The isolation and culture of BMSCs were performed as previously described with some modifications. Briefly, the female Sprague–Dawley (SD) rats (3–4 weeks old, 80–100 g) were euthanized by CO₂ asphyxiation. The marrow was flushed out from tibias and femora, and the isolated cells were cultured in low glucose Dulbecco’s modified Eagle’s medium (L-DMEM) supplemented with 10% fetal bovine serum, penicillin/streptomycin, and 4 ng/ml bFGF. The round cells began to adhere within 24 h after incubation, and the nonadherent cells were removed by changing the medium 48 h after incubation. The cells were then trypsinized and sub-cultured when reaching 80–90% of confluence. The cells between passages 3 and 6 were used in the following experiments.

The BMSCs of the third passage were used to detect the surface antigen markers. Adherent cells were harvested and resuspended in PBS and then incubated with FITC-labeled anti-CD90 (0.06 µg/test, Invitrogen, 11-0900-81) and PE-labeled anti-CD45 antibodies (0.25 µg/test, Invitrogen, 12-0461-82) in the dark at 4°C for 30 min. Unstained cells and isotype controls (0.06 µg/test, Invitrogen, 11-4724-81) served as a negative control. After staining, the expression of cell surface markers was analyzed by flow cytometry (Beckman).

To confirm the differentiation capability of isolated BMSCs, the third passage of cells was induced to differentiate into adipocytes, osteoblasts, and chondrocytes using specific induction media (RASMX-90031, RASMX-9004, RASMX-90021, Cyagen Biosciences Inc.). The BMSCs were seeded onto six-well culture plates at a density of 2×10⁴ cells/cm² and cultured to 100% confluence. The culture medium was changed to adipocytic and osteogenic induction medium. After 21–28 days of induction, Oil Red O staining was carried out to detect lipid droplets, and Alizarin Red S staining was performed to detect the calcifying nodules.

BMSCs were transduced with a lentiviral vector expressing both GFP and firefly luciferase (Fluc) purchased from Jikai Co. (Shanghai, China). Then, 72 h after transduction, the transduction efficiency was evaluated by analyzing the GFP-positive cells under fluorescence microscopy, and the infected cells were sorted by flow cytometry sorting.

The animal study was reviewed and approved by the Animal Ethics Committee of the Sixth Affiliated Hospital, Sun Yat-sen University (No.20190307-002). The female SD rats were purchased from Beijing Vital River Laboratory Animal Technology Co. and bred at a temperature of 23–25°C with a 12 h light/dark cycle. After 1-week adaptive feeding, vaginal smears of the rats were performed to determine the estrous cycle, and only those with normal estrous cycles were chosen for subsequent experiments. A total of 48 rats were randomly assigned to four groups: normal, injured, intra-arterial, and intrauterine. The surgical procedure was performed under isoflurane anesthesia. For all rats, except for those in the normal group, the abdomen was open, and both sides of the uterus were carefully picked out to inject 500 μl of 95% ethanol in a sterile condition. The ethanol was aspirated out after 5 min; then, the uterine cavity was rinsed with physiological saline twice. For the intra-arterial and intrauterine groups, 1×10⁶ GFP/Luc-labeled BMSCs resuspended in 100 and 20 μl of PBS, respectively, were transplanted unilaterally after modeling. The injured group did not undergo any therapy. For each group and each time point of examination, a size corresponding to three animals was used (Figure 1).

To detect the distribution of BMSCs comprehensively, the rats were imaged continuously by PerkinElmer IVIS Spectrum at 1, 2, 4, and 8 days after BMSC treatment. Briefly, the rats received a single intraperitoneal injection of 150 mg/kg firefly D-luciferin potassium salt (MB1834, Meilun Biotechnology Co. LTD.). The photographic and luminescent images were recorded by the camera 15 min after injection. For further comparison, the luminescent signal was recorded as total flux (average photos per second, p/s), and the region of interest (ROI) of constant size and position among different individuals was manually selected to quantify the luminescent intensity.

To further determine the precise location, quantity, and differentiation of administered BMSCs, immunofluorescence analysis was performed. The paraffin-embedded uteri slides were dewaxed in xylene and rehydrated through graded ethanol concentrations; then, antigen retrieval was performed by the microwave method in citrate solution. The uteri sections were incubated with anti-GFP mouse monoclonal antibody overnight at 4°C, followed by cy3-conjugated goat anti-mouse IgG secondary antibody (1:500, E031610-01, EarthOx) for 1 h in the dark, and stained with DAPI (meilunbio) at 10 μg/ml for 5 min. After that, the sections were examined under a laser scanning confocal microscope (Leica SP8). For differentiation detection, double staining was carried out. The sections were incubated with anti-GFP mouse antibody and anti-vimentin rabbit antibody (1:200, ABCAM, ab52942).

