p38α MAPK-Mediated Redox Regulation of Transglutaminase 2 Drives Microvascular Leakage in Diabetic Retinas

Abstract

Microvascular leakage is an early hallmark of diabetic retinopathy (DR), but the redox-dependent mechanisms underlying this dysfunction remain unclear. Here, we investigated whether p38α mitogen-activated protein kinase (MAPK) activates transglutaminase 2 (TGase2) through reactive oxygen species (ROS) generation, thereby promoting hyperglycemia-induced vascular permeability in diabetic retinas. In human retinal endothelial cells (HRECs), vascular endothelial growth factor (VEGF), which is elevated under hyperglycemic conditions, activated both p38α MAPK and TGase2. VEGF-induced TGase2 activation was inhibited by the p38 MAPK inhibitor SB203580 or by p38α MAPK siRNA. Similarly, VEGF-stimulated TGase2 activity in non-diabetic mouse retinas was blocked by knockdown of either p38α MAPK or TGase2. In diabetic retinas, hyperglycemia-increased ROS production and TGase2 activity were reduced by SB203580 or p38α MAPK siRNA, but not by the TGase inhibitor cystamine, indicating upstream ROS-dependent regulation. The antioxidant Trolox also suppressed TGase2 activation in VEGF-treated HRECs and diabetic retinas. Functionally, knockdown of p38α MAPK or TGase2 preserved vascular endothelial (VE)-cadherin integrity and attenuated cytoskeletal remodeling in HRECs and diabetic retinas, resulting in reduced microvascular leakage. These findings identify a redox-dependent p38α MAPK–TGase2 axis as a key mediator of retinal vascular permeability in DR and highlight this pathway as a potential therapeutic target for maintaining vascular integrity.

1. Introduction

Diabetic retinopathy (DR) is the most common microvascular complication of chronic hyperglycemia and remains a leading cause of blindness among working-age adults [1,2]. DR progresses from early non-proliferative to advanced proliferative stages [3,4]. Non-proliferative DR is characterized by pericyte loss, acellular capillary formation, microaneurysms, and microvascular leakage [5,6], whereas proliferative DR is marked by pathological neovascularization and diabetic macular edema, ultimately leading to vision loss through vitreous hemorrhage and retinal detachment [1,7]. A clinical hallmark of DR is microvascular leakage, resulting from vascular endothelial growth factor (VEGF)-mediated disassembly of adherens junctions and disruption of the blood-retinal barrier, leading to macular edema and vision loss [7,8,9]. Therefore, preventing microvascular leakage at early stages is critical for halting DR progression; however, the underlying molecular mechanisms remain incompletely understood.

Transglutaminase 2 (TGase2) is a member of the transglutaminase family that catalyzes Ca²⁺-dependent protein cross-linking through the transamidation of glutamine and lysine residues [1,10]. TGase2 is ubiquitously expressed and functions as a multifunctional enzyme, acting as a transamidase, serine/threonine kinase, protein disulfide isomerase, and GTPase [11,12]. It has been implicated in the pathogenesis of diverse diseases, including celiac disease, neurodegenerative disorders, cancers, inflammatory diseases, and fibrosis [1,10,13,14]. Increasing evidence also identifies TGase2 as a key mediator of diabetic vascular complications, including DR, nephropathy, pulmonary disease, and cardiovascular disease [1,7,8,15].

In the diabetic retina, microvascular leakage is primarily driven by hyperglycemia-induced VEGF upregulation [1,8,16]. Elevated VEGF activates TGase2 via reactive oxygen species (ROS) generation, which sequentially inhibits AMP-activated protein kinase (AMPK) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), ultimately leading to vascular leakage via stress fiber formation and vascular endothelial (VE)-cadherin disruption [7,8]. Despite these observations, the molecular mechanism by which hyperglycemia activates TGase2 in the retina remains poorly understood.

p38 mitogen-activated protein kinase (MAPK), a subfamily of the MAPK family, is a serine/threonine protein kinase activated by various environmental and intracellular stresses, including inflammatory cytokines, infection, and oxidative stress [17]. p38 MAPK phosphorylates a broad range of substrates, including protein kinases (e.g., MAPK-activated protein kinase and GSK3β), transcription factors, cell-cycle regulators, and cell-death mediators [18,19,20]. Owing to its broad substrate specificity, p38 MAPK plays a central role in numerous pathophysiological processes, including inflammation, cancer, neurodegeneration, cardiovascular dysfunction, and metabolic diseases such as non-alcoholic fatty liver disease [17,21,22,23,24].

p38 MAPK is also involved in the pathogenesis of diabetic complications, including cardiomyopathy, nephropathy, and retinopathy [24,25,26]. In diabetic cardiomyopathy, hyperglycemia induces p38 MAPK phosphorylation, whereas pharmacological inhibition or genetic deletion of p38 MAPK improves cardiac function in experimental models [3]. Inhibition of p38 MAPK attenuates left ventricular dysfunction by suppressing hyperglycemia-induced pro-inflammatory cytokine expression in the diabetic heart [27], and deletion of the predominant isoform p38α prevents apoptosis and pathological remodeling in the diabetic myocardium [28]. Similarly, in the kidney, hyperglycemia activates p38 MAPK [26], and its inhibition reduces high glucose-induced apoptosis and inflammation in human renal proximal tubular epithelial cells [29]. In the diabetic retina, hyperglycemia activates p38 MAPK, and its inhibition attenuates early DR pathology, including acellular capillary formation and pericyte loss [24,30,31]. Despite these findings, the specific mechanisms by which p38 MAPK contributes to diabetes-induced microvascular dysfunction remain poorly defined, particularly with respect to microvascular leakage.

