Hypoxia-Induced Extracellular Vesicles Derived from Human Umbilical Cord Mesenchymal Stem Cells Regulate Macrophage Polarization and Enhance Angiogenesis to Promote Diabetic Wound Healing

The First Affiliated Hospital of Guangxi Medical University granted ethical approval (Approval Number: 2025-E0711, Approval date: 28 March 2025) for the successful acquisition of HUCMSCs. Human umbilical cord samples were collected from full-term healthy donors via cesarean section at The First Affiliated Hospital of Guangxi Medical University, and collection was implemented based on the principles of aseptic technique. The previous study has been taken as reference for extraction, culture and identification of HUCMSCs [12]. HSFs (AW-CH0154, Anweisci, Shanghai, China) and HUVECs (AW-CH0165, Anweisci) were both cultured in high-glucose medium (BL304A, Biosharp, Hefei, China), supplemented with additional glucose to achieve the final glucose concentration of 35 mmol/L. Macrophages were differentiated from human monocyte leukemia cells (AW-CH0359, Anweisci) using phorbol ester (P6741, Solarbio, Beijing, China). M1 macrophages were induced by lipopolysaccharide (L8880, Solarbio) for 48 h. RPMI 1640 medium (BL303A, Biosharp). These cells were maintained in a humidity-controlled incubator containing 5% CO₂ at 37 °C.

Primary extracted HUCMSCs were used for subsequent experiments after passaging to the 5th generation. When HUCMSCs reached a density of approximately 80%, we changed the medium to a serum-free medium. The n-EVs group was incubated continuously at 37 °C for 48 h under normoxic conditions of 5% carbon dioxide (CO₂), 21% oxygen (O₂), and 74% nitrogen (N₂), while the hy-EVs group was cultured at 37 °C for 48 h with 5% CO₂, 3% O₂, and 92% N₂, and the culture was controlled by a hypoxic incubator (MIC101, COLLRUN), a dual flow meter (DFM3002, COLLRUN), and a modular oxygen monitor (MOM5003, COLLRUN) to precisely control the 3% O₂ concentration. After 48 h of incubation, the medium was successfully collected and aliquoted into 50 mL centrifuge tubes for density gradient centrifugation. The first centrifugation was performed for 10 min at 300× g, then centrifuged for 20 min at 2000× g to remove cellular debris. Following this, we passed the supernatant via a 0.22 μm filter to clear nanoscale contaminants. The volume of the supernatant from the initial centrifugation was approximately 200 mL, and the resulting supernatant underwent ultracentrifugation at 4 °C for 70 min at 120,000× g, repeated twice. These isolated n-EVs and hy-EVs were meticulously resuspended in 100 µL of Phosphate-Buffered Saline (PBS), these preparations could be used directly in experiments or kept at −80 °C for later applications. Nanoparticle tracking analysis (NTA, (ZetaView PMX 110, Particle Metrix, Meerbusch, Germany)) was employed to comprehensively characterize the distribution of sizes for n-EVs and hy-EVs. Transmission electron microscopy (TEM, (H7650, HITACHI, Tokyo, Japan)) was employed to characterize their microstructure. Western blot analysis was conducted to measure major biomarkers associated with EVs. The examined markers included the positive markers CD63, CD9, CD81, and TSG101, as well as the negative marker Calnexin.

N-EVs and hy-EVs were labeled with Cell Plasma Membrane Red Fluorescence Staining Kit (PKH26, Beyotime, Shanghai, China). To begin with, the fluorescent dye working solution was used to label the n-EVs and hy-EVs, followed by washing with PBS, centrifugation, and complete removal of unbound dyes. The HUVECs were co-incubated with the fluorescently labeled n-EVs and hy-EVs for 6 h to observe their uptake of these fluorescently labeled n-EVs and hy-EVs. First, tissue samples were fixed for 15 min with 4% paraformaldehyde. After the fixation procedure, we used Actin-Tracker Green-488 (C2201S, Beyotime) for an additional duration of 10 min to visualize the organization of the cytoskeleton. After that, DAPI (C1006, Beyotime) was used to stain the nuclei for 1 min. Following the staining procedures, we used PBS to wash any unbound reagents before imaging. In the end, fluorescence images were taken with a confocal laser scanning microscope (CLSM).

