Human pericytes from the placenta (PromoCell, Heidelberg, Germany), human primary retinal microvascular endothelial cells (HRMECs; ACBRI, Kirkland, WA, USA), and human brain astrocytes (ACBRI) were maintained in M199 medium (HyClone, Logan, UT, USA) containing 20% fetal bovine serum (FBS), Dulbecco’s modified eagle medium (DMEM; Thermo Fisher Scientific, Waltham, MA, USA) containing 10% FBS, and pericyte medium containing growth factors (PromoCell), respectively. HMO6, an immortalized human microglial cell line, was obtained as described in a previous study (22) and maintained in DMEM containing 10% FBS. The cells were incubated at 37°C in a 5% CO₂ atmosphere.

Recombinant human IL-4, TNF-α, and mouse IL-4 enzyme-linked immunosorbent assay (ELISA) kits were purchased from R&D Systems (Minneapolis, MN, USA). Human IL-1β, IL-6, IL-23, TNF-α, arginase-1 (Arg-1), IL-10, and insulin-like growth factor-1 (IGF-1) ELISA kits were purchased from Thermo Fisher Scientific. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), AS1517499 (a selective signal transducer and activator of transcription 6 [STAT6] inhibitor), wortmannin (a PI3K inhibitor that suppresses Akt pathway activation), PD98059 (a MEK1 inhibitor that blocks the extracellular signal-regulated kinase 1/2 [Erk1/2] pathway), streptozotocin (STZ), Evans blue dye, bovine serum albumin (BSA), glucose, and mannitol were purchased from Millipore-Sigma (St. Louis, MO, USA). Other reagents and antibodies used were: anti-STAT6, anti-STAT5, anti-Akt, anti- Erk1/2, anti-cleaved caspase-3, anti-Bcl-2-associated X protein (Bax), anti-B-cell lymphoma 2, anti-Bcl-extra-large (Bcl-xL), anti-phosphorylated STAT6 (anti-phospho-STAT6), anti-phospho-STAT5, anti-phospho-Akt, and anti-phospho-Erk1/2 (Cell signaling Technology [CST], Danvers, MA, USA); anti-β-tubulin and peroxidase-conjugated secondary antibodies (Santa Cruz Biotechnology, Dallas, TX, USA); anti-zonula occludens-1 (ZO-1) and anti-occludin (Thermo Fisher Scientific).

Cell viability was measured using the MTT assay kit (Millipore-Sigma). Briefly, 5 × 10³ cells/well were plated in 96-well plates for 48 h and treated with IL-4 (50 ng/mL) or TNF-α (100 mg/mL) for 48 h under normal glucose (5 mmol/L glucose), high mannitol (20 mmol/L mannitol and 5 mmol/L glucose), and high glucose (25 mmol/L glucose) conditions. After treatment, 10 μL of MTT solution (5 mg/mL) was added to each well containing 100 μL of culture medium, resulting in a final MTT concentration of 0.45 mg/mL. The cells were incubated for an additional 3 h at 37°C. Subsequently, the supernatant was carefully removed, and the resulting formazan crystals were solubilized in 100 μL of dimethyl sulfoxide (DMSO). Absorbance was measured at 570 nm using a microplate reader.

The apoptotic effect was assessed using the Annexin-V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) double-staining assay, following protocols supplied with Muse AnnexinV & Dead Cell Assay kit (Millipore-Sigma). Approximately 3 × 10⁵ cells were treated with IL-4 or TNF-α for 48 h under normal glucose, high mannitol, and high glucose conditions. After incubation, cells were collected in 1 mL of medium containing 1% FBS. Subsequently, the cell suspension (100 μL) was mixed with FITC AnnexinV/PI and vortexed gently for 5 s. The mixture was then incubated for 20 min at room temperature and analyzed using the Muse cell analyzer (Millipore-Sigma). Flow cytometry data were visualized as dot plots, with untreated cells serving as the control. All experiments were performed in triplicate.