In this study, we sought to elucidate the molecular mechanism by which hyperglycemia activates TGase2 in the diabetic retina. We hypothesized that p38α-mediated regulation of TGase2 contributes to hyperglycemia-induced microvascular leakage in DR. Our findings demonstrate that p38α is required for hyperglycemia-induced TGase2 activation through ROS generation in diabetic retinas, leading to vascular leakage via stress fiber formation and VE-cadherin disassembly. Collectively, these results identify the p38α–TGase2 signaling axis as a critical redox mediator of hyperglycemia-induced microvascular leakage in DR.

2. Materials and Methods

2.1. Cell Culture

Human retinal endothelial cells (HRECs) were obtained from Cell Systems (Kirkland, WA, USA) and cultured as previously described [1]. Cells between passages 8 and 12 were used and maintained on 2% gelatin-coated plates in M199 medium supplemented with 20% fetal bovine serum (FBS), 3 ng/mL basic fibroblast growth factor, 100 U/mL penicillin, and 100 µg/mL streptomycin (Gibco, Grand Island, NY, USA) in a humidified incubator at 37 °C with 5% CO₂. For experiments, cells were incubated for 6 h in low-serum medium containing 1% FBS and antibiotics.

2.2. Measurement of ROS Generation and In Situ TGase Activity in HRECs

ROS generation was measured using CellROX™ Green and dihydroethidium (DHE; Thermo Fisher Scientific, Waltham, MA, USA), as previously described [15]. Briefly, labeled cells on coverslips were mounted in a perfusion chamber and imaged by confocal microscopy (K1-fluo; Nanoscope Systems, Daejeon, Republic of Korea). ROS levels were quantified at the single-cell level from 30 randomly selected cells across four microscopic fields per experiment (n = 4).

In situ TGase activity was measured by confocal microscopy [8]. Cells were incubated with 1 mmol/L 5-(biotinamido)pentylamine (BAPA; Sigma-Aldrich, St. Louis, MO, USA) for 1 h, fixed with 3.7% formaldehyde for 30 min, and permeabilized with 0.2% Triton X-100 for 30 min. After incubation with fluorescein isothiocyanate (FITC)-conjugated streptavidin (1:200; MilliporeSigma, Burlington, MA, USA) for 1 h, cells were imaged by confocal microscopy. TGase activity was quantified by measuring fluorescence intensity in 30 randomly selected cells in four microscopic fields per experiment (n = 4).

2.3. Transfection with siRNA in HRECs

HRECs were transfected with 100 nmol/L scrambled control, human p38α MAPK (p38α)-specific, or human TGase2-specific siRNA (Santa Cruz Biotechnology, Dallas, TX, USA) using siLentFect™ lipid reagent (Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions [8]. Cells were incubated with siRNA–lipid complexes for 24 h, followed by replacement with fresh culture medium. Transfected cells were cultured for an additional 24 h before experiments.

2.4. Western Blot Analysis

Protein extracts from HRECs and mouse retinas were prepared using lysis buffer as previously described [7], separated by SDS-PAGE, and transferred onto polyvinylidene fluoride membranes. Membranes were incubated with monoclonal antibodies against p38, phosphorylated p38, β-actin, and transglutaminase 2 (TGase2) (1:2000; Cell Signaling Technology, Danvers, MA, USA), followed by horseradish peroxidase-conjugated secondary antibodies. Protein bands were visualized using a ChemiDoc imaging system (Bio-Rad Laboratories) [5].

2.5. Endothelial Monolayer Permeability Assay in HRECs

In vitro permeability assay was performed as previously described [8]. HRECs were seeded onto gelatin-coated Transwell inserts (0.4 µm pore size; Costar, Corning, NY, USA) and cultured for 6 days to establish confluent monolayers. Cells were then incubated with 1 mg/mL 40-kDa FITC-dextran (MilliporeSigma) for 60 min, and fluorescence in the lower chamber was measured using a microplate spectrofluorometer (Molecular Devices, Sunnyvale, CA, USA).

2.6. Visualization of Actin Filaments and VE-Cadherin in HRECs

Actin filaments and VE-cadherin were stained as previously described [8]. For actin filament staining, cells were fixed with 3.7% formaldehyde, permeabilized with 0.2% Triton X-100, and incubated with Alexa Fluor 488–phalloidin (1:200; Molecular Probes, Eugene, OR, USA) for 1 h. For VE-cadherin staining, cells were fixed, permeabilized, and incubated overnight at 4 °C with an anti-VE-cadherin monoclonal antibody (1:200; Santa Cruz Biotechnology), followed by FITC-conjugated goat anti-mouse secondary antibody (1:200; Sigma-Aldrich). Samples were imaged by confocal microscopy.

2.7. Generation of Diabetic Mice and Intravitreal Injection

Six-week-old male C57BL/6 mice were obtained from DBL (Eumseong, Republic of Korea) and housed in groups of six per filtered-top cage under specific pathogen-free conditions in a temperature-controlled facility (22 °C) with a 12 h light/dark cycle. Mice had free access to food and water and were allowed to acclimatize to the housing conditions for at least 4 days before experimental procedures. After acclimatization, mice were randomly assigned to normal or diabetic groups.

Diabetes was induced by a single intraperitoneal injection of streptozotocin (150 mg/kg body weight; MilliporeSigma) freshly prepared in 100 mmol/L citrate buffer (pH 4.5) [7]. Mice with fasting blood glucose concentrations ≥ 19 mmol/L, polyuria, and glucosuria were considered diabetic. Body weights and blood glucose levels were monitored weekly.