To study cellular growth, an equal number of cells were inoculated into each well of a 96-well microplate. N-EVs and hy-EVs were added to the medium at doses of 0, 5, 10, 15, 20, and 25 μg/mL, and cells were incubated. After 24 h of cell incubation, the medium was carefully aspirated and washed gently with PBS. Immediately thereafter, 200 μL of freshly prepared CCK-8 working solution was dispensed into every well. The cells were incubated for 1 h at 37 °C. The density of light absorption in every well was finally detected at 450 nm by a spectrophotometer.

An equal number of cells were plated into the 48-well plate and separated into three groups: the control (PBS) group, the n-EVs group, and the hy-EVs group. After 24 h, we removed the medium and washed it once with PBS. Then we added the Calcein AM/PI assay working solution (C2015S, Beyotime), followed by incubation for 20 min at 37 °C. We used PBS to wash and then the cells were observed and photo-documented via a microscope.

To assess the capacity of cell migration, trypsin was used to digest the cells and subsequently resuspended in serum-free medium. The upper compartment of a 12-well Transwell chamber was seeded with 200 µL of the prepared cell suspension. At the same time, the 800 μL DMEM medium of different components was introduced into the lower chamber for the nutrient supply of the cells. We used a cotton bud to gently discard the cells from the upper chamber after 24 h of incubation. The filter membrane was cleansed using PBS to get rid of any non-adherent cells. Following that, a 10 min fixation was performed on the cells using 4% paraformaldehyde. After a 15 min staining period with crystal violet, the cells were rinsed with PBS. A light microscope was employed to analyze and count the cells in the lower chamber.

Wound healing was simulated by culturing cells in 6-well plates in vitro. Once the cells reached full confluence, a straight scratch was made across the monolayer. PBS was used to briefly rinse the cells, and the prepared medium was added according to different treatment groups. Photographs were acquired at spaced intervals of 0 h, 12 h and 24 h to study the wound healing and cell migration. The extent of the wound region was quantitatively measured by the ImageJ software (v1.53t), and the formula ((D0 − Dt)/D0 × 100%) was used. Here, D0 refers to the size of the wound at the beginning, and Dt refers to the wound's residual size at the set time.

HUVECs were chosen as the research subjects. First, 100 μL of pre-cooled ABW^® Matrigel (0827045, ABW, Shanghai, China) at 4 °C was carefully added to the wells of a pre-chilled 48-well plate. The plate was subsequently left in an incubator to polymerize the Matrigel for 1 h at 37 °C. After that, HUVECs resuspended in the medium were inoculated onto the surface of the polymerized gel, and groups (control, n-EVs, and hy-EVs) were set up with the same number of cells in each group. After incubating the plate for 6 h at 37 °C, the generation of lumen formations on gel was seen and recorded under the optical microscope. ImageJ software was employed to process and analyze all images.

Following the various treatments, the cells were fixed using immunostaining fixative (P0098, Beyotime) for 10 min. Then cells were permeabilized using the immunostaining permeabilizing solution (P0096, Beyotime) for 10 min. Finally, the cells were blocked in immunostaining blocking solution (P0102, Beyotime) for 30 min. Overnight incubation of primary antibodies was performed at 4 °C. The cells were co-incubated with fluorescently labeled secondary antibody (A0516 and A0562, Beyotime) for 1 h. Afterward, DAPI (C1006, Beyotime) was used to stain the cells for 1 min. Finally, the cells were visualized and captured with the microscope. The primary antibodies used were CD86 (26903-1-AP, Proteintech, Wuhan, China), CD206 (18704-1-AP, Proteintech), HIF-1α (BM4083, BOSTER, Wuhan, China), CD31 (11265-1-AP, Proteintech), and VEGFA (19003-1-AP, Proteintech).

