Macrophage extracellular traps amplify retinal endothelial anoikis via the S1P–S1PR axis in diabetic retinopathy

Abstract

Objective

To investigate the pathogenic role of macrophage extracellular traps (METs) in proliferative diabetic retinopathy (PDR), focusing on endothelial dysfunction and inflammatory activation.

Methods

Transcriptomic data from PDR patients were first analyzed to identify METs-associated pathways. Anoikis-related subgroups were defined using the median gene set variation analysis (GSVA) enrichment score of the anoikis gene set, and enriched pathways were selected for subsequent validation. In vitro, diabetic retinopathy models were established using high-glucose and METs stimulation. Human retinal microvascular endothelial cells (HRMECs) were assessed for functional alterations and apoptosis. Rescue assays employed the AKT agonist SC79, the endocytosis inhibitor Dynasore, and DNase I. In vivo, streptozotocin (STZ)-induced diabetic mice received intravitreal METs with or without pharmacological interventions to evaluate vascular leakage, inflammatory responses, and neovascularization.

Results

Transcriptomic analysis revealed a strong association between METs-induced endothelial injury and activation of anoikis pathways, with sphingolipid signaling significantly enriched in the anoikis-high subgroup. In vitro, METs disrupted endothelial barrier integrity, induced ROS accumulation and mitochondrial damage, and activated anoikis and inflammatory signaling, accompanied by FAK dephosphorylation. These effects were partially reversed by SC79, DNase I, and sphingolipid-pathway inhibition. METs were internalized by HRMECs via endocytosis, triggering downstream signaling. In vivo, intravitreal METs induced vascular leakage, inflammatory cytokine elevation, and neovascularization, whereas inhibition of the SPHK1/S1P pathway (SKI-II) significantly mitigated these pathological changes.

Conclusion

METs promote retinal vascular dysfunction by inducing endothelial anoikis and inflammatory activation through FAK/AKT and SPHK1/S1P/S1PR2/NF-κB signaling. Targeting METs-triggered lipid signaling may offer new therapeutic insights for diabetic retinopathy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07770-6.

Introduction

Diabetic retinopathy (DR) is one of the most severe microvascular complications of diabetes and, in advanced cases, can lead to irreversible blindness [1, 2]. Clinically, DR is classified into non-proliferative (NPDR) and proliferative (PDR) stages [3]. PDR is characterized by pathological neovascularization, vitreous hemorrhage, and tractional retinal detachment, constituting the primary cause of vision loss [4]. Although anti-VEGF therapies have proven effective in suppressing aberrant neovascularization in some patients, accumulating clinical evidence indicates considerable heterogeneity in therapeutic responses and high recurrence rates among PDR patients [5]. This suggests that the pathogenesis of PDR is not solely VEGF-driven, highlighting the need to further elucidate its underlying immuno-inflammatory mechanisms.

Recent research increasingly recognizes DR not merely as a vascular disease but as a neurovascular-immune disorder characterized by synergistic damage [6]. Under persistent hyperglycemia, enhanced local oxidative stress, endothelial dysfunction, disruption of the blood-retinal barrier (BRB), and remodeling of the inflammatory microenvironment collectively contribute to DR initiation and progression [7–9]. Among these, chronic low-grade inflammation within the retina is considered a key driver of early-stage damage [10]. During inflammation, innate and adaptive immune cells—including macrophages, microglia, and T cells—are aberrantly recruited and activated in the retinal milieu, where they continuously secrete proinflammatory cytokines such as TNF-α, IL-1β, and IL-6 [10]. These mediators exacerbate endothelial apoptosis, increase vascular permeability, and contribute to macular edema formation.

Against the backdrop of advancing inflammation research, increasing attention has been paid to a novel effector function of innate immune cells: extracellular traps (ETs). Initially discovered in neutrophils, ETs are composed of chromatin, histones, and granule-derived enzymes (e.g., MPO and NE), and function to entrap and neutralize pathogens as part of the innate immune defense [11]. More recently, it has been revealed that macrophages, upon specific inflammatory stimuli, can also release similar structures known as macrophage extracellular traps (METs) [12]. Emerging evidence suggests that ETs may contribute to the progression of diabetic complications by inducing tissue injury, promoting immune activation, and triggering cell death [13, 14]. However, the expression patterns, pathological functions, and molecular mechanisms of METs within the neovascular microenvironment of PDR retinas remain largely undefined.

In light of this, the present study aims to systematically investigate the expression characteristics and functional roles of METs in PDR, with a particular focus on evaluating whether METs contribute to retinal disease progression by modulating endothelial permeability, activating inflammatory signaling pathways, and inducing anoikis. This research is expected to identify critical nodes in the immunopathological cascade of PDR and may provide novel theoretical foundations and therapeutic targets for immunomodulatory strategies in diabetic retinopathy.

Methods

Bioinformatics analysis

In this study, transcriptomic datasets related to diabetic retinopathy (DR), including GSE60436 and GSE221521, were downloaded from the Gene Expression Omnibus (GEO) database. These datasets included retinal tissue samples from both DR patients and healthy controls. Raw expression matrices were extracted using the GEOquery package in R (version 4.2.1), followed by background correction and normalization using the limma package (version 3.62.2). The anoikis-related gene set used in this study was derived from a previously published comprehensive collection of regulated cell death (RCD) signatures, which integrated 18 RCD modalities from multiple databases and literature sources [15]. Specifically, 29 curated anoikis signatures reported in that study were merged to generate a unified anoikis gene set for downstream analyses. Gene set variation analysis (GSVA) was then performed to calculate the anoikis pathway enrichment score for each sample. Samples were subsequently stratified into anoikis-high and anoikis-low groups using the median GSVA score as the cutoff.

