Endostatin-expressing endometrial mesenchymal stem cells inhibit angiogenesis in endometriosis through the miRNA-21-5p/TIMP3/PI3K/Akt/mTOR pathway

The results of EMSC isolation and characterization were described in our prior study. To construct EMSCs-Endo that overexpress endostatin, we infected EMSCs with recombinant viruses that contained human Endo cDNA. Microscopic observations revealed that the EMSCs-Endo had a characteristic fibroblast spindle morphology and a radial arrangement, which were comparable to the morphological characteristics of EMSCs and other MSCs noted in alternative references ( Figure 1A ). Endo was positively expressed in the EMSCs-Endo but was negatively expressed in the EMSCs and EMSCs-GFP ( Figure 1B ). To determine OCT4 expression in EMSCs-Endo, we used human bone marrow mesenchymal stem cells (hBMSCs) as a positive control and human skin fibroblasts (HSFs) as a NC. RT-PCR demonstrated the presence of OCT4 in the EMSCs, EMSCs-Endo, and EMSCs-GFP ( Figure 1C ). In addition, flow cytometric detection revealed that the EMSCs-Endo were positive for CD90, CD73, CD166, and CD105 but negative for CD45, CD34, and CD14, thus confirming their MSC phenotypes ( Figure 1D ). These findings further revealed that EMSCs-Endo exhibited expression of HLA-ABC (MHC-I) and no expression of HLA-DR (MHC-II) ( Figure 1D ), suggesting the low immunogenicity of EMSCs-Endo. Characterization of endostatin-expressing EMSCs. (A) The morphology of cells. (A-1) Morphology of EMSCs-Endo; (A-2) Morphology of EMSCs; (A-3) Morphology of EMSCs-GFP. Scale bar = 1000 μm. (B) Endostatin expression in EMSCs-Endo. The endoplasmid was used as a positive control. GAPDH served as the internal control. Endo mRNA was positively expressed in the EMSCs-Endo cohort but was negatively expressed in the EMSCs and EMSCs-GFP cohorts. (C) Expression of the stem cell-specific marker OCT4 in EMSCs-Endo. Human bone marrow mesenchymal stem cells were used as a positive control, and human skin fibroblasts were used as a negative control. GAPDH served as the internal control. OCT4 mRNA was positively expressed in EMSCs, EMSCs-Endo, and EMSCs-GFP. (D) Immunophenotypes of EMSCs-Endo. Flow cytometry analysis revealed that EMSCs-Endo positively expressed CD73, CD90, CD105, and CD166 but negatively expressed CD14, CD34, and CD45. The results also revealed that the EMSCs-Endo expressed HLA-ABC (MHC class I) but not HLA-DR (MHC class II). (E) Expression of endostatin in the cell supernatant. Approximately 463.88 ± 6.00 pg/mL Endo could be produced by 1 × 10 5 EMSCs-Endo within 48 hours. (F) Expression of endostatin in cell supernatants from P1 to P5. The levels of Endo expression were stable from P1 to P5 EMSCs-Endo (456.88 ± 15.72 to 472.47 ± 19.86 pg/mL). The data are presented as the mean ± SD of n  = 3 independent experiments. Statistical analysis was performed by 1-way ANOVA with the Bonferroni post hoc correction (E and F), ** P  < 0.01. As detected by a Human Endostatin ELISA Kit, 463.88 ± 6.00 pg/mL Endo was produced by 1 × 10 5 EMSCs-Endo 48 hours after infection. The expression of Endo was markedly higher than that in the EMSCs-GFP (5.69 ± 3.78 pg/mL), the EMSCs (6.77 ± 3.35 pg/mL), and the control (4.17 ± 1.27 pg/mL) group ( Figure 1E ). After the P1–P5 EMSCs-Endo were cultured for 48 hours, constant Endo expression was detected via ELISA. The levels of Endo expression were stable from P1 to P5 EMSCs-Endo (456.88 ± 15.72 to 472.47 ± 19.86 pg/mL, P  > 0.05) ( Figure 1F ). Endo-expressing EMSCs were successfully prepared and used in subsequent research. To assess the effect of EMSCs-Endo on endometriosis in vitro, the migratory capacity of cocultured HUVECs was determined via wound healing assays. IPP 6.0 software was used to evaluate the difference in the wound area at 0 hour and 12 hours and to determine the percentage of the initial wound area. The migration of HUVECs was significantly lower in the EMSCs-Endo group (21.67% ± 7.64%) and Endo group (26.33% ± 4.51%) than in the EMSCs group (66.67% ± 5.03%), the EMSCs-GFP group (74.67% ± 14.50%), and the control group (69.33% ± 9.02%) ( P  0.05) ( Figure 2A and B ). Effect of EMSCs-Endo on endometriosis in vitro. (A) Treatment with EMSCs-Endo inhibited the migration of HUVECs. HUVECs were cocultured with EMSCs, EMSCs-Endo, EMSCs-GFP, or Endo. Then, the HUVECs were scraped, and images were taken at 0 hour and 12 hours after scraping. Scale bar = 200 μm. (B) Compared with those in the control group, the migratory capacity of HUVECs in the EMSCs-Endo group was significantly lower. There was no significant difference between the EMSCs-Endo group and the Endo group. (C) Treatment with EMSCs-Endo reduced the angiogenic capacity of HUVECs. After coculture, a tube formation assay was performed on the HUVECs, and images were taken after 6 hours. Scale bar = 400 μm. (D) Compared with the other treatments, treatment with EMSCs-Endo significantly reduced the angiogenic capacity of HUVECs. (E) EMSCs-Endo suppressed the proliferation of HUVECs. HUVECs were cocultured with EMSCs, EMSCs-Endo, EMSCs-GFP, or Endo, after which a CCK-8 assay was performed to assess the proliferation of the HUVECs. Compared with those in the other groups, the proliferative capacity of HUVECs in the EMSCs-Endo group and Endo group were lower. There was no significant difference between the EMSCs-Endo group and the Endo group. (F) EMSCs-Endo-induced apoptosis in cocultured HUVECs. HUVECs were cocultured with EMSCs, EMSCs-Endo, EMSCs-GFP, or Endo. (G) Comparison of the percentage of apoptotic cells between each group. EMSCs-Endo and Endo significantly induced the apoptosis of cocultured HUVECs. There was no significant difference between the EMSCs-Endo group and the Endo group. The data are presented as the mean ± SD of n  = 3 independent experiments. Statistical analysis was performed by 1-way ANOVA with Bonferroni post hoc correction (B, D, E, and G), * P  < 0.05, ** P  < 0.01, *** P  < 0.001. The therapeutic effect of EMSCs-Endo on endometriosis in vitro was determined by evaluating the angiogenic capacity of HUVECs via tube formation assays. ImageJ software was used to measure the correlation of the central tendency of tube lengths across groups. The findings demonstrated that the mean length of tube formation in HUVECs after coculture with EMSCs-Endo (40.3% ± 11.0%) was markedly lower than that in the EMSCs group (87.5% ± 11.2%) ( P  < 0.001), the EMSCs-GFP group (86.5% ± 10.4%) ( P  < 0.001), the control group (100% ± 14.2%) ( P  < 0.001), and the Endo group (63.2% ± 10.2%) ( P  < 0.05) ( Figure 2C and D ). To evaluate the therapeutic effects of EMSCs-Endo on inhibiting HUVECs proliferation, we cocultured EMSCs-Endo with HUVECs in a Transwell system, and a CCK-8 assay was performed to detect HUVECs proliferation. The findings suggested