Cells were collected and lysed in radioimmunoprecipitation assay (RIPA) buffer supplemented with a protease inhibitor cocktail. Equal amounts of protein (30 μg per lane) were loaded onto 4–20% gradient SDS-polyacrylamide gels for electrophoresis and subsequently transferred to nitrocellulose membranes. The membranes were blocked for 1 h at room temperature in Tris-buffered saline containing 0.1% Tween-20 (TBST) and 5% non-fat dry milk, and then incubated overnight at 4°C with primary antibodies in the same blocking buffer (TBST with 5% non-fat dry milk). The following primary antibodies were used: anti-phospho-STAT6 (1:1000, CST; #9361), anti-STAT6 (1:1000, CST; #5397), anti-phospho-Akt (1:1000, CST; #4060), anti-Akt (1:1000, CST; #4691), anti-phospho-Erk1/2 (1:1000, CST; #9101), anti-Erk1/2 (1:1000, CST; #9102), anti-cleaved caspase-3 (1:1000, CST; #9664), anti-Bax (1:1000, CST; #2772), anti-Bcl-2 (1:1000, CST; #2870), anti-Bcl-xL (1:1000, CST; #2764), anti-ZO-1 (1:500, Thermo Fisher Scientific; #61-7300), and anti-Occludin (1:500, Thermo Fisher Scientific; #71-1500). β-tubulin (1:2000, Santa Cruz Biotechnology; #sc-5274) was used as a loading control.

After washing three times with TBST (10 min each), the membranes were incubated for 1 h at room temperature with horseradish peroxidase (HRP)-conjugated secondary antibodies (1:2000, Santa Cruz Biotechnology; #sc-2005 or #sc-2004) diluted in TBST containing 5% non-fat dry milk.

Protein bands were visualized using an HRP-based enhanced chemiluminescence substrate (Thermo Fisher Scientific) and exposed to film. Densitometric analysis of the bands was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Total RNA was extracted from cultured cells and tissues using the RNeasy Plus Mini kit (Qiagen, Hilden, Germany), following the manufacturer’s protocol. Briefly, 1 μg of total RNA was reverse-transcribed into cDNA using murine leukemia virus reverse transcriptase in the presence of 2.5 μM oligo-dT primers and 1 mM dNTPs. qRT-PCR was carried out using SYBR Green master mix (Applied Biosystems, CA, USA) on an Applied Biosystems StepOnePlus Real-Time PCR system. Each reaction was carried out in a final volume of 20 μL, containing 2 μL of synthesized cDNA, 10 μL of 2× SYBR Green master mix, and 0.4 μM of each primer (forward and reverse). The thermal cycling conditions were as follows: initial denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. A melting curve analysis was performed after amplification to confirm the specificity of the PCR products. The primer sequences used in this study are listed in Table 1 . Each sample was run in triplicate, and relative gene expression levels were calculated using the 2^–ΔΔCt method. Expression values were normalized to Actb (mouse) or ACTB (human) as internal reference genes.

The Transwell inserts (Corning, NY, USA) were inverted, and then pericytes (5 × 10⁴ cells/cm²) or HMO6 (5 × 10⁴ cells/cm²) were seeded on the bottom side of the inserts for 2 h to form an in vitro blood-retina barrier model. After the pericytes or HMO6 were attached, the inserts were inverted, and HRMECs (1 × 10⁵ cells/cm²) were seeded on the top side, followed by incubation for 24 h. Both cells were exposed to normal or high glucose conditions, with or without IL-4, TNF-α, wortmannin, or AS1517499 for 48 h. To assess endothelial permeability, 0.5% Evans blue dye was added to the upper chamber and incubated for 1 h at 37°C. The optical density of the medium collected from the bottom chamber was measured spectrophotometrically at 650 nm using a microplate reader (Tecan, Mannedorf, Switzerland).

All animal experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of Kangwon National University.