Differential gene expression analysis between groups was performed using the limma package, with the thresholds set at |log2 fold change| > 1 and adjusted P-value < 0.05. The resulting differentially expressed genes were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using the clusterProfiler package (version 4.14.6). Significance thresholds were defined as P < 0.05 and false discovery rate (FDR) < 0.25. To assess the activity of the sphingolipid signaling pathway across samples, gene set variation analysis (GSVA) was performed using the GSVA algorithm (version 2.0.7).

Animal model

Six-week-old male C57BL/6J mice were purchased from GemPharmatech Co., Ltd. (Nanjing, China) and maintained under specific pathogen-free (SPF) conditions in the animal facility. After one week of acclimatization, a DR model was induced. Mice in the experimental group were fed a high-fat diet (HFD, 45% kcal from fat) for 4–6 weeks, followed by daily intraperitoneal injections of low-dose streptozotocin (STZ; 30–40 mg/kg/day, dissolved in 0.1 mol/L citrate buffer, pH 4.5) for five consecutive days. After STZ administration, mice were maintained on the HFD throughout the experimental period. Fasting blood glucose levels were measured weekly, and mice with fasting glucose consistently above 16.7 mmol/L for three consecutive measurements were considered diabetic. Successfully modeled mice were further maintained for an additional 3–5 months to induce DR-like retinal pathology.

To mimic the local accumulation of METs in the retina under pathological conditions, intravitreal injections were performed under stereomicroscopy. Mice were anesthetized and fixed in a surgical position, and a 33-gauge microsyringe was used to gently penetrate the sclera and deliver 1–2 µL of sterile phosphate-buffered saline (PBS) containing 1–5 µg of BCA-quantified METs protein into the vitreous cavity. According to the experimental design, injections were repeated every 7–14 days for a total duration of 4–8 weeks. The SPHK1/2 inhibitor SKI-II (Sigma-Aldrich, dissolved in DMSO) was administered intraperitoneally at a dose of 50 mg/kg, 2 h prior to each METs intravitreal injection. Control mice received an equal volume of DMSO vehicle and intravitreal PBS injection. After each injection, antibiotic eye drops were applied to prevent infection. Body weight, blood glucose levels, and ocular symptoms were monitored throughout the study.

At the end of the experiment, mice were euthanized and eyeballs were collected from each group. Samples were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 5 μm thickness. Hematoxylin and eosin (H&E) staining was performed to assess retinal morphology and neovascularization. The area of retinal neovascularization was quantified using ImageJ software. All animal procedures were performed in accordance with national ethical guidelines and approved by the Animal Ethics Committee of Shanghai Tenth People’s Hospital, School of Medicine, Tongji University.

Electroretinogram (ERG) recording

Electroretinography was performed using a biological signal acquisition system (RETIport, Roland, Germany). Mice were dark-adapted for 12 h prior to testing, followed by anesthesia. A drop of corneal lubricant was applied to prevent drying, and a contact lens–type gold wire electrode was placed on the cornea to record scotopic b-wave responses. The amplitudes of b-waves under dark-adapted conditions were measured and analyzed to assess retinal function.

Assessment of vascular permeability by evans blue dye

Vascular leakage was assessed using the Evans blue dye extravasation method. Mice were injected via the tail vein with Evans blue dye (45 mg/kg). After 1 h of circulation, eyeballs were harvested, fixed in formalin, embedded, and sectioned for imaging. Retinal vascular leakage was evaluated by visual inspection and quantified by optical density analysis using ImageJ software.

Western blot analysis

Cells or tissues from different treatment groups were lysed using RIPA buffer containing protease and phosphatase inhibitors (Beyotime, P1046) on ice for 30 min, followed by centrifugation at 12,000 rpm for 15 min. Supernatants were collected and quantified using a BCA assay kit (Thermo Fisher Scientific, 23225). Equal amounts of protein were mixed with 5× SDS loading buffer, denatured at 100 °C for 5 min, and separated by SDS-PAGE. Proteins were then transferred onto PVDF membranes (Millipore), blocked in 5% non-fat milk for 1 h at room temperature, and incubated with primary antibodies overnight at 4 °C. After washing, membranes were incubated with HRP-conjugated secondary antibodies (1:5000) for 1 h at room temperature. Bands were visualized using ECL reagents (Thermo Fisher) and quantified using ImageJ. Protein levels were normalized to β-actin. The detailed information on the primary antibodies is provided in Supplementary Table 1.

Immunofluorescence (IF) staining

HRMECs were seeded on coverslips and subjected to Control, METs, or METs+Dynasore treatment. Cells were fixed with 4% paraformaldehyde for 10 min, permeabilized, and blocked with 5% BSA for 1 h. Primary antibodies were incubated overnight at 4 °C, followed by Alexa Fluor 488-conjugated secondary antibodies (1:500) for 1 h at room temperature. Nuclei were stained with DAPI. Images were captured using a confocal microscope.

Cell culture and treatments

Human retinal microvascular endothelial cells (HRMECs) were cultured in medium supplemented with 10% fetal bovine serum, endothelial growth supplements, and 1% penicillin–streptomycin at 37 °C with 5% CO₂. To model diabetic hyperglycemia, cells were exposed to 25 mM D-glucose (high-glucose, HG), while the normal-glucose control contained 5.5 mM D-glucose; an osmotic control was included by adding 19.5 mM mannitol to control medium. For METs stimulation, purified METs were added at a final DNA concentration of 500 ng/mL and incubated for 24 h. Rescue experiments involved pretreatment with the AKT agonist SC79, the endocytosis inhibitor Dynasore, or DNase I for 1 h prior to HG + METs stimulation.

MitoTracker staining

Cells were incubated with MitoTracker^® Green FM (YEASEN, 40742ES50, 100nM) and MitoTracker^® Red CMXRos (YEASEN, 40741ES50, 100 nM) for 30 min to label mitochondrial mass and membrane potential, respectively. After washing with PBS, cells were observed under a fluorescence microscope (Leica DMi8). Green signal indicates mitochondrial content; red reflects mitochondrial membrane integrity.