that the proliferation of the HUVECs cocultured with EMSCs-Endo (83.68% ± 5.72%) was markedly lower than that of the EMSCs group (96.98% ± 2.59%), the EMSCs-GFP group (98.03% ± 5.02%), and the control group (100.00% ± 4.22%) ( P   0.05) ( Figure 2E ). Flow cytometry was used to detect the apoptotic population of HUVECs via Annexin V/PI staining to examine the effect of EMSCs-Endo on endometriosis in vitro. As shown in Figure 2F , EMSCs-Endo caused an increase in late apoptosis (Annexin V+/PI+) in cocultured HUVECs. The proportion of apoptotic HUVECs increased to 22.15% ± 2.48% in the presence of EMSCs-Endo ( P   0.05) ( Figure 2F and G ). In contrast, the levels of apoptosis in the cocultures of HUVECs in the EMSCs, the EMSCs-GFP, and the control group exhibited markedly lower levels of apoptosis (10.65% ± 2.14%, 9.55% ± 1.15%, and 8.90% ± 2.50%, respectively) ( P  > 0.05) ( Figure 2F and G ). We searched for endometriosis microarrays in the GEO database ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE105765 ) and performed differential expression analysis of miRNAs. Similarly, miRNA-21-5p expression was markedly greater in the endometria of endometriosis patients than in those of normal individuals ( P  < 0.05) ( Figure 3A ). Then, we performed validation at the clinical sample level. The findings demonstrated that miRNA-21-5p levels were markedly greater in the eutopic endometria and ectopic lesions of endometriosis patients than in those of normal individuals ( Figure 3B ) ( P  < 0.001). TIMP3 is a target gene of miR-21-5p, and the expression of the miR-21-5p/TIMP3/PI3K/Akt/mTOR was affected by EMSCs-Endo in vitro. (A) Bioinformatic analysis of the differential expression of miR-21-5p in endometriosis. (B) The levels of miR-21-5p were significantly elevated in the eutopic and ectopic tissues of patients with endometriosis. (C) The TIMP3 3ʹ-UTR contains miR-21-5p binding sites. (D) Dual-luciferase reporter assays confirmed TIMP3 as a target gene of miR-21-5p. (E) The relative expression of miR-21-5p in cocultured HUVECs from each group. (F) Expression analysis of TIMP3/PI3K/Akt/mTOR in cocultured HUVECs from each group by Western blotting. (G-J) Quantification of protein levels of TIMP3 (G), p-PI3K (H), p-Akt (I), p-mTOR (J). The data are presented as the mean ± SD of n  = 3 independent experiments. Statistical analysis was performed via 1-way ANOVA with Bonferroni post hoc correction (B, D, E, and G-J), * P  < 0.05, *** P  < 0.001. MiRNAs can function by regulating target genes. As predicted by TargetScan, miRNA-21-5p and TIMP3 complement each other ( Figure 3C ). To confirm the interaction between these 2 components, we conducted a dual-luciferase assay. Compared to the NC, the miRNA-21-5p mimics notably reduced TIMP3-WT reporter activity but had no effect on the TIMP3-MUT ( Figure 3D ). These findings indicate that miRNA-21-5p can directly target TIMP3. The levels of miRNA-21-5p in HUVECs cocultured with EMSCs-Endo were analyzed via qRT-PCR. We found that miRNA-21-5p in HUVECs was downregulated in the EMSCs-Endo group and Endo group compared to the other groups ( Figure 3E ). Moreover, the expression levels of TIMP3/p-PI3K/p-Akt/p-mTOR in HUVECs cocultured with EMSCs-Endo were assessed. TIMP3 expression in HUVECs was higher in the EMSCs-Endo group and Endo group than in the other groups ( P  < 0.001). Compared to the Endo group, the EMSCs-Endo group showed no significant difference ( Figure 3F and G ). In contrast, the levels of p-PI3K ( P  < 0.001), p-Akt ( P  < 0.001), and p-mTOR ( P  < 0.05) were lower in the HUVECs from the EMSCs-Endo group and Endo group than in those from the other groups. The EMSCs-Endo group did not show a significant difference from the Endo group ( Figure 3F , H - J ). To evaluate the potential of EMSCs-Endo to generate tumors or elicit substantial side effects in recipients, we subcutaneously injected 1 × 10 7 EMSCs, EMSCs-Endo, or EMSCs-GFP into 5-week-old nude mice ( n  = 5/group). Initially, we observed small, firm swellings resembling nonspecific inflammation within the first 3 days posttransplantation, which gradually resolved. After 45 days, the mice were euthanized. Throughout the experimental period, no noticeable weight loss or indications of deteriorating health were noted. Moreover, no discernible solid tumors were detected surrounding the injection sites in any of the groups of mice ( Figure 4A ). Pathological examination of hematoxylin-eosin (HE)-stained slides verified the lack of tumor generation ( Figure 4B-D ). These findings indicate that both EMSCs and those modified to express Endo proteins show promise in terms of safety for transplantation therapies. EMSCs-Endo tumorigenicity analysis and effects of EMSCs-Endo on endometriotic lesion growth in a nude mouse model. (A) Macroscopic observation of subcutaneous tumor formation in each group of nude mice. (A-1) Absence of tumor formation from inoculated EMSCs. (A-2) Absence of tumor formation by inoculated EMSCs-Endo. (A-3) Absence of tumor formation from inoculated EMSCs-GFP. (B) HE staining of the scapular subcutaneous tissue of nude mice revealed the absence of tumor formation in the EMSCs group. (B-1) HE staining of the EMSCs injection site. (B-2) HE staining of the saline injection site. (C) HE staining of nude mouse scapular subcutaneous tissue showing the absence of tumor formation in the EMSCs-Endo group. (C-1) HE staining of the EMSCs-Endo injection site. (C-2) HE staining of the saline injection site. (D) HE staining of nude mouse scapular subcutaneous tissue showing the absence of tumor formation in the EMSCs-GFP group. (D-1) HE staining of the EMSCs-GFP injection site. (D-2) HE staining of the saline injection site. Scale bar = 50 μm. (E) Endometriotic lesions in nude mice from various groups. (F) Growth curve of endometriotic lesions. EMSCs-Endo compared with the control, * P  < 0.01, *** P  < 0.001. Endo compared with the control, ## #P  < 0.001. (G) Comparison of endometriotic lesion volumes in nude mice across different groups. * P  < 0.05. The data are presented as mean ± SD. Statistical analysis was performed by 1-way ANOVA with the Bonferroni post hoc correction (F and G). In a nude mouse model of endometriosis, prior to the initial tail vein injection of MSCs, there were no obvious differences in lesion volume among the 5 groups. After MSCs injection, the growth rate of subcutaneous endometriotic lesions markedly decreased in the EMSCs-Endo group and Endo group compared with the EMSCs group, the EMSCs-GFP group, and the control group. Additionally, the volume of endometriotic lesions in EMSCs-Endo group began to decrease compared with that of the corresponding control group on day 13, and the volume of lesions was gradually reduced after day 22 ( Figure 4F ). Three days after the last injection, the