Human dermal microvascular endothelial cells (HMEC-1) were obtained from the American Type Culture Collection (ATCC, CRL-3243). The cells were cultured in MCDB 131 complete medium (Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Nanjing Wisent Biotechnology Co., Ltd.), 10 mM L-glutamine (MedChemExpress), 10 ng/ml recombinant human epidermal growth factor (EGF; MilliporeSigma; Merck KGaA), 1 μg/ml hydrocortisone (MilliporeSigma; Merck KGaA) and 1% penicillin-streptomycin (Beyotime Biotechnology). The cells were maintained at 37°C in a humidified incubator with 5% CO₂. The medium was changed every 2-3 days, and cells were passaged upon reaching 80-90% confluency.

For transfection, cells in optimal growth conditions were used. miR-222-3p mimic (Guangzhou RiboBio Co., Ltd.) and si-APLN (Shanghai Sangon Biotech Co., Ltd.) were transfected using an in vitro miRNA/siRNA transfection reagent (Yisheng Biotechnology Co., Ltd.). Transfection of the APLN overexpression plasmid (Shanghai Sangon Biotech Co., Ltd.) was performed using Lipofectamine™ 2000 transfection reagent (Thermo Fisher Scientific Inc.). Following the manufacturer's instructions, transient transfection was performed when the cell confluency reached 50%. For each transfection, 2.5 μg nucleic acid was used. The nucleic acid was incubated with the transfection reagent for 10 min at room temperature, and the transfection mixture was then added to the cells. Subsequent experiments were performed 48 h following transfection. HMEC-1 cells were randomly divided into the following groups: Normal control (NC), MGO intervention (MGO, MilliporeSigma; diluted with PBS to concentrations of 50, 100, 200, 400, 600, 800 and 1,000 μmol/l), MGO group (400 μmol/l), MGO + miR-con (MGO-treated HMEC-1 cells transfected with negative control miR-con), MGO + miR-222-3p (MGO-treated HMEC-1 cells transfected with miR-222-3p mimic), MGO + miR-222-3p + oe-APLN group (miR-222-3p mimic and APLN overexpression plasmid were co-transfected into MGO-treated HMEC-1 cells), MGO + miR-222-3p + Apelin-13 group [miR-222-3p mimic was transferred into MGO-treated HMEC-1 cells and 50 ng/ml Apelin-13 (MedChemExpress) was supplemented], MGO + si-APLN (MGO-treated HMEC-1 cells transfected with si-APLN), and MGO + si-NC (MGO-treated HMEC-1 cells transfected with negative control si-NC). The specific sequences of the miR-222-3p mimic and its negative control (miR-con) are proprietary to the supplier and are not disclosed. The siRNA sequences are listed in Table SI.

RT-qPCR was performed to measure the expression levels of miR-222-3p and APLN. Total RNA was isolated from cells, animal tissues and human tissues using TRIzol reagent (Thermo Fisher Scientific, Inc.). RNA concentration was quantified using a spectrophotometer (NanoDrop; Thermo Fisher Scientific, Inc.). According to the manufacturer's protocol, reverse transcription and quantitative PCR for miR-222-3p were performed using the miRNA reverse transcription and fluorescence quantitative detection kit (cat. no. E22003, Shanghai GenePharma Co., Ltd.). The PCR thermocycling conditions for miR-222-3p were as follows: An initial denaturation at 95°C for 3 min, followed by 40 cycles of 95°C for 12 sec and 62°C for 40 sec, with fluorescence signal acquisition at the 62°C step. Similarly, mRNA reverse transcription and amplification were conducted using the mRNA reverse transcription and fluorescence quantitative detection kit (cat. nos. 22107 and 22204, Shanghai Tolo Gang Biotechnology Co., Ltd.). The PCR thermocycling conditions for mRNA were as follows: An initial denaturation at 95°C for 30 sec, followed by 40 cycles of 95°C for 10 sec and 60°C for 30 sec, with fluorescence signal acquisition at the 60°C step. Relative expression levels of miR-222-3p and APLN were calculated using the comparative threshold cycle method (2^−ΔΔCq) (27), implementing U6 snRNA and GAPDH as endogenous reference genes in a triplicate experimental design, and average values were used for analysis. The primer sequences are listed in Table SII.