To evaluate the effects of pharmacological inhibitors or gene silencing on diabetes-induced molecular events, five weeks after streptozotocin injection, diabetic mice were anesthetized with 3% isoflurane and intravitreally injected with 2 μL of PBS, 500 μmol/L SB203580, 2 μmol/L Trolox, or 50 mmol/L cystamine (Sigma-Aldrich) for 1 day (n = 10 mice per group), or with 2 μL of PBS or 65 μmol/L control, TGase2, or p38α siRNA for 2 days (n = 4 mice per group; Supplementary Figure S1).

To assess the effects of gene silencing on VEGF-induced p38 expression and phosphorylation and TGase activity in non-diabetic mice, age-matched normal mice were anesthetized with 3% isoflurane and intravitreally injected with 2 μL of PBS or 65 μmol/L control, TGase2, or p38α siRNA for 1 day, followed by intravitreal injection of 2 μL of 50 μg/mL VEGF for 1 day (n = 3 mice per group). Sample sizes were determined based on previous studies employing similar experimental designs [5,7].

All procedures conformed to the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Ethics Committee of Kangwon National University (approval no. KW-241218-2).

2.8. Measurement of In Vivo TGase Activity in Mouse Retinas

In vivo TGase transamidating activity in whole-mount retinas was determined by confocal microscopy [5]. Mice were deeply anesthetized, and 48 µL of 100 mmol/L BAPA was injected into the left ventricle. Retinas were dissected in the Maltese cross configuration, permeabilized with 0.2% Triton X-100 for 30 min, incubated with Alexa 647–isolectin B4 (1:500; Thermo Fisher Scientific) to visualize blood vessels, and probed with FITC-conjugated streptavidin (1:200, v/v) for 1 h.

In vivo TGase activity in retinal sections was quantified by confocal microscopy as described [7]. Retinal cryosections (10 µm) were incubated with 1 mmol/L BAPA for 1 h, fixed with 3.7% formaldehyde for 20 min, and permeabilized with 0.2% Triton X-100. Sections were sequentially incubated with FITC-conjugated streptavidin (1:200, v/v) for 1 h, Alexa 647–isolectin B4 (1:500) for 2 h, and 1 µg/mL 4′,6-diamidino-2-phenylindole (DAPI) (MilliporeSigma) for 10 min. Stained samples were observed by confocal microscopy, and TGase activity was quantified by measuring fluorescence intensity (n = 6 eyes).

2.9. Measurement of VEGF Levels by ELISA in Mouse Retinas

VEGF levels were determined using a commercial ELISA kit (MMV00; R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s instructions [5]. Retinal lysates, prepared in lysis buffer [7], were centrifuged at 17,600× g for 15 min at 4 °C, and VEGF concentrations were determined from the supernatants using a microplate spectrophotometer (Epoch; BioTek, Winooski, VT, USA).

2.10. Immunofluorescence in Mouse Retinal Sections

Expression of p38, phosphorylated p38, and VEGF was visualized by immunofluorescence [5]. Retinal cryosections were fixed with 4% paraformaldehyde for 30 min, and permeabilized with 0.2% Triton X-100 for 20 min. Sections were incubated overnight at 4 °C with polyclonal antibodies against p38 and phosphorylated p38 (1:200; Cell Signaling Technology) or a monoclonal antibody against VEGF (1:200; Abcam, Cambridge, UK). After washing, the sections were probed with Alexa 546–conjugated goat anti-rabbit IgG (1:200; Invitrogen, Carlsbad, CA, USA) for 2 h and counterstained with 1 µg/mL DAPI for 10 min. Stained samples were visualized by confocal microscopy, and expression levels were quantified by measuring fluorescence intensities (n = 6 eyes).

2.11. Measurement of ROS Generation in Mouse Retinas

Retinal cryosections were incubated with 5 µmol/L DHE for 30 min at 37 °C. Stained sections were examined by confocal microscopy, and ROS levels were quantified by measuring fluorescence intensities (n = 6 eyes) [7].

2.12. Visualization of Actin Filaments and VE-Cadherin in Whole-Mount Mouse Retinas

Actin filaments and VE-cadherin were visualized in whole-mount retinas [5]. For actin filament staining, enucleated eyeballs were fixed with 4% paraformaldehyde for 45 min, and retinas were dissected in the Maltese cross configuration. Retinas were permeabilized with 1.0% Triton X-100 for 1 h and incubated with Alexa Fluor 488 phalloidin (1:200) for 2 h. For VE-cadherin, eyeballs were sequentially fixed with 4% paraformaldehyde for 45 min and acetone for 3 min at −20 °C, and retinas were dissected in the Maltese cross configuration. After permeabilization with 1.0% Triton X-100 for 4 h, retinas were incubated overnight at 4 °C with a monoclonal anti–VE-cadherin antibody (1:200; BD Pharmingen, San Diego, CA, USA) and probed with Alexa Fluor 647–conjugated goat anti-rat IgG (1:300; Invitrogen). Actin filaments and VE-cadherin in the superficial vascular plexus of whole-mount retinas were visualized by confocal microscopy.

2.13. Measurement of Vascular Leakage in Mouse Retinas

Vascular leakage was assessed by fundus fluorescein angiography and FITC-dextran angiography. For fundus fluorescein angiography, mice were anesthetized with 3% isoflurane, and pupils were dilated using a small drop of Mydrin-P (Santen Pharmaceutical, Osaka, Japan). Mice were intraperitoneally injected with fluorescein sodium (100 μL/20 g body weight; 20 mg/mL in PBS; Alcon, Fort Worth, TX, USA). After applying eye gel (Samil Pharmaceutical, Seoul, Republic of Korea), fundus fluorescein angiography images were acquired using the Phoenix MICRON IV system (Phoenix MICRON, Bend, OR, USA), and vascular leakage was quantified by measuring fluorescence intensities (n = 6 eyes).

