Study on the Regulatory Mechanism of GLP-1 Receptor Agonist Liraglutide in Diabetic Corneal Epithelial Wound Healing Authors: Xin Liu, Hui Lin, Yuting Shao, Jiaqi Shen, Yanlong Bi Published: 2026 ABSTRACT PurposeThe purpose of this study was to elucidate the effects and mechanisms of liraglutide in repairing diabetic corneal epithelial injury, explore its therapeutic advantages, and provide experimental evidence for clinical application.MethodsGLP-1R expression in mouse corneal epithelium was detected by Western blotting, qPCR, and immunofluorescence. Streptozotocin-induced diabetic mice and 30 mM high-glucose (HC) cell models were established. Epithelial wound healing, viability, proliferation, migration, and apoptosis were evaluated by fluorescein staining, TUNEL, CCK-8, EdU, and scratch assay. Bioinformatics was used to screen differentially expressed genes. Reactive oxygen species (ROS), redox indicators, and key proteins in Keap1-Nrf2-ARE, PI3K/Akt, and MAPK pathways were measured. PI3K inhibitors were applied to verify the pathway function, and macrophage polarization and recruitment were further assessed.ResultsGLP-1R was mainly located in the deep basal cells of corneal epithelium. Liraglutide improved cell viability, promoted proliferation and migration, and inhibited apoptosis. It restored redox balance and alleviated oxidative stress via activating Nrf2/HO-1/NQO-1 pathway, and enhanced cell functions through PI3K/Akt pathway; these effects were abolished by PI3K inhibition. Moreover, liraglutide reduced aberrant macrophage infiltration, promoted anti-inflammatory macrophage polarization, and decreased pro-inflammatory factors, which was associated with MAPK pathway downregulation.ConclusionsLiraglutide exerts multi-mechanism protective effects on diabetic corneal epithelium. It regulates Nrf2 and PI3K/Akt pathways to attenuate oxidative stress and improve cell functions, and modulates macrophage polarization and recruitment via MAPK pathway to relieve inflammation, thereby promoting corneal epithelial injury repair. Diabetes mellitus is a category of endocrine disorders caused by insufficient insulin secretion or insulin resistance, characterized by abnormally elevated blood glucose levels. It has now become one of the most prevalent and severe chronic diseases worldwide. Projections indicate that by 2050, the number of people affected by diabetes globally may reach 1.31 billion.1 The World Health Organization further estimates that by 2030, diabetes will rise to become the seventh leading cause of death worldwide.2 The primary causes of disability and mortality in patients with diabetes stem from the various complications it induces, which include two major categories: macrovascular diseases and microvascular diseases.3 Beyond these widely documented systemic complications that affect multiple organ systems, diabetes also exerts a significant impact on ocular health, among which ocular surface disease is a clinically highly common related manifestation.4 Diabetic corneal epithelial lesion (DCEL) is a frequent diabetes-related ocular surface complication. The primary pathological features of DCEL include a prolonged regeneration cycle and delayed healing after corneal epithelial injury, which may lead to persistent epithelial defects, recurrent corneal erosion, or delayed wound healing. These conditions can cause ocular discomfort, foreign body sensation, eye pain, photophobia, and blurred vision, and may compromise visual function and quality of life in affected individuals.5 However, present therapeutic treatment for DCEL relies mostly on symptomatic support approaches, such as artificial tear lubrication and corneal protective agent covering, with no particular intervention medications available to promote epithelium repair function at the pathological mechanism level. This therapeutic status is inconsistent with the clinical needs of patients with DCEL, which is a common and troublesome complication that impairs ocular surface health and visual quality. Further investigation into novel therapeutic targets and effective intervention strategies is therefore warranted. The integrity and healing efficiency of the corneal epithelium are determined by the proliferation and migration capacity of epithelial cells, the dynamic balance of the cellular redox system, and the homeostatic regulation of the local inflammatory microenvironment.6,7 On one hand, excessive accumulation of reactive oxygen species (ROS) induced by high glucose (HG) disrupts the stability of the Keap1-Nrf2-ARE oxidative stress defense pathway, leading to intracellular lipid peroxidation and increased DNA damage, which in turn suppresses epithelial cell activity and promotes apoptotic cell death.8 On the other hand, sustained hyperglycemia inhibits the activation of critical cell survival and repair-related signaling pathways, including the PI3K/Akt pathway, thereby reducing the proliferation and migration of corneal epithelial cells and directly impairing the tissue repair process after injury.9 Furthermore, diabetic corneal injury is frequently accompanied by abnormal recruitment and imbalanced polarization of macrophages. Excessive infiltration of pro-inflammatory M1 macrophages leads to the release of inflammatory factors such as TNF-α and IL-6, further exacerbating epithelial cell damage and amplifying local inflammatory responses. Conversely, the insufficient abundance of anti-inflammatory M2 macrophages compromises effective tissue repair and active inflammation resolution. This disturbance of the local inflammatory microenvironment represents another key pathological mechanism underlying the delayed healing of DCELs.10–12 Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have emerged as important therapeutic agents in the field of metabolic disorders in recent years. Their therapeutic value extends well beyond their hypoglycemic effects. Numerous studies have shown that GLP-1RAs can attenuate myocardial remodeling in the cardiovascular system, reduce glomerulosclerosis in the kidneys, and promote repair after nerve injury in the nervous system by activating GLP-1 receptors (GLP-1Rs). Their major mechanisms of action include anti-inflammatory effects, anti-oxidative stress effects, and the regulation of cell survival signaling pathways.13,14 It is worth noting that the expression and function of GLP-1Rs in ocular surface tissues have not been fully clarified. Although several studies have suggested that GLP-1RAs may improve diabetic retinopathy,15 it remains necessary to investigate whether GLP-1RAs can promote corneal epithelial wound healing by regulating epithelial cell function and improving the local inflammatory microenvironment. The specific molecular mechanisms, especially the GLP-1R–mediated signaling pathways related to oxidative stress and inflammatory regulation, still require further exploration. Accordingly, the present study selected liraglutide, a widely used GLP-1RA, as the research agent. A series of systematic experiments were performed using a streptozotocin (STZ)-induced diabetic mouse corneal epithelial injury model and a HG-cultured corneal epithelial cell model. The present study aimed to achieve the following objectives. First, to verify the expression and localization of GLP-1R in mouse corneal epithelium and to clarify the cytological basis for the action of liraglutide. Second, to assess the effects of liraglutide on the viability, proliferation, migration, and apoptosis of diabetic corneal epithelial cells, and to clarify its direct regulatory role in the epithelial repair process. Third, to