Glycolysis-Histone Lactylation Crosstalk Drives TXNIP-NLRP3-Mediated PANoptosome Assembly and PANoptosis Activation Underlying Diabetic Retinopathy Pathogenesis

Abstract

Diabetic retinopathy (DR), a major cause of vision loss in adults, involves aberrant metabolism and inflammation. This study investigated the interplay between glycolysis, histone lactylation, and PANoptosis in DR using human retinal pigment epithelial (RPE) cells under high glucose and diabetic mouse models. Results demonstrated a positive feedback loop where enhanced glycolysis increased histone lactylation, which in turn further promoted glycolysis. This cycle activated the expression of thioredoxin interacting protein (TXNIP) and NOD-like receptor thermal protein domain associated protein 3 (NLRP3), leading to PANoptosome formation and triggering PANoptosis, a coordinated cell death pathway contributing to DR pathology. Crucially, experiments manipulating TXNIP expression (via RNAi or overexpression) confirmed its central role in linking histone lactylation to NLRP3 activation and PANoptosome assembly. Importantly, inhibiting glycolysis or downregulating TXNIP successfully reduced histone lactylation, suppressed PANoptosome formation, and alleviated PANoptosis. These findings establish that the glycolysis-histone lactylation axis, mediated by TXNIP/NLRP3 signaling, drives PANoptosis in RPE cells through PANoptosome formation, playing a critical role in DR development. Targeting this specific pathway presents a promising new therapeutic strategy for diabetic retinopathy.

Keywords: PANoptosis; diabetic retinopathy; glycolysis; histone lactylation; thioredoxin interacting protein.

FIGURE 1

Massive RPE death impacts the progression of diabetic retinopathy (DR). (A) Representative images of immunofluorescence (IF) staining in retinal sections from diabetic and control mice (n = 3). Nuclei were counterstained with DAPI. Scale bar: 100 µm. (B) Representative images illustrating leukostasis in retinal vasculature, assessed by FITC‐conjugated concanavalin A perfusion labeling followed by flat‐mount staining. Scale bar: 100 µm. (C) Representative micrographs of retinal vasculature visualized using periodic acid–Schiff (PAS) staining in diabetic and age‐matched non‐diabetic mice. Arrows indicate acellular capillaries. Scale bar: 25 µm. (D and E) Quantification of dihydroethidium (DHE) fluorescence intensity (D) and dichlorofluorescein (DCF) fluorescence (E) in retinal sections as indicators of superoxide and reactive oxygen species (ROS) levels, respectively. Data represent mean ± SD from five biological replicates (n = 5). Statistical significance was determined by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (F) Quantification of electroretinography (ERG) a‐ and b‐wave amplitudes under scotopic conditions to assess retinal function. Data are presented as mean ± SD (n = 3). One‐way ANOVA with Bonferroni post hoc test was used. *p < 0.05. (G) Live/dead viability/cytotoxicity assay performed using calcein AM (green fluorescence for live cells) and EthD‐III (red fluorescence for dead cells) (n = 3). Scale bar: 100 µm. (H) Assessment of cell viability (left) and cytotoxicity (right) following 48‐h treatment of ARPE‐19 cells with 25 mM high glucose, in combination with apoptosis inhibitor Z‐VAD‐FMK (VAD, 25 µM), necroptosis inhibitor necrostatin (Nec, 20 µM), ferroptosis inhibitor ferrostatin‐1 (Fer, 10 µM), pyroptosis inhibitors Ac‐DMPD‐CMK/DMLD‐CMK (DMPD/DMLD, 20 µM), disulfiram (Dis, 1 µM), or autophagy inhibitor 3‐methyladenine (3‐Me, 10 µM). Data represent mean ± SD from three independent experiments (n = 3). Statistical analysis was performed using one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control.

FIGURE 2

ARPE‐19 cells undergo PANoptosis with interconnected apoptosis, pyroptosis, and necroptosis. (A) Representative images of YP1‐positive (green, DNA damage indicator) and propidium iodide (PI)‐positive (red, membrane disruption) staining in ARPE‐19 cells after high glucose treatment. Nuclei were counterstained with DAPI. Scale bar = 100 µm. (B) Detection and quantification of reactive oxygen species (ROS) levels using DCFDA fluorescent probe. Data represent mean ± SD from three biological replicates (n = 3). One‐way ANOVA with Bonferroni correction was applied. *p < 0.05. (C) Flow cytometry analysis and quantification of annexin V/PI‐stained cells to evaluate apoptosis and cell death. Data are shown as mean ± SD (n = 3). Statistical significance was calculated using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (D) Representative western blot images showing expression and cleavage status of caspase‐3, ‐7, ‐8, ‐9; phosphorylated and total MLKL; phosphorylated and total RIP3; GSDME and cleaved GSDME; GSDMD and cleaved GSDMD in ARPE‐19 cells after 48 h of high glucose treatment. β‐Actin served as loading control. Data represent mean ± SD from three independent experiments (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05. (E) Western blot analysis of phosphorylated MLKL, total MLKL, caspase‐3, cleaved caspase‐3, GSDME, and cleaved GSDME expression in ARPE‐19 cells transfected with siRNA targeting caspase‐3, GSDME, or MLKL. Protein levels were normalized to β‐actin. Data represent mean ± SD from three independent experiments (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05.