animals were euthanized through cervical dislocation. The sizes of the endometriotic lesions in each group were compared, and the results demonstrated that the lesions in the EMSCs-Endo group were markedly smaller than those in the EMSCs group, the EMSCs-GFP group, and the control group ( P   0.05) ( Figure 4E and G ). Additionally, we obtained similar results in a BALB/c mouse model of endometriosis by means of allogeneic transplantation. Endometriosis was induced in 2 recipient mice through the injection of endometrial fragments from one donor mouse into the peritoneal cavity. In these mice, the mean volume of the lesions of EMSCs-Endo-treated mice (8.580 ± 6.275 mm 3 ) was significantly smaller than that of the EMSCs-treated mice (35.684 ± 13.314 mm 3 ) ( P  < 0.05), EMSCs-GFP-treated mice (35.804 ± 11.618 mm 3 ) ( P  < 0.05), and NS-treated mice (36.274 ± 7.809 mm 3 ) ( P   0.05) ( Supplementary Figure S1A and B ). To verify that the therapeutic effects of EMSCs-Endo were caused by the Endo content in endometriotic lesions, we detected Endo by immunohistochemistry in lesions. Tissues were harvested 3 days after the final injection. In the mice that received EMSCs-Endo, a large amount of Endo was found at the lesion sites. Compared with that in the EMSCs group, the EMSCs-GFP group, the control group, and the Endo group, the expression of Endo in the lesions of EMSCs-Endo group was significantly higher ( Figure 5A ). The therapeutic effect of EMSCs-Endo on endometriosis in vivo. (A) Immunohistochemical staining and comparison of endostatin expression in endometriotic lesions of each group. Scale bar = 25 μm. (B) Immunohistochemical staining and comparison of VEGF expression in endometriotic lesions from each group. Scale bar = 25 μm. (C) Representative images of areas with the highest density of microvessels in each group. Scale bar = 100 μm. The data are presented as mean ± SD. Statistical analysis was performed by 1-way ANOVA with the Bonferroni post hoc correction. * P  < 0.05, ** P  < 0.01, *** P  < 0.001. Endostatin concentrations in the serum of nude mice were measured by means of a specific ELISA kit; compared with those in the EMSCs, EMSCs-GFP, and control groups, the concentration of endostatin in the EMSCs-Endo group was significantly higher ( P  < 0.001). However, no significant difference was found between the EMSCs-Endo and Endo groups ( Supplementary Figure S1C ). To explore the potential association between the inhibition of endometriotic lesion growth by EMSCs-Endo and the suppression of angiogenesis, we assessed the expression of VEGF and the MVD for angiogenesis analysis via VEGF and CD34 antibodies, respectively. The paraffin sections used were obtained from the same lesions utilized for Endo detection. Ten random high-magnification fields were selected for each section, and the mean optical density values of VEGF protein expression were calculated by means of IPP6.0. Statistical analysis revealed that the VEGF level was obviously lower in the EMSCs-Endo group (0.0214 ± 0.0034) than those in the EMSCs group (0.0277 ± 0.0038), the EMSCs-GFP group (0.0290 ± 0.0041), and the control group (0.0280 ± 0.0036) ( P  < 0.01). Compared with that in the Endo (0.0240 ± 0.0027) group, the VEGF level in the EMSCs-Endo group was also obviously lower ( P   0.05) ( Figure 5B ). The MVD counting approach proposed by Weidner and coworkers 33 was employed. As shown in Figure 5C , at 100× magnification, fewer countable blood vessels were observed in the lesions of the EMSCs-Endo-engrafted group (21 ± 3) than in those of the control group (33 ± 5) ( P  < 0.01), the EMSCs group (32 ± 5) ( P  < 0.01), the EMSCs-GFP group (34 ± 4) ( P  < 0.01), and the Endo group (30 ± 3) ( P   0.05) ( Figure 5C ). The TIMP3/PI3K/Akt/mTOR expression levels in the lesions of the mice exposed to EMSCs-Endo were analyzed by immunohistochemistry and Western blotting. TIMP3 expression was higher in the EMSCs-Endo and Endo groups compared to the other groups ( P  < 0.001). In contrast, the levels of p-mTOR, p-Akt, and p-PI3K were lower in the EMSCs-Endo and Endo groups than in the other groups. There was no significant difference between the EMSCs-Endo group and Endo group ( P  > 0.05) ( Figure 6A , B , and D ). Moreover, the miRNA-21-5p levels in the lesions of the mice were analyzed via qRT-PCR. We observed that the miRNA-21-5p level in the ectopic lesions was lower in the EMSCs-Endo ( P  < 0.001) and Endo groups ( P   0.05) ( Figure 6C ). The expression of miR-21-5p/TIMP3/p-PI3K/p-Akt/p-mTOR in mouse endometriotic lesions. (A) Expression analysis of TIMP3/p-PI3K/p-Akt/p-mTOR in the endometriotic lesions of each group by Western blotting. (B) Quantification of protein levels of TIMP3, p-PI3K, p-Akt, and p-mTOR. (C) Relative expression of miR-21-5p in the endometriotic lesions of each group. (D) Expression analysis of TIMP3/p-PI3K/p-Akt/p-mTOR in the endometriotic lesions of each group by means of immunohistochemistry. Scale bar = 25 μm. The data are presented as the mean ± SD. Statistical analysis was performed by means of 1-way ANOVA with the Bonferroni post hoc correction. * P  < 0.05, ** P  < 0.01, *** P  < 0.001. TIMP3 was overexpressed by transfection of an adenoviral vector into HUVECs. TIMP3 silencing was performed by si-TIMP3 transfection into HUVECs. The overexpression and silencing of TIMP3 were confirmed by qRT-PCR and Western blot analysis. TIMP3 expression was significantly upregulated in HUVECs-OE-TIMP3 compared with HUVECs and HUVECs-OE-NC ( Supplementary Figure S2A and B ). Additionally, TIMP3 expression was significantly lower in the HUVECs-si-TIMP3 group than in the HUVECs and HUVECs-si-NC groups ( Supplementary Figure S3A and B ). Wound healing and tube formation assays were used to verify the effect of EMSCs-Endo on the migration and angiogenic capacity of HUVECs-OE-TIMP3 ( Supplementary Figure S2C and E ). The results showed that, compared with those of normal HUVECs, the migration distance and the total length of tube formation in all groups of HUVECs-OE-TIMP3 significantly decreased ( P  < 0.001), whereas there were no significant differences in the migration distances and total tube formation lengths observed among the HUVECs-OE-TIMP3 groups ( Supplementary Figure S2D and F ). Additionally, CCK-8 and flow cytometry assays were performed to evaluate the impact of EMSCs-Endo on the proliferation and apoptosis levels of HUVECs-OE-TIMP3 ( Supplementary Figure S2G and H ). After overexpression of TIMP3, proliferation and apoptosis in each HUVECs-OE-TIMP3 group was lower and higher, respectively, than those in the corresponding HUVECs ( P  < 0.001). Nonetheless, there were no significant differences among the HUVECs-OE-TIMP3 groups ( Supplementary Figure S2G and I ). After transfection with TIMP3 siRNA, a series of experiments, as