Total protein was extracted from cells, mouse wound tissues, and human tissue specimens using RIPA lysis buffer (Beyotime Biotechnology), with the subsequent quantification of protein concentrations conducted through bicinchoninic acid assay. Following denaturation by boiling, 15 μg of protein samples were loaded into 12% SDS-PAGE gel wells for electrophoresis to separate proteins based on their molecular weight. The proteins were then transferred onto polyvinylidene fluoride (PVDF) microporous membranes. Following membrane transfer, the membranes underwent blocking in 5% (w/v) non-fat dairy protein solution (25°C, 2 h). The membranes were then incubated overnight (16-18 h) cold incubation (4°C) with primary antibodies diluted at pre-determined working concentrations, as follows: Rabbit anti-β-actin (1:10,000; cat. no. R380624, Chengdu Zen-Bioscience Co., Ltd.), rabbit anti-APLN (1:2,000; cat. no. BD-PT5163, Biodragon; https://www.biodragons.com/cn/goods/goodsView?GoodsId=23366&Catalog=), rabbit anti-IL-1β (1:1,000; cat. no. 16806-1-AP, Proteintech Group, Inc.), anti-IL-6 (1:1,000; cat. no. 21865-1-AP, Proteintech Group, Inc.) and anti-IL-8 (1:1,000; cat. no. 27095-1-AP, Proteintech Group, Inc.). Following overnight incubation (4°C), the membranes underwent sequential Tris-buffered saline-Tween 20 (TBST, Shanghai Sangon Biotech Co., Ltd., cat. no. B548105) rinses (3×8 min) prior to 2 h of exposure (25°C) to horseradish peroxidase-conjugated species-matched secondary antibodies (1:5,000; cat. no. 511203, Zen-Bioscience). The membranes were then rinsed with TBST three times again. Chemiluminescent signal acquisition was performed using a Tanon 5200 chemiluminescence imaging system (Tanon Science & Technology Co., Ltd.), followed by densitometric quantification via Image J software (version 1.8.0; National Institutes of Health) with β-actin normalization for relative protein expression determination.

Cell viability was assessed using the Cell Counting kit-8 (CCK-8; Labgic Technology Co., Ltd.). Following cell digestion, cells from each group were resuspended, and the density was adjusted. Cells were plated at 5×10³/well density in 96-well microplates, with at least six replicates per group. Following the designated intervention, the CCK-8 reagent was diluted with serum-free medium at a 10:1 ratio to prepare the CCK-8 working solution, and 100 μl of this solution were added to each well. Following 1 h of incubation under standard culture conditions (37°C, 5% CO₂), optical density measurements were conducted at 450 nm using a multimode plate reader (Tecan Group Ltd.).

Cellular apoptosis was evaluated via dual-fluorochrome labeling using the Annexin V-FITC/PI Apoptosis Detection kit (cat. no. K2003, APExBIO Technology LLC) according to the manufacturer's protocols. Following designated intervention, the old culture medium was collected to terminate digestion. Adherent cells underwent enzymatic dissociation using EDTA-free trypsinization. The cell suspension was subjected to sequential centrifugation (300 × g, 5 min, 4°C) and dual PBS rinses at 4°C, followed by resuspension in 500 μl binding buffer (cat. no. K2003, APExBIO Technology LLC). Fluorescent staining was achieved by the addition of 5 μl Annexin V-FITC and 5 μl propidium iodide (PI) working solutions, with subsequent light-protected incubation (25°C, 10 min). Immediate flow cytometric analysis was performed using the Beckman Coulter CytoFLEX LX platform.