Tube formation assay

Matrigel (Corning) was thawed on ice and added to pre-chilled 96-well plates (50 µL/well), incubated at 37 °C for 30 min to solidify. HRMECs (5 × 10⁴ cells/well) suspended in ECM medium were seeded onto Matrigel and treated as indicated. After 8 h incubation, tube-like structures were imaged (Olympus IX73), and branch points were quantified using ImageJ.

ROS detection

Cells were treated for 24 h, washed with PBS, and incubated with DCFH-DA (10 µM, 1:1000 in serum-free medium, Beyotime) for 30 min at 37 °C in the dark. Excess dye was washed off, cells were trypsinized, resuspended in PBS, and analyzed using a BD FACSCalibur flow cytometer. ROS levels were quantified using FlowJo and presented as fold change.

Cell permeability assay

HRMECs were seeded onto Transwell inserts and treated with Control, METs, or METs + inhibitors. After 24 h, FITC-Dextran (70 kDa, 1 mg/mL, YEASEN, 61224ES08) was added to the upper chamber and incubated for 2 h. Fluorescence in the lower chamber was measured and used to calculate relative permeability.

ELISA for cytokine quantification

Levels of TNF-α and IL-6 in cell culture supernatants were measured using commercial ELISA kits (MULTI SCIENCES, EK182, EK1153) according to the manufacturer’s instructions.

Nitric oxide (NO) assay

NO production was determined by measuring nitrite (NO₂⁻) levels in supernatants using a Griess reagent kit (Beyotime, S0021S), following the manufacturer’s protocol. Absorbance was normalized to total protein content.

Quantitative real-time PCR (RT-qPCR)

Total RNA was extracted using TRIzol reagent and reverse transcribed into cDNA with HiScript III RT SuperMix (Vazyme). qPCR was performed using ChamQ SYBR qPCR Master Mix with the following cycling conditions: 95 °C for 30 s, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s. GAPDH was used as the internal control. Relative expression levels were calculated using the 2^-ΔΔCt method. All reactions were performed in technical triplicates. The primer sequences used for qPCR are detailed in Supplementary Table 2.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software (version 10.1.2). Data are presented as mean ± standard deviation (SD). Before applying parametric tests, the Shapiro–Wilk test was used to assess data normality, and the Levene test was used to evaluate homogeneity of variances. For comparisons between two groups, normally distributed data with equal variances were analyzed using two-tailed Student’s t-tests; otherwise, the Mann–Whitney U test was applied. For comparisons among three or more groups, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was used when normality and homoscedasticity assumptions were met; otherwise, the Kruskal–Wallis test with Dunn’s multiple-comparison test was performed. All experiments were independently repeated at least three times, and a p-value < 0.05 was considered statistically significant.

Results

METs induce PDR-like retinal pathology via activation of anoikis signaling

By analyzing the GSE221521 dataset, which includes peripheral blood samples from healthy controls, diabetic patients without retinopathy (DM group), and patients with diabetic retinopathy (DR group), we first assessed whether METs are elevated in the clinical setting. The results revealed a marked increase in circulating MPO–DNA complexes in DR patients, whereas no such elevation was observed in healthy individuals or diabetic patients without retinopathy (Fig. 1A). This finding indicates that METs are robustly activated in the systemic circulation of patients with PDR, providing direct clinical evidence supporting their involvement in human disease.

Fig. 1.

METs induce diabetic retinopathy (DR) -like retinal pathology. In this figure, “DR” denotes mice with streptozotocin (STZ)–induced diabetic retinopathy. (A) MPO–DNA complex levels in peripheral blood from healthy controls, diabetic patients without retinopathy (DM), and patients with diabetic retinopathy (DR) based on the GSE221521 cohort. (B) Venn diagram showing the overlap of differentially expressed genes (DEGs) between human DR transcriptome dataset (GSE60436) and METs-treated retinal tissues. (C) Heatmap representing hierarchical clustering of overlapping DEGs. (D) GSEA illustrating enriched signaling pathways in METs-exposed retinas. (E-F) Body weight (E) and fasting blood glucose levels (F) recorded in mice across experimental groups from 8 to 24 weeks. (G-H) Representative ERG waveforms (G) and quantification of b-wave amplitude (H). (I) Assessment of retinal vascular leakage by Evans blue dye extravasation. (J) Quantification of serum METs level via combined MPO-DNA ELISA and qPCR. (K) Serum levels of TNF-α and IL-6 determined by ELISA. (L-M) mRNA levels of inflammatory and angiogenic factors in retinal tissues measured by qPCR. (N) H&E staining of retinal sections with retinal thickness analysis. (O-P) Serum concentrations of soluble Fas ligand (sFasL) and soluble Fas (sFas) measured by ELISA. (Q) Western blot analysis of retinal protein levels including CD68, cleaved-PARP, cleaved caspase-3, total caspase-3, phosphorylated FAK (p-FAK), total FAK, integrin β1, and β-actin as internal control

Having established this clinical relevance, we next explored the potential mechanistic role of METs in PDR from a transcriptomic perspective. By integrating human retinal transcriptome data from PDR patients (GSE60436) with METs-related differentially expressed genes (DEGs) obtained from METs-treated retinal tissues, we identified a total of 2,468 overlapping genes (Fig. 1B-C). Subsequent GSEA revealed significant enrichment of the anoikis signaling pathway in PDR samples (Fig. 1D). These observations suggest that METs may compromise endothelial adhesion and structural stability, thereby promoting ECM-detachment–associated cell death and accelerating disease progression in PDR.