described above, were conducted again to validate the effects of EMSCs-Endo on HUVECs-si-TIMP3. The results of the wound healing assay showed that, compared with those of the normal HUVECs, the HUVECs-si-TIMP3 groups of the control, the EMSCs, and the EMSCs-GFP, the migration distance of HUVECs-si-TIMP3 cocultured with EMSCs-Endo or Endo significantly decreased. There were no significant differences between the EMSCs-Endo and Endo groups or between the normal HUVECs and the control, EMSCs, and EMSCs-GFP groups ( Supplementary Figure S3C and D ). The tube formation assay results indicated that there were no significant differences in angiogenic capacity among the groups ( Supplementary Figure S3E and F ). The CCK-8 assay result revealed that the proliferation of the HUVECs-si-TIMP3 cocultured with EMSCs-Endo or Endo was markedly lower than that of the EMSCs group, the EMSCs-GFP group, the control group, and the normal HUVECs ( P  < 0.001). Additionally, the control, EMSCs, and EMSCs-GFP groups had lower levels of proliferation than did the normal HUVECs ( P  < 0.05). However, there was no significant difference between the EMSCs-Endo group and the Endo group ( Supplementary Figure S3G ). The results of flow cytometry showed that the HUVECs-si-TIMP3 cocultured with EMSCs-Endo or Endo had exhibited higher levels of apoptosis than did the corresponding control HUVECs cells. Moreover, the control, EMSCs, and EMSCs-GFP groups had higher levels of apoptosis than did the normal HUVECs ( Supplementary Figure S3H and I ). The TIMP3/p-PI3K/p-Akt/p-mTOR expression levels in the HUVECs-OE-TIMP3 group of cells cocultured with EMSCs-Endo were analyzed by Western blot analysis. TIMP3 expression in the HUVECs-OE-TIMP3 groups cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, and Endo was higher than that in the normal HUVECs group. Conversely, the levels of p-PI3K, p-Akt, and p-mTOR in the HUVECs-OE-TIMP3 groups cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, and Endo were lower than those in the normal HUVECs group. However, there were no significant differences among the EMSCs, EMSCs-GFP, EMSCs-Endo, and Endo groups ( Supplementary Figure S2J and K ). TIMP3 expression in the HUVECs-si-TIMP3 groups cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo was much lower than that in the normal HUVECs group. Conversely, the expression levels of p-PI3K, p-Akt, and p-mTOR in HUVECs-si-TIMP3 groups cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, and Endo were higher than those in the normal HUVECs group. However, there were no significant differences among the EMSCs, EMSCs-GFP, EMSCs-Endo, and Endo groups ( Supplementary Figure S3J and K ).

Human endometrial tissues were collected from female patients who underwent laparoscopy for idiopathic infertility at the same institute. Ectopic endometrial tissue was acquired from endometriosis patients undergoing laparoscopy. Surgery was planned to coincide with the advanced proliferative phase, and the collection of endometrial tissues took place during the surgical procedure. Patients refrained from undergoing hormone treatments for a minimum of 3 months before the scheduled surgery. This study adhered to the principles outlined in the Helsinki Declaration and received approval from the Ethics Committee of the same institute (approval number: 2022JS30). The procedures were executed in strict compliance with the approved regulations and guidelines. Informed written consent was obtained from all the participants prior to their inclusion. All cell protocols and studies were approved by the Ethical Committee of the same institute and followed the 2016 ISSCR Guidelines for Stem Cell Research and Clinical Translation. We utilized female BALB/c nude mice (aged 5 weeks and weighing 18–20 g) procured from Shanghai SLAC Laboratory Animal (identification number 2013001809469) and female BALB/c mice (8 weeks old and 23–28 g) from Charles River Laboratories (identification number 230729241100018458). The animals were maintained under a 12-hour light/dark cycle and were exposed to sterile water and food, allowing them to acclimate to specific pathogen-free conditions before experimentation. The animal procedures adhered strictly to the recommendations outlined in the Guide for the Care and Use of Laboratory Animals of the same institute. The study protocol received approval from the Animal Ethical Committee of the First Affiliated Hospital of Harbin Medical University (approval number: 2019019). Every effort was made to minimize any potential discomfort or distress experienced by the animals. Dr. Zhibin Peng generously provided the HUVECs, which were maintained in RPMI 1640 medium containing 1% penicillin/streptomycin and 10% FBS. Human bone marrow mesenchymal stem cells (hBMSCs) and human skin fibroblasts (HSFs), also provided by Dr. Zhibin Peng, were maintained in DMEM/F12 medium containing 1% penicillin/streptomycin and 10% FBS. Additionally, Dr. Xiaojun Wang kindly provided the 293T cells, which were maintained in high-glucose DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. The cells were grown at 37 °C and 5% CO 2 in the aforementioned media. Endostatin-expressing endometrial MSCs were isolated and expanded from the human endometrium according to our previous study. 31 In brief, endometrial tissue was immersed in ice-cold DMEM/F12 supplemented with 10% FBS and 1% penicillin/streptomycin. The tissue specimens were finely minced and enzymatically dissociated in type III collagenase (300 µg/mL) and type I deoxyribonuclease (40 µg/mL) for 2 hours at 37 °C. Following digestion, the cell suspension was filtered through a 70-µm sieve. The filtrate was subsequently centrifuged at 140 g for 10 minutes. The cells were then grown in the aforementioned media at 37 °C and 5% CO 2 . The medium was replaced every 3 days to eliminate nonadherent cells. The morphological features of EMSCs from various passages were examined according to our previous study. 