The principle of the cell scratch assay is to evaluate coordinated cell migration driven by intercellular adhesion and matrix contact through the measurement of wound closure rates. It is primarily suitable for studies related to skin wound healing and epithelial repair (28-30). In the present study, when the cell density reached ~80-90%, two parallel lines were drawn on the bottom of the culture plate, followed by the application of two additional lines on the cell surface using a 200-μl pipette tip, perpendicular to the parallel lines at the plate bottom. Post-PBS rinsing (2X), serum-depleted growth medium was introduced for controlled migration studies under standard incubation. The scratch wound was photographed at 0 and 24 h using the MShot digital imaging system (Guangzhou Mingmei Photoelectric Technology Co., Ltd.), with cell motility quantified as follows: Relative migration rate (%)=[(0 h scratch distance -24 h scratch distance)/0 h scratch distance] ×100%.

In the tube formation assay protocol, 50 μl Matrigel matrix was uniformly dispensed into 96-well culture plates and allowed to solidify at 37°C with 5% CO₂ for 60 min. Following density calibration to 3×10⁵ cells/ml, 100 μl of the standardized cell suspension were dispensed into each well using calibrated micropipettes. Following 4 h of incubation (37°C), tube formation was observed under the microscope and photographed using the MShot digital imaging system (Guangzhou Mingmei Photoelectric Technology Co., Ltd.). The total number of branch points and total tube length were quantified using ImageJ software (version 1.8.0; National Institutes of Health).

The wild-type plasmid vector (APLN 3'-UTR-WT) and the mutant plasmid vector (APLN 3'-UTR-Mut) were constructed (General Bio Co., Ltd.). The two plasmids were co-transfected with either miR-222-3p mimic or miR-con using Lipo8000™ Transfection Reagent (cat. no. C0533, Beyotime Biotechnology) according to the manufacturer's instructions. For each transfection, 2.5 μg nucleic acid was used. The transfection mixture was incubated at room temperature for 10 min and then added to the cells. The groups were classified as follows: miR-222-3p + APLN-WT, miR-con + APLN-WT, miR-222-3p + APLN-Mut, and miR-con + APLN-Mut. Following 48 h of transfection, 100 μl Firefly luciferase detection reagent were added to each well according to the dual luciferase reporter gene assay kit protocol (cat. no. RG027, Beyotime Biotechnology). Firefly luciferase activity was measured using a multimode plate reader (Tecan Group Ltd.) after mixing. Subsequently, 100 μl Renilla luciferase detection reagent (cat. no. RG027, Beyotime Biotechnology) were added, and Renilla luciferase activity was detected after mixing.

Intracellular ROS levels were measured using a ROS Assay kit (cat. no. S0033S Beyotime Biotechnology). Following washing with PBS, the cells were incubated with DCFH-DA working solution (included with the kit) diluted 1:1,000 in serum-free medium at 37°C for 30 min in the dark, with gentle mixing every 5-10 min. The cells were then washed three times with serum-free medium to remove extracellular probe. For nuclear staining, cells were incubated with Hoechst 33342 (cat. no. C1029, Beyotime Biotechnology; diluted 1:100 in serum-free medium) at 37°C for 15 min in the dark. Following incubation, the cells were washed three times with PBS. Fluorescence images were captured using a fluorescence microscope (Olympus Corporation; three random fields per group), and green fluorescence intensity indicated cellular ROS levels.