To experimentally validate the pathogenic effects of METs in vivo, we established a PDR mouse model via intravitreal injection of METs. Mice subjected to a high-fat, high-glucose diet combined with STZ served as a positive control. METs administration did not significantly affect body weight or blood glucose levels (Fig. 1E-F), but ERG recordings revealed a significant reduction in b-wave amplitude (Fig. 1G-H), indicating impaired retinal function. METs exposure also led to increased vascular permeability in the retina (Fig. 1I) and elevated levels of MPO-DNA complexes in the serum (Fig. 1J), confirming the successful systemic induction and activation of extracellular traps. Additionally, METs-treated mice displayed pronounced local and systemic inflammatory responses. Levels of TNF-α and IL-6 were elevated in the serum (Fig. 1K), and the expression of pro-inflammatory cytokines (TNF-α, IL-6, IL-1β, CCL2) and the angiogenic factor VEGF were significantly upregulated in retinal tissues (Fig. 1L-M). H&E staining revealed increased retinal thickness and disrupted structural integrity in the METs group, resembling the pathological features observed in the DR control group (Fig. 1N).

At the mechanistic level, METs treatment markedly increased serum concentrations of sFasL and sFas (Fig. 1O-P), indicating activation of apoptosis-related pathways. Western blot analysis of retinal tissues demonstrated elevated expression of CD68, cleaved PARP, and cleaved caspase-3, along with ECM-associated proteins such as phosphorylated FAK and Integrin β1 (Fig. 1Q). Collectively, these results demonstrate that macrophage-derived METs induce PDR-like retinal pathology by disrupting vascular barrier integrity, amplifying local inflammation, and triggering anoikis. These findings underscore the potential pathogenic role of METs in the immune-mediated mechanisms of diabetic retinopathy.

METs induce endothelial anoikis and mitochondrial dysfunction

Having established that METs compromise retinal vascular barrier integrity in vivo, we next investigated the underlying mechanisms at the endothelial cell level. Compared with controls, METs treatment significantly increased endothelial permeability (Fig. 2A), suppressed NO production (Fig. 2B), and promoted the cytoplasmic release of Cyt-C from mitochondria (Fig. 2C). These effects were comparable to those observed under HG conditions, suggesting that METs may trigger mitochondria-dependent apoptotic pathways.

Fig. 2.

SC79 reverses METs-induced vascular barrier disruption, mitochondrial dysfunction, and anoikis. (A-C) Assessment of endothelial permeability (A), nitric oxide (NO) content (B), and cytosolic Cytochrome C (Cyto-C) levels (C) following high glucose (HG) or METs stimulation. (D-E) Cell survival rate (D) and sFas concentration (E) in hRMECs exposed to METs with or without SC79, MCC950 or GSK872 treatment. (F) Permeability of hRMECs under indicated treatments.(G-J) Quantification of NO content (G), cytosolic Cyto-C (H), sFas concentration (I), and superoxide dismutase (SOD) activity (J) across groups. (K) Flow cytometry analysis of intracellular ROS levels using DCFH-DA probe. (L-M) MitoTracker Red and Green staining for mitochondrial membrane potential (L) and mitochondrial morphology (M) observation. (N) Transmission electron microscopy (TEM) imaging of mitochondrial ultrastructure and quantification of mitochondrial length. (O) Endothelial tube formation assay using Matrigel to assess angiogenic capacity. (P) Immunofluorescence staining of PARP (green) and nuclei (DAPI, blue) in hRMECs. (Q) Western blot analysis of cleaved caspase-3, cleaved-PARP, phosphorylated FAK (p-FAK), ET-1, and VEGF expression levels; β-actin served as loading control. Quantification shown on the right

To delineate the key regulatory mechanisms, METs-treated cells were co-incubated with either the AKT agonist SC79, the NLRP3 inflammasome inhibitor MCC950, or the RIPK3 inhibitor GSK872. CCK-8 assays revealed that only SC79 significantly rescued cell viability among the tested compounds (Fig. 2D). Consistently, ELISA analysis showed that SC79 was the only treatment that markedly reduced METs-induced sFas release, whereas MCC950 and GSK872 had no significant effect (Fig. 2E).

Based on these screening results, further validation experiments focused on SC79. In METs-exposed cells, SC79 significantly restored endothelial barrier integrity (Fig. 2F). Functional and biochemical assays further demonstrated that SC79 restored NO production (Fig. 2G), suppressed both Cyt-C release and sFas upregulation (Fig. 2H-I), enhanced SOD activity (Fig. 2J), and mitigated ROS accumulation (Fig. 2K), collectively indicating improved endothelial function, preserved mitochondrial integrity, and reduced oxidative stress. Morphological analyses further supported these findings. MitoTracker staining demonstrated that METs treatment reduced mitochondrial membrane potential and disrupted mitochondrial distribution, whereas SC79 restored fluorescence intensity and normal mitochondrial architecture (Fig. 2L-M). Transmission electron microscopy confirmed that SC79 attenuated METs-induced mitochondrial swelling and fragmentation, preserving mitochondrial length and membrane integrity (Fig. 2N).

Functionally, SC79 significantly suppressed METs-induced endothelial tube formation (Fig. 2O). At the molecular level, immunofluorescence and Western blot analyses confirmed that SC79 downregulated cleaved Caspase-3 and cleaved PARP (Fig. 2P-Q), and restored phosphorylated FAK while reducing endothelin-1 (ET-1) and VEGF expression—key mediators of adhesion, vascular tone, and angiogenesis (Fig. 2Q).

To determine whether this signaling axis is dependent on the structural integrity of METs, we performed enzymatic degradation using DNase I. This intervention effectively reversed METs-induced increases in endothelial permeability, oxidative stress, and mitochondrial dysfunction (Figure S1A-J), and also mitigated tube formation capacity and apoptotic activation (Figure S1K-M). Overall, these results demonstrate that activation of the AKT signaling pathway provides robust protection against METs-induced anoikis, at least in part by restoring FAK-mediated adhesion signaling, preserving mitochondrial function, attenuating oxidative stress, and modulating pro-apoptotic responses.