31 In brief, the expression of the stem cell-specific marker octamer-binding transcription factor 4 (OCT4) was determined by reverse transcription polymerase chain reaction (RT-PCR). Cell-surface biomarkers, such as CD90, CD73, CD166, CD105, CD45, CD34, CD14, HLA-DR, and HLA-ABC, were evaluated by flow cytometry. The osteogenic, adipogenic, and chondrogenic differentiation capacities of the EMSCs in vitro were assessed in specific conditioned media. The puc57-Endo plasmid was obtained from Genscript Corporation and was amplified and extracted from Escherichia coli . pTY-CMV-eGFP, which is based on HIV-1, harbors the eGFP reporter gene under regulation via the CMV promoter. This plasmid was generously supplied by Dr. Xiaojun Wang. For the generation of pTY-CMV-Endos, eGFP was replaced with an Endo. For lentivirus-Endo and lentivirus-GFP, 293T cells were transfected with the transfer plasmid pTY-CMV-Endo/pTY-CMV-eGFP or the packaging plasmids vsvg and psPAX2 with Lipofectamine 2000 (Invitrogen, USA). The transfected cells were grown at 37 °C and 5% CO 2 for 48 hours to facilitate the generation of the recombinant lentivirus. Following incubation, the recombinant virus was collected, titrated, purified, and subsequently utilized to infect EMSCs for another 48-hour period. EMSCs expressing Endo (referred to as EMSCs-Endo) and EMSCs expressing GFP (EMSCs-GFP) were selected on the basis of their resistance to puromycin after retroviral infection. Morphological observation and flow cytometric analysis were also conducted to assess the characteristics of the EMSCs-Endo. The OCT4 and Endo mRNA levels were determined via RT-PCR assays. The level of Endo expressed by the EMSCs-Endo was detected via ELISA (Abcam, UK) following the manufacturer’s protocols. For flow cytometry analysis, the cells were harvested and suspended in PBS at 2 × 10 4 cells/20 µL. Then, the cells were incubated with specific antibodies, including mouse IgG1/FITC, mouse IgG1/PE, CD105/FITC, CD90/FITC, CD14/PE, CD73/FITC, CD45/FITC, CD34/FITC, HLA-DR/PE, and HLA-ABC/FITC (BD, USA), for 20 minutes at room temperature in the dark. Following incubation, the cells were rinsed with PBS, followed by centrifugation (140 g, 5 minutes). After resuspension in 300 µL of PBS, the cells were subjected to analysis with a flow cytometer (BD FACSCalibur). The data obtained were processed with CellQuestPro software. RNA extraction was performed with TRIzol, followed by subsequent removal of genomic DNA. RT-PCR was conducted with a PCR instrument (Eppendorf) as follows: 42 °C for 1 hour; 94 °C for 15 minutes; 35 cycles at 94 °C for 30 seconds, 57 °C for 30 seconds, and 72 °C for 30 seconds; and 72 °C for 10 minutes. The PCR products were visualized on a 1% agarose gel. The primer sequences utilized in the amplification process are provided in Table 1 . Primer pairs for each gene. Adenovirus vectors containing coding sequences for human TIMP3 or the scrambled control were purchased from GeneChem (Shanghai, China). HUVECs were transfected with an adenoviral vector (HUVECs-OE-TIMP3) or a scrambled control vector (HUVECs-OE-NC) at a multiplicity of infection according to the standard procedure. The siRNA against TIMP3 was used to knock down TIMP3 expression in cultured HUVECs. The siRNAs for TIMP3 (si-TIMP3) (HUVECs-si-TIMP3) and the negative control (NC) (HUVECs-si-NC) were purchased from GeneChem (Shanghai, China). The transfection efficacy was verified by qRT-PCR and Western blot analysis 3 days after transfection. A transwell culture system was used to prevent cell–cell contact. The effect of EMSCs-Endo on endometriosis was assessed by detecting the proliferative capacity of HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or 50 μg/mL recombinant human endostatin (Endo; Shandong Xiansheng Maidejin Biological Pharmaceutical Co., Ltd., China). EMSCs, EMSCs-GFP, and EMSCs-Endo were grown in a Transwell chamber at 5 × 10 4 cells/well. HUVECs were grown in 12-well plates at 3 × 10 5 cells/well at 37 °C and 5% CO 2 for 12 hours. Then, the cells in the Transwell chambers were cocultured with HUVECs at 37 °C and 5% CO 2 for 24 hours. To detect the proliferation of HUVECs, a CCK-8 assay was conducted. All experiments were carried out in triplicate. The in vitro effect of EMSCs-Endo on endometriosis was assessed by measuring the migratory capacity of HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo (50 μg/mL). Cells were seeded in accordance with the previously mentioned protocol. To evaluate the migratory capacity of HUVECs, wound healing assays were also conducted. After 24 hours of coculture, the HUVECs had achieved 100% confluence. The cell surface was scratched to create wounds. Three crossing areas were photographed by means of a phase contrast microscope. Then, the HUVECs were cocultured again for 12 hours, after which the 3 areas were photographed. Each experiment was conducted 3 times for validation. The effect of EMSCs-Endo on endometriosis was determined by testing the angiogenic ability of HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo (50 μg/mL). Cells were seeded according to previously described methods. To evaluate the angiogenic capacity of HUVECs, tube formation assays were performed. Briefly, after 48 hours of coculture, HUVECs were harvested via trypsinization and resuspended in the supernatant of each group at a density of 2 × 10 4 cells/mL. The cells were subsequently seeded onto Matrigel-coated plates (BD Biosciences, USA) and grown at 37 °C and 5% CO 2 . After a 6-hour incubation period, cellular imaging was conducted with a digital camera connected to an inverted microscope. Each experiment was replicated 3 times. The total lengths of the tubes were measured by the ImageJ (National Institutes of Health, NIH, USA). The in vitro effect of EMSCs-Endo on endometriosis was determined by evaluating the apoptosis of HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo (50 μg/mL) for 48 hours. Cell seeding was performed according to the aforementioned protocol. The apoptosis of HUVECs was evaluated. Cellular staining with Annexin V-FITC/PI (BD, USA) and flow cytometric analysis were performed according to the manufacturer’s instructions. Each assay was replicated 3 times. To predict the presence of hsa-miRNA-21-5p binding sites on the tissue inhibitor of metalloproteinase 3 (TIMP3), TargetScan ( https://www.targetscan.org/vert_80/ ) was utilized to identify the binding sites of the TIMP3 mRNA 3ʹ-UTR to the mature sequence of the miRNA hsa-miRNA-21-5p. We conducted a luciferase assay using 293T cells. In brief, we constructed 2 Luc-TIMP3 3ʹ-UTR constructs: one containing the potential binding sequence (TIMP3 WT) and the other containing the truncated potential binding sequence (TIMP3 MUT). The cells were seeded in 24-well plates and cotransfected with the Luc-TIMP3 3ʹ-UTR reporter vector (100 ng), TK (40 ng) or the miRNA-21-5p mimic (30 n m ) overnight. Then, luciferase activity was quantified by means of the Dual-Glo Luciferase assay system. We used the TRIzol method to isolate total RNA from HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo (50 μg/mL). cDNA synthesis was performed through reverse transcription with either random primers or miRNA-specific stem-loop primers. The Bulge-Loop miRNA qRT-PCR Primer Sets, designed specifically for miRNA-21-5p by RiboBio (Guangzhou, China), included one RT primer and a pair of qRT-PCR primers for each set. The cDNA samples were subsequently subjected to RT-PCR analysis on a 7500 RT-PCR system, with U6 serving as the internal control. Western blotting analysis was performed as described previously. 