FIGURE 3

Activation of TXNIP/NLRP3 signaling promotes PANoptosis in cells. (A) Representative Western blot images of TXNIP and NLRP3 in retinal tissues from diabetic and control mice. β‐Actin was used as loading control. Data represent mean ± SD from three independent experiments (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05. (B) Representative western blot images of TXNIP and NLRP3 expression in ARPE‐19 cells treated with high glucose. Data represent mean ± SD from three independent experiments (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05. Representative images depicting TUNEL‐positive apoptotic cells (C), EthD‐III‐positive pyroptotic cells (D), and PI‐positive necroptotic cells (E) in ARPE‐19 cells treated with high glucose alone or in combination with siTXNIP. Nuclei were stained with DAPI. Scale bar = 100 µm. Data represent mean ± SD from three independent experiments (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05. (F) Co‐immunoprecipitation (Co‐IP) analysis showing interaction between TXNIP and NLRP3 in ARPE‐19 cells treated with high glucose. Whole‐cell lysates were immunoprecipitated with anti‐TXNIP antibody and analyzed by Western blotting with anti‐NLRP3. IgG was used as negative control. Densitometric quantification of co‐IP band intensity is shown.

FIGURE 4

NLRP3 interacts with NLRP12 and cell death molecules to drive PANoptosis. (A and B) NC and Nlrp3 knockout ARPE‐19 cells were either left unstimulated (0 h) or treated with high glucose for 48 h and immunostained for NLRP3. Nuclei were counterstained with DAPI. A broader field of view for NC cells at 48 h is shown (A), and quantification of NLRP3‐positive aggregates is provided (B). Scale bar = 10 µm. Data represent mean ± SD (n = 3). One‐way ANOVA with Bonferroni post hoc test was used. *p < 0.05. (C) NC and Nlrp3 knockout ARPE‐19 cells were treated with high glucose for 48 h and co‐stained for NLRP3, ASC, caspase‐8 (CASP8), and RIPK3. Representative images of cells containing co‐localized puncta are shown. Enlarged views of selected regions are displayed on the right. Scale bar = 10 µm. (D) Immunoblot (IB) analysis of ASC, TXNIP, NLRP3, CASP8, and RIPK3 following immunoprecipitation (IP) with IgG control or anti‐ASC antibodies in NC and Nlrp3 knockout ARPE‐19 cells after high glucose treatment. Densitometric quantification is shown. (E) Representative images showing immunohistochemistry (IHC)‐stained retinal sections from wild‐type (WT) and Nlrp3 knockout mice 2 weeks after intraperitoneal injection of PBS or streptozotocin (STZ). Staining for cleaved caspase‐3, GSDMD‐N, and phosphorylated MLKL was performed. Scale bar = 100 µm. (F) Western blot analysis of NLRP3 expression in retinal tissues from WT mice treated with PBS or STZ for 2 weeks. β‐Actin served as loading control. Data represent mean ± SD from five biological replicates (n = 5). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (G) Representative images of ZO‐1 immunofluorescence staining to assess retinal vascular integrity in WT and Nlrp3 knockout mice following PBS or STZ treatment. Scale bar = 100 µm. (H) Western blot analysis of RIPK3, CASP3, and GSDME expression levels in retinal tissues from WT and Nlrp3 knockout mice at 2 weeks post‐PBS or STZ injection. β‐Actin served as internal control. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control.

FIGURE 5

Aberrant glycolysis elevates histone lactylation in RPE cells. (A) Representative western blot images showing changes in HIF1A and glycolytic pathway proteins (HK2, PFKL, ALDOA, PKM2, LDHA) in murine RPE (mRPE) cells after high glucose treatment. β‐Actin was used as loading control. Data represent mean ± SD from three biological replicates (n = 3). One‐way ANOVA with Bonferroni correction was applied. *p < 0.05 versus control. (B and C) Quantitative detection of glucose uptake, pyruvate levels, lactate production, ATP concentration, and enzymatic activities (hexokinase [HK], phosphofructokinase [PFK], aldolase [ALDO], pyruvate kinase [PKM], lactate dehydrogenase [LDH]) in ARPE‐19 cells using commercial assay kits. Data represent mean ± SD from three biological replicates (n = 3). One‐way ANOVA with Bonferroni correction was applied. *p < 0.05 versus control. (D) Representative immunohistochemistry (IHC) images detecting HIF1A, PKM2, LDHA, SLC16A3 (MCT4), and histone lactylation (H3K18la) levels in retinal tissues from diabetic and control mice. Scale bar: 100 µm. (E) Representative western blot images assessing global histone lactylation levels in ARPE‐19 cells after exposure to high glucose. Histones were extracted and probed with anti‐H3K18la antibody. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control.