METs amplify endothelial injury via activation of the sphingolipid signaling pathway

While the FAK/AKT signaling axis was implicated as a key upstream regulator of METs-induced anoikis in endothelial cells, we further investigated whether downstream metabolic-inflammatory amplifiers might synergistically aggravate endothelial damage. To this end, transcriptomic data from PDR patients were stratified based on the expression levels of anoikis-related genes, followed by pathway enrichment analysis of the DEGs. Notably, the sphingolipid signaling pathway was significantly enriched in the anoikis-high subgroup (Fig. 3A-B), suggesting a potential amplifying role in METs-induced endothelial dysfunction. In this pathway, sphingosine is phosphorylated by sphingosine kinases (SPHK1/2) to generate sphingosine-1-phosphate (S1P), which is exported by Spns2 and binds to cell surface S1P receptors (S1PRs), thereby activating downstream inflammatory and survival pathways (Fig. 3C). Experimental validation showed that METs stimulation significantly upregulated the expression of SPHK1, SPHK2, S1PR2, and S1PR3 (Fig. 3D-E). Concurrently, NF-κB activation, a key downstream target of the S1P–S1PR axis, was also observed, as evidenced by increased phosphorylation of p65 and reduced levels of IκBα (Fig. 3F). Notably, these changes were attenuated by DNase I treatment. Immunofluorescence analysis further revealed that SPHK1 and S1PR2 expression was enhanced and redistributed to the perinuclear and membrane regions upon METs stimulation (Fig. 3G-H), indicating active signaling compartmentalization. Thus, these findings support an integrated mechanistic model in which METs are first recognized via integrins, triggering FAK-dependent adhesion stress and mitochondrial injury. Subsequently, sphingolipid metabolism is upregulated, leading to S1P-mediated inflammatory and apoptotic amplification.

Fig. 3.

METs activate the sphingolipid signaling pathway and promote anoikis through the S1P–S1PR axis. (A) KEGG enrichment analysis of anoikis-related differentially expressed genes (GSE221521); sphingolipid signaling pathway is highlighted. (B) GSVA analysis of sphingolipid pathway activity between low- and high-anoikis groups. (C) Schematic illustration of the S1P–S1PR signaling axis and key metabolic enzymes (SPHK1, SPHK2, S1PR1-3, Spns2). (D) qPCR analysis of SPHK1, SPHK2, and S1PR1 mRNA expression in different groups. (E) Western blot analysis and quantification of SPHK1, SPHK2, S1PR2, and S1PR3 protein levels. (F) Western blot analysis and quantification of NF-κB pathway proteins (p-p65, p65, and IκBα). (G-H) Immunofluorescence staining of SPHK1 (G) and S1PR2 (H) in hRMECs from different treatment groups. DAPI was used for nuclear staining

Inhibition of the S1P–S1PR axis reverses METs-induced endothelial injury

To validate the critical role of the SPHK–S1P–S1PR axis in METs-mediated endothelial dysfunction, we employed two pharmacological inhibitors: the SPHK inhibitor SKI-II and the pan-S1PR antagonist FTY720. Both agents significantly attenuated METs-induced increases in endothelial permeability (Fig. 4A) and restored NO production (Fig. 4B). In parallel, they reduced Cyt-C release (Fig. 4C), suppressed the lipid peroxidation marker 4-HNE (Fig. 4D), and downregulated pro-inflammatory cytokines TNF-α and IL-6 (Fig. 4E), suggesting key roles in regulating apoptosis, oxidative stress, and inflammatory responses. Furthermore, increased SOD activity (Fig. 4F) and decreased ROS levels (Fig. 4G) confirmed their protective effects on redox homeostasis. MitoTracker staining and transmission electron microscopy demonstrated that blockade of SPHK/S1PR signaling effectively restored mitochondrial fluorescence intensity (Fig. 4H), reduced the proportion of fragmented mitochondria (Fig. 4I-J), increased mitochondrial quantity (Fig. 4K), and elevated mitochondrial membrane potential (Fig. 4L), indicating alleviation of mitochondrial structural and functional damage.

Fig. 4.

Inhibition of the S1P–S1PR axis alleviates METs-induced retinal endothelial barrier dysfunction, oxidative stress, mitochondrial injury, and anoikis. (A-F) Quantification of relative permeability (A), NO content (B), Cyt-C level (C), sFas concentration (D), and inflammatory cytokines TNF-α and IL-6 (E), as well as SOD activity (F) in hRMECs from different treatment groups (Control, METs, METs+SKI-II, METs+FTY720). (G) Flow cytometry analysis and quantification of intracellular ROS levels using DCFH-DA probe. (H) Mitochondrial membrane potential and morphology were assessed by MitoTracker Red and Green staining (top and second rows). Mitochondrial ultrastructure was visualized by transmission electron microscopy (third row). Capillary-like structure formation was evaluated by tube formation assay on Matrigel (bottom row). (I-J) Quantification of mitochondrial intensity (I) and classification of mitochondrial morphology into intermediate or fragmented types (J). (K) TEM images of hRMECs displaying mitochondrial ultrastructure, and quantification of mitochondrial length. (L) Tube formation assay showing representative images and quantification of branch point numbers to assess angiogenic capacity. (M-N) Western blot analysis and protein quantification of cleaved caspase-3, cleaved PARP, p-FAK, and components of the S1P–S1PR signaling axis (SPHK1, SPHK2, S1PR2, S1PR3), as well as p65 in response to SKI-II (M) or FTY720 (N) intervention. β-actin served as loading control

At the mechanistic level, Western blot analysis revealed that SKI-II significantly downregulated cleaved caspase-3 and cleaved PARP, while reversing METs-induced dephosphorylation of FAK and reducing the expression of SPHK1/2 and S1PR2/3 (Fig. 4M). Similarly, FTY720 reduced the protein levels of these markers and additionally inhibited the expression of p65, a core component of the NF-κB signaling pathway (Fig. 4N), suggesting coordinated involvement of this axis in METs-induced apoptosis and inflammation. Collectively, these results indicate that the SPHK–S1P–S1PR axis functions as a central metabolic–signaling amplifier in METs-driven endothelial injury. Its downstream interaction with FAK and NF-κB pathways facilitates the propagation of oxidative stress, inflammatory responses, and apoptotic programs, highlighting its potential as a therapeutic target in diabetic retinal vascular pathology.