Eight-week-old male C57BL/6 mice were obtained from Central Laboratory Animal Inc. (Seoul, South Korea). Only male mice were used to avoid variability caused by the estrous cycle, which can influence inflammatory responses and vascular permeability. This approach ensured consistency in evaluating the effects of IL-4 in DR. Diabetes was induced by fasting the mice for 6 hours, followed by a single intraperitoneal injection of STZ (180 mg/kg) dissolved in 10 mmol/L citrate buffer (pH 4.5). Age-matched control mice received citrate buffer alone. Blood glucose levels were measured 72 h after injection via tail vein sampling using a glucometer. Mice with glucose levels >300 mg/dL were considered diabetic and included in subsequent experiments. Although multiple low-dose STZ injections are commonly used to model autoimmune Type 1 diabetes, in this study, a single high-dose protocol was selected to achieve rapid and consistent hyperglycemia, which is widely used in mechanistic studies of DR. Animals were monitored daily and sacrificed by CO&₂ inhalation at 3 months post-STZ injection to evaluate retinal pathophysiology during the chronic phase of hyperglycemia. Immediately after euthanasia, eyes were enucleated, and retinas were carefully isolated on ice for subsequent qRT-PCR, ELISA, and western blotting.

HMO6 microglial cells were treated with IL-4 (50 ng/mL) alone or in combination with AS1517499 (1 μM), wortmannin (1 μM), or PD98059 (25 μM) for 48 h. After treatment, culture supernatants were collected and centrifuged at 3,000 rpm for 10 min at 4°C to remove detached cells and cellular debris. The resulting supernatants were carefully transferred to fresh tubes and used as microglia-conditioned media for subsequent experiments involving pericytes or endothelial cells. No additional ultracentrifugation or extracellular vesicle isolation was performed.

Statistical analyses were performed using GraphPad Prism (version 9; GraphPad Software, San Diego, CA, USA). For comparisons involving more than two groups or multiple factors, one-way or two-way analysis of variance (ANOVA) followed by Tukey’s post-hoc tests was applied. An unpaired two-tailed Student’s t-test assuming unequal variances was used for comparisons between two groups. A P-value of < 0.05 was considered statistically significant. Quantitative data are presented as mean ± standard deviation (SD).

We first investigated the expression profiles of various interleukins in the retinas of STZ-induced diabetic mice. The level of Il4 mRNA was significantly reduced in the retinas of diabetic mice compared to those in those of control mice, whereas the mRNA levels of Il1b, Il6, and IL12b were significantly elevated ( Figure 1A ). In contrast, Il10, Il13, and Il18 expression did not show significant differences between diabetic and control retinas. Next, we performed ELISA and western blot analyses to assess IL-4 expression and the activation status of its downstream signaling molecules, such as STAT6, STAT5, Akt, and Erk1/2. Expression of IL-4 was significantly reduced ( Figure 1B ), along with a significant reduction in the phosphorylation of STAT6, Akt, and Erk1/2 in STZ-induced diabetic retinas ( Figures 1C, D ). However, the expression of phospho-STAT5 did not show significant alteration.

MTT and annexin-V/PI flow cytometric analyses revealed that IL-4 restored the high glucose- and TNF-α-induced reductions in cell viability of pericytes but did not affect the cell viability of HRMECs, astrocytes, and HMO6 ( Figure 2A ). Similarly, IL-4 reduced high glucose- and TNF-α-induced pericyte apoptosis but did not affect the apoptosis of HRMECs, astrocytes, and HMO6 ( Figure 2B ). Next, we performed western blot analysis to evaluate the effect of IL-4 on the pericyte apoptotic pathway. High glucose and TNF-α treatment increased the expression levels of cleaved caspase-3 and pro-apoptotic Bax proteins and decreased those of anti-apoptotic proteins, including Bcl-2 ( Figures 2C, D ). However, treatment with IL-4 reversed these high glucose- and TNF-α-induced changes, restoring the levels of cleaved caspase-3, Bax, and Bcl-2 in pericytes ( Figures 2C, D ).