For FITC-dextran angiography, anesthetized mice were injected with 500-kDa FITC–dextran (375 mg/kg; MilliporeSigma) into the left ventricle and allowed to circulate for 5 min [5]. Eyes were enucleated, fixed with 4% paraformaldehyde for 2 h, and retinas were dissected in the Maltese cross configuration. Vascular leakage in the superficial layer of whole-mount retinas was visualized by confocal microscopy and quantified by measuring fluorescence intensities of extravasated FITC–dextran in four microscopic fields per retina (n = 6 eyes).

2.14. Statistical Analysis

Data were analyzed using OriginPro 2015 software (OriginLab, Northampton, MA, USA) and are expressed as the mean ± standard deviation of three, four, or six independent experiments. Statistical significance was determined using one-way analysis of variance (ANOVA) with Holm–Sidak’s multiple comparisons test. A p value < 0.05 was considered statistically significant.

3. Results

3.1. p38 MAPK and TGase Contribute to Hyperglycemia-Induced Microvascular Leakage in Diabetic Retinas

To investigate the roles of p38 MAPK and TGase in hyperglycemia-induced microvascular leakage, we intravitreally injected the p38 inhibitor SB203580 or the TGase inhibitor cystamine into the eyes of diabetic mice and assessed retinal vascular permeability by FITC–dextran angiography and fundus fluorescein angiography (Figure 1A). Compared with non-diabetic controls, mice with five weeks of diabetes exhibited increased food and water intake, body weight loss, and severe hyperglycemia (Figure 1B,C). Hyperglycemia markedly induced retinal microvascular leakage, which was significantly reduced by treatment with either SB203580 and cystamine (Figure 1D).

Figure 1. p38 MAPK and TGase contribute to hyperglycemia-induced vascular leakage in HRECs and diabetic mouse retinas. (A–F) Diabetic C57BL/6 mice were intravitreally injected with 2 μL of 500 μmol/L SB203580 (SB) or 50 mmol/L cystamine (Cys) for 24 h. Transglutaminase (TGase) activity, vascular leakage, and vascular endothelial growth factor (VEGF) expression were assessed in the retinas. (A) Schematic illustration of diabetic mouse generation and in vivo experimental design. (B,C) Body weight (B) and blood glucose levels (C) were monitored weekly (n = 6). (D) Vascular leakage visualized by fluorescein Isothiocyanate (FITC)-dextran angiography (scale bar, 100 μm) and fundus fluorescein angiography (scale bar, 300 μm). (E) VEGF expression (green) in retinal sections visualized by immunofluorescence with 4′,6-diamidino-2-phenylindole (DAPI) nuclear counterstain (blue). Scale bar, 100 μm. (F) Quantification of VEGF expression by ELISA (n = 6). (G) In vitro endothelial monolayer permeability assay using 40-kDa FITC-dextran in human retinal endothelial cells (HRECs) (n = 4). HRECs were incubated with 20 μmol/L SB or 50 μmol/L Cys and then treated with 10 ng/mL VEGF for 90 min. Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; NS, non-significant; *** p < 0.001.

The involvement of p38 and TGase in vascular leakage was further examined using an endothelial monolayer permeability assay in HRECs treated with VEGF, a key mediator of retinal vascular leakage [8]. VEGF expression was significantly higher in diabetic retinas than in non-diabetic controls (Figure 1E,F). VEGF treatment significantly increased monolayer permeability in HRECs, whereas this effect was suppressed by SB203580 and cystamine (Figure 1G). Together, these findings indicate that both p38 and TGase contribute to hyperglycemia-induced microvascular leakage in diabetic retinas.

3.2. p38α MAPK Mediates VEGF-Induced TGase2 Activation in HRECs and Mouse Retinas

To explore the role of p38α in VEGF-induced TGase2 activation, we first examined the effects of SB203580 and cystamine on VEGF-induced p38 phosphorylation and TGase activity in HRECs. VEGF stimulation induced time-dependent p38 phosphorylation, reaching a maximum at 10 min (Figure 2A,B). This phosphorylation was inhibited by SB203580, but not by cystamine (Figure 2C,D), indicating that TGase is not required for VEGF-induced p38 activation. Immunofluorescence analysis yielded consistent results.

Figure 2. p38α MAPK mediates VEGF-induced TGase2 activation in HRECs. (A–D) HRECs were incubated with 10 ng/mL vascular endothelial growth factor (VEGF) for the indicated times or 10 min in the presence of 20 μmol/L SB203580 (SB) or 50 mmol/L cystamine (Cys). (A,B) Time course of VEGF-induced p38 MAPK (p38) phosphorylation. (A) Representative Western blot. (B) Densitometric quantification of phosphorylated p38 (p-p38) normalized to total p38 (n = 4). (C,D) Effects of SB and Cys on VEGF-induced p38 phosphorylation. (C) Representative Western blot. (D) Densitometric quantification of p-p38 (n = 4). (E,F) Effects of SB or Cys on VEGF-induced transglutaminase (TGase) activation. Human retinal endothelial cells (HRECs) were treated with 10 ng/mL VEGF for 120 min in the presence of 20 μmol/L SB or 50 μmol/L Cys. (E) In situ TGase activity visualized by confocal microscopy. Scale bar, 20 μm. (F) Quantification of TGase activity by fluorescence intensity (n = 4). (G–I) Inhibition of TGase activity by TGase2 siRNA. HRECs were transfected with 100 nmol/L control (Ctrl) or TGase2 siRNA and treated with 10 ng/mL VEGF for 120 min. (G) Western blot of TGase 2 expression. (H) Densitometric quantification of TGase 2 expression normalized to β-actin (n = 3). (I) In situ TGase activity quantified by fluorescence intensity (n = 4). (J–L) Inhibition of TGase activity by p38α MAPK siRNA. HRECs were transfected with 100 nmol/L Ctrl or p38 siRNA, and treated with 10 ng/mL VEGF for 120 min. (J) Western blot of p38 expression. (K) Densitometric quantification of p38 expression normalized to β-actin (n = 3). (L) Quantification of in situ TGase activity by fluorescence intensity (n = 4). Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. NS, non-significant; *** p < 0.001.