32 Briefly, total protein was extracted from HUVECs cocultured with EMSCs, EMSCs-GFP, EMSCs-Endo, or Endo (50 μg/mL) with RIPA buffer (Beyotime Biotechnology, Beijing, China). The extracted proteins were then separated via SDS-PAGE and subsequently transferred onto PVDF membranes. Next, each membrane was blocked overnight with 5% nonfat milk and then incubated with antibodies against TIMP3 (Proteintech Group, China), p-Akt (CST, USA), Akt (CST, USA), p-PI3K (Proteintech Group, China), PI3K (Proteintech Group, China), p-mTOR (Proteintech Group, China), mTOR (CST, USA), and GAPDH (Proteintech Group, China) at 4° C overnight. After being rinsed with TBST, each membrane was exposed to HRP-linked secondary antibodies for 60 minutes. Visualization and quantification of the protein bands were performed with enhanced chemiluminescence and Image-Pro Plus v6.0 (IPP6.0) (Media Cybernetics, USA), respectively. Fifteen mice were randomly allocated into 3 groups, namely, the EMSCs group, the EMSCs-GFP, and the EMSCs-Endo group, each consisting of 5 mice. Tumorigenicity models were constructed by subcutaneously injecting normal saline (NS) (0.2 mL) containing 1 × 10 7 cells/mL from each group into the right scapular tissue, while the left side received the same volume of NS as a NC. Tumor development in the 3 groups was recorded daily. After 45 days, all the animals were euthanized via cervical dislocation. Both the right and left subcutaneous scapular tissues were subsequently harvested and subjected to macroscopic examination and H&E staining to evaluate tumor generation. Human ectopic endometrium was procured from individuals diagnosed with ovarian endometriosis following the previously outlined procedure. The tissues were cultured for no more than 2 hours in DMEM/F12 at 4 °C and then rinsed with sterile NS before injection into nude mice. Twenty-four mice were subcutaneously injected with NS (0.2 mL) containing an ectopic endometrial fragment (0.5 cm³) in the dorsal region. After 7 days, when the endometriotic lesions became palpable, the mice were randomly assigned to 1 of 5 groups by simple randomization ( n  = 6 per group): the EMSCs, the EMSCs-GFP, the EMSCs-Endo, the Endo and the control group. EMSCs (1 × 10 7 /mL), EMSCs-GFP (1 × 10 7 /mL), and EMSCs-Endo (1 × 10 7 /mL) were intravenously injected into the tail vein. The mice in the Endo group received an injection of 20 mg/kg Endo. The mice in the control group were injected with NS. Injections were administered every 3 days. Tail tip blood samples were collected and stored 24 hours after each injection. The animals were euthanized 3 days after the seventh injection was administered. The ectopic endometriotic lesion size was determined, and the lesion volume was measured as follows: 1/6π ab 2 ( a : the lesion’s long axis; b : the lesion’s short axis). Endometriotic lesion tissues and mouse serum were collected from each group. For qRT-PCR, the collected tissues were frozen in liquid nitrogen until RNA extraction, and the expression of miRNA-21-5p was determined via qRT-PCR as described previously. For the immunohistochemical analysis, the harvested tissues were fixed in paraformaldehyde (10%) for 24 hours at RT, followed by dehydration via a sequential alcohol concentration gradient. Paraffin-embedded tissue sections were affixed onto microscope slides, dewaxed, and rehydrated prior to antigen retrieval. The slides were then exposed to primary antibodies (anti-endostatin, anti-VEGF, anti-TIMP3, anti-p-mTOR, anti-p-PI3K, and anti-p-Akt) (BIOSS, China) at RT for 1 hour, followed by rinsing with PBST. The slides were subsequently stained with secondary antibodies (ZSGB-BIO, China) against #1 and #2 for 20 and 30 minutes, respectively. Afterward, the sections were subjected to rinsing, DAB staining, and hematoxylin counterstaining. Finally, the samples were dehydrated and visualized under a microscope. All the data were analyzed with Image-Pro Plus v6.0. Furthermore, the microvessel density (MVD) was assessed in these tumors. MVD was assessed using anti-CD34 (BIOSS, China) staining, and positivity was evaluated via light microscopy. 33 In brief, regions on the slides with the highest concentration of microvessels were selected, and each microvessel was counted at 100× magnification. The viable vessels included isolated luminal microvascular structures and immunoreactive endothelial cells. For the Western blot analysis, proteins were extracted from the endometriotic lesions of each group. The procedures for Western blot analysis and the use of primary antibodies were performed as previously described. To determine the concentration of endostatin in the serum of nude mice, we conducted an ELISA experiment following the manufacturer’s protocols. The BALB/c mouse model of endometriosis was adapted from the model previously described by Sanchez. 34 In summary, 1 week prior to the surgical procedure, the donor mice were administered estradiol (E2) at a dosage of 3 μg per mouse. On the designated day 0, the donor mice were euthanized, and the uterus was extracted. The uterine horns were then separated, and the myometrium was excised through scraping. The remaining endometrial tissue was subsequently cut into small pieces with scissors. These fragments, obtained from the isolated uterine horns, were weighed and mixed with saline containing ampicillin at a concentration of 1 mg/mL. Endometriosis was experimentally induced by injecting equivalent amounts of minced endometrial tissue from a BALB/c donor mouse directly into the peritoneum of 2 BALB/c recipient mice. To support the continued development of the lesions, each recipient mouse was given daily injections of E2. After 7 days, the mice were randomly assigned to 5 groups by simple randomization ( n  = 4 per group): the EMSCs, the EMSCs-GFP, the EMSCs-Endo, the Endo, and the control group. The experimental protocol for caudal intravenous therapy was performed as described above. The animals were euthanized 3 days after the seventh injection. All the ectopic endometriotic lesions were collected from each mouse. The lesion size was measured as described above. All the experimental data are presented as the mean ± standard deviation. All analyses were performed with SPSS 22.0 statistical software (Softonic, San Francisco, CA, USA) and GraphPad Prism software (version 8.0; GraphPad Inc., La Jolla, CA, USA). First, normality analysis was conducted on the data. Data that followed a normal distribution were analyzed by means of t -tests or 1-way ANOVA, while nonparametric tests were conducted for nonnormally distributed data. Comparisons among multiple groups were analyzed by means of 1-way ANOVA, followed by the Bonferroni post hoc correction or Dunnett’s T3 test for pairwise comparisons, as appropriate. A P- value < 0.05 was considered to indicate statistical significance.