determine whether liraglutide ameliorates HG-induced oxidative stress by regulating the Keap1-Nrf2-ARE pathway and whether it restores epithelial cell function by activating the PI3K/Akt pathway. Fourth, to investigate the effects of liraglutide on macrophage recruitment and polarization in injured corneal regions, and to illustrate the molecular mechanisms by which it regulates the inflammatory microenvironment, particularly its association with the MAPK pathway. The innovations and academic significance of this study are summarized as follows. First, this study systematically demonstrates the role of GLP-1RA in DCEL wound healing, supplementing the understanding of GLP-1R function in ocular surface tissues. Second, this study reveals the multi-target molecular mechanism by which liraglutide regulates DCEL from the perspectives of epithelial cell self-repair capacity and local inflammatory microenvironment homeostasis, offering a new reference for mechanistic studies of DCEL. Third, this study provides experimental evidence supporting the potential clinical application of GLP-1RA in the management of DCEL, which may help address the current lack of targeted therapeutic drugs for DCEL and shows meaningful translational medical value. Materials and Methods Materials GLP-1R (B4ZY91), Bax (T40051), Bcl2 (T40056), Histone H3, PI3K (T40115), p-PI3K (T40116), Akt1/2/3 (T55561), p-Akt (T40067), JNK1/2/3 (T40073F), p-JNK1/2/3 (T40074F), ERK1/2 (T40071), p-ERK1/2 (T40072), p38 MAPK (T55600F), p-p38 MAPK (TP56391), and CD206 (TU313804) were from Abmart Pharmaceutical Technology Co., Ltd. (Shanghai, China). Nrf2 (16396-1-AP), Keap1 (10503-2-AP), HO-1 (10701-1-AP), NQO-1 (11451-1-AP), CD206 (18704-1-AP), and β-actin (20536-1-AP) were obtained from Proteintech (Wuhan, China). TUNEL assay kit, CCK-8 kit, and STZ were purchased from E&G Bio (Shanghai, China). The ROS Assay Kit, SOD Assay Kit, malondialdehyde (MDA) Assay Kit, and the GSH/GSSG Assay Kit were purchased from Beyotime (Shanghai, China). Liraglutide was purchased from AbMole (USA). The EdU test kit was purchased from Ya Mei Biotechnology (Shanghai, China) and Beyotime (Shanghai, China). Animals and Grouping C57BL/6 male mice aged 6 to 8 weeks were purchased from Shanghai Shengchang Biotechnology Co., Ltd. All animals were housed in an SPF facility under controlled environmental conditions, including a relative humidity of 38% to 40%, a temperature of 22°C to 23°C, and a 12-hour light/dark cycle. All animal procedures were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the guidelines of the School of Medicine, Tongji University. The experimental protocol was approved by the Animal Ethics Committee of Tongji Hospital Affiliated to Tongji University (approval number: 2021-DW-005). Mice were randomly assigned to three groups (N = 15 per group). The normal control (NC) group received topical sterile PBS administration three times daily after corneal epithelial curettage. The diabetic model (DM) group was given topical sterile PBS administration three times daily after corneal epithelial curettage. The DM + liraglutide group received topical ocular administration of liraglutide (400 µg/kg/day,16 dissolved in sterile PBS) three times daily immediately following corneal epithelial debridement. Establishment of a Mouse Model of Corneal Epithelial Mechanical Injury To establish a corneal epithelial mechanical injury model, mice received a single intraperitoneal injection of STZ solution at a dose of 120 mg/kg to induce diabetes, whereas control mice were injected with an equal volume of citrate-sodium citrate buffer. After confirmation of successful diabetic model establishment, all mice were anesthetized with 1% pentobarbital sodium, and 0.5% proparacaine eye drops were applied for topical ocular anesthesia. A 2.5‑mm diameter area of central corneal epithelium was removed unilaterally in each mouse using an Algerbrush II Corneal Rust Ring Remover. Corneal epithelial defect areas were evaluated at 0, 24, and 48 hours after injury using 0.25% sodium fluorescein staining and photographed under a slit lamp microscope. The proportion of the residual epithelial defect area was quantified using ImageJ software. Immunofluorescence Staining After euthanasia, mouse corneas were carefully isolated, embedded, and preserved at low temperature for frozen sectioning. Tissue sections were permeabilized with 0.1% Triton X‑100 and blocked at room temperature for 1 hour. Sections were then incubated with primary antibodies overnight at 4°C. On the following day, sections were rinsed with PBS 3 times for 10 minutes each. Corresponding secondary antibodies were applied and incubated for 1 hour at room temperature while protected from light. After 3 additional washes with PBS (10 minutes each), sections were stained with DAPI for 3 minutes at room temperature and rinsed again with PBS 3 times. Finally, sections were mounted with anti‑fluorescence quenching medium and observed under a fluorescence microscope. Quantitative Real-Time Polymerase Chain Reaction Total RNA was extracted from corneal tissue using TRIzol reagent. One microgram of total RNA was used for the reverse transcription reaction. Complementary DNA was synthesized using a FastKing gDNA Dismissioning RT SuperMix Kit. Relative mRNA expression levels of target genes were quantified by quantitative real-time polymerase chain reaction with SYBR Green Supermix. Data were analyzed using the 2−ΔΔCT method. The sequences of the primers used in this study are listed in the Table. Table.Primer Sequences for Quantitative Real-Time PCRGenesPrimer Sequence, 5ʹ to 3ʹ GLP-1R ReverseATCAAAGGTCCGGTTGCAGAAForwardTCAGAGACGGTGCAGAAATGG CD86 ReverseCCAGTTGGTAACAATGCCATGTForwardCAGCACGGACTTGAACAACCReverseCTCCACGAAACAGCATCTGA iNOS ForwardTGTCTGCAGCACTTGGATCAGReverseAAACTTCGGAAGGGAGCAATG TNF-α ForwardAGCCGATGGGTTGTACCTTGReverseATAGCAAATCGGCTGACGGT Arg1 ForwardTGGGTGACTCCCTGCATATCTReverseTTCCATCACCTTGCCAATCC CD206 ForwardTTCAGCTATTGGACGCGAGGReverseGAATCTGACACCAGCGGAA IL-10 ForwardTGCTAACCGACTCCTTAATGCAReverseTTCTCACCCAGGGAATTCAAA β-actin ReverseCCAGTTGGTAACAATGCCATGTForwardGGCTGTATTCCCCTCCATCG Primer Sequences for Quantitative Real-Time PCR TUNEL Staining TUNEL staining was performed to detect cell apoptosis in frozen corneal sections. Sections were allowed to equilibrate at room temperature and rinsed with PBS to remove residual optimum cutting temperature (OCT) compound. Samples were fixed in 4% paraformaldehyde for 30 minutes and incubated with 20 µg/mL Proteinase K for 10 minutes, followed by gentle rinsing with PBS. Sections were then incubated with 1 × equilibration buffer for 30 minutes. The TdT reaction mixture was prepared and added to the sections, followed by incubation at 37°C for 60 minutes in a humidified chamber away from light. After 3 washes with PBS (5 minutes each), sections were stained with DAPI for 3 minutes and rinsed again with PBS 3 times. Slides were mounted with anti-fade mounting medium and observed under a fluorescence microscope. The number of apoptotic cells was quantified using ImageJ software. Cell Culture Immortalized mouse corneal epithelial cells were obtained from Saibaikang (Shanghai) Biotechnology Co., Ltd. with the product number iCell-0093a. Cells were divided into three groups and cultured in different media. The NC group was maintained in low glucose DMEM medium containing 5.5 mM glucose, supplemented with 10% fetal bovine serum and 1% streptomycin-penicillin mixture. The HG group was cultured in HG DMEM medium containing 30 mM glucose, also supplemented with 10% fetal bovine serum and 1% streptomycin-penicillin