Eight-week-old male C57BL/6 mice weighing 20–26 g were procured from Jie Si Jie Laboratory Animal Corp (Shanghai, China). The mice were kept in an SPF environment, following a standard light/dark cycle of 12 h, with ad libitum access to food and water. The study was conducted at Nanchang University, China, following institutional guidelines and authorized by the Animal Ethics Committee, with an ethical institution number of SYXK 2021-0001, and an animal experimentation ethics number of NCULAE-20230610002. The welfare of the mice was ensured following the updated 2011 eighth edition of the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health.

Mice were subjected to oral administration of dapagliflozin (1 mg/kg/day, Sigma-Aldrich Co. St. Louis, MO) or saline for 2 weeks. Subsequently, ten-week-old mice were anesthetized by inhalation of isoflurane. The surgical site was prepared by shaving and cleaning the fur with ethanol, and the femoral artery was isolated from the saphenous nerve and vein. Ligation of the femoral artery was conducted at the proximal profunda femoris and epigastric arteries and distal popliteal fossa, while the right limb served as a control. The surgical site was then sutured.

After ischemic surgery, laser Doppler perfusion imaging (PeriScan PIM 3, Perimed) was performed on days 0, 3, 7, 14, and 21 to evaluate blood flow in the hindlimbs. The mice were anesthetized with isoflurane (2%–4%) and maintained on a heating pad at 37°C for 3 minutes. The blood flow was displayed using different colored pixels corresponding to changes in laser frequency. The perfusion ratio between the ischemic and non-ischemic hindlimbs was calculated by determining the mean pixel value.

Tissue damage and hindlimb functionality were evaluated in mice subjected to the hindlimb ischemia (HLI) model on the 14th day postoperatively. The severity of necrosis was assessed using a scoring system, wherein a score of 0 denoted the absence of necrosis in the ischemic limb, a score of 1 indicated necrosis limited to the toe, a score of 2 represented necrosis extending to the dorsum of the foot, a score of 3 corresponded to necrosis extending to the crus, and a score of 4 indicated necrosis extending to the mid-tibia or complete limb necrosis. To evaluate limb motor function, a scoring system was employed. Mice exhibiting no usage of the limb were assigned a score of 1. Mice demonstrating limb usage but no utilization of the foot were assigned a score of 2. Restricted foot usage led to a score of 3, while mice displaying foot usage but no toe movement received a score of 4. Mice with unrestricted limb usage were assigned the highest score of 5 (Ma et al., 2021).

The gastrocnemius muscles were collected on day 21 post-surgery and sliced into 8-mm-thick paraffin-embedded sections. They were dewaxed and rehydrated before undergoing an antigen retrieval process. Subsequently, the sections were incubated with antibodies against a-SMA (1:100 dilution, ab5694, Abcam, Cambridge, United Kingdom) and CD31 (1:1000 dilution, ab182981, Abcam, Cambridge, United Kingdom) overnight at 4°C. After several washes with TBST, the sections were treated with secondary HRP-labeled antibodies (1:200 dilution) for 2 h at room temperature. Finally, the CD31 and a-SMA staining was semiquantitatively evaluated using ImageJ software. Two or four fields of view were examined in each section and the mean capillary density or arterioles density values for each field of view were recorded.

To quantify perivascular fibrosis in the gastrocnemius muscles, a fixation step using 4% paraformaldehyde was followed by paraffin embedding. Masson trichrome staining was conducted to visualize the fibrosis, and a digital imaging system was utilized to measure the fibrotic area. This was then represented as a percentage of the total tissue area.

The paraffin-embedded sections were incubated with primary antibodies for CD68 (1:50 dilution, ab283654, Abcam, Cambridge, United Kingdom), iNOS (1:200 dilution, ab178945, Abcam, Cambridge, United Kingdom), and CD206 (1:1000 dilution, ab64693, Abcam, Cambridge, United Kingdom) overnight at 4°C, followed by treatment with appropriate secondary antibodies (1:400 dilution) and counterstaining with DAPI. The labeled cells were examined under fluorescence microscopy.