The 2′,7′-dichlorodihydrofluorescein diacetate fluorescent probe (DCFH-DA, Beyotime) was used to quantify ROS levels in HSFs and HUVECs. A 12-well plate was used to seed the cells, and then the cells were washed gently after 24 h of incubation. Afterwards, ROS staining was carried out. We used CLSM to take fluorescence photos of HSFs and HUVECs, then used ImageJ software to measure and quantify the signals.

The cells were harvested from the 6-well plate, with all procedures performed under ice-cold conditions. First, the cells were centrifuged at 4 °C for 10 min at 4000× g. The supernatant was carefully removed, and the pellet of cells was retained. The samples were then lysed in RIPA lysis buffer (P0013B, Beyotime) and protein inhibitor (P1005, Beyotime) for 30 min to ensure effective lysis. After being lysed, the sample was centrifuged at 4 °C for 10 min at 12,000× g. Following the separation by centrifugation, the supernatant was harvested. The obtained supernatant was mixed with SDS-PAGE loading buffer (P0015, Beyotime) at a predetermined ratio and heated for 10 min at 100 °C. The samples were stored at −20 °C for subsequent experiments. Subsequently, SDS-PAGE electrophoresis buffer (P0014D, Beyotime) was used for electrophoresis. Then, the membranes (ISEQ00010, Merck Millipore, Billerica, MA, USA) were employed for protein transfer for 20 min, followed by blocking them with blocking buffer (P0023B, Beyotime) for another 20 min. After overnight incubation of primary antibodies at 4 °C, secondary antibody (A0208, Beyotime) was applied for 1 h. The target protein band was detected using the ECL reagent kit (P0018M, Beyotime), and the gray value was measured using ImageJ to represent the protein expression level. The internal reference was used as the loading control to normalize the data. The primary antibodies included HIF-1α (BM4083, BOSTER), CD31 (11265-1-AP, Proteintech), VEGFA (19003-1-AP, Proteintech), and Alpha Tubulin (14555-1-AP, Proteintech).

All animal-related protocols were executed following the specified guidelines established by the Animal Experiment Center of Guangxi Medical University (Approval Number: 202505004, Approval date: 17 June 2025). The experimental models consisted of Sprague-Dawley (SD) male rats, eight-week-old SD rats were housed under standard laboratory animal conditions and were acclimatized to the new environment for 1 week. Following an overnight fasting period, all rats received an intraperitoneal administration of streptozotocin (S8050, Solarbio) at 50 mg/kg to induce diabetes. Blood glucose level was measured daily until stabilized after one week. All animals with serum glucose concentrations more than 16.67 mmol/L were used in the study. A diabetic skin defect model with a diameter of approximately 10 mm was prepared on the back of rats following intraperitoneal anesthesia using 2% pentobarbital (P3761, Sigma-Aldrich, St. Louis, MO, USA) at a dose of 200 μL/kg. The rats were randomly split into three groups, which were the negative control group (100 μL PBS by subcutaneous injection, n = 6), the n-EVs group (100 μL n-EVs at 20 μg/mL by subcutaneous injection, n = 6), and the hy-EVs group (100 μL hy-EVs at 20 μg/mL by subcutaneous injection, n = 6). A digital camera monitored the advancement of wound healing at days 0, 3, 6, 9 and 12 post-injury. The wound area was quantitatively estimated at each time point using ImageJ software. Euthanasia using overdose anesthesia after completion of experiments.

On the 12th day post-treatment, the wound tissues from rat skin were collected and fixed. After 48 h of fixation, they were dehydrated, embedded in paraffin, and sectioned for future analysis. Afterwards, hematoxylin–eosin (HE) staining was performed for general histopathological features. We performed Masson's trichrome staining to quantify the amount of collagen present in the wound tissues.