mixture. All cells were incubated in a humidified incubator at 37°C with 5% CO2. CCK-8 Solution Cells were seeded in 96-well plates at a density of 2.0 × 104 cells/mL and incubated overnight. The following day, 10 µL of CCK-8 solution was added to each well. After incubation of the plates in an incubator for 2 hours, absorbance at 450 nm was measured with a microplate reader and cell viability was calculated. EdU Cell Proliferation Assay Cells were seeded in 12‑well plates and incubated overnight before treatment. EdU working solution was added and incubated for 2 hours. After removing the culture medium, cells were fixed with 1 mL fixing solution at room temperature for 15 minutes and washed 3 times for 5 minutes each. Cells were then treated with 1 mL permeabilization solution for 10 minutes at room temperature and washed twice for 5 minutes each. Then, 200 µL Click reaction solution was added and incubated in the dark for 30 minutes, followed by 3 washes. The 1 × Hoechst 33342 was applied and incubated in the dark for 10 minutes, and the cells were washed another 3 times. Images were captured under a fluorescence microscope and analyzed using ImageJ software. Cell Scratch Wound Healing Assay Cells were cultured in 6-well plates until they reached confluence. A vertical scratch was made across the cell monolayer using a 200 µL pipette tip. The scratched cells were gently washed with PBS to remove detached cells, and the medium was replaced with fresh low-serum medium (supplemented with 1% FBS). The cells were then returned to the incubator for continued culture. Microscopic images were captured at the corresponding time points, and the scratch healing rate was calculated using ImageJ software, following the formula: (initial scratch area-remaining scratch area)/initial scratch area × 100%. Detection of Oxidative Stress Indicators Cells were cultured in 24‑well plates and treated accordingly prior to assay. For intracellular ROS detection, cells were incubated with 10 µM DCFH‑DA in serum‑free medium at 37°C for 20 minutes, washed 3 times with PBS, and visualized under a fluorescence microscope; fluorescence intensity was quantified using ImageJ software. For MDA measurement, treated cells were lysed on ice for 20 minutes and centrifuged at 12,000g for 10 minutes, and supernatants were collected for protein quantification. A 0.37% TBA stock solution was prepared and used to freshly prepare MDA working solution; 100 µL of sample or standard was mixed with 200 µL working solution, heated at 100°C for 15 minutes, cooled, centrifuged at 1000g for 10 minutes, and measured at 532 nm. For SOD activity, cells were lysed in SOD sample buffer and centrifuged at 12,000g for 3 to 5 minutes; freshly prepared WST‑8/enzyme and reaction starter solutions were mixed with samples in 96‑well plates, incubated at 37°C for 30 minutes, and absorbance was read at 450 nm, with activity calculated and normalized to protein content. For GSH and GSSG analysis, cells were lysed with protein removal reagent M, freeze‑thawed twice, and centrifuged at 10,000g for 10 minutes; for GSSG detection, lysates were treated with GSH scavenging reagents. Serially diluted GSSG standards were used for standard curves, and detection solutions containing DTNB, glutathione reductase, and NADPH were freshly prepared. Reactions were performed at 25°C with absorbance recorded at 412 nm. Total glutathione was calculated as 2 × GSSG (without scavenging), GSH as total glutathione minus 2 × measured GSSG, and the GSH/GSSG ratio was determined accordingly. Co-training System Corneal epithelial cells were inoculated in a 6-well plate, cultured in an incubator (37°C, 5% CO2) until nearly 100% of the cells were fused, and scratch damage was made in step 2.11, and then macrophages (purchased from the Cell Bank of the Chinese Academy of Sciences, Catalog No.: SCSP-5036) were inoculated in a 0.4-µm pore size Transwell chamber, and then the chamber was placed in a 6-well plate with corneal epithelial cells scratch damage, and cultured for 24 hours under different treatment conditions. Cell Invasion Assay To perform the cell invasion assay, corneal epithelial cells were seeded in 24-well plates and cultured in an incubator at 37°C with 5% CO₂ until they reached nearly 100% confluence. Macrophages were seeded at a density of 20,000 cells per well into Transwell inserts with 8-µm pores that had been precoated with Matrigel. These inserts were then placed into the 24-well plates containing confluent corneal epithelial cells, followed by co-culture under different treatment conditions for 24 hours. After the incubation period, the inserts were removed, excess culture medium was discarded, and the cells were fixed with 4% paraformaldehyde at room temperature for 15 minutes. Following 3 washes with PBS, the cells were stained with crystal violet for 20 minutes, rinsed again with PBS, and the non-invading cells on the upper surface of the inserts were gently wiped away using a moist cotton swab. The inserts were then observed and photographed under a microscope, and the results were quantified using ImageJ software. Extraction of Cytoplasmic and Nuclear Proteins Cytoplasmic and nuclear proteins were extracted using specific extraction reagents, with 1 mM PMSF freshly added to the reagents immediately before use. Cells were first washed with pre-cooled PBS, then scraped and centrifuged to collect cell pellets. For cytoplasmic protein extraction, 200 µL of cytoplasmic extraction reagent A containing PMSF was added to 20 µL of cell pellets, followed by vortexing for 5 seconds and incubation on ice for 15 minutes. Subsequently, 10 µL of reagent B was added, vortexed for 5 seconds, chilled on ice for 1 minute, re-vortexed, and centrifuged at 12,000g at 4°C for 5 minutes; the supernatant was collected as cytoplasmic proteins. For nuclear protein extraction, the residual supernatant was removed from the cell pellets after cytoplasmic protein collection. Then, 50 µL of nuclear extraction reagent containing PMSF was added to the pellets, vortexed for 15 to 30 seconds, and incubated on ice for 30 minutes with intermittent vortexing every 1 to 2 minutes. Finally, the mixture was centrifuged at 12,000g at 4°C for 10 minutes, and the supernatant containing nuclear proteins was collected. Western Blot Total protein was extracted from target tissues or cells, and the protein concentration was quantified using the BCA method. Protein samples were subjected to SDS-PAGE electrophoresis on 10% polyacrylamide gels, followed by transfer onto PVDF membranes. The PVDF membranes were blocked with 5% skim milk overnight, then incubated with the corresponding primary antibodies at 4°C overnight. After incubation, the membranes were washed thoroughly and then incubated with secondary antibodies at room temperature for 2 hours. Finally, protein bands were visualized using a protein imaging system, and the band intensity was quantified and analyzed using ImageJ software. Bioinformatics Analysis The original sequencing data (SRP329135) was downloaded from the Sequence Read Archive (SRA) database, converted into a gene expression matrix through quality control and filtering, and differentially expressed genes were analyzed using the DESeq2 package to |log2FC| > 0.585 and adjusted P < 0.05 as the standard, the differential gene volcano map was constructed. The differential genes were enriched and analyzed by Metascape. Statistical Analysis SPSS 20.0 and GraphPad Prism 9.0 were used for statistical analysis and chart production. All data are presented as mean ± standard deviations. The two-tailed t-test