Bone-marrow-derived macrophages (BMDMs, 1 × 10⁶ cells/each well in 6-well plate) were collected from 6 to 8-week-old mice via flushing of the femur and tibia, and were cultured in DMEM (Gibco, New York, United States) medium supplemented with 100 ng/mL M-CSF (Sigma-Aldrich Co. St. Louis, MO) for 5 days. After 36 h of treatment with dapagliflozin, macrophages were subjected to an exposure of 100 ng/mL LPS (Sigma-Aldrich Co. St. Louis, MO) for 12 h, following which they were deemed the conditioned medium (CM) group. Macrophages that were exposed to 100 ng/mL LPS for 12 h without any prior treatment were categorized as Con group macrophages. The resulting CM from these macrophage cells was used to replace the medium for the subsequent incubation of HUVECs (Cyagen Biosciences).

To assess cell proliferation, HUVECs were exposed to a CM produced by macrophage cells for 36 h, with medium changes every 12 h. After pretreatment, the HUVECs were subjected to hypoxia for 36 h, and a 2-h treatment with 20 μM 5-ethynyl-2′-deoxyuridine (EdU, Beyotime, Shanghai, China) was performed at 37°C to measure cell proliferation. The nuclei of the proliferated cells were stained green, and the nuclei of all cells were counterstained with DAPI. The quantification of cell proliferation was determined by calculating the ratio of EdU + cells to the total number of cells using ImageJ.

Prior to performing the scratch assay and Transwell Migration Assay, HUVECs were treated with a CM for 36 h. Cells were seeded into 6-well plates and grown to 95% confluence, and then scratch wound was induced in a monolayer of cells using a p10 pipette tip, followed by cultivation in a sterilized hypoxic incubator for 48 h using a serum-free medium. The healing process of the wound was assessed by calculating the reduction in the wound gap utilizing the ImageJ software.

After the treatment, the HUVECs (1×10⁵/each well) were suspended in serum-free DMEM and were seeded into Transwell inserts (Corning, NJ), followed by exposure to a hypoxic environment for 8 h. A solution consisting of 5% serum in DMEM was utilized as the attractant, which was placed in the lower chamber of the Transwell insert. A 4% formaldehyde fixative and crystal violet staining were applied to the cells that migrated to the bottom surface of the membrane, whereas the cells remaining in the upper compartment were removed using cotton swabs.

For the tube formation assay, HUVECs were subjected to the same treatment conditions used for the proliferation assay for 36 h. The 24-well plates and tips were precooled at −20°C before the start of the experiments, and the plates were coated with Matrigel (BD Biosciences, New Jersey, United States) and incubated at 37°C for 30 min until the Matrigel solidified. The cells (1×10⁵/each well) were then added to the plate and incubated in a sterilized hypoxic incubator for 6 h. The resulting capillary-like structures were visualized using a microscope and quantified using ImageJ software.

Protein from the cells or tissue was extracted using an ice-cold RIPA buffer, and the concentration of proteins was quantified by a BCA kit (Invitrogen, California, United States). Afterward, around 20 μg of protein was boiled for 5 min at 98°C and subjected to SDS-polyacrylamide gel electrophoresis using either a 12% or 15% gel. The separated proteins were transferred onto PVDF (Millipore, Massachusetts, United States) membranes, followed by a blocking step using a 5% BSA (Solarbio Science & Technology, Beijing, China) solution for 120 min at room temperature. Primary antibodies were incubated with the PVDF membranes overnight at 4°C, followed by washing with TBST and incubation with secondary antibodies. Finally, the protein on the PVDF membranes was visualized using enhanced chemiluminescence (ECL Invitrogen, California, United States) solution. Anti-IL-1β (1:1000 dilution, 29530-1-AP, Proteintech, Wuhan, China), anti-IL-6 (1:1000 dilution, 12912, CST Danvers, MA, United States), anti-TNF-α (1:1000 dilution, 17590-1-AP, Proteintech, Wuhan, China), anti-iNOS (1:1000 dilution, ab178945, Abcam, Cambridge, United Kingdom), anti-Arg-1 (1:1000 dilution, 93668, CST Danvers, MA, United States), anti-phosphorylated-NF-κB (1:1000 dilution, 3034, CST Danvers, MA, United States), anti-NF-κB (1:1000 dilution, 3033, CST Danvers, MA, United States), anti-phosphorylated-IκBα (1:1000 dilution, 2895, CST Danvers, MA, United States), anti-IκBα (1:1000 dilution, 4,814, CST Danvers, MA, United States), and anti-β-actin (1:1000 dilution, 4,970, CST Danvers, MA, United States) antibodies were utilized in the experiment.