Endometriosis is a common gynecological disease, and its clinical manifestations include pain and infertility, which can profoundly impact a woman’s mental and physical health, as well as her overall quality of life. 35 Notably, 40% of infertile women have endometriosis, and up to 50% of patients with chronic pelvic pain also have endometriosis. 36 Currently, the mechanisms underlying endometriosis are not fully understood 37 ; thus, the issue of endometriosis recurrence has not been completely resolved by the available clinical treatments. 38 Current treatment methods are aimed primarily at addressing infertility or pelvic pain, the 2 main symptoms, and are limited to surgery, hormonal therapy, and analgesics. These approaches often result in numerous side effects and rarely offer long-term relief. 39 In addition, endometriosis is a nonmalignant, persistent gynecological condition that has various similarities to invasive cancer, such as invasion, metastasis, and recurrence. The limitations of current therapies stem from the absence of an effective therapeutic approach to address its underlying pathological processes. New methods that can eliminate ectopic lesions, reduce side effects, and prevent recurrence are urgently needed. Since angiogenesis is an important factor in the occurrence of endometriosis, antiangiogenic treatments, such as endostatin, which is considered the most effective inhibitor of microvessel growth, might be effective. 40 , 41 However, given its disadvantages as a protein preparation, the establishment of a lesion-specific targeted gene transfer system is a key factor in successful treatment. Efficacy, safety, and the establishment of a reliable delivery system are primary concerns in gene therapy when targeting lesion tissues. 42 MSCs are widely used in basic and clinical research. They have been shown to reduce the inflammatory response 43 and have emerged as highly essential candidates for cell therapy owing to their minimal immunogenicity, nontumourigenicity, and inherent affinity for tumor sites or inflamed tissues. 23 , 44 , 45 While there has been extensive research on the use of EMSCs in regenerative medicine, 46 , 47 to our knowledge, their use as a gene therapy vector is relatively underreported. During the last decade, the endometrium has become a novel source of MSCs. 26 , 48 In our previous study, we successfully isolated endometrial mesenchymal stem cells from women without endometrium-associated disease. 31 Adult stem cells are rare, undifferentiated cells found in mature tissues and organs. They are challenging to identify because of their scarcity, lack of unique morphological characteristics, and absence of specific markers. These cells are defined by their functional traits, including significant self-renewal capacity, high proliferation potential, and the ability to differentiate into various lineages. 49 , 50 In our research, several experiments were conducted to evaluate the characteristics of Endo-expressing EMSCs as mesenchymal stem cells, as well as their stability and safety. In our study, EMSCs-Endo exhibited a high proliferative capacity. The presence of OCT4, a marker of cellular pluripotency, 51 suggested that the EMSCs-Endo cultured in our study displayed stem cell characteristics. The RT-PCR results revealed that the EMSCs-Endo were positive for Endo, and the ELISA data revealed that the EMSCs-Endo could secrete Endo into the supernatant, which demonstrated that human Endo cDNA was successfully transferred and that Endo-expressing EMSCs were prepared. According to the flow cytometry results, the EMSCs-Endo were positive for CD90, CD73, CD166, and CD105, the major MSC biomarkers, but lacked CD45, CD34, and CD14, indicating that their phenotypes did not change. Furthermore, the expression of the MHC class I molecule HLA-ABC in EMSCs-Endo cells, coupled with the absence of HLA-DR expression, indicates low immunogenicity. These results suggest that EMSCs-Endo retain the characteristics of MSCs and can be utilized safely in vivo without apparent immune rejection. Additionally, considering the potential tumorigenic risk associated with EMSCs and EMSCs-Endo, we conducted further experiments to assess the possibility of tumor formation by EMSCs. Nude mice that were administered up to 1 × 10 7 EMSCs or EMSCs-Endo showed no apparent signs of tumorigenesis during the entire 45-day period. Pathological analysis also confirmed the absence of tumorigenesis in the injected subcutaneous tissues. Thus, we conclude that EMSCs-Endo are safe and do not induce tumor formation. Nevertheless, the long-term potential of EMSCs to cause tumors warrants further investigation. Endometriosis is a noncancerous condition characterized by features such as neovascularization, adhesion, infiltration, migration, and recurrence. 52 Since angiogenesis is necessary for the development and progression of endometriosis, 40 antiangiogenic treatments might constitute targeted therapies for endometriosis. 41 Our findings indicated that EMSCs modified to express Endo were successful in suppressing endometriosis in both cell and animal models of endometriosis. When HUVECs were cocultured with EMSCs-Endo, we observed a significant reduction in proliferation, migration, and angiogenesis compared with those of the controls. The proliferation and migration of HUVECs contribute not only to lesion formation but also to endometriosis metastasis. Since HUVECs form newborn blood vessels, suppressing HUVECs may inhibit angiogenesis. 53 Thus, EMSCs-Endo, which can regulate HUVECs, may be effective in the treatment of endometriosis in vitro. The obvious therapeutic effects on angiogenesis observed in vitro verified that the administration of Endo by EMSCs-Endo was effective; however, its impact on the intricate in vivo milieu remains unclear. As described previously, we assessed the effects of therapy on the basis of indicators that embody the characteristics of endometriosis, including lesion size and angiogenesis. Our study demonstrated that EMSCs-Endo were effective at suppressing ectopic lesions in a nude mouse model of endometriosis. In mice treated with EMSCs-Endo, the secreted Endo protein was exclusively detected to assemble at lesion sites, displaying a distribution pattern similar to that of EMSCs, as demonstrated in our previous work. 31 Moreover, the EMSCs-Endo group exhibited a notable reduction in lesion size. Given these immunological factors, we conducted treatment experiments by injecting EMSCs-Endo into a BALB/c allogeneic transplantation mouse model of endometriosis. The results indicated that EMSCs-Endo could also inhibit the growth of endometriotic lesions. This significant therapeutic outcome underscores the efficacy of EMSCs-Endo in suppressing lesions, confirming the efficient targeted delivery of Endo by EMSCs-Endo, even within the intricate in vivo milieu. VEGF serves as a pivotal mediator of angiogenesis and is implicated in fostering neovascularization by promoting the attachment and invasion of endometrial cells to the peritoneal surface, thus potentially contributing to the development of endometriosis. 54 Previous research has indicated that elevated levels of VEGF promote the survival of ectopic tissue within the peritoneal cavity, consequently fostering the development of endometriotic lesions. 