was used for comparison between the two groups, 1-way ANOVA was used for comparison of differences between multiple groups, and 2-way RM ANOVA and Tukey's multiple comparison test were used for comparison of differences between groups at different time points. P < 0.05 was statistically significant. Quantitative analysis was performed by investigators blinded to the experimental groups. Results GLP-1RA Relieves Diabetic Corneal Epithelial Damage Establishment of Diabetic Mouse Model and Corneal Localization of GLP-1R To investigate the role of liraglutide in repairing diabetic corneal epithelial injury, a diabetic mouse model was first established by intraperitoneal injection of STZ. At 16 weeks after model establishment, NC mice showed normal body weight (29.45 ± 1.47g) and blood glucose levels (7.92 ± 0.66 mmol/L). In contrast, diabetic mice presented significantly decreased body weight (21.83 ± 2.54g, P < 0.0001) and markedly elevated blood glucose concentrations (28.33 ± 3.04 mmol/L, P < 0.0001; Figs. 1A, 1B). Corneal localization of GLP-1R and the effect of liraglutide on corneal epithelial regeneration and apoptosis in diabetic mice. (A, B) Body weight and blood glucose levels of normal and diabetic mice at 16 weeks after model establishment (N = 6). (C) Localization of GLP-1R in the corneal epithelium of normal mice, detected by immunofluorescence staining (N = 3). (D) The mRNA expression level of GLP-1R in mouse corneal epithelial cells and RAW 264.7 cells, measured by qPCR (N = 3). (E, F) Protein expression level of GLP-1R in mouse corneal epithelial cells and RAW 264.7 cells, determined by Western blot (N = 3). (G, J) Corneal epithelial repair at different time points after injury, assessed by sodium fluorescein staining (N = 5). (H, I) Apoptosis level in mouse corneal epithelium at 48 hours after injury, detected by TUNEL staining (N = 3, scale bar = 100 µm). CEC, mouse corneal epithelial cells. Compared with the corresponding group, ns indicates no significant statistical difference, **P < 0.01, ****P < 0.0001. To identify the expression and localization of GLP-1R in the cornea, mRNA and protein levels were measured by quantitative real-time PCR and Western blot analysis, respectively, and immunofluorescence staining was used to observe its spatial distribution. The results verified the expression of GLP-1R in mouse corneal tissue, with prominent localization in the basal cells of the deep corneal epithelium (Figs. 1C–F). With GLP‑1R‑positive RAW 264.7 macrophages as a reference,17 GLP‑1R expression in mouse corneal epithelium was further confirmed at both transcriptional and translational levels. Effects of Liraglutide on Corneal Epithelial Regeneration and Apoptosis in Diabetic Mice Fluorescein sodium staining was performed at 0 hours, 24 hours, and 48 hours after corneal epithelial debridement to evaluate corneal epithelial wound healing (Figs. 1G, 1J). At 24 hours post-injury, DM mice exhibited significantly larger corneal epithelial wound defects compared with NC mice (P < 0.01). In contrast, no significant difference in wound size was observed between diabetic mice treated with liraglutide (DM + liraglutide group) and NC mice (P > 0.05). At 48 hours post-injury, neither DM nor DM + liraglutide mice showed statistically significant differences in wound defects when compared with NC mice (P > 0.05). However, all DM mice still had residual patchy epithelial defects, with a larger average wound area than the other two groups. In contrast, two of five mice in the DM + liraglutide group achieved complete corneal epithelial healing, whereas the remaining three of five mice only had punctate defects, and their average wound area was closer to that of NC mice. To explore the effect of liraglutide on corneal epithelial apoptosis in diabetic mice, TUNEL staining was performed at 48 hours post-injury to detect apoptotic cells under different treatment conditions (Figs. 1H, 1I). Few TUNEL-positive cells were observed in the corneal epithelium of NC mice. In comparison, DM mice showed a significant increase in the number of TUNEL-positive cells compared with NC mice (P < 0.01), with positive staining detected in the epithelial, stromal, and endothelial layers of the cornea. The DM + liraglutide group exhibited a significant reduction in TUNEL-positive cells compared with the DM group (P < 0.01), and no significant difference was found between the DM + liraglutide group and NC group (P > 0.05). Notably, the reduction in TUNEL-positive cells was more pronounced in the epithelial and stromal layers. Effects of Liraglutide on Viability, Proliferation, Migration, and Apoptosis of Corneal Epithelial Cells Under High Glucose Culture To further investigate the effects and underlying mechanisms of liraglutide on the diabetic corneal epithelium, in vitro experiments were performed. For the establishment of HG-induced corneal epithelial cell injury models, glucose concentrations of 25 mM or 30 mM have been reported in previous studies.18 In the present study, CCK-8 assay was used to optimize the glucose concentration. The results showed that 30 mM glucose significantly reduced cell viability compared with 5 mM glucose (Fig. 2A), so 30 mM was selected for the subsequent HG cell models. For the optimization of liraglutide concentration, CCK-8 assay revealed that cell viability in the HG group was notably lower than that in the NC group. Treatment with liraglutide at concentrations ranging from 10 nM to 500 nM restored cell viability in a dose-dependent manner, with 200 nM liraglutide exhibiting the most significant effect (P < 0.0001). Notably, treatment with 1000 nM liraglutide did not exert adverse effects on cell viability. Therefore, 200 nM was chosen as the optimal concentration for further experiments (Fig. 2B). Effects of liraglutide on viability, proliferation, migration, and apoptosis of corneal epithelial cells cultured under high-glucose conditions. (A) Effect of different glucose concentrations on the viability of mouse corneal epithelial cells after 24 hours of culture (N = 6). (B) Influence of different liraglutide concentrations on the viability of mouse corneal epithelial cells after 24 hours of incubation (N = 3). (C, D) Proliferation levels of mouse corneal epithelial cells under different culture conditions, detected by EdU assay (N = 3). (E, F) Migration capacity of mouse corneal epithelial cells under various culture conditions, evaluated by scratch wound assay (N = 3). (G–I) Expression levels of apoptosis-related proteins in mouse corneal epithelial cells under different culture conditions, determined by Western blot (N = 3). Compared with the corresponding group, ns indicates no significant statistical difference, *P < 0.05, **P < 0.01, ***P < 0.001. Given that epithelial wound healing is closely associated with cell proliferation and migration, EdU staining was used to assess cell proliferation. Compared with the NC group, corneal epithelial cell proliferation in the HG group was significantly inhibited (P < 0.001). Liraglutide treatment partially reversed this inhibitory effect (P < 0.05 versus the HG group; Figs. 2C, 2D). A scratch wound healing assay was performed to evaluate cell migration. At 6 hours, 12 hours, and 24 hours post-scratch, cell migration in the HG group was significantly suppressed compared with the NC group, with the most prominent difference observed at 12 hours (P < 0.001). Liraglutide treatment partially restored cell migration, and the migration level at 24 hours was comparable to that in the NC group (P > 0.05 versus the NC group; Figs. 2E, 2F). To confirm the direct