Paraffin-embedded tissue sections from mouse peri-wound tissues (5-μm thick) were routinely deparaffinized in xylene and rehydrated through graded ethanol. Heat-induced antigen retrieval was performed in citrate buffer (pH 6.0) under high pressure for 2 min. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide at room temperature for 20 min. After washing with PBS, the sections were incubated with rabbit anti-CD31 antibody (cat. no. ab182981, 1:1,000, Abcam) and rabbit anti-VEGFA antibody (1:400; cat. no. GB15165, Wuhan Servicebio Technology Co., Ltd.) at 37°C for 60 min. Following washes with PBS, the sections were incubated with an HRP-conjugated goat anti-rabbit IgG secondary antibody (1:2,000; cat. no. ab205718, Abcam) at 37°C for 20 min. Following further PBS washes, immunoreactivity was visualized using 3,3'-diaminobenzidine (DAB; cat. no. ZLI-9018, Zsbio) substrate, and the sections were counterstained with hematoxylin (cat. no. BA-4041, Baso Diagnostics Inc.) at room temperature for 1 min, dehydrated through graded ethanol (cat. no. A500737, Shanghai Sangon Biotech Co., Ltd.), cleared in xylene (cat. no. A530011, Shanghai Sangon Biotech Co., Ltd.) and mounted with neutral resin (cat. no. C0173, Beyotime Biotechnology). Images were acquired using a light microscope (Nikon Corporation). Brown-yellow granular staining was considered positive.

DFU tissues were obtained from patients with type 2 diabetes undergoing debridement or amputation (DFU group, n=12; 8 males and 4 females; age range, 50-78 years; median age, 61 years). Non-diabetic chronic lower limb ulcer tissues were collected from non-diabetic patients with chronic wounds as the controls (skin ulcer control, SUC group, n=12; 7 males and 5 females; age range, 45-72 years; median age, 58 years). All tissue specimens were collected at the First Affiliated Hospital of Anhui Medical University (Hefei, China) between June, 2023 and December, 2023. Written informed consent was obtained from all participants, and the study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Anhui Medical University (Ethics batch number P20210039). Following surgical excision, tissue samples were immediately rinsed with sterile saline to remove blood, snap-frozen in liquid nitrogen, and stored at −80°C for RNA and protein extraction.

The study design involving animal models received ethical validation (approval no. LLSC20201040) from the Animal Care and Use Committee of Anhui Medical University (Hefei, China), with operational protocols strictly conforming to their guidelines. Male C57BL/6 mice (6-8 weeks old; median age, 7 weeks; n=70 males, 0 females; GemPharmatech Co., Ltd.) were intraperitoneally injected with streptozotocin (STZ, 50 mg/kg; cat. no. B2001, Beijing Boaigang Biotechnology Co., Ltd.) for 5 consecutive days to induce diabetes. Fasting blood glucose (FBG) levels were measured 2 weeks thereafter, and mice with FBG levels ≥16.7 mmol/l on two consecutive measurements were diagnosed as diabetic. Thereafter, FBG was monitored twice weekly to confirm persistent hyperglycemia. All diabetic mice were maintained in the hyperglycemic state for 4 weeks prior to use in subsequent experiments. Prior to wound creation, FBG and body weight were measured in all mice, and both parameters were monitored and recorded throughout the wound healing experiment (post-operative day 0 to day 8). The exclusion criteria were defined as follows: (i) FBG <16.7 mmol/l at any monitoring point during the experiment; (ii) >20% loss of initial body weight after wounding; (iii) severe health issues caused by hyperglycemia or infection, such as persistent lethargy, piloerection, or markedly reduced activity. Mice meeting any of these criteria were excluded from the final data analysis. No animals met these exclusion criteria during the experiment. For the APLN overexpression groups, AAV-oe-APLN vectors (1.0×10¹¹ vector genomes/mouse) were administered by multi-point injection into the dorsal skin 3 weeks prior to wound creation. Both normal and diabetic mice were anesthetized with 1% sodium pentobarbital (50 mg/kg), and a full-thickness excisional wound (6 mm in diameter) was created on the dorsal midline. According to the experimental design, the following reagents were injected around the wound at a dose of 2.5 nmol per wound, every other day, for a total of four injections: miR-222-3p agomiR, agomiR negative control (agomiR-NC), miR-222-3p antagomiR and antagomiR negative control (antagomiR-NC) (Guangzhou RiboBio Co., Ltd.). The specific sequences of these products are proprietary to the supplier and are not disclosed. The groups were as follows: Ctrl + agomiR-NC, Ctrl + agomiR-222-3p, DM + agomiR-NC, DM + agomiR-222-3p, DM + agomiR-222-3p + oe-APLN, DM + antagomiR NC, and DM + antagomiR-222-3p (n=10 per group). On post-operative days 4 and 8, the mice were first deeply anesthetized with 1% sodium pentobarbital (50 mg/kg) administered intraperitoneally, and were then immediately euthanized by cervical dislocation. Death was confirmed by the cessation of respiration and bilateral pupillary dilation. Peri-wound tissues within ~2 mm of the wound margin were then collected. Of note, one portion of the tissue was snap-frozen in liquid nitrogen and stored at −80°C for RNA and protein analyses, and the remainder was fixed in 4% paraformaldehyde (cat. no. BL539A, Biosharp Life Sciences; at room temperature for 48 h), embedded in paraffin and processed for histological observation. The wound area was quantified using ImageJ software (version 1.8.0; National Institutes of Health), and the relative healing rate was calculated using the following formula: Relative healing rate (%)=[(day 0 wound area - day 8 wound area)/day 0 wound area] ×100%.