The present investigation employed GraphPad Prism 8 software to perform statistical analysis, with data presented as mean ± SD. The t-tests or non-parametric tests were conducted to compare the means of the two groups. Moreover, one-way ANOVA was utilized to compare the means among multiple groups, and the Bonferroni post hoc test was performed to evaluate the significance between the groups. The significance was determined at a threshold of p < 0.05.

To assess the effect of dapagliflozin on hindlimb ischemic disease, C57BL/6 mice were pretreated with dapagliflozin for 2 weeks, and then the HLI model was induced by ligating the unilateral femoral artery of mice. Laser Doppler imaging assessed perfusion recovery on days 1, 3, 7, 14, and 21 after the operation (Figure 1A). The results showed a significant improvement in blood flow recovery in the dapagliflozin-treated group after the surgical procedure compared to the PBS group (Figure 1B; C, Two-way ANOVA, p < 0.05). In addition, Tissue damage and limb motor function were scored and statistically analysed by unpaired t-test in both groups, which revealed that the gangrene and limb necrosis scores were strikingly higher in the PBS-treated groups than in the DAPA-treated group (HLI + PBS vs. HLI + DAPA = 2.33 vs. 1.54, unpaired t-test, p < 0.05) (Figure 1D; E), but there was no remarkable difference in limb motor function scores (Figure 1F).

The gastrocnemius muscle tissue of mice was then collected on the 21st day after the operation to evaluate the effect of dapagliflozin on angiogenesis. Immunohistochemical analysis was used to evaluate the expression of CD31 and α-SMA in muscle tissue as characteristic markers for capillary and arteriole. The results showed that the capillary (HLI + PBS vs. HLI + DAPA = 217.97 vs. 308.03, one-way ANOVA, p < 0.01) (Figure 2A; B) and arteriole (HLI + PBS vs. HLI + DAPA = 18.23 vs. 27.48, one-way ANOVA, p < 0.05) (Figure 2C; D) densities in ischemic muscles were significantly increased in the HLI + DAPA group mice than in those administered the PBS control. Moreover, a Masson’s trichrome staining of gastrocnemius muscles was carried out to assess the degree of fibrosis (Figure 2E), and the results revealed that the fibrotic area was remarkably smaller in the HLI + DAPA group on day 21 after the operation (HLI + PBS vs. HLI + DAPA = 2.12% vs. 1.25%, one-way ANOVA, p < 0.05) (Figure 2F).

Macrophages play a critical role in angiogenesis after an ischemic attack (Fu et al., 2023). We evaluated macrophage infiltration in mouse gastrocnemius muscle tissue collected on day five after HLI. Analysis of immunofluorescence staining showed no significant differences in total macrophage numbers defined by CD68-positive between the PBS and dapagliflozin-treated groups (Supplementary Figures S1A, B). Interestingly, the expression levels of CD206 (the marker of M2 macrophages) were significantly higher in ischemic muscle from DAPA-treated mice than in PBS-treated mice (HLI vs. HLI + DAPA = 21.78% vs. 32.92%, unpaired t-test, p < 0.05) (Figure 3A; B). Correspondingly, iNOS (the marker of M1 macrophages) immunofluorescent staining showed significantly lower M1 macrophage density in the HLI + DAPA group (HLI vs. HLI + DAPA = 20.06% vs. 13.54%, Mann–Whitney test, p < 0.01) (Figure 3C; D). These results were further confirmed by Western blot analysis (Figure 3E), which showed that dapagliflozin treatment significantly decreased iNOS expression (HLI vs. HLI + DAPA = 0.86 vs. 0.62, unpaired t-test, p < 0.05) and increased the protein expression level of Arg-1 (HLI vs. HLI + DAPA = 0.75 vs. 1.08, unpaired t-test, p < 0.01), a marker highly expressed in M2 macrophages (Figure 3F; G).