regulatory effect of liraglutide on corneal epithelial cell apoptosis, which was consistent with the in vivo findings, Western blot analysis was used to detect the expression of apoptosis-related proteins. In the HG group, the expression of BAX was upregulated, whereas that of Bcl2 was downregulated, leading to an increased BAX/Bcl-2 ratio (P < 0.01). Additionally, the level of cleaved caspase-3 was significantly elevated (P < 0.05). Liraglutide treatment normalized the expression of these proteins, and no significant differences were observed compared with the NC group (P > 0.05; Figs. 2G–I). Investigation of the Mechanism of GLP-1RA in Alleviating Diabetic Corneal Epithelial Damage Bioinformatics Analysis Identifies Key Dysregulated Pathways in Diabetic Corneal Epithelial Injury In vitro and in vivo experiments demonstrated that liraglutide alleviates diabetic corneal epithelial injury by enhancing cell viability, promoting cell proliferation and migration, and reducing cell apoptosis. To further explore the underlying molecular mechanisms, transcriptional profile differences in the corneal epithelium during wound repair were analyzed between normal and diabetic mice. Using the criteria of |log2FC| > 0.585 and adjusted P < 0.05, 182 upregulated genes (marked in red) and 123 downregulated genes (marked in blue) were identified (Fig. 3A). Enrichment analysis of these differentially expressed genes showed that the downregulated genes were predominantly enriched in ROS metabolic processes, tissue remodeling, and the regulation of PI3K/Akt signaling pathway (Fig. 3B). These findings suggest that the downregulation of these functional pathways plays important roles in the impairment of diabetic corneal epithelial repair. Bioinformatics analysis reveals key dysregulated pathways in diabetic corneal epithelial injury, and experiments validate the regulatory effects of liraglutide on oxidative stress-related molecules. (A) Volcano plot illustrating differentially expressed genes in the central 2.5 mm corneal epithelium at 24 hours after debridement, comparing normal mice with diabetic mice (red dots = upregulated genes; blue dots = downregulated genes; and gray dots = non-differentially expressed genes). (B) Enrichment analysis of downregulated differentially expressed genes using Metascape, displaying the top 20 enriched terms. (C) Immunofluorescence staining showing the expression level of NOX4 in the mouse corneal epithelium under different treatment conditions (N = 3). (D, E) Western blot analysis detecting the expression level of NOX4 in the mouse corneal epithelium under different treatment conditions (N = 3). (F) Fluorescence microscopy images showing the level of ROS in corneal epithelial cells under different conditions (green fluorescence indicates ROS levels, N = 3). Compared with the corresponding group, *P < 0.05, **P < 0.01. Our previous studies have confirmed the role of liraglutide in promoting tissue remodeling by enhancing cell proliferation and migration during diabetic corneal epithelial repair. Therefore, subsequent investigations will focus on the regulatory effects of liraglutide on ROS metabolism and PI3K/Akt signaling pathway in the corneal epithelium. Liraglutide Alleviates Excessive Oxidative Stress in Diabetic Mouse Corneal Epithelium Oxidative stress is a key driver of diabetes and its complications, and previous bioinformatics analysis also indicated that abnormal ROS metabolic processes play an important role in the impairment of diabetic corneal epithelial repair. NOX4, a major NADPH oxidase, has been identified as a key therapeutic target for diabetic complications in multiple studies.19–21 Given the instability of direct ROS detection in tissues, NOX4 was used as an indirect indicator of corneal ROS levels in this study. Immunofluorescence results (Fig. 3C) showed that the fluorescence intensity of NOX4 in the corneal epithelium of diabetic mice was significantly higher than that in NC mice, whereas liraglutide treatment effectively attenuated this abnormal increase. Western blot analysis (Figs. 3D, 3E) further confirmed that corneal epithelial cells cultured under HG conditions exhibited markedly elevated NOX4 expression, which was significantly downregulated by liraglutide treatment. To explore the regulatory role of liraglutide in oxidative stress of corneal epithelial cells, ROS staining and redox system indices were measured under different treatment conditions. The results (Fig. 3F, Figs. 4A–F) demonstrated that HG culture significantly increased intracellular ROS and MDA levels, whereas liraglutide treatment notably reduced excessive ROS accumulation and MDA content. Additionally, compared with the NC group, HG conditions significantly decreased superoxide dismutase (SOD) activity and the glutathione (GSH)/oxidized glutathione (GSSG) ratio, and these reductions were reversed by liraglutide treatment. These findings indicate that liraglutide mitigates oxidative stress-induced damage in diabetic corneal epithelial cells, which is closely associated with its regulation of the intracellular redox system. Effects of liraglutide on oxidative stress-related indicators and the Keap1-Nrf2-ARE pathway. (A) Quantitative fluorescence analysis of ROS staining to assess the level of intracellular reactive oxygen species. (B–F) Detection results of SOD, MDA, GSH, GSSG, and the GSH/GSSG ratio in cells from different treatment groups (N = 3). (G–K) Western blot analysis of the expression levels of proteins related to the Keap1-Nrf2-ARE pathway in mouse corneal epithelial cells under different culture conditions (N = 3). Compared with the corresponding group, ns indicates no significant statistical difference; *P < 0.05, **P < 0.01, ***P < 0.001. Subsequent investigations focused on the specific mechanism underlying this regulatory effect. Nuclear factor erythroid 2-related factor 2 (Nrf2) is a key transcription factor governing oxidative stress responses and plays an important role in inducing antioxidant defense. Moreover, the Keap1-Nrf2-ARE pathway has been shown to regulate corneal epithelial wound repair.22 Western blot analysis of proteins related to the Keap1-Nrf2-ARE pathway (Figs. 4G–K) revealed that compared with the NC group, HG culture significantly reduced nuclear Nrf2 expression, whereas liraglutide effectively reversed the HG-induced inhibition of Nrf2 nuclear translocation. Keap1 expression was significantly higher in the HG group than in the NC group, and this abnormal upregulation was suppressed by liraglutide. Furthermore, HG culture significantly decreased the expression of HO-1 and NQO-1–major functional proteins regulated by Nrf2- and liraglutide treatment restored their expression to near-normal levels. Collectively, these results demonstrate that liraglutide inhibits excessive oxidative stress by upregulating the Nrf2/HO-1/NQO-1 signaling pathway. Liraglutide Ameliorates Diabetic Mouse Corneal Epithelial Injury by Upregulating the PI3K/Akt Signaling Pathway Based on previous bioinformatics findings, we investigated whether liraglutide modulates the PI3K/Akt signaling pathway, a key pathway involved in the repair of diabetic corneal epithelium. Immunofluorescence staining results showed that the expression of phosphorylated Akt (p-Akt) in the corneal epithelium of diabetic mice was significantly reduced, whereas liraglutide intervention partially restored p-Akt expression levels (Fig. 5A). Western blot analysis was further performed to evaluate changes in the PI3K/Akt pathway under different experimental conditions (Figs. 5B–D). The results indicated that HG culture