All data were analyzed using GraphPad Prism 9.0 software. The normality of data distribution was assessed using the Shapiro-Wilk test. Normally distributed data are presented as the mean ± standard deviation (X̅ ± S). Comparisons between two groups were performed using an independent samples t-test, and comparisons among multiple groups were performed using one-way analysis of variance (ANOVA) followed by Tukey's post hoc test for pairwise comparisons. Non-normally distributed data are expressed as median (interquartile range) [M (P25, P75)] and were compared using the Mann-Whitney U test for two groups or the Kruskal-Wallis test with Dunn's multiple comparison test for multiple groups. All experiments were independently repeated at least three times. All statistical tests were two-sided, and P<0.05 was considered to indicate a statistically significant difference.

The glycolytic intermediate, MGO, is recognized as a contributor to vascular complications in diabetes. In a previous study, 400 μmol/l MGO was employed to establish an in vitro diabetic wound healing model in human skin keratinocytes (31). In the present study, HMEC-1 cells were exposed to increasing concentrations of MGO (50, 100, 200, 400, 600, 800 and 1,000 μmol/l). CCK-8 assay revealed that MGO inhibited cell proliferation in a concentration-dependent manner (Fig. 1A). At 400 μmol/l, MGO induced a marked reduction in proliferation, while maintaining sufficient cell viability for subsequent functional assays. Thus, 400 μmol/l MGO was used in subsequent experiments. To further validate the pathophysiological relevance of the selected MGO concentration, ROS production and pro-inflammatory cytokine expression in HMEC-1 cells treated with 400 μmol/l MGO were assessed. The results revealed that intracellular ROS levels were significantly elevated in the MGO-treated group compared with the control group (Fig. 1B). Moreover, MGO treatment markedly upregulated the expression levels of IL-1β, IL-6 and IL-8 (Fig. 1C-E). These results demonstrated that 400 μmol/l MGO effectively induced oxidative stress and inflammatory responses in HMEC-1 cells, supporting its suitability for use in subsequent experiments as a relevant pathophysiological stimulus. The present study then investigated whether MGO treatment affects miR-222-3p expression. RT-qPCR revealed that the miR-222-3p levels were significantly elevated in the MGO-treated cells compared with the untreated controls (Fig. 1F). This finding indicates that MGO-induced cellular stress is accompanied by the upregulation of miR-222-3p expression.