VEGF stimulation also significantly increased TGase activity, which was suppressed by both SB203580 and cystamine (Figure 2E,F), suggesting that p38 MAPK mediates VEGF-induced TGase activation. Neither SB203580 nor cystamine effected basal TGase activity. To confirm that TGase2 is the predominant isoform responsible, HRECs were transfected with TGase2-specific siRNA. Knockdown of TGase2 expression (Figure 2G,H) markedly inhibited VEGF-induced TGase activity, whereas scrambled siRNA had no effect (Figure 2I), confirming that TGase2 is the main contributor to VEGF-induced TGase activity.

We next examined the role of p38α using p38α-specific siRNA. Transfection with p38α siRNA effectively reduced p38 expression (Figure 2J,K), confirming that p38α is the predominant isoform expressed in HRECs. Knockdown of p38α significantly attenuated VEGF-induced TGase activity, while control siRNA had no effect (Figure 2L). These findings demonstrate that p38α is required for VEGF-induced TGase2 activation in HRECs.

To extend these findings in vivo, we evaluated the effects of intravitreal injection of p38α- and TGase2-specific siRNAs on VEGF-induced p38 phosphorylation and TGase2 activation in non-diabetic mouse retinas. p38α siRNA effectively reduced p38 expression, whereas TGase2 or control siRNAs did not (Figure 3A,B). Consistent with in vitro results, VEGF-induced p38 phosphorylation was inhibited by p38α siRNA but not by TGase2 siRNA (Figure 3C,D). VEGF also increased TGase2 activity, which was suppressed by both p38α and TGase2 siRNAs (Figure 3E,F). Collectively, these data demonstrate that p38α MAPK is essential for VEGF-induced TGase2 activation in both HRECs and mouse retinas, whereas TGase2 does not contribute to VEGF-induced p38 phosphorylation.

Figure 3. p38α MAPK siRNA inhibits VEGF-induced TGase2 activation in mouse retinas. Non-diabetic C57BL/6 mice were intravitreally injected with 2 μL of 65 μmol/L control (Ctrl), p38α MAPK (p38α), or TGase2 siRNA for 24 h, followed by 2 μL of 50 μg/mL vascular endothelial growth factor (VEGF). p38 expression, phosphorylated p38 (p-p38), and in vivo transglutaminase (TGase) activity were analyzed by immunofluorescence in retinal sections. (A,B) Suppression of p38 expression by p38α siRNA. (A) Representative fluorescence images of p38 expression. Scale bar, 100 μm. (B) Quantification of p38 expression by fluorescence intensity (n = 6). (C,D) Inhibition of p38 phosphorylation by p38α siRNA. (C) Visualization of p-p38 (red) with 4′,6-diamidino-2-phenylindole (DAPI) nuclear counterstain (blue). Scale bar, 100 μm. (D) Quantification of p-p38 by fluorescence intensity (n = 6). (E,F) Inhibition of TGase activity by p38α or TGase2 siRNA. (E) In vivo TGase activity (green) and blood vessels (red, isolectin B4) visualized by confocal microscopy with DAPI nuclear counterstain (blue). Scale bar 100 μm. (F) Quantification of TGase activity by fluorescence intensity (n = 6). Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; NS, non-significant; *** p < 0.001.

3.3. p38α MAPK Mediates Hyperglycemia-Induced TGase2 Activation in Diabetic Retinas

To determine the role of p38α in TGase2 activation under hyperglycemic conditions, we examined the effects of intravitreal injection of SB203580 and cystamine on p38 phosphorylation and TGase2 activity in diabetic retinas. Western blot analysis revealed that hyperglycemia increased p38 phosphorylation, which was inhibited by SB203580, but not by cystamine (Figure 4A,B). Hyperglycemia also enhanced TGase2 activity in whole-mount diabetic retinas, and this effect was suppressed by both SB203580 and cystamine (Figure 4C). Consistently, both inhibitors markedly attenuated hyperglycemia-induced TGase2 activity in retinal sections (Figure 4D,E).

Figure 4. Role of p38α MAPK in hyperglycemia-induced TGase2 activation in diabetic mouse retinas. (A–E) Diabetic C57BL/6 mice were intravitreally injected with 2 μL of 500 μmol/L SB203580 (SB) or 50 mM cystamine (Cys), and p38 phosphorylation and transglutaminase (TGase) activity were analyzed in the retinas. (A,B) Effects of SB or Cys on hyperglycemia-induced p38 phosphorylation analyzed by Western blot (A) and quantified by densitometry (n = 4) (B). (C–E) Effects of SB or Cys on hyperglycemia-induced TGase2 activation in whole-mount retinas or retinal sections. (C) In vivo TGase activity (green) and blood vessels (red, isolectin B4) in whole-mount retinas visualized by confocal microscopy. Scale bar, 100 μm. (D,E) In vivo TGase activity in retinal sections was visualized by confocal microscopy (D) and quantified by fluorescence intensity (n = 6) (E). Scale bar, 100 μm. (F–I) Diabetic mice were intravitreally injected with 2 μL of 65 μmol/L Ctrl or p38α siRNA, and then p38 expression, p-p38, and TGase activity were analyzed in retinal sections. (F) Visualization of p38 and p-p38 (red) by immunofluorescence with 4′,6-diamidino-2-phenylindole (DAPI) nuclear counterstain (blue). Scale bar, 100 μm. (G,H) Quantification of p38 (G) and p-p38 (H) by fluorescence intensity (n = 6). (I) Quantification of in vivo TGase activity (n = 6). Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. NS, non-significant; ** p < 0.01; *** p < 0.001.