significantly downregulated the phosphorylated forms of PI3K (p-PI3K) and Akt (p-Akt) compared with the NC group, whereas liraglutide treatment effectively reversed this downregulation and restored the expression of p-PI3K and p-Akt. Effect of liraglutide on the PI3K/Akt pathway in diabetic corneal epithelial injury. (A) Immunofluorescence staining was used to detect the expression level of p-Akt in the mouse corneal epithelium under different treatment conditions (N = 3, scale bar = 50 µm). (B–D) Western blot analysis was performed to determine the expression levels of proteins involved in the PI3K/Akt signaling pathway in mouse corneal epithelial cells under different culture conditions (N = 3). (E) The effect of PI3K inhibitor on the viability of mouse corneal epithelial cells was evaluated (N = 3). (F, G) EdU assay was used to detect the effect of PI3K inhibitor on the proliferative capacity of mouse corneal epithelial cells (N = 3). (H, I) Scratch wound assay was conducted to evaluate the effect of PI3K inhibitor on the migratory capacity of mouse corneal epithelial cells (N = 3). (J–L) Western blot analysis was used to detect the expression levels of proteins related to the PI3K/Akt signaling pathway and apoptotic proteins in mouse corneal epithelial cells under different culture conditions (N = 3). Compared with the corresponding group, ns indicates no significant statistical difference; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. To preliminarily explore the role of the PI3K/Akt pathway, we used the classical PI3K inhibitor LY294002. CCK-8 assay results demonstrated that LY294002 significantly abolished the liraglutide-induced improvement in the viability of corneal epithelial cells cultured under HG conditions (Fig. 5E). EdU staining analysis showed that LY294002 reversed the liraglutide-mediated enhancement of cell proliferation in HG-cultured cells (Figs. 5F, 5G). Scratch assay results revealed that LY294002 inhibited the liraglutide-promoted migration of corneal epithelial cells exposed to HG conditions (Figs. 5H, 5I). Additionally, Western blot analysis confirmed that LY294002 not only suppressed the liraglutide-induced upregulation of p-PI3K and p-Akt but also abrogated the liraglutide-mediated downregulation of apoptosis-related proteins (Figs. 5J–L). To verify the specificity of the PI3K/Akt pathway, we further performed experiments using Alpelisib, a selective PI3Kα inhibitor, considering the potential off-target effects of LY294002. Concentration screening confirmed that 1 µM Alpelisib significantly reduced the phosphorylation of Akt without obvious cytotoxicity (Figs. 6A–D). Consistent with the results obtained using LY294002, Alpelisib reversed the effects of liraglutide on cell viability (Fig. 6E), proliferation (Figs. 6F, 6H), migration (Figs. 6G, 6I), and the expression of p-Akt, Bax, and Bcl-2 (Figs. 6J–M) under HG conditions. These results indicate that the PI3K/Akt pathway mediates the protective effects of liraglutide in corneal epithelial cells. Effects of selective PI3K inhibitor Alpelisib on liraglutide-treated corneal epithelial cells. (A) CCK8 was used to detect the effect of different concentrations of Alpelisib on corneal epithelial cell viability (N = 3). (B–D) Western blot was used to detect the effect of different concentrations of Alpelisib on PI3K/AKT pathway (N = 3). (E) CCK8 was used to detect the effect of 1 µM Alpelisib on the protective effect of liraglutide cells (N = 3). (F, H) EdU was used to detect the effect of Alpelisib on the proliferation of liraglutide (N = 3). (G, I) The effect of Alpelisib on the pro-migration effect of liraglutide was detected by scratch experiment (N = 3). (J–M): Western blot was used to detect the effect of liraglutide on BAX, Bcl2, and p-AKT after Alpelisib treatment (N = 3). Compared with the corresponding group, ns indicates no significant statistical difference; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. GLP-1RA Regulates Macrophage Recruitment and Polarization During Diabetic Corneal Epithelial Injury Liraglutide Regulates Macrophage Migration and Polarization During Diabetic Corneal Epithelial Wound Repair Immunofluorescence staining was used to detect macrophage polarization at 48 hours post corneal epithelial injury under different treatment conditions (Figs. 7A, 7B). The results showed that the macrophage-specific marker F4/80 was predominantly expressed in the corneal stroma. Compared with the NC group, the DM group exhibited a significant increase in F4/80 expression in the corneal stroma, whereas liraglutide treatment notably reduced F4/80 expression compared with the DM group. For CD206, a marker of M2 macrophages that was also mainly localized in the corneal stroma at 48 hours post-injury, its expression was lower in the DM group than in the NC group. In contrast, liraglutide intervention increased CD206 expression in the DM + liraglutide group compared with the DM group (statistical analysis of the average number of CD206-positive cells is shown in Fig. 7C). Liraglutide regulates macrophage polarization during the repair of diabetic corneal epithelial injury. (A, B) Immunofluorescence staining was used to detect macrophage recruitment and polarization at 48 hours after mouse corneal epithelial injury under different treatment conditions (N = 3). (C) Statistical analysis of the average number of CD206-positive cells (N = 3). (D, E) The qPCR analysis was performed to evaluate macrophage polarization at 48 hours after mouse corneal epithelial injury under different treatment conditions (N = 3). (F, G) Western blot analysis of CD206 protein expression in corneal tissues from the three groups (N = 3). Compared with the corresponding group, ns indicates no statistically significant difference; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The qPCR was further performed to quantify the expression levels of M1 and M2 macrophage-related markers (Figs. 7D, 7E). At 48 hours post-injury, compared with the NC group, the DM group showed significantly higher mRNA levels of M1-related factors (CD86, iNOS, and TNF-α) and significantly lower mRNA levels of M2-related factors (CD206, Arg-1, and IL-10, all P < 0.05). Liraglutide treatment significantly decreased the mRNA expression of M1-related factors (P < 0.05) and increased that of M2-related factors. Notably, the mRNA levels of M2-related factors in the liraglutide-treated group showed no significant difference from those in the NC group (P > 0.05). In addition, Western blot results also demonstrated that the protein level of CD206 was markedly decreased in the DM group compared with the NC group, and liraglutide intervention significantly upregulated the protein expression of CD206 (Figs. 7F, 7G). Liraglutide Regulates Macrophage Invasion and Polarization in a Co-Culture System of Corneal Epithelial Injury Subsequently, these phenomena were validated in vitro using an established co-culture system. First, a Transwell assay was performed to assess macrophage invasion under different treatment conditions (Figs. 8A, 8B). Following 24 hours of co-culture with injured corneal epithelial cells, macrophages exhibited varying degrees of invasive activity. Specifically, the invasive capacity of macrophages in the HG group was significantly stronger than that in the NC group, whereas the HG + liraglutide group showed a marked reduction in macrophage invasion (both P < 0.01). Morphologically, macrophages that invaded in the HG group were mostly flat and fried egg-shaped, whereas those in the NC and HG + liraglutide groups were predominantly slender and spindle-shaped. Liraglutide