To further explore the role of miR-222-3p in MGO-treated HMEC-1 cells, we transfected a miR-222-3p mimic to overexpress miR-222-3p. Transfection efficiency was confirmed by RT-qPCR (Fig. 2A). miR-222-3p expression was significantly higher in the MGO + miR-222-3p group than in the MGO + miR-con group, confirming successful transfection. Subsequently, cell function was then evaluated by CCK-8 assay for cell proliferation, flow cytometry for apoptosis, scratch assay for migration ability, and tube formation assay for angiogenic capacity (Fig. 2B-E). Compared with the NC group, the MGO group exhibited a significantly reduced proliferation, increased apoptosis, impaired migration and diminished tube formation, as reflected by fewer total branch points and a shorter total tube length. Furthermore, compared with the MGO + miR-con group, the MGO + miR-222-3p group exhibited further reductions in proliferation, migration and tube formation, along with a further increase in apoptosis. These results indicate that the overexpression of miR-222-3p exacerbates MGO-induced injury in microvascular endothelial cells.

Through the analysis of the ENCORI/starBase database (http://starbase.sysu.edu.cn/), a nucleotide sequence within the 3'UTR of APLN mRNA with the potential to bind miR-222-3p was identified, suggesting APLN as a downstream target of miR-222-3p. To examine this, miR-222-3p was first overexpressed in HMEC-1 cells by transfection with a miR-222-3p mimic. The results of RT-qPCR and western blot analysis revealed that the overexpression of miR-222-3p significantly reduced both the APLN mRNA and protein levels compared with the miR-con group (Fig. 3A-C). Given that MGO treatment upregulated miR-222-3p expression, the present study then determined whether MGO can similarly suppress APLN expression. Indeed, in the MGO-treated HMEC-1 cells, both the APLN mRNA and protein levels were markedly decreased relative to the NC group (Fig. 3D-F), consistent with endogenous miR-222-3p upregulation under diabetic-like stress conditions. This inverse association was further validated in clinical specimens. The expression levels of miR-222-3p and APLN were simultaneously examined in tissues from patients with DFU and tissues from individuals with non-diabetic chronic lower limb ulcer (SUC). Compared with the SUC tissues, DFU tissues exhibited a significantly elevated expression of miR-222-3p and markedly reduced mRNA and protein levels of APLN (Fig. 3G-J), demonstrating a negative association between the two molecules in human diseased tissues. A dual-luciferase reporter assay was then performed to confirm the direct interaction. Co-transfection of the miR-222-3p mimic with the wild-type APLN 3'UTR reporter plasmid resulted in a marked decrease in luciferase activity compared with the miR-con group (Fig. 3K and L), whereas no significant difference in luciferase activity was observed between the two groups when the mutant plasmid was co-transfected. Collectively, these findings confirm that APLN is a direct downstream target of miR-222-3p, and this regulatory axis is conserved from in vitro MGO-induced stress to human DFU pathology.

Having confirmed that APLN is directly targeted and downregulated by miR-222-3p, the present study then wished to examine whether the loss of APLN itself is sufficient to drive the endothelial dysfunction caused by miR-222-3p. To this end, APLN was silenced in MGO-treated HMEC-1 cells using siRNA. RT-qPCR and western blot analysis were performed to assess the APLN mRNA and protein levels (Fig. 4A-C). Compared with the MGO + si-NC group, APLN mRNA and protein expression was significantly reduced in the MGO + si-APLN group, confirming the successful knockdown of APLN. Functional assays, including CCK-8 assay, flow cytometry, scratch assay and tube formation assay, were used to evaluate endothelial cell proliferation, apoptosis, migration and tube formation (Fig. 4D-G). In comparison with the MGO + si-NC group, the MGO + si-APLN group demonstrated a decline in cell proliferation activity, an elevation in the number of apoptotic cells, a reduction in cell migration capacity, and a decrease in both the total number of branch points and the total tube length of tubule formation. These data demonstrate that the silencing of APLN mimics the detrimental effects of miR-222-3p, reinforcing APLN as a critical functional mediator downstream of miR-222-3p.