To investigate the signaling mechanism underlying IL-4-induced pericyte survival in diabetic retinas, we focused on pathways previously reported to be activated by IL-4 and associated with cell survival, including STAT6, Erk1/2, and Akt (30–32). Under high glucose conditions, IL-4 treatment significantly increased the phosphorylation of STAT6, Akt, and Erk1/2 in pericytes ( Figures 3A, B ). TNF-α treatment alone markedly reduced Akt phosphorylation without affecting that of STAT6 or Erk1/2 ( Figures 3A, B ). Co-treatment with IL-4 and TNF-α inhibited the IL-4-induced increase in phospho-Akt, while that of STAT6 and Erk1/2 remained unaffected ( Figures 3A, B ). To further evaluate the involvement of these pathways in IL-4-mediated pericyte survival, we used specific pharmacological inhibitors–AS1517499, wortmannin, and PD98059 completely prevented IL-4-induced STAT6 phosphorylation, IL-4-induced Akt phosphorylation, and IL-4-induced Erk1/2 phosphorylation, respectively, in pericytes ( Supplementary Figures 1 - 3 ). Of these, wortmannin abolished IL-4-induced increase in pericyte survival ( Figure 3C ), decrease in cleaved caspase-3 and pro-apoptotic Bax levels, and increase in anti-apoptotic Bcl-2 levels in pericytes ( Figures 3D, E ).

Next, we evaluated the effects of IL-4 on microglial inflammatory responses under diabetic conditions by analyzing mRNA and protein levels of representative pro-inflammatory cytokines (IL-1β, IL-6, IL-23, and TNF-α) and anti-inflammatory or tissue-supportive factors (arginase-1 [Arg-1], IL-10, and IGF-1) in HMO6 microglial cells using qRT-PCR and ELISA. IL-4 treatment alone significantly decreased the expression of pro-inflammatory cytokines and increased the expression of Arg-1, IL-10, and IGF-1 under both normal glucose and high mannitol conditions, indicating an inherent immunomodulatory role independent of hyperglycemic stress ( Figures 4A, B ). Exposure to high glucose conditions induced a pro-inflammatory phenotype in HMO6 cells, characterized by elevated IL-1β, IL-6, IL-23, and TNF-α, and reduced levels of Arg-1, IL-10, and IGF-1. Co-treatment with IL-4 under high glucose conditions reversed these changes, restoring gene and protein expression levels toward those observed under normal glucose conditions ( Figures 4A, B ).

To identify the intracellular pathways responsible for IL-4-mediated immunoregulation, we assessed the expression of phosphorylated and total STAT6, Akt, and Erk1/2 using western blot analysis. IL-4 increased STAT6, Akt, and Erk1/2 phosphorylation in HMO6 cells under both normal and high glucose conditions, indicating that IL-4-driven activation of these signaling pathways is independent of extracellular glucose concentration ( Figures 5A, B ). In contrast, high glucose alone does not alter phosphorylation levels of these proteins, suggesting that the glucose-induced pro-inflammatory shift does not act through direct modulation of STAT6, Akt, or Erk1/2 activity. Instead, IL-4 appears to act as an external regulator capable of overriding glucose-induced changes in microglial function by selectively activating these signaling cascades. To further delineate the signaling pathway mediating immunomodulatory effects of IL-4, we used pharmacological inhibitors targeting each pathway: AS1517499 (STAT6 inhibitor), wortmannin (PI3K/Akt inhibitor), and PD98059 (Erk1/2 inhibitor). Each inhibitor selectively blocked IL-4-induced phosphorylation of its corresponding target ( Figure 5C ; Supplementary Figure 4A ). Only AS1517499 abolished the IL-4-mediated suppression of pro-inflammatory cytokines and enhancement of Arg-1, IL-10, and IGF-1 expression under high glucose conditions ( Figure 5D ; Supplementary Figure 5A ).

IL-4 reduced the secretion of TNF-α and IL-1β, known to induce pericyte apoptosis from HMO6 microglial cells under high glucose conditions ( Figures 4A, B ). Therefore, we evaluated whether IL-4 confers indirect protection to pericytes by modulating microglial signaling.