Sections were dewaxed, hydrated, antigenically repaired, and blocked for immunohistochemical staining. After an overnight exposure to the primary antibodies, the sections were subsequently exposed to the secondary antibodies for one hour, and counterstained with DAPI, visualized and imaged with a microscope. For immunofluorescence staining, fluorescently labeled secondary antibodies were utilized during the incubation step. All other procedures were similar to those used for immunohistochemistry staining except that the samples were protected from light throughout the process. The primary antibodies included CD31 (11265-1-AP, Proteintech), VEGFA (19003-1-AP, Proteintech), TNF-α (17590-1-AP, Proteintech), CD86 (26903-1-AP, Proteintech), and CD206 (18704-1-AP, Proteintech).

We used GraphPad Prism 8.0 software for all statistical analysis, and the data in this study were repeated at least three times independently and expressed as mean ± standard deviation (SD). The independent Student's t-test was used to compare the two groups. For three or more groups, one-way analysis of variance (ANOVA) was applied. A p-value < 0.05 was used to define statistical significance in all evaluations, indicating a significant difference between groups.

We took HUCMSCs from human umbilical cord tissue to grow them in culture. Flow cytometric analysis was used for the assessment of expression of specific surface markers of HUCMSCs. Flow cytometric analysis demonstrated that the expression rates of surface markers in HUCMSCs were 100% for CD73+, 99.8% for CD90+, and 100% for CD105+ (Figure 2a). In addition, we analyzed the expression levels of surface markers associated with monocytes, B lymphocytes, hematopoietic stem cells, and other hematopoietic cells. The results showed that the expression of CD11b+, CD19+, CD34+, CD45+ and HLA-DR+ was negligible, with positive rates of 0.066%, 0.43%, 0.29%, 0.047% and 0.41% (Supplementary Figure S1), which met the identification requirements for HUCMSCs. These results showed that HUCMSCs were successfully extracted from umbilical cord tissue. Thereafter, we induced HUCMSCs by normoxic and hypoxic conditions, respectively, and we found that there was no significant difference in morphology between the two cells after 48 h of culture (Supplementary Figure S2). After that, we used ultracentrifugation to extract and purify n-EVs and hy-EVs from HUCMSCs. NTA showed that the average particle sizes of n-EVs and hy-EVs were approximately 146.17 ± 1.54 nm and 150.67 ± 4.48 nm, respectively, which confirmed that both n-EVs and hy-EVs were within the expected size range (Figure 2b). Through TEM observation, the n-EVs and hy-EVs exhibited a unique cup-shaped structure and possessed an intact vesicle membrane (Figure 2c). Furthermore, Western blot analysis verified the high expression of the major markers CD63, CD9, and TSG101 in the purified n-EVs and hy-EVs, while Calnexin was negatively expressed (Figure 2d). Overall, the analysis revealed that the n-EVs and hy-EVs were successfully extracted. To verify whether HUVECs can take up n-EVs and hy-EVs, we used PKH26 to label n-EVs and hy-EVs, then co-incubated with HUVECs. After co-incubation for 6 h, the results of immunofluorescence showed that both n-EVs and hy-EVs were successfully taken up by HUVECs. HUVECs that took up hy-EVs exhibited stronger fluorescence signals, which surrounded the nucleus (Figure 2e). The findings suggested that HUVECs were able to take up hy-EVs more easily, which might potentially enhance their functional efficacy in biological processes.