These findings were further confirmed using intravitreal injection of p38α-specific siRNA. p38α siRNA significantly reduced p38 expression (Figure 4F,G) and suppressed hyperglycemia-induced p38 phosphorylation (Figure 4F,H), whereas hyperglycemia itself did not alter total p38 levels. Importantly, p38α siRNA also inhibited hyperglycemia-induced TGase2 activation (Figure 4I). Together, these results demonstrate that p38α is essential for hyperglycemia-induced TGase2 activation in diabetic retinas.

3.4. P38α MAPK Activates TGase2 Through ROS Generation in HRECs and Diabetic Retinas

To investigate the mechanism by which p38α activates TGase2, we examined the effects of SB203580, the ROS scavenger Trolox, and cystamine on VEGF-induced ROS generation in HRECs using CellROX™ Green, DHE, and MitoSOX red assays. VEGF stimulation significantly increased intracellular ROS levels, which was inhibited by SB203580 or Trolox (Figure 5A,B). Similar results were obtained using DHE and MitoSOX red staining (Figure 5A,C,D). VEGF-induced lipid peroxidation was also inhibited by SB203580 or Trolox (Figure 5E). Transfection with p38α-specific siRNA significantly attenuated VEGF-induced ROS generation, whereas control siRNA had no effect (Figure 5F).

Figure 5. p38α MAPK activates TGase2 through ROS generation in HRECs. Human retinal endothelial cells (HRECs) were treated with 10 ng/mL vascular endothelial growth factor (VEGF) for 10 min in the presence of 20 μmol/L SB203580 (SB), 0.5 μmol/L Trolox, or 50 μmol/L cystamine (Cys), or were transfected with 100 nmol/L Ctrl or p38α siRNA. (A–E) Effects of SB, Trolox, and Cys on VEGF-induced ROS generation and lipid peroxidation. (A) ROS generation visualized using CellROX™ Green or dihydroethidium (DHE). Scale bars, 20 μm. (B–D) Quantification (n = 4) of ROS generation using CellROX™ Green (B), DHE (C), or MitoSOX red (D). (E) Quantification of malondialdehyde (MDA) levels. HRECs were treated with 10 ng/mL VEGF for 4 h in the presence of the inhibitors. (F) Inhibition of ROS generation by p38α siRNA. (G,H) Effects of SB and Trolox on VEGF-induced p38 phosphorylation. p-p38 was analyzed by Western blot (G) and quantified by densitometry (n = 4) (H). (I) Quantification of in situ transglutaminase (TGase) activity (n = 4). Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. NS, non-significant; ** p < 0.01; *** p < 0.001.

In contrast, VEGF-induced p38 phosphorylation was inhibited by SB203580, but not by Trolox (Figure 5G,H), indicating that ROS generation occurs downstream of p38α activation. VEGF-induced TGase2 activity was also suppressed by SB203580 or Trolox (Figure 5I), demonstrating that intracellular ROS are required for TGase2 activation. Notably, cystamine did not affect VEGF-induced ROS generation and lipid peroxidation (Figure 5A–E). Collectively, these findings indicate that p38α MAPK activates TGase2 through ROS generation in HRECs.

To validate these findings in vivo, we examined the effects of intravitreal injection of SB203580, Trolox, or cystamine on hyperglycemia-induced ROS generation in diabetic retinas. Hyperglycemia significantly increased retinal ROS levels, as detected by DHE staining, and this increase was suppressed by SB203580 or Trolox, but not by cystamine (Figure 6A,B). Intravitreal injection of p38α-specific siRNA also inhibited hyperglycemia-induced ROS generation (Figure 6C). Furthermore, hyperglycemia-induced TGase2 activation was prevented by both SB203580 and Trolox (Figure 6D,E). These results demonstrate that p38α MAPK activates TGase2 through ROS generation in diabetic retinas.

Figure 6. p38α MAPK activates TGase2 through ROS generation in diabetic mouse retinas. Diabetic C57BL/6 mice were intravitreally injected with 2 μL of 500 μmol/L SB203580 (SB), 2 μmol/L Trolox, or 50 mmol/L cystamine (Cys), or with 2 μL of 65 μmol/L Ctrl or p38α siRNA. Retinal ROS generation and transglutaminase (TGase) activity were then assessed. (A,B) Effects of SB and Cys on hyperglycemia-induced ROS generation. (A) Visualization of ROS generation using dihydroethidium (DHE). Scale bar, 100 μm. (B) Quantification of ROS generation by fluorescence intensity (n = 6). (C) Inhibition of hyperglycemia-induced ROS generation by p38α siRNA (n = 6). (D,E) Inhibitory effects of SB and Trolox on hyperglycemia-induced TGase activity, visualized by confocal microscopy (D) and quantified by fluorescence intensity (n = 6) (E). Scale bar, 100 μm. Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; NS, non-significant; *** p < 0.001.

3.5. The p38α–TGase2 Axis Is Essential for Hyperglycemia-Induced Stress Fiber Formation, VE-Cadherin Disassembly, and Microvascular Leakage in Diabetic Retinas

To determine the functional significance of p38α-mediated TGase2 activation in vascular leakage, we examined the effects of p38α- and TGase2-specific siRNAs on VEGF-induced stress fiber formation and VE-cadherin disassembly in HRECs. VEGF stimulation induced prominent stress fiber formation, which was markedly suppressed by both p38 and TGase2 siRNAs, but not by control siRNA (Figure 7A). VEGF-induced stress fiber formation was also inhibited by SB203580 and cystamine. VEGF treatment disrupted VE-cadherin localization at cell–cell junctions, indicative of adherens junction disassembly, and this effect was reversed by knockdown of either p38α or TGase2 (Figure 7A).