55 Hence, we investigated the impact of EMSCs-Endo on angiogenesis by assessing the levels of VEGF. Our findings revealed a substantial decrease in VEGF levels within endometriotic lesions in nude mice treated with EMSCs-Endo, compared to those in the other stem cell groups and the control group, suggesting a potential mechanism through which EMSCs-Endo may mitigate angiogenesis in endometriosis. Furthermore, we assessed the impact of EMSCs-Endo on angiogenesis by analyzing the MVD within endometriotic lesions. Our investigation revealed a notable decrease in blood vessel density in the treated group, compared to the control group, indicating significant angiogenic inhibition. It is worth mentioning that, compared to the Endo group, the EMSCs-Endo group exhibited higher levels of Endo in endometriotic lesions and lower levels of VEGF and MVD. These results indicate that EMSCs-Endo has a superior effect on inhibiting angiogenesis compared with recombinant protein formulations of endostatin. This may be due to the fact that, with the same injection intervals, EMSCs-Endo is able to specifically target and express Endo at the lesion site, resulting in higher and more stable concentrations of Endo within the lesion. Our observations align with findings from a previous study utilizing a bladder tumor model, further supporting the notion that EMSCs-Endo are effective at suppressing angiogenesis 56 and demonstrating that EMSCs-Endo therapy contributes to a decrease in vascularization. This reduced angiogenesis observed in vivo confirms the efficacy of EMSCs-Endo-targeted therapy for treating endometriosis. Multiple studies have shown that miR-21-5p promotes angiogenesis in various disease processes. 57-59 On the basis of our previous National Natural Science Foundation project, as well as the search results from the GEO database and clinical sample test outcomes, we selected miRNA-21-5p as the target for this study. Our experiments revealed that EMSCs-Endo can inhibit angiogenesis by regulating the miRNA-21-5p/TIMP3/PI3K/Akt/mTOR signaling pathway, suggesting that treatment with EMSCs-Endo is a promising strategy for treating endometriosis and preliminarily elucidating its mechanism. However, it should be noted that there are still some limitations in this study. First, after constructing EMSCs-Endo, we need to conduct more comprehensive research on their characteristics, such as gene expression profiling sequencing and cell proliferation, migration, and invasion assays, among others. Therefore, in future research, we will elaborate on their characteristics more specifically. Second, in the in vivo toxicity experiments involving EMSCs-Endo, we analyzed the tumorigenicity of EMSCs, EMSCs-Endo, and EMSCs-GFP. Since infertility is one of the main symptoms of endometriosis, the formation of endometrial stromal vessels not only affects the pathogenesis of endometriosis but also impacts patient fertility. Therefore, in future studies, we also need to investigate the effects of EMSCs-Endo on fertility and placental development. Finally, this study focused solely on the mechanism of angiogenesis in endometriosis, since the pathogenesis of endometriosis is related to immunity and inflammation, and considering the immunoregulatory function of human mesenchymal stem/stromal cells, 60 we will also study the role of EMSCs-Endo in treating endometriosis through immunomodulation in future research. We hope to be able to elaborate more comprehensively on the mechanism of EMSCs-Endo in treating endometriosis.

In conclusion, our study demonstrated that EMSCs could be used as excellent carriers of drugs for targeted therapy for endometriosis. We first demonstrated that EMSCs are safe and effective after being genetically engineered to express Endo, with the aim of treating endometriosis. EMSCs-Endo were able to home to the lesion site in an endometriosis xenograft mouse model and displayed therapeutic efficacy. Our data suggest that the ability of EMSCs-Endo to restrict endometriosis can be attributed to the inhibition of angiogenesis via the regulation of the miRNA-21-5p/TIMP3/PI3K/Akt/mTOR signaling pathway. Further investigations are needed to determine the optimal cell dose and administration routes, as well as how to prolong the duration of therapeutic efficacy. Future studies should explore the efficacy and safety of EMSCs-Endo gene therapy in patients, with the aim of promoting the translational application of EMSCs-Endo. This study may provide a potential strategy for the treatment of endometriosis.

Endometriosis (EMs) is a prevalent gynecological disease affecting approximately 10% of female patients. It manifests as the presence of extrauterine active endometrial tissue, leading to persistent pelvic discomfort, dyspareunia, dysmenorrhea, and fertility challenges among women of reproductive age. 1 , 2 Despite being benign, endometriosis shares many traits with malignant conditions, including infiltration, migration, and recurrence. 3 Conventional approaches include hormonal therapies, invasive surgical interventions, or a combination thereof. 4 Although these treatments have varying degrees of efficacy, the risk of recurrence remains notably high 5 because the pathogenesis of this disease is still unclear, and treatments that target the pathogenesis are lacking. Research indicates that regardless of the treatment approach, approximately 50% of patients experience symptom recurrence within 5 years, and many patients need to undergo surgery again. 6 Angiogenesis is an important step in the development of endometriosis, 7 therefore, antiangiogenic therapy may be a novel option for treating endometriosis. 8 Endostatin (Endo), a 22-kDa polypeptide originating from a carboxyl-terminal segment of type XVIII collagen, is one of the most effective endogenous inhibitors of angiogenesis and specifically inhibits newborn blood vessels. 9 , 10 Many studies have shown its advantages in inhibiting angiogenesis in cancers. 11-13 However, its application in the treatment of endometriosis has rarely been reported. Nevertheless, disadvantages such as a high effective dose, unstable blood drug levels, the need for multiple injections, increased toxicity, and high price limit the clinical application of recombinant protein formulations of endostatin. 14 Gene therapy may be a better method because of its predominance. Therefore, we assume that a stable drug delivery method involving the transfer of the endostatin gene into endometriotic lesions might be a preferable strategy for endometriosis therapy. Recently, stem cells, especially mesenchymal stem cells (MSCs), have been investigated as important candidates for cell therapy. 15-18 MSCs are a type of adult nonhematopoietic stem cell that can be harvested from diverse tissues and organs, including the umbilical cord, adipose tissue, amniotic fluid, and bone marrow; additionally, these cells self-renew and proliferate at high levels, undergo multidifferentiation, and are characterized by their lack of oncogenicity and low immunogenicity. 19-22 MSCs can migrate to inflammatory sites, such as ectopic lesions and tumors, and are considered promising carriers for delivering treatments across a spectrum of diseases. 22 , 23 In previous research examining the antitumor properties of modified MSCs, our studies showed that amniotic fluid-derived MSCs expressing IFNα can migrate toward and inhibit tumors derived from HeLa cells in a mouse model. 24 However, limited attention has been given to endometrial MSCs (EMSCs). The isolation of EMSCs was initially accomplished by Chan and coworkers in 2004. 25 MSCs, which are widely recognized and utilized within the endometrium, were identified by Gargett and coworkers in 2009. 26 EMSCs exhibit all the defining features of MSCs. Growing evidence supports their potential utility in regenerative medicine and cancer treatment. 27 , 28 Additionally, EMSCs can serve as immune modulators, mitigating inflammation and influencing the formation of blood vessels crucial for tissue regeneration. 29 , 30 Owing to their unique characteristics, EMSCs may be a feasible cell source for targeted therapy. In our previous studies, we extracted EMSCs from endometrial tissue and identified their biological characteristics. Furthermore, we validated their tropism for endometriosis through in vitro and in vivo experiments. 31 Therefore, we propose to investigate the role and mechanism of targeted gene therapy with endostatin-expressing EMSCs (EMSCs-Endo) for endometriosis and to offer a new possibility for treating endometriosis.