modulates macrophage invasion and polarization via the ERK/p38 MAPK pathway in a high-glucose corneal epithelial co-culture system. (A, B) Macrophage invasion capacity was assessed in the co-culture system under different treatment conditions. (C, D) Macrophage polarization status was detected in the co-culture system under different treatment conditions (N = 3). (E–H) Western blot analysis was conducted to determine the expression levels of MAPK signaling pathway-related proteins in macrophages within the co-culture system under different treatment conditions (N = 3). Compared with the corresponding group, ns indicates no significant statistical difference, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Furthermore, qPCR was used to detect macrophage polarization under different experimental conditions (Figs. 8C, 8D). After 24 hours of co-culture with injured corneal epithelial cells, the HG group showed significantly higher mRNA levels of M1 macrophage-related factors (CD86 and iNOS) and significantly lower mRNA levels of M2 macrophage-related factors (CD206 and Arg-1) compared with the NC group (all P < 0.001). Liraglutide treatment significantly decreased the mRNA expression of M1-related factors and increased that of M2-related factors (all P < 0.05), which was consistent with the results obtained from the in vivo experiments. Liraglutide Regulates Macrophage Polarization via the ERK and p38 MAPK Signaling Pathways To further explore whether the MAPK signaling pathway is involved in the regulation of macrophages by liraglutide during corneal epithelial wound repair, Western blot analysis was used to detect the expression of MAPK pathway-related proteins (ERK, p38 MAPK, and JNK) in macrophages under different treatment conditions in the co-culture system (Figs. 8E–H). After 24 hours of co-culture with injured corneal epithelial cells, the phosphorylation levels of ERK, p38 MAPK, and JNK proteins in macrophages of the HG group were significantly higher than those in the NC group (all P < 0.01). In contrast, liraglutide treatment (in the HG + liraglutide group) significantly reduced the phosphorylation levels of ERK and p38 MAPK proteins in macrophages (P < 0.05), whereas no significant difference was observed in the phosphorylation levels of JNK protein between the HG + liraglutide group and other groups (P > 0.05). Discussion DCEL remains a clinically intractable problem, attributed to its complex pathological mechanisms, which are characterized by impaired epithelial self-repair capacity, dysregulated oxidative stress, and a disrupted inflammatory microenvironment. This study systematically demonstrates that liraglutide, a GLP-1RA, exerts multi-targeted protective effects on DCEL. The underlying mechanisms involve the regulation of corneal epithelial cell function, modulation of the Keap1-Nrf2-ARE and PI3K/Akt signaling pathways, as well as the orchestration of macrophage polarization. These findings provide novel insights into the translational potential of GLP-1RAs for the clinical treatment of DCEL. GLP-1R Expression in Corneal Epithelium: A Functional Basis for Liraglutide-Mediated Repair The integrity and efficiency of corneal epithelial wound healing depend heavily on the proliferation, migration, and differentiation of basal cells at the wound margin.23 These cells migrate centripetally as a cohesive sheet to cover epithelial defects, followed by proliferation to restore the epithelial layers and the formation of new hemidesmosomes for stable anchorage.24 A critical prerequisite for liraglutide's regulatory effects is the presence of its specific cognate receptor, GLP-1R, in the target tissue. Our study confirmed that GLP-1R is expressed in the mouse corneal epithelium, with predominant localization in the basal cells of the deep epithelial layer. This distribution is consistent with the functional role of these basal cells as the “repair reserve” of the corneal epithelium,25 strongly suggesting that GLP-1R may directly regulate the proliferation and migration of basal cells-two core processes that are disrupted in DCEL. Consistent with this hypothesis, our in vivo and in vitro results demonstrated that liraglutide effectively reverses the HG-induced impairment of corneal epithelial cell function. In DM mice, corneal epithelial wound healing was delayed at 24 hours post-injury, with persistent patchy defects and larger residual wound areas observed at 48 hours. In contrast, liraglutide treatment not only accelerated wound healing (with 2 out of 5 mice in the DM + liraglutide group achieving complete epithelial closure at 48 hours) but also improved the quality of healing (only punctate defects were observed in the remaining mice, with the average wound area comparable to that of the NC group; see Figs. 1G, 1J). In vitro, liraglutide restored the viability, proliferation (assessed by EdU staining), and migration (evaluated by scratch assay) of corneal epithelial cells cultured under HG conditions, while simultaneously inhibiting excessive apoptosis (confirmed by TUNEL staining and Western blot detection of Bax/Bcl-2/cleaved caspase-3 expression; see Figs. 2A–I). These findings are consistent with previous studies on GLP-1RAs in other tissues: liraglutide accelerates skin wound healing in normal mice,26 promotes Schwann cell activity and myelination,27 and enhances the function of endothelial progenitor cells in patients with diabetes.28 Collectively, these data confirm that liraglutide directly targets corneal epithelial cells via GLP-1R to restore their intrinsic repair capacity, which is a key initial step in alleviating DCEL. Notably, TUNEL staining revealed that apoptotic cells in diabetic corneas were distributed across the epithelial, stromal, and endothelial layers, whereas liraglutide treatment predominantly reduced apoptosis in the epithelial and stromal layers (see Figs. 1H, 1I). This observation highlights the efficacy of topical liraglutide administration, as its therapeutic effects are most pronounced at the drug-tissue interface, supporting its clinical applicability for DCEL treatment. Aberrant apoptosis is a well-documented driver of DCEL: prior studies have linked endoplasmic reticulum stress (via caspase-12 activation)29 and imbalanced IL-1β signaling30 to increased epithelial cell apoptosis and delayed wound healing. Liraglutide's anti-apoptotic effect adds to the growing body of evidence that GLP-1RAs modulate cell survival in diabetic complications, for example, via the SIRT1/Nrf2/HO-1 pathway in corneal epithelial cells31 or by inhibiting endoplasmic reticulum stress in cardiomyocytes.32 Liraglutide Alleviates Oxidative Stress via the Keap1-Nrf2-ARE Pathway and Restores Cell Function Through PI3K/Akt Activation Bioinformatics analysis of transcriptional profiles in diabetic corneal epithelium identified two key downregulated pathways, including ROS metabolism and PI3K/Akt signaling (see Figs. 3A, 3B). These observations are consistent with the central roles of oxidative stress and impaired cell survival signaling in the pathogenesis of DCEL. At physiological levels, ROS act as essential second messengers to mediate PI3K/Akt activation and support epithelial repair.33 However, excessive ROS accumulation under HG conditions disrupts intracellular redox homeostasis, triggers lipid peroxidation and DNA damage, and ultimately leads to cell death.34 Our results verified this abnormal state, as diabetic mice and corneal epithelial cells cultured in HG medium showed elevated ROS and MDA levels, accompanied by reduced SOD activity and a decreased GSH/GSSG ratio (see Fig. 