Hematoxylin–eosin staining (H&E) was used to analyze the tissue structure. The uteri were fixed in 4% paraformaldehyde (PFA) solution overnight and embedded in paraffin, cut into 5-μm sections, and then stained with H&E.

To examine the collagen deposition in the injured endometrium, Masson’s trichrome staining was carried out. The extent of fibrosis among different groups, defined as the areas occupied by collagen fiber relative in proportion to the entire endometrium, was quantified by the software Image Pro Plus 6.0.

For immunohistochemistry staining (IHC), the slides were dewaxed in xylene and rehydrated through graded ethanol concentrations. Antigen retrieval was performed in citrate solution, and endogenous peroxidase activity was blocked in 3% H₂0₂ for another 15 min. The sections were incubated with anti-cytokeratin (1:200, Abcam, ab7763), anti-vimentin (1:200, Abcam, ab52942), anti-VEGF (1:200, Abcam, ab1316), and anti-ki67 (1:200, Abcam, ab16667) antibodies overnight at 4°C, followed by goat anti-mouse IgG or goat anti-rabbit IgG secondary antibody for 60 min, and then stained with DAB substrate kit (ZSGB-BIO, China) and hematoxylin in accordance with the manufacturer’s protocol.

The expression of cytokeratin, vimentin, VEGF, LIF, and integrin β3 was analyzed by Western blotting. Briefly, equal amounts of proteins extracted from different groups were separated by 10% SDS-PAGE electrophoresis and transferred to PVDF membranes. The membranes were incubated with anti-cytokeratin (1:1,000 Abcam, ab7763), anti-vimentin (1:1,000, Abcam, ab52942), anti-VEGF (1:1,000, Abcam, ab1316), anti-TGF-β1 (1:1,000, Abcam, ab215715), and anti-ki67 (1:1,000, Abcam, ab16667) antibodies overnight at 4°C after blocking in 5% BSA and then incubated with HRP-conjugated goat anti-mouse or anti-rabbit secondary antibody at room temperature for 1 h on the second day. β-actin was used as a loading control.

All data were expressed as mean ± SD, and statistical analysis was performed by SPSS version 24.0. Student’s t-test or one-way ANOVA were applied to analyze the data. The differences were considered statistically significant when p < 0.05. GraphPad Prism 8.0 software was used to draw scientific graphs.

The isolated BMSCs displayed typical fibroblast morphology after purification (Figures 2A,B). Flow cytometry analysis was carried out to confirm the immunophenotype of BMSCs at passage 3. The expression of CD90 and CD45 was 99.7 and 0.56%, respectively (Figure 2C), showing specific immunophenotypes of BMSCs. In addition, BMSCs exhibited the capacity to differentiate into adipocytes and osteoblasts after induction (Figure 2D). Taken together, the BMSCs used in the following experiments match the MSC criteria.

The distribution of differently administered BMSCs by bioluminescence imaging and immunofluorescence analysis showed that the survival and retention time of BMSCs were longer in the intrauterine group.

The BMSC recruitment was monitored at different time intervals (immediately following transplantation, 2, 4, and 8 days post transplantation) by an in vivo imaging system (Figures 3A,B). When BMSCs were infused into the lumen of the right horn directly, the luminescent signal remained distinct in the abdomen with no signals in other organs such as lungs in the first 4 days post treatment, indicating the precise and concentrated distribution of BMSCs when administered locally. However, in the subsequent 4 days, the signal underwent a rapid decrease and could barely be detected by bioluminescence imaging.

While infusing BMSCs through the ipsilateral iliac artery, the results were comparable. There were no significant differences in the uterine cavity between two different treating groups, and the signal was strong in the first 4 days and disappeared when tested at 8 days post treatment. However, the total fluorescence intensity was stronger following local administration (7.98×10⁵ and 6.02×10⁵ at day 2 and 5.83×10⁵ and 2.74×10⁵ at day 4. Figure 3C). It was assumed that a small number of cells entered the peripheral circulation when administered via the artery. Furthermore, we found that two individuals in the intra-artery group had signals in their right feet, and there were no signals detected in the abdomen (Supplementary Figure S3). We supposed that it was probably due to the anatomical differences in the diameter ratio of the internal iliac artery and external iliac artery, the major feeding artery of the lower limb. These individuals were excluded from subsequent analysis.