To determine whether APLN functionally mediates miR-222-3p-induced endothelial injury, rescue experiments were performed. Apelin-13, the most active subtype produced by the enzymatic cleavage of the APLN-encoded precursor protein, was used for exogenous supplementation. Previous studies have demonstrated that Apelin-13 activates the human APJ receptor with a half-maximal effective concentration (EC₅₀) of ~0.3 nM and exerts concentration-dependent pro-proliferative, anti-apoptotic and endothelial protective effects within the 1-1,000-nM range (32-35). Accordingly, in the present study, a concentration of 50 ng/ml (~32 nM) Apelin-13 was selected, which far exceeds the receptor activation threshold, is sufficient to saturate the APJ receptor, and falls well within the established bioactive window. In parallel, APLN expression was restored at the genetic level by transfecting an APLN overexpression plasmid (oe-APLN). The successful overexpression was confirmed by RT-qPCR and western blot analysis, which revealed markedly elevated APLN mRNA and protein levels in the oe-APLN group relative to the control (Fig. 5A-C).

In the MGO-treated HMEC-1 cells, the overexpression of miR-222-3p significantly reduced cell proliferation, migration and tube formation, and markedly increased apoptosis compared with the control. Notably, both exogenous Apelin-13 supplementation and APLN overexpression significantly reversed these endothelial dysfunctions, as evidenced by restored proliferation, migration and tube formation, as well as reduced apoptosis (Fig. 5D-G). The highly consistent effects of the two rescue strategies collectively indicate that APLN is a critical downstream target required for miR-222-3p-mediated impairment of microvascular endothelial function.

To further evaluate the impact of miR-222-3p on skin wound healing in vivo, both gain- and loss-of-function experiments were performed in normal and diabetic mice. miR-222-3p agomiR and its negative control (agomiR-NC) were injected locally into the peri-wound area of the dorsal skin in both diabetic and normal mice. The results revealed that, compared with the respective agomiR-NC groups, the miR-222-3p levels in the peri-wound skin were significantly elevated in the agomiR-222-3p groups of both diabetic and normal mice, while APLN expression was markedly reduced (Fig. 6A-D). Wound healing analysis demonstrated that, relative to their agomiR-NC controls, the agomiR-222-3p groups exhibited significantly delayed wound closure in both the diabetic and normal mice, indicating that miR-222-3p overexpression alone is sufficient to inhibit skin wound healing. To further determine whether this inhibitory effect is mediated by APLN downregulation, a rescue group in diabetic mice was established, in which APLN was overexpressed in the peri-wound skin via AAV vector prior to agomiR-222-3p treatment. The results revealed that APLN overexpression effectively reversed the delayed wound healing caused by agomiR-222-3p (Fig. 6E and F).

To assess whether silencing endogenous miR-222-3p promotes diabetic wound healing, DM + antagomiR-NC and DM + antagomiR-222-3p groups were established, in which miR-222-3p antagomiR was injected locally into the peri-wound area of diabetic mice. Compared with the antagomiR-NC group, the antagomiR-222-3p group exhibited significantly decreased peri-wound miR-222-3p levels, an elevated expression of APLN, and markedly accelerated wound closure, confirming the negative regulatory role of miR-222-3p in wound repair from the loss-of-function perspective (Fig. 6A-F).

To provide histopathological evidence for the underlying mechanisms, immunohistochemical staining for VEGFA and CD31 was performed on peri-wound tissues from all groups. In diabetic mice, the agomiR-222-3p group exhibited a reduced VEGFA and CD31 expression compared with the control group, whereas the antagomiR-222-3p group exhibited the opposite pattern. In normal mice, agomiR-222-3p similarly led to downregulation of VEGFA and CD31. Notably, in the rescue experiment, APLN overexpression significantly restored the VEGFA and CD31 levels, providing histopathological evidence that APLN is a key downstream effector through which miR-222-3p impairs wound angiogenesis (Fig. 7A-C). Taken together, these gain-and loss-of-function results demonstrate that miR-222-3p impairs skin wound healing in vivo by inhibiting angiogenesis.