Neutrophil extracellular traps exacerbate blood-brain barrier disruption in neuromyelitis optica spectrum disorder

Abstract

Neuromyelitis optica spectrum disorder (NMOSD) is an autoimmune inflammatory disease of the central nervous system (CNS