Since the signals were barely detected in either group after 8 days post treatment, the immunofluorescence analysis was then performed to further assess the survival and retention time of BMSCs in the two groups. As expected, the proportion of GFP-positive cells in the injured horns showed an overall similar declining trend from 8 to 28 days after therapy (Figure 3D). However, there was a significant decrease of rate in the intrauterine group, with an average of 5.67 and 4/HPF GFP-positive cells located in the injured endometrium at day 8 in the intrauterine and intra-arterial groups, respectively. At the 28th day post treatment, GFP staining remained low in the basalis layer of the endometrium in the intra-arterial group, and there were no GFP fluorescence signals that could be detected in the intrauterine group (Figures 3E,F).

We then compared the precise distribution and differentiation of the BMSCs delivered in two strategies by immunofluorescence analysis to explore the causes of difference in the BMSC retention time. The murine uterus tissue consists of endometrial (glandular and luminal) epithelium, stroma, and myometrium. We observed nearly all the GFP-positive cells localized in the stroma but not in the epithelium or myometrium. In addition, in terms of specific localization patterns, there was significantly more positive staining in the basal layer of the endometrium close to the glands and vessels than in the outer layer of the endometrium in the intra-arterial group (Figure 4A). Furthermore, in the intra-arterial infusion group, some GFP-positive cells were located around the vessels in the surrounding ligaments of the uterus (Figure 4B), confirming the feasible targeted delivery of BMSCs by our method.

To further explore the fate of transplanted cells, uterine sections from two groups at 12 days after therapy were double-stained by anti-GFP antibody and anti-vimentin or anti-VEGF antibody and examined under fluorescence microscopy, considering their specific distribution in the basal layer of the endometrium. Some of the stromal cells were both EGFP- and vimentin-positive, suggesting that MSCs were not only recruited but also began to differentiate into other cell types to restore the endometrium (Figures 4C,D). However, the proportion of recruited BMSCs was quite small. It is hard to capture GFP signals in the uterine sections at the 28th day post treatment, indicating that MSCs most likely do not exert their therapeutic effects through cellular replacement of the injured endometrium.

BMSCs administered differently were equally effective in reducing collagen deposition in different BMSC administered groups; VEGF and integrin β3 were higher expressed when transplanted through the artery.

As shown in Figure 5, the modeling rats had a relatively thinner endometrium, and the intact structure was damaged with a remarkably decreased number of vessels and glands and increased interstitial edema. Some of the epithelial cell layers even became discontinuous, confirming the successful modeling as previously reported.

The H&E staining showed significantly increased endometrial thickness in both intra-arterial and intrauterine infusion groups compared to the model group, and the density of blood vessels and glands was also increased. However, in some individuals, the morphology of luminal epithelial cells remained flattened even at the 28th day after BMSC transplantation as compared with columnar epithelial cells in the normal control group, reflecting the incomplete endometrial repair by BMSCs. We then compared the morphological discrepancies of different groups at a consecutive time after BMSC treatment by H&E staining, Masson’s trichrome staining, immunohistological analysis, and Western blotting.

The results showed increased fibrosis after modeling and no obvious change in the collagen deposition in the 12 days post therapy in both groups, despite the mitigation of other injury indicators. At the 28th day post treatment, a comparable reduction in tissue fibrosis was observed in the two groups (Figure 5B). These histological results were consistent with the results of the Western blot, in which the expression of TGF-β1 only showed a slight change on the 28th day after therapy (Figure 6B). No obvious differences were observed when comparing these protein expressions in these two groups at consecutive time after therapy. Thus, we concluded that BMSCs administered differently were equally effective in reducing collagen deposition by regulating the TGF-β signaling pathway.

To further compare the therapeutic effects of differently administered BMSCs, we evaluated the expression of cytokeratin, vimentin, PCNA, LIF, integrin β3, and VEGF by immunohistochemical analysis and Western blotting. Cytokeratin and vimentin, the markers of epithelial and stromal cells, showed an increase in the two BMSC-infused groups (Figure 6A), and the protein expression of PCNA, the cell proliferation marker, increased significantly after BMSC infusion, which indicates the regeneration of both epithelial and stromal cells after BMSC therapy (Figure 6A).

Integrin β3 and LIF are the markers of endometrial receptivity and essential for successful embryo implantation. Our results demonstrated an increased expression of integrin β3, suggesting that BMSCs not only promoted endometrial renewal but also improved endometrial receptivity. Another interesting finding was the differential expression of VEGF and integrin β3 in the two groups at the same time point (Figures 6C,D). VEGF and integrin β3 were much higher in the intra-arterial infusion group at day 12 post treatment, implying that BMSCs promoted angiogenesis by upregulating related cytokines when transplanted through the ipsilateral iliac artery.