FIGURE 6

Histone lactylation activates TXNIP transcription in RPE cells. (A) RT‐qPCR analysis of TXNIP mRNA expression in ARPE‐19 and mRPE cells after 48 h of high glucose treatment. β‐Actin was used as internal reference gene. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was determined by one‐way ANOVA followed by Bonferroni post hoc test. *p < 0.05 versus untreated control. (B) ChIP‐qPCR analysis of DNA fragments immunoprecipitated from ARPE‐19 cells using an H3K18la‐specific antibody, followed by qPCR with primers targeting the TXNIP promoter region. Data represent mean ± SD from three biological replicates (n = 3). One‐way ANOVA with Bonferroni correction was applied. *p < 0.05 versus control. (C) ChIP‐qPCR assessment of H3K18la enrichment at the TXNIP promoter in ARPE‐19 cells treated with glycolysis inhibitors (4 mM 2‐deoxyglucose [2‐DG], 8 mM oxamate), or transfected with siRNA against EP300. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (D and E) Representative western blot images showing TXNIP protein expression in ARPE‐19 and mRPE cells treated with high glucose for varying durations (D) and with different concentrations of exogenous lactate (E). β‐Actin served as loading control. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was determined by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (F) Representative western blot images illustrating TXNIP expression levels in ARPE‐19 and mRPE cells treated with increasing concentrations of glycolytic inhibitors (2DG and oxamate). β‐Actin served as loading control. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (G) Representative Western blot images displaying TXNIP expression in ARPE‐19 and mRPE cells treated with oxamate or supplemented with N‐acetyl‐l‐alanine (Nala). β‐Actin was used as internal control. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control.

FIGURE 7

ERG recruits EP300 to form a complex upregulating histone lactylation at the TXNIP Promoter region. (A) Protein–protein interactions between ERG, EP300, and other co‐factors predicted using the BioGRID database. (B and C) Prediction of transcription factors binding to the TXNIP promoter region using the JASPAR database. Putative ERG binding sites are indicated. (D) Representative images of fluorescence in situ hybridization (FISH) showing subcellular localization of TXNIP mRNA in ARPE‐19 cells. Scale bar: 10 µm. (E) qPCR analysis of TXNIP mRNA expression changes following lentiviral‐mediated knockdown or overexpression of ERG and EP300 in ARPE‐19 cells. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (F–G) Pull‐down assays analyzing direct interaction between ERG and EP300 in ARPE‐19 cell lysates. (H) Representative images of proximity ligation assay (PLA) demonstrating endogenous interaction between ERG and EP300 in ARPE‐19 cells. Scale bar: 10 µm. (I) ChIP‐qPCR assessment of ERG and EP300 occupancy at the TXNIP gene promoter in ARPE‐19 cells. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was determined by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (J) ChIP‐qPCR analysis of ERG, EP300, and H3K18la occupancy at the TXNIP promoter in ARPE‐19 cells stably transfected with siERG or siEP300. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (K) Representative western blot images showing protein expression levels of ERG and EP300 in stably transfected ARPE‐19 cells. (L) Representative western blot images illustrating expression of PANoptosis‐related proteins (caspase‐3, GSDMD, and pMLKL) in stably transfected ARPE‐19 cells. β‐Actin served as loading control. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control.

FIGURE 8

Glycolysis regulates RPE PANoptosis via TXNIP. (A) Quantification of DCF fluorescence intensity indicating ROS levels in ARPE‐19 cells treated with glycolysis inhibitor 2DG and/or TXNIP‐targeting RNAi delivered via adeno‐associated virus (AAV). Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (B) Quantification of DHE staining reflecting superoxide levels in RPE cells under the same treatments. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was determined by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (C) Representative images and quantification of immunofluorescence (IF) staining for cleaved caspase‐3, GSDMD‐N, and pMLKL in retinal sections from mice treated with 2DG and/or AAV‐TXNIP‐RNAi. Scale bar: 100 µm. Data represent mean ± SD from three biological replicates (n = 3). Statistical analysis was conducted using one‐way ANOVA with Bonferroni post hoc test. *p < 0.05 versus control. (D) Quantification of ERG a‐ and b‐wave amplitudes under scotopic conditions in mice treated with 2DG and/or AAV‐TXNIP‐RNAi. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (E) Live/Dead viability/cytotoxicity assay using calcein AM (live cells, green) and EthD‐III (dead cells, red) in ARPE‐19 cells treated as above. Scale bar: 100 µm. (F) Representative western blot images showing alterations in PANoptosis‐related protein expressions (caspase‐3, GSDMD, and pMLKL) in RPE cells following treatment with 2DG and/or AAV‐mediated TXNIP knockdown. β‐Actin was used as internal control. Data represent mean ± SD from three biological replicates (n = 3). Statistical significance was assessed by one‐way ANOVA with Bonferroni correction. *p < 0.05 versus control. (G) Schematic illustration of the proposed mechanistic pathway linking glycolysis, histone lactylation, ERG‐EP300 complex formation, TXNIP upregulation, and subsequent activation of PANoptosis in RPE cells during diabetic retinopathy.