Next, mouse bone marrow-derived macrophages (BMDMs) were utilized in vitro experiments. The protein expression levels of iNOS and Arg-1 in BMDMs were evaluated following treatment with dapagliflozin at varying concentrations and treatment times. Western blot analysis demonstrated that the protein expression of Arg-1 was upregulated as dapagliflozin treatment time (Supplementary Figures S2A, B) and concentration (Supplementary Figures S2C, D) increased, while the expression of iNOS was downregulated. Based on these results, an effective concentration of 10 μM and a treatment time of 36 h was used in subsequent experiments (Figure 3H). The results showed that dapagliflozin significantly decreased the protein expression of iNOS (LPS vs. LPS + DAPA = 1.29 vs. 0.83, unpaired t-test, p < 0.05) and increased the expression of Arg-1 (LPS vs. LPS + DAPA = 1.02 vs. 1.49, unpaired t-test, p < 0.05) in LPS-stimulated BMDMs (Figure 3I; J).

Given the key role of the NF-κB signaling pathway in macrophage polarization (Takeda et al., 2011), Western blot analysis was performed to detect the expression and phosphorylation states of NF-κB and IκB-α (Figure 4A). Quantification of the Western blot results indicated that the protein expression of total NF-κB and total IκB-α did not differ between groups, whereas ischemic injury significantly enhanced the phosphorylation of both molecules, which were attenuated by dapagliflozin administration (p-NF-κB/t-NF-κB: HLI vs. HLI + DAPA = 1.45 vs. 1.08; p-IκB-α/t-IκB-α: HLI vs. HLI + DAPA = 1.24 vs. 0.98, one-way ANOVA, p < 0.05) (Figure 4B; C). Additionally, inflammatory factors such as IL-1β, IL-6, and TNF-α were measured by Western blot (Figure 4D), and their expression levels were significantly lower in the HLI + DAPA group than in the HLI group (Figure 4E; F, G, one-way ANOVA, p < 0.05).

To further investigate the mechanism by which dapagliflozin promotes angiogenesis, we treated HUVECs under hypoxic conditions with dapagliflozin and observed changes in vascular endothelial cell function in vitro. However, we found no significant differences in the migration (Supplementary Figures S3A, B), proliferation (Supplementary Figures S3C, D), and tube formation ability (Supplementary Figures S3E, F, D) of HUVECs between the dapagliflozin-treated and PBS groups.

We hypothesize that dapagliflozin could indirectly regulate the function of endothelial cells by promoting M2 macrophage polarization. BMDMs were then exposed to DAPA in vitro and its conditioned medium was collected, which was subsequently added to HUVECs under hypoxic conditions. Subsequently, scratch assay, transwell, 5-Ethynyl-2′-deoxyuridine (EdU), and tube formation assays were performed. Our scratch test revealed that the CM from DAPA-treated macrophages enhanced the migration ability of HUVECs compared with that from Con group macrophages (Con vs. CM = 28.21% vs. 37.02%, unpaired t-test, p < 0.01) (Figure 5A; B). Similarly, the CM from DAPA-treated macrophages increased the number of migrated HUVECs in the lower compartment of the transwell chamber (Con vs. CM = 1.00 vs. 1.19, unpaired t-test, p < 0.01) (Figure 5C; D). Furthermore, DAPA-treated macrophage CM increased the number of EdU-positive cells (Con vs. CM = 23.83% vs. 29.49%, unpaired t-test, p < 0.05) (Figure 5E; F), indicating its potential to promote HUVEC proliferation. Consistent with these findings, our tube formation assays (Figure 5G) demonstrated that the CM from macrophages treated with DAPA increased the total tube length (Con vs. CM = 1.00 vs. 1.10, unpaired t-test, p < 0.05) (Figure 5H) and the number of branches (Con vs. CM = 1.00 vs. 1.08, unpaired t-test, p < 0.05) (Figure 5I).