The wound healing process involves several cell types including HSFs that are critical functional cells in this process [27]. First, we investigated the impact of hy-EVs on HSFs' functions. We used the CCK-8 assay to assess the role of hy-EVs on the proliferation of HSFs. The data revealed that hy-EVs promoted the proliferation of HSFs in a dose-dependent manner under high-glucose conditions (Figure 3a). Specifically, the proliferative effect of hy-EVs increased with concentration in the range of 0–20 μg/mL and plateaued at the concentration of 20 μg/mL, and the proliferative effect did not increase significantly with increasing concentration of hy-EVs. Thus, we concluded that 20 μg/mL was the optimal concentration for further study. Meanwhile, we compared the effect of the Control group and n-EVs group on the proliferation of HSFs at the same concentration. We found that hy-EVs had a stronger proliferative effect on HSFs at the same concentration (Figure 3b). Live–dead cell staining indicated that hy-EVs had good biocompatibility and verified their proliferative effect on HSFs (Figure 3c). We performed a Transwell assay to assess the migration ability. The study revealed that hy-EVs considerably enhanced the migration ability of HSFs compared to the control and n-EVs groups at 24 and 48 h (Figure 3d,e). Additionally, the scratch assay results indicated that the rate of wound healing in the control and n-EVs groups was markedly lower than the hy-EVs group at 24 and 48 h, indicating that hy-EVs could greatly promote the migration of HSFs (Figure 3f,g). Our findings demonstrated that hy-EVs were essential for enhancing the function of HSFs, which may aid in enhancing diabetic wound healing.

It is widely recognized that angiogenesis serves as a pivotal factor in diabetic wound healing, and promoting angiogenesis continues to be a significant challenge in diabetic wounds [28]. Therefore, we tested the impact of hy-EVs on the capabilities of HUVECs. Similarly, we used the CCK-8 assay to assess the impact of hy-EVs on proliferation of HUVECs. Consistent with the results of the study on HSFs, hy-EVs markedly enhanced the proliferation of HUVECs in a dose-dependent manner (Figure 4a). The optimal concentration of hy-EVs for promoting the proliferation of HUVECs was 20 μg/mL. At the same concentration, hy-EVs exhibited stronger proliferative effects on HUVECs contrasted with both the control group and the n-EVs group (Figure 4b). Live–dead cell staining further corroborated the good biocompatibility of hy-EVs and the proliferative effect on HUVECs (Figure 4c). The Transwell assay validated the proliferative impact of hy-EVs on the migration of HUVECs (Figure 4d,e). Additionally, we used the scratch wound healing experiment to assess the effect of hy-EVs on HUVECs. Similarly to the findings of the study on HSFs, the scratch healing rate of HUVECs was lowest in the control group, followed by the n-EVs group, while the scratch healing rate was significantly increased by hy-EVs treatment (Figure 4f,g). Given the positive correlation between angiogenesis and wound healing, we further investigated whether hy-EVs could promote angiogenesis in vitro. The results of our study showed that the highest number of lumens formed in the hy-EVs group, although n-EVs treatment increased lumen formation, it was markedly lower than the hy-EVs group, while the control group exhibited the fewest lumens (Figure 4h–j). The results confirmed the ability of hy-EVs to stimulate angiogenesis in vitro. Collectively, these findings revealed that hy-EVs had significant characteristics of promoting the proliferation, migration, and angiogenesis of HUVECs, thereby contributing to the promotion of wound healing in diabetes.

Recent studies have shown that immune imbalance in diabetic wounds constitutes a critical factor influencing the recovery progression of such wounds [29]. Based on this, we further explored the impact of hy-EVs on macrophages. The results showed that the expression of CD86 (a marker for M1 macrophages) was markedly reduced after hy-EVs treatment, while the expression of CD206 (a marker for M2 macrophages) was notably increased, indicating that hy-EVs can modulate the transition of macrophages from M1 (pro-inflammatory) to M2 (anti-inflammatory) phenotype (Figure 5a,b), thereby regulating the immunological microenvironment associated with diabetic wounds.