METs are internalized by endothelial cells via endocytosis and mediate mitochondrial dysfunction and anoikis

To elucidate how endothelial cells recognize and internalize METs, we performed immunofluorescence assays and observed that METs treatment led to intracellular and membrane-associated accumulation of MPO-positive structures. This was accompanied by F-actin disorganization and stress fiber disruption, both of which were significantly alleviated by the endocytosis inhibitor Dynasore (Fig. 5A), suggesting that METs are taken up via endocytic mechanisms and trigger cellular stress responses. Furthermore, subcellular localization analysis revealed that METs promoted membrane enrichment of S1PR2, a key receptor in downstream inflammatory signaling, whereas Dynasore treatment abrogated this translocation (Fig. 5B). At the functional level, Dynasore markedly reversed METs-induced endothelial barrier dysfunction, as evidenced by reduced permeability and restored NO production (Fig. 5C-D). In parallel, ROS accumulation was suppressed and SOD antioxidant activity was enhanced (Fig. 5E-F). Dynasore also reduced the expression of pro-inflammatory cytokines TNF-α and IL-6, inhibited cytochrome c release, and decreased sFas levels (Fig. 5G-H), collectively indicating its anti-inflammatory and anti-apoptotic effects.

Fig. 5.

Dynasore alleviates METs-induced mitochondrial dysfunction and endothelial injury. (A) Immunofluorescence staining of MPO (green) and phalloidin (red) in hRMECs to evaluate the uptake of METs and cytoskeletal integrity. (B) Immunofluorescence staining of S1PR2 (green) in hRMECs with nuclear counterstaining (DAPI, blue). (C-D) Quantification of vascular permeability (C) and NO levels (D) in conditioned hRMECs. (E) Flow cytometry analysis of intracellular ROS levels using ROS-FCS probe. (F-G) Quantification of SOD activity (F) and inflammatory cytokine expression (TNF-α, IL-6) by ELISA (G). (H) Quantification of Cyt-C release and sFasL levels in hRMECs. (I) Evaluation of mitochondrial integrity and angiogenic function in hRMECs. (J) Quantification of mitochondrial membrane potential using MitoTracker Red staining; (K) Classification of mitochondrial morphology (intermediate vs fragmented) based on MitoTracker Green; (L) Measurement of mitochondrial length by transmission electron microscopy (TEM); (M) Capillary-like structure formation assessed by tube formation assay on Matrigel. (N) Western blot analysis of cleaved caspase-3, cleaved-PARP, p-FAK, FAK, Bax, Bcl-2, and β-actin in hRMECs

With respect to mitochondrial homeostasis, MitoTracker staining and ultrastructural analysis revealed that Dynasore restored mitochondrial activity, reduced the proportion of fragmented mitochondria, and preserved cristae integrity following METs exposure (Fig. 5I-M). These results suggest that inhibition of METs uptake protects against mitochondrial dysfunction and prevents activation of mitochondrial-dependent apoptotic pathways. Additionally, METs-induced angiogenic activity was suppressed by Dynasore, indicating its functional impact on endothelial remodeling. At the molecular level, Western blot analysis confirmed that Dynasore reversed METs-induced downregulation of phosphorylated FAK, while also reducing the levels of cleaved Caspase-3 and the pro-apoptotic protein Bax. Conversely, it increased the expression of the anti-apoptotic protein Bcl-2 (Fig. 5N). Altogether, our findings demonstrate that endothelial cells internalize METs via endocytosis, which in turn triggers oxidative stress, mitochondrial dysfunction, and anoikis, ultimately contributing to vascular barrier disruption.

Inhibition of the sphingolipid pathway attenuates inflammation and vascular dysfunction in vivo

To further validate the pathogenic role of sphingolipid signaling in DR, we evaluated the therapeutic efficacy of the SPHK1/2 inhibitor SKI-II in both the STZ-induced DR model and the intravitreal METs injection model. SKI-II administration did not affect body weight or blood glucose levels, yet robustly restored ERG b-wave amplitudes and markedly reduced vascular leakage and retinal edema in both models (Fig. 6A-F; Fig. S2A-D).

Fig. 6.

SKI-II treatment attenuates retinal vascular dysfunction, inflammation, and anoikis in DR mouse model. (A-B) Longitudinal monitoring of body weight (A) and fasting blood glucose levels (B) in Sham, DR, and DR+SKI-II groups from 8 to 24 weeks. (C-D) ERG waveform recording (C) and quantification of B-wave amplitude (D). (E) Quantification of retinal vascular permeability. (F) Representative H&E staining of retinal cross-sections (left) and quantification of retinal vascular lumen area (right). (G) Levels of serum MPO–DNA complexes and retinal METs markers (citH3, MPO) assessed by ELISA. (H) Measurement of SOD activity. (I) ELISA quantification of serum pro-inflammatory cytokines TNF-α and IL-6. (J) Serum levels of sFas and sFasL detected by ELISA. (K) mRNA expression levels of IL-6, IL-1β, TNF-α, CCL2, and VEGF in retinal tissues, analyzed by qPCR. (L) Western blot detection of cleaved caspase-3, full-length caspase-3, cleaved PARP, p-FAK, FAK, Bax, Bcl-2, and β-actin in retinal tissues. (M) Quantification of apoptosis- and FAK pathway-related proteins from (L). (N) Western blot analysis and quantification of SPHK1, SPHK2, S1PR2, S1PR3, p-p65, and p65 in retinal tissue. (O) Western blot analysis and quantification of VEGF, VEGFR2, ET-1, and Integrin β1 expression in retina

At the inflammatory and oxidative levels, SKI-II diminished MPO–DNA complex accumulation, lowered circulating METs, restored SOD activity, and suppressed both systemic and retinal cytokine production (TNF-α, IL-6, IL-1β, CCL2) in the STZ and METs models alike (Fig. 6G-K; Fig. S2E-I). This striking cross-model concordance indicates that sphingolipid blockade not only restrains METs formation but also mitigates the downstream inflammatory injury they provoke.

