragmentation energy from 20 to 59 corresponds to ion mobility 1/K0 = 0.6 to 1.6 Vs/cm2). DIA data were analyzed with DIA-NN2.0. For LFQ of GDF-9, log2-transformed LFQ intensities obtained from DIA-NN2.0 were normalized, and the relative expression between control EVs and GOEs groups was compared using Student’s t test (58). Flow cytometry To comprehensively characterize the viability and cellular responses of GCs following induction with cisplatin and treatment with EVs, flow cytometry analysis was conducted using a Beckman Coulter flow cytometer (59). This analytical approach allowed for the assessment of GCs live status, apoptosis occurrence, mechanical damage, and cell death dynamics. Specifically, the annexin V–fluorescein isothiocyanate/propidium iodide apoptosis kit (Proteintech, PF00005-100 T) was used to discern apoptotic and necrotic cell populations. This methodological framework enabled the quantitative evaluation of cellular responses to cisplatin induction and subsequent EVs treatment, providing crucial insights into the mechanisms underlying therapeutic interventions aimed at ameliorating ovarian dysfunction (60). DO model in vitro establishment Upon extraction from the animals, rat ovaries were promptly immersed in cold, sterile PBS supplemented with 1% gentamicin (Aladdin, G302822). Subsequently, the ovaries were sectioned into smaller fragments and subjected to enzymatic dissociation in 2% collagenase IV (Gibco, 17104019) solution for a duration of 1 hour to facilitate cellular dissociation. Termination of the dissociation process was achieved by adding an equivalent volume of culture medium containing 20% FBS. Following centrifugation, the supernatant was carefully aspirated, and the resulting cell pellets were resuspended in culture medium supplemented with 10% FBS. The suspended cells were then seeded into T-175 culture flasks and maintained in a standard cell culture environment at 37°C with 5% CO2. Medium replacement was conducted every 24 hours to ensure optimal cellular growth and viability. After a 5-day incubation period, the cultured cells were subjected to flow cytometric analysis to characterize the GCs. Cisplatin (MCE, HY-17394), used as the inducing agent, was administered at a concentration determined on the basis of preliminary cell count experiments. Total RNA-seq Ovarian samples were obtained from both DO rats and control rats and promptly immersed in cold, sterile PBS to maintain RNA integrity. Total RNA extraction was carried out using TRIzol reagent (Thermo Fisher Scientific, 15596018) according to the manufacturer’s instructions. The quality and quantity of extracted RNA were evaluated by a NanoDrop ND-1000 spectrophotometer (NanoDrop, Wilmington, DE, USA) and an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). Rat DO model induction Animal care and experimental procedures adhered strictly to the guidelines outlined by the Animal Welfare Ethics Committee of Beijing Hospital. The DO rat model was induced in female Sprague-Dawley rats aged 4 to 8 weeks. This was achieved through the administration of cisplatin at a dosage of 2.5 mg kg−1 day−1 for three consecutive days. Control rats, on the other hand, received injections of saline solution. Therapeutic regimen EVs and GOEs were administered via localized intraperitoneal injection, a route selected on the basis of the anatomical positioning of the ovaries within the abdominal cavity, thereby allowing efficient delivery to the target organ. Each treatment DO rat received 1 × 106 EV or GOEs particles suspended in 10 μl of sterile saline. Body weight and ovarian value Following euthanasia, rats within each experimental cohort were weighed to ensure consistency in subsequent analyses. Subsequent to euthanasia, bilateral ovaries were meticulously dissected, with the removal of the bicornuate uterus and adipose tissue to facilitate clear isolation. The isolated ovaries were then carefully positioned on a pad for standardized measurements and photographic documentation. To ascertain ovarian mass accurately, each ovary was individually weighed using a precision balance instrument. Ovarian index Ovarian index was calculated through the relative weight of ovary against body weight. It can be calculated as the following formula (24)ovarian index=ovarian wet weight (mg)body weight (g)×100% Follicle analysis Histological sections of ovarian tissue were meticulously examined under a microscope equipped with scanning and analysis software, namely Slide Viewer (2.6.0). Using established criteria from prior research conducted by our team, follicles at various developmental stages, including primordial, primary, secondary, and antral, were systematically classified (61). Follicles were classified on the basis of established morphological criteria: PMF were identified by the presence of an oocyte surrounded by a single layer of flattened squamous GCs; primary follicles exhibited a single layer of cuboidal GCs; SCF contained more than two layers of GCs; and antral follicles were characterized by the formation of a fluid-filled cavity (antrum). For quantification, serial ovarian sections were obtained, and every 10th section was selected for analysis to avoid double counting. Three independent investigators conducted blinded counts of follicles at each developmental stage, and the average number per stage was calculated across the examined sections to ensure accuracy and reproducibility. The resultant datasets were subjected to rigorous statistical analysis using GraphPad Prism software (9.5.0), enabling robust interpretation and inference generation. Estrous cycle detection Vaginal smears were collected each morning by gently flushing the vaginal opening with saline using a pipette. The collected fluid was transferred to a glass slide, air-dried, and stained with hematoxylin and eosin. Proestrus is defined as equal leukocytes and elongated nuclear epithelium. Estrus is defined as most large cornified epithelial cells. Metestrus is defined as similar number of leukocytes, elongated nuclear epithelium, and large cornified epithelial cells. Diestrus is defined as leukocytes mostly (8). The stages were recorded daily to assess cycle regularity and duration, providing insight into reproductive status and ovarian function. Fertility rescue Sperm embolus was observed 24 hours after mating, and pregnancy was confirmed 15 to 20 days later. After delivery, litter size and pup health were recorded. To evaluate pup viability, four functional parameters were assessed: body movement, sucking reflex, righting reflex, and skin color. Movement was evaluated by observing limb and head activity; the sucking reflex was tested by stimulating the mouth region; the righting reflex was assessed by placing pups on their backs and recording their ability to return to a prone position; and skin color was examined as an indicator of oxygenation. Each parameter was scored, and an overall viability score was calculated (62). IHC and IF The tissue slices were deparaffinized and dehydrated through a series of solvent treatments. Antigen retrieval was then performed to enhance immunostaining efficiency (63). After antigen retrieval, the slices were washed with PBS and incubated at room temperature in a 3% H2O2 PBS buffer for 10 min to block endogenous peroxidase activity, followed by thorough washing. Following this, the tissue sections were incubated at 4°C for 16 hours with primary antibodies targeting ACVR1B (Proteintech, 83026-6-RR), FSHR (Invitrogen, PA5-50963), or SMAD2 (Invitrogen, 51-1300). Then, the sections were washed and then incubated at room temperature for corresponding secondary antibodies. Visualization of bound antibodies was achieved using either 3,3′-diaminobenzidine (AbMole, M13777) or 4′,6-diamidino-2-phenylindole (Solarbio, ID2250). Subsequently, a counterstain was applied, and the stained sections were analyzed using ImageJ software for quantitative analysis (64). In vivo imaging system We administered the treatment and biodistribution of EVs in the body. These tags emit light when exposed to specific wavelengths, which is captured by the in vivo imaging system’s sensitive camera. Data analysis software then processes the captured light to create visual representations of the spatial distribution and intensity of the signal (65). Statistical analysis Statistical analysis used GraphPad Prism software (Version 9.5). Data are presented as means ± SD from at least three independent experiments, consistently yielding comparable results. For comparisons between two groups, a two-tailed, unpaired Student’s t test was used. One-way analysis of variance (ANOVA) followed by the Bonferroni post hoc test was conducted for comparisons involving more than three groups. Statistical significance was defined as P < 0.05 (59).