Recent research has pointed out that excessive ROS accumulation will significantly impair the regular wound recovery process [30]. Therefore, we further evaluated whether hy-EVs can lessen the ROS levels. The outcomes revealed that the fluorescence intensity of ROS was highest in the Control group and that hy-EVs treatment could effectively inhibit the production of ROS. In contrast, n-EVs treatment showed a relatively limited ability to scavenge ROS (Figure 5c,d). This indicated that hy-EVs can effectively scavenge ROS, which may be the key mechanism for relieving excessive oxidative stress in diabetic wounds.

In addition, to verify whether hypoxia activates HIF-1α, we first evaluated the effect of hypoxia induction on HIF-1α expression in HUCMSCs, and we found that HIF-1α was highly expressed in hypoxia-induced HUCMSCs after 48 h, whereas normoxic cultured HUCMSCs were low in HIF-1α expression (Figure S3). Our previous experimental results have confirmed the in vitro angiogenic properties of hy-EVs. Therefore, we evaluated whether hy-EVs promote angiogenesis by activating HIF-1α and assessed the effects of different treatments on the expression levels of VEGFA and CD31. Immunofluorescence results showed that the expression levels of HIF-1α, VEGFA and CD31 were significantly up-regulated after treatment with hy-EVs (Figure 5e,f). Consistent with the results of immunofluorescence experiments, protein blotting experiments further confirmed that hy-EVs significantly increased the expression levels of HIF-1α, VEGFA and CD31 (Figure 5g,h). This indicated that hy-EVs promoted angiogenesis by activating the HIF-1α pathway.

In summary, our study demonstrated that hy-EVs can play multiple roles, such as inducing a phenotypic shift from M1-type to M2-type in macrophages, attenuating oxidative stress, exerting anti-inflammatory effects, and stimulating angiogenesis. All these factors were crucial in facilitating diabetic wound healing.

Modeling the diabetic rats was performed so that we could investigate the function of hy-EVs during wound healing. Following this, a cutaneous wound about 10 mm in diameter was inflicted on the rats' dorsal area, and a subcutaneous injection of 100 μL of 20 μg/mL hy-EVs was performed on the wound edge every two days (Figure 6a). Wound healing was investigated with the aid of macroscopic images taken on days 0, 3, 6, 9 and 12 post-modeling. Throughout the experiment, it was observed that the hy-EVs group had the fastest rate of wound healing, while the Control group was the slowest (Figure 6b–d). Notably, treatment with n-EVs could indeed significantly promote the regenerative process of wounds in diabetes. However, compared with the hy-EVs group, the healing process was significantly delayed, which fully indicated that hy-EVs played an important role in facilitating diabetic wound healing. Histological analysis by HE staining also confirmed these findings (Figure 6e,f). On day 12, the wounds in the hy-EVs group were nearly fully healed, and new hair follicles were visible. In contrast, the control group and the n-EVs group both showed relatively limited skin thickness and hair follicle regeneration. The findings from Masson's trichrome staining demonstrated that the hy-EVs group exhibited the most abundant collagen deposition (Figure 6g,h), indicating that hy-EVs can promote collagen deposition and consequently enhance diabetic wound healing. Additionally, we measured the expression levels of angiogenesis-associated factors like VEGFA and CD31. Immunohistochemical staining of excised tissues indicated that hy-EVs notably upregulated the expression levels of VEGFA (Figure 7a,b) and CD31 (Figure 7c,d) in vivo to promote angiogenesis in diabetic wounds. Meanwhile, we also detected the expression of the inflammatory factor TNF-α (Figure 7e,f), which was markedly decreased in the hy-EVs group, indicating that hy-EVs significantly reduced inflammation in the wound. In the hy-EVs group, we found that CD206 expression was highest (Figure 7g,h), while CD86 expression was lowest (Figure 7i,j). This suggested that hy-EVs can be used to modulate the immune response and facilitate wound healing by modulating the shift from M1 to M2 macrophage polarization. Collectively, these findings emphasized the important role of hy-EVs in enhancing diabetic wound healing.