At the molecular level, SKI-II consistently attenuated anoikis—evidenced by reduced cleaved caspase-3, cleaved PARP, and Bax—and restored the expression of p-FAK and Bcl-2 (Fig. 6L-M; Fig. S2J), suggesting that sphingolipid signaling governs METs-induced anoikis. Notably, SKI-II reversed the upregulation of SPHK1/2, S1PR2/3, and NF-κB p65 induced by either METs or hyperglycemia, and simultaneously suppressed key vascular-functional proteins, including VEGFR2, ET-1, and integrin β1, across both models (Fig. 6N-O; Fig. S2K-L). Taken together, the parallel rescue effects observed in these mechanistically distinct DR models converge on a unifying pathogenic axis: METs activate the sphingolipid–S1PR–NF-κB signaling cascade to drive endothelial dysfunction and anoikis, whereas SKI-II effectively interrupts this METs-dependent pathological circuitry, thereby establishing the sphingolipid–S1PR axis as a central therapeutic target in DR.

Discussion

DR is a leading cause of blindness in individuals with diabetes, characterized by vascular barrier dysfunction, pathological neovascularization, and chronic inflammatory activation. Although anti-VEGF therapy has shown clinical efficacy in a subset of patients, its overall effectiveness is constrained by high recurrence rates and marked inter-individual variability in therapeutic response. These limitations underscore the urgent need to explore the immuno-inflammatory underpinnings of DR and identify more precise and effective intervention targets. In this study, we systematically demonstrated the pathogenic role of METs in PDR, revealing a previously unrecognized mechanism by which METs impair vascular integrity and amplify retinal inflammation. Our findings show that METs can be actively internalized by HRMECs, subsequently suppressing the FAK/AKT signaling pathway and triggering anoikis, a form of detachment-induced apoptosis. Additionally, METs further exacerbate cellular injury through activation of the S1P–S1PR–NF-κB pathway, ultimately leading to mitochondrial dysfunction and BRB disruption (Fig. 7). The reversal of these effects by pharmacological blockade of S1P synthesis or S1PR signaling highlights the potential of targeting the “METs–Integrin–S1P–FAK” axis in treating PDR.

Fig. 7.

Upon intravitreal injection, METs are endocytosed by hRMECs, leading to cytoskeletal disorganization and mitochondrial dysfunction. Mechanistically, METs inhibit the FAK/AKT pathway while activating the S1P–S1PR2 signaling axis, thereby promoting oxidative stress and anoikis, a form of detachment-induced apoptosis. These cellular events culminate in increased vascular permeability, inflammation, and impaired angiogenesis, which are hallmarks of PDR

Previous studies have extensively explored the role of NETs in chronic inflammatory diseases such as diabetes, atherosclerosis and pulmonary fibrosis [16–18]. However, little is known about the function of METs in diabetic microvascular complications. Our transcriptomic analysis and animal experiments revealed that intravitreal injection of METs induced retinal vascular injury, upregulation of proinflammatory cytokines, and functional decline, mimicking PDR phenotypes without affecting blood glucose or body weight. These observations suggest that METs function as a novel damage-associated molecular pattern (DAMP) that can independently promote DR progression, irrespective of hyperglycemia. More importantly, our study is the first to identify activation of the anoikis signaling pathway in retinal endothelial cells under diabetic conditions, providing a new perspective on endothelial stress responses in retinal microangiopathy.

Anoikis is a form of programmed cell death triggered by the detachment of cells from the ECM and plays a vital role in various physiological and pathological processes, including vascular development, tumor metastasis, and chronic inflammation [19, 20]. In the retinal microvascular system, ECM-dependent adhesion is essential for maintaining endothelial cell polarity, barrier integrity, and survival [21]. Loss of ECM attachment initiates intracellular signaling events involving FAK, Src, and RhoA, which ultimately activate caspase-dependent apoptotic cascades [22–24]. In this study, we provide the first evidence that METs induce anoikis through multiple converging pathways. Morphologically, METs stimulation disrupted cytoskeletal organization, as evidenced by F-actin fragmentation and impaired cellular architecture, thereby compromising cell-ECM adhesion. At the molecular level, METs markedly reduced the expression of phosphorylated FAK, while increasing cleaved Caspase-3 levels, promoting the release of cytochrome c, and enhancing ROS accumulation, all of which are hallmarks of anoikis. Importantly, METs-induced anoikis appears to be driven by cooperative signaling pathways. In addition to suppression of the FAK/AKT signaling pathway, lipid metabolism–related signals, particularly the S1P–S1PR2 pathway, were also involved in regulating this form of cell death. These findings suggest that METs trigger anoikis via combined mechanical stress–driven cytoskeletal disruption and metabolic stress–induced inflammatory signaling, highlighting a multifaceted regulatory framework in diabetic retinal microvascular injury.

The involvement of the SPHK–S1P–S1PR axis in METs-induced endothelial injury further supports the emerging view that lipid metabolism and innate immune signaling are closely intertwined. Sphingosine-1-phosphate (S1P), a bioactive lipid generated by SPHK1 and SPHK2, has been widely implicated in the regulation of vascular permeability, leukocyte adhesion and migration, and pro-inflammatory cytokine release [25–28]. For instance, recent evidence suggests that S1PR2 promotes mitochondrial fission and activates the NLRP3 inflammasome, thereby exacerbating cardiac inflammation and pyroptotic cell death in ischemia–reperfusion injury [29]. Moreover, S1P derived from exosomes has been reported to induce angiogenesis through the S1PR1–AKT–FN1 axis, contributing to diabetic wound healing, which expands the functional spectrum of S1P in vascular biology [30].

Despite the well-documented involvement of S1P signaling in multiple diseases, its specific association with METs and PDR has remained unexplored. In the present study, we systematically characterized the activation of the SPHK1/2–S1P–S1PR2 signaling cascade under the pathological influence of METs. This included upregulation of SPHK1/2, enhanced membrane localization of S1PR2, and activation of the NF-κB pathway. More importantly, we demonstrated that this pathway activation is dependent on the structural integrity of METs. Enzymatic degradation of METs with DNase I effectively inhibited signaling activation and downstream cellular damage, suggesting that the DNA scaffold of METs functions as a pattern recognition ligand. Upon cellular uptake, this scaffold initiates a cascade of metabolic and inflammatory responses. This mechanism is highly consistent with the recently proposed concept of extracellular DNA-driven immunometabolic reprogramming [31, 32] and implies that METs, in addition to acting as structural immune stressors, may also serve as functional modulators of lipid signaling. These findings provide new theoretical insights and potential intervention targets for understanding immune–metabolic crosstalk during PDR progression.

Mechanistically, we provide compelling evidence that endocytosis is the primary route through which METs exert their deleterious effects. METs were taken up by HRMECs and formed MPO-positive aggregates in the cytoplasm, accompanied by cytoskeletal disruption, mitochondrial fragmentation, and ROS accumulation. These changes were significantly mitigated by the endocytosis inhibitor Dynasore, which not only restored FAK activation but also reduced sFas expression. This finding refines our understanding of how extracellular traps interact with vascular cells. Unlike NETs, which have been reported to signal primarily through surface Toll-like receptors [33, 34], METs appear to exert their cytotoxic effects predominantly via internalization, providing a mechanistic explanation for their pronounced capacity to induce vascular dysfunction.

Notably, our study integrates structural immune stimuli, signaling pathway activation, metabolic reprogramming, and cellular outcome into a unified model. We delineated a complete cascade from METs internalization to FAK/AKT inhibition, lipid metabolic amplification via the SPHK/S1PR axis, transcriptional regulation through NF-κB, and execution of cell death via anoikis. Furthermore, we emphasized the central role of mitochondria as a hub for METs-induced stress, characterized by ROS production, cytochrome c release, and disruption of mitochondrial morphology. These findings highlight the importance of mitochondrial homeostasis in endothelial fate decisions under chronic inflammatory conditions. Our in vivo data further support the therapeutic potential of targeting the SPHK–S1P–S1PR axis. Pharmacological inhibition with SKI-II or FTY720 not only ameliorated METs-induced endothelial dysfunction at the cellular level but also restored retinal vascular integrity, reduced oxidative stress and pro-apoptotic markers, and improved retinal electrophysiological function in mouse models. These results provide preclinical evidence that this signaling pathway may serve as an actionable target for immunometabolic intervention in PDR.

Despite the strengths of this study, several limitations should be acknowledged. Although pharmacological modulation of the S1P axis with SKI-II or FTY720 ameliorated METs-induced vascular leakage, inflammatory activation, oxidative stress, and apoptotic signaling, these inhibitors have known off-target effects—including alterations of broader lipid metabolites (SKI-II) [35], systemic immunosuppression (FTY720) [36], and inhibition of dynamin-dependent endocytosis (Dynasore) [37]. Moreover, translating systemic S1P modulators into intravitreal therapies poses challenges related to safety, dosing, and pharmacokinetics, even though SKI-II did not affect body weight or glucose levels in our mouse models (Fig. 6A-B). Thus, the therapeutic relevance of this pathway should be interpreted cautiously, and future studies using genetic models or more selective agents will be required to validate pathway specificity.

Importantly, the METs–S1P–FAK axis identified here complements rather than replaces existing VEGF-targeted therapies. This pathway may represent an additional pathological layer that operates alongside VEGF-dependent mechanisms, offering potential opportunities for combination strategies or for treating patient subgroups with suboptimal responses to anti-VEGF agents. Future work leveraging human PDR tissues and clinically relevant ocular delivery systems will be essential for determining the translational feasibility of targeting METs-associated immunometabolic signaling.

Conclusion

In summary, our findings support the role of METs as pathogenic contributors to PDR progression through two major mechanisms: the induction of endothelial anoikis via suppression of the FAK/AKT pathway and the amplification of inflammatory signaling through the S1P–NF-κB axis. By delineating the METs–FAK/AKT and METs–S1P–NF-κB pathways, this study provides a model that suggests how METs-driven immunometabolic signaling may influence retinal vascular dysfunction. Although these conclusions are based on in vitro and mouse models, they highlight the potential of targeting METs-triggered lipid signaling as a therapeutic strategy. These results also offer complementary insight to existing VEGF-centered paradigms, indicating that METs-associated pathways may represent an additional layer of regulation in diabetic retinal disease.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Ying Song, Xin Liao, Hui Li, and Min Li conceived, designed and supervised the whole study; Ying Song operated the experiment; Xin Liao performed the analyses and audited the data; Ying Song, Le Feng, Guodong Liu, and Fang Wang wrote the manuscript; Xin Liao, Hui Li, and Min Li revised the manuscript. All authors provided critical comments and approved the final manuscript.

Funding

This study was funded by the National Natural Science Foundation of China (81500727), the Clinical Research Center of Shanghai Tenth People’s Hospital (YNCR2C004), and Program for Research-oriented Physician of Shanghai Tenth People’s Hospital (2023YJXYSB009). All funding was acquired by Min Li.

Data availability

The data analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All experiments were approved by the Institutional Animal Care and Use Committee of Shanghai Tenth People’s Hospital, School of Medicine, Tongji University.

Consent for publication

Not applicable.

Competing interests

The authors report no declarations of interest.