To assess this, pericytes were cultured with conditioned media collected from HMO6 cells treated with IL-4, alone or in combination with pathway-specific inhibitors. Conditioned media from IL-4-treated HMO6 cells significantly reduced pericyte apoptosis under high glucose conditions ( Figure 6A ). This reduction was abolished when IL-4 was co-administered with AS1517499 in HMO6 cells ( Figure 6A ). In contrast, co-treatment with wortmannin or PD98059 did not affect the IL-4-induced protective effect ( Figure 6A ). Consistently, western blot analysis showed that conditioned media from IL-4-treated HMO6 cells decreased cleaved caspase-3 and pro-apoptotic Bax levels and increased Bcl-2 levels in pericytes ( Figures 6B, C ). These effects were reversed when IL-4 was combined with AS1517499 but not with wortmannin or PD98059.

Next, we assessed changes in endothelial permeability under high glucose conditions using co-cultures of pericytes and HRMECs or HMO6 and HRMECs to evaluate the effect of IL-4 on endothelial barrier integrity. First, we investigated whether IL-4 directly influences endothelial permeability in DR. High glucose alone did not affect endothelial permeability or the expression of ZO-1 and occludin in HRMECs ( Supplementary Figures 6A–C ). However, TNF-α significantly increased endothelial permeability and reduced ZO-1 and occludin expression under both normal and high glucose conditions, suggesting a dominant, glucose-independent effect of TNF-α on barrier disruption ( Supplementary Figures 6A–C ). IL-4 treatment did not directly alter endothelial permeability or affect ZO-1 and occludin expression in HRMECs ( Supplementary Figures 6A–C ), indicating that IL-4 did not directly influence endothelial cell permeability in the diabetic retina model.

In our previous study, we confirmed that preventing pericyte apoptosis under DR conditions prevents the increase in endothelial permeability by preventing the decrease in ZO-1 expression but not occludin (5). Therefore, we speculated that IL-4 might prevent the increase in endothelial permeability by preventing the decrease in ZO-1 expression in DR when co-cultured with pericytes and HRMECs. Our findings showed that in pericyte-HRMEC co-cultures, IL-4 prevented the increase in endothelial permeability and decrease in ZO-1 expression in HRMECs under high glucose conditions and TNF-α exposure ( Figures 7A–C ). IL-4 also exerted a significant protective effect against TNF-α–induced endothelial permeability and ZO-1 reduction even under normal glucose conditions, although the effect was more pronounced under high glucose ( Figures 7A–C ). This finding suggests that the barrier-protective function of IL-4 is not strictly glucose-dependent but may be amplified in a diabetic-like environment. However, IL-4 did not affect occludin expression levels in these co-cultures ( Figures 7B, C ). Furthermore, the addition of wortmannin to the pericyte-HRMEC co-cultures restored the effect of IL-4 on endothelial permeability and ZO-1 expression in HRMECs ( Supplementary Figures 7A–C ).

Next, we explored whether the regulatory effects of IL-4 on microglial inflammatory states could influence endothelial permeability. Under high glucose, microglia upregulated TNF-α, which disrupted endothelial barrier function by downregulating ZO-1 and occludin ( Figures 4A, B ; Supplementary Figures 6A–C ). Co-culture of HRMECs with IL-4-treated HMO6 cells under hyperglycemic conditions preserved both ZO-1 and occludin expression and suppressed the increase in endothelial permeability ( Figures 7D–F ). However, inhibition of the STAT6 pathway using AS1517499 abrogated the protective effects of IL-4, indicating that the immunoregulatory function of IL-4 is STAT6-dependent in microglia ( Supplementary Figures 8A–C ). Furthermore, IL-4 significantly reduced TNF-α expression in HMO6-HRMEC co-cultures under high glucose, while STAT6 inhibition reversed this effect ( Supplementary Figure 8D ).