3F, Figs. 4A–F), all of which represent typical indicators of severe oxidative stress. The Keap1-Nrf2-ARE pathway constitutes the primary endogenous antioxidant defense system, and its dysregulation is recognized as a key feature of diabetic complications.35 As a core transcription factor, Nrf2 is normally sequestered in the cytoplasm by its repressor protein Keap1. Upon activation, Nrf2 translocates into the nucleus and initiates the transcription of downstream antioxidant genes including HO-1 and NQO-1.36 In the present study, HG exposure upregulated Keap1 expression and suppressed Nrf2 nuclear translocation, resulting in reduced expression of HO-1 and NQO-1 (see Figs. 4G–K). These findings are consistent with previous reports in diabetic nephropathy35 and injured corneal epithelial cells.31 Liraglutide effectively reversed this inhibitory effect by downregulating Keap1, enhancing nuclear accumulation of Nrf2, and restoring the expression of HO-1 and NQO-1, thereby re-establishing redox balance and attenuating ROS-mediated cellular damage. This protective mechanism is supported by earlier studies demonstrating that GLP-1RAs activate Nrf2 signaling to exert neuroprotective effects against ischemic stroke37 and inhibit NLRP3 inflammasome activation, an important inflammatory mediator in DCEL.38 In parallel, liraglutide activated the PI3K/Akt pathway, which serves as a critical regulator of cell survival, proliferation, and migration during corneal epithelial repair.39 HG conditions markedly reduced the phosphorylation levels of PI3K and Akt (p-PI3K and p-Akt; see Figs. 5A–D), whereas liraglutide treatment restored their expression. To confirm the functional requirement of PI3K/Akt signaling in the protective actions of liraglutide, the PI3K inhibitor LY294002 and Alpelisib were applied. Inhibition of PI3K significantly abolished the beneficial effects of liraglutide, including improved cell viability, proliferation, and migration, as well as suppressed apoptosis (see Figs. 5E–L, Fig. 6). This observation is highly relevant given the extensive crosstalk between the PI3K/Akt pathway and oxidative stress. Excessive ROS production induced by HG exposure inactivates PI3K/Akt signaling,40 whereas antioxidant treatment such as N-acetylcysteine can reactivate this pathway to facilitate corneal wound healing.41 Therefore, the dual regulatory effects of liraglutide on oxidative stress and the PI3K/Akt pathway generate synergistic protection against DCEL by targeting two inter-related pathological mechanisms. Liraglutide Modulates Macrophage Recruitment and Polarization via ERK/p38 MAPK Signaling to Improve the Inflammatory Microenvironment The inflammatory microenvironment represents a third critical determinant of DCEL healing. In normal wound repair, macrophages undergo a phenotypic switch from pro-inflammatory M1 to anti-inflammatory M2 subtypes. Specifically, M1 macrophages clear cellular debris and secrete pro-inflammatory cytokines such as TNF-α and iNOS in the early stages of healing, whereas M2 macrophages resolve inflammation and promote tissue repair through the release of factors including IL-10 and Arg-1.42,43 In diabetic conditions, this phenotypic switch is impaired, with excessive M1 macrophage infiltration and insufficient M2 polarization prolonging inflammation and delaying corneal epithelial repair.44 Our in vivo results confirmed this inflammatory imbalance. At 48 hours post-injury, diabetic corneas showed increased macrophage recruitment, as indicated by the macrophage marker F4/80, and reduced M2 polarization, marked by the M2-specific marker CD206. Concurrently, the expression of M1-related genes (CD86, iNOS, and TNF-α) was upregulated, while that of M2-related genes (CD206, Arg-1, and IL-10) was downregulated (see Figs. 7A–G). Liraglutide treatment effectively reversed these pathological changes: it reduced abnormal macrophage infiltration, promoted M2 polarization with CD206 levels comparable to those in the NC group, and normalized the expression of M1 and M2-related cytokines. These in vivo observations were validated in vitro using an established co-culture system. HG conditions enhanced macrophage invasion and promoted M1 polarization, whereas liraglutide treatment reduced macrophage invasion and shifted macrophages toward an M2 phenotypic profile. This phenotypic shift was evidenced by the slender, spindle-shaped morphology of macrophages-a characteristic feature of M2 cells45 and corresponding alterations in cytokine expression patterns (see Figs. 8A–D). Mechanistically, we identified the ERK and p38 MAPK pathways as key mediators of liraglutide's anti-inflammatory effects. MAPK signaling is aberrantly activated in diabetic complications, where it amplifies oxidative stress-induced inflammatory responses.46 Our Western blot results showed that HG conditions increased the phosphorylation levels of ERK, p38 MAPK, and JNK in macrophages, whereas liraglutide specifically reduced the phosphorylation of ERK and p38 MAPK (see Figs. 8E–H). This finding is consistent with previous studies demonstrating that GLP-1RAs inhibit MAPK signaling to attenuate inflammation in diabetic nephropathy35 and hepatic ischemia-reperfusion injury.47 This inhibition of ERK and p38 MAPK signaling likely contributes to liraglutide's regulation of macrophage polarization, as MAPK activation is required for M1 macrophage differentiation and its suppression favors M2 polarization.48 Translational Implications and Future Directions This study's findings hold important translational value. First, it identifies the GLP-1R as a novel therapeutic target in the corneal epithelium, filling the gap in our understanding of GLP-1R function on the ocular surface. Second, liraglutide exerts a multi-mechanistic effect by targeting epithelial cell function, oxidative stress, and inflammation-addressing the complex pathogenesis of DCEL and overcoming the limitations of current symptomatic treatments. Finally, the efficacy of topical liraglutide, as evidenced by its layer-specific anti-apoptotic effects, supports its potential as a noninvasive clinical intervention. Although this study highlights liraglutide's efficacy in DCEL through multi-mechanistic regulation, further research can enhance its translational relevance. First, the current work focuses on topical administration of liraglutide; future studies may explore alternative delivery systems, such as sustained-release hydrogels, to optimize drug bioavailability at the corneal surface. Second, this study evaluates liraglutide as a single agent; subsequent research could investigate its synergistic effects with standard clinical treatments, such as artificial tears, to improve therapeutic outcomes for DCEL. Third, clinical trials are essential to translate preclinical findings into clinical practice. Future efforts should validate the safety and efficacy of topical liraglutide in patients with diabetes with corneal defects and refine dosing protocols. Conclusions Liraglutide exerts a protective effect against DCEL through three interconnected mechanisms: (1) directly enhancing the proliferation, migration, and survival of corneal epithelial cells via GLP-1R; (2) alleviating oxidative stress by activating the Keap1-Nrf2-ARE pathway and restoring PI3K/Akt signaling; and (3) improving the inflammatory microenvironment by inhibiting the ERK/p38 MAPK pathway to regulate macrophage recruitment and polarization. These findings position GLP-1 receptor agonists as promising candidates for DCEL therapy and advance our understanding of metabolic drug repurposing for ocular surface diseases.