Figure 7. p38α MAPK-TGase2 axis mediates hyperglycemia-induced stress fiber formation and VE-cadherin disassembly. (A) Effects of p38α and TGase2 siRNA on vascular endothelial growth factor (VEGF)-induced stress fiber formation and VE-cadherin disassembly in human retinal endothelial cells (HRECs). HRECs were treated with 10 ng/mL VEGF for 60 min in the presence of 100 nmol/L Ctrl, p38α, or TGase2 siRNA, and visualized by confocal microscopy. Scale bars, 10 μm. (B) Effects of p38α and TGase2 siRNA on hyperglycemia-induced stress fiber formation and VE-cadherin disassembly in diabetic retinas. Diabetic mice were intravitreally injected with 2 μL of 65 μmol/L Ctrl, p38α, or TGase2 siRNA. Stress fiber formation and VE-cadherin disassembly in the superficial vascular plexus of whole-mount retinas were visualized by confocal microscopy. Scale bars, 100 μm.

To corroborate these findings in vivo, we examined the effects of intravitreal injection of p38α- or TGase2-specific siRNAs on hyperglycemia-induced cytoskeletal remodeling and junctional integrity in the superficial vascular plexus of whole-mount diabetic retinas. Hyperglycemia induced stress fiber formation, which was significantly attenuated by either p38α or TGase2 siRNA (Figure 7B). Hyperglycemia also disrupted VE-cadherin, and this disassembly was prevented by both siRNAs, whereas control siRNA had no effect (Figure 7B).

Finally, we assessed whether the p38α–TGase2 axis contributes to retinal microvascular leakage in vivo. Hyperglycemia-induced vascular leakage, evaluated by FITC–dextran angiography and fundus fluorescein angiography, was significantly reduced by intravitreal injection of either p38α or TGase2 siRNA, but not by control siRNA (Figure 8A–C). Collectively, these results demonstrate that hyperglycemia induces stress fiber formation, VE-cadherin disassembly, and microvascular leakage through activation of the p38α–TGase2 axis in diabetic retinas.

Figure 8. p38α MAPK-TGase2 axis is essential for hyperglycemia-induced vascular leakage in diabetic retinas. (A–C) Diabetic C57BL/6 mice were intravitreally injected with 2 μL of 65 μmol/L Ctrl, p38α, or TGase2 siRNA, and vascular leakage was assessed. (A) Vascular leakage visualized by fluorescein Isothiocyanate (FITC)-dextran angiography (scale bar, 100 μm) and fundus fluorescein angiography (scale bar, 300 μm). (B,C) Quantification (n = 6) of vascular leakage by fluorescence intensity of FITC-dextran (B) and fluorescein (C). (D) Schematic model depicting the role of the p38 MAPK-TGase2 axis in hyperglycemia-induced vascular leakage. Statistical significance was determined using one-way ANOVA with Holm–Sidak’s multiple comparisons test. *** p < 0.001.

4. Discussion

DR is the most common microvascular complication of diabetes and a leading cause of blindness worldwide [1,2]. Despite its high prevalence, effective therapeutic strategies remain limited. DR progresses from early non-proliferative to advanced proliferative stages, with microvascular leakage representing a defining clinical hallmark [3,9]. Preventing vascular leakage during the early stages is therefore critical for halting disease progression; however, the underlying molecular mechanisms remain incompletely understood. In this study, we identify a novel mechanism in which the p38α–TGase2 signaling axis plays a pivotal role in hyperglycemia-induced microvascular leakage in the diabetic retina (Figure 8D). Specifically, p38α activates TGase2 through ROS generation, leading to stress fiber formation and VE-cadherin disassembly, ultimately resulting in vascular leakage.

TGase2 has been implicated in a wide range of diabetic microvascular and macrovascular complications, including diabetic aortic dysfunction, retinal vascular leakage, pulmonary disease, and glomerular endothelial dysfunction [1,7,8,15]. In the diabetic retina, VEGF upregulation under hyperglycemic conditions has been shown to activate TGase2 via sequential increases in intracellular Ca²⁺ and ROS, thereby promoting microvascular leakage through stress fiber formation and adherens junction disruption [7,8]. TGase2 activation also contributes to hyperglycemia-induced AMPK dephosphorylation, leading to VE-cadherin disassembly and FITC-dextran extravasation [7]. In HRECs, high glucose–induced TGase2 activation suppresses GAPDH activity through AMPK inhibition, further facilitating VE-cadherin disruption and endothelial hyperpermeability [7]. In addition, TGase2 mediates VEGF-induced adherens junction disassembly through a pathway involving c-Src and β-catenin [8]. Collectively, these findings establish TGase2 as a central mediator of hyperglycemia-induced vascular leakage in the diabetic retina, acting through multiple downstream signaling pathways, including AMPK and GAPDH regulation and c-Src- and β-catenin-dependent junctional remodeling.

Despite these advances, the upstream mechanisms by which hyperglycemia activates TGase2 have remained unclear. Here, we demonstrate that p38α functions as a critical upstream regulator of TGase2, activating it through ROS generation and promoting stress fiber formation and VE-cadherin disassembly in diabetic mouse retinas. Notably, our data indicate that intracellular Ca²⁺ is not involved in VEGF-induced p38 phosphorylation, suggesting that the p38α–TGase2 axis represents a distinct pathway from Ca²⁺-dependent TGase2 activation. Thus, p38α–TGase2 signaling constitutes a key mechanism by which hyperglycemia promotes vascular leakage in DR, at least in part through downstream regulation of AMPK, GAPDH, and c-Src.

Further delineation of the upstream signaling events that engage the p38α–TGase2 axis under hyperglycemic conditions will be important. Previous studies suggest that the Tiam1–Rac1 pathway and apoptosis signal-regulating kinase 1 (ASK1) may contribute to p38α activation in diabetic retinas [32,33]. The Tiam1–Rac1 axis mediates high glucose–induced p38 MAPK activation through NADPH oxidase 2, leading to mitochondrial dysfunction and endothelial apoptosis [32]. ASK1, a well-established activator of p38 MAPK [34], has also been implicated in ER stress-driven retinal endothelial apoptosis in diabetic rat models [32]. However, whether VEGF engages the Tiam1–Rac1 pathway or ASK1 during hyperglycemia-induced microvascular leakage remains unresolved.

p38 MAPK phosphorylates a broad range of substrates, including protein kinases (e.g., protein kinase C and p21-activated kinase 6), transcription factors (e.g., Fos and c-Jun), and regulatory proteins involved in cell-cycle control and DNA/RNA binding [17,18,20]. Consequently, dysregulated p38 MAPK signaling has been implicated in numerous physiological and pathological processes, including immune regulation, neuronal function, cardiomyocyte activity, and metabolism [17,19,35]. Aberrant p38 MAPK activity contributes to diseases such as inflammation, neurodegeneration, cardiovascular disorders, and metabolic dysfunction [17,19,21,22,23]. Moreover, p38 MAPK has been implicated in diabetic complications, including cardiomyopathy, nephropathy, and retinopathy [24,25,26].

Given that vascular dysfunction, particularly vascular leakage, is a central feature of diabetic complications [36,37], understanding the role of p38 MAPK in this process is crucial. Although p38 MAPK has been linked to acellular capillary formation and pericyte loss [24], its specific role in hyperglycemia-induced vascular leakage has not been clearly defined. In this study, we demonstrate that p38 MAPK is a key mediator of vascular leakage in diabetic retinas. The p38 MAPK inhibitor SB203580 suppressed VEGF-induced endothelial permeability in HRECs and hyperglycemia-induced vascular leakage in diabetic retinas. Consistently, p38α-specific siRNA attenuated hyperglycemia-induced vascular leakage by suppressing stress fiber formation and adherens junction disassembly. Given the central role of vascular leakage in diabetic complications [1,15,37], p38 MAPK likely represents a critical mediator of microvascular dysfunction. Further studies are warranted to determine its involvement in other microvascular complications, such as diabetic nephropathy, neuropathy, and pulmonary disease.

Among p38 MAPK family members, p38α emerged as the predominant isoform mediating hyperglycemia-induced vascular leakage in diabetic retinas. p38α was initially identified as a 38 kDa polypeptide undergoing tyrosine phosphorylation in response to lipopolysaccharide stimulation and was later recognized as the molecular target of pyridinyl imidazole compounds such as SB203580 [17]. Although the four mammalian p38 MAPK isoforms—p38α, p38β, p38γ, and p38δ—share substantial sequence homology, they differ in tissue distribution, substrate specificity, and pharmacological sensitivity [17,18,19]. In the present study, SB203580, which selectively inhibits p38α and p38β, suppressed VEGF-induced endothelial permeability and hyperglycemia-induced vascular leakage. Importantly, p38α-specific siRNA effectively inhibited VEGF- or hyperglycemia-induced TGase2 activation, stress fiber formation, and adherens junction disassembly in both HRECs and mouse retinas. These findings identify p38α as a pivotal mediator of diabetic microvascular dysfunction, although the contributions of other p38 MAPK isoforms warrant further investigation.

DR is increasingly recognized as a disorder of the neurovascular unit, characterized by both neurodegeneration and microvascular abnormalities [6,38,39]. Notably, although TGase2 plays a critical role in retinal vascular leakage, it is not involved in VEGF-induced neuropathological alterations in the diabetic retina [40]. Hyperglycemia induces reactive gliosis, neuronal excitotoxicity, and ganglion cell apoptosis through mechanisms involving VEGF and ROS; however, these neurodegenerative processes do not require TGase2 [40]. Thus, in diabetic retinas, microvascular dysfunction and neurodegeneration are regulated by distinct mechanisms, although both are driven by VEGF-dependent oxidative stress. Increasing evidence suggests that neurodegeneration may precede or contribute to microvascular abnormalities during early DR [6,38], underscoring the need to elucidate the interplay between neuronal and vascular pathology for comprehensive therapeutic intervention.

Several limitations of this study should be acknowledged. First, our mechanistic insights were derived from HRECs and diabetic mouse models; validation in human retinal tissues will be essential to establish translational relevance. Second, although the Tiam1–Rac1 axis and ASK1 are plausible upstream regulators of p38α–TGase2 signaling, their precise contributions remain to be defined. Third, it is unclear whether this signaling axis contributes to other diabetes-associated microvascular or macrovascular complications. Finally, given the need for sustained inhibition of retinal vascular leakage with minimal adverse effects, the development of pharmacological agents that selectively target the p38α–TGase2 axis will be critical.

5. Conclusions

We demonstrate that p38α-mediated regulation of TGase2 plays a pivotal role in hyperglycemia-induced microvascular leakage in diabetic retinas. p38α is required for TGase2 activation through ROS generation, leading to stress fiber formation, VE-cadherin disassembly, and vascular leakage. These findings identify the p38α–TGase2 signaling axis as a critical mediator of diabetic retinal microvascular dysfunction and a potential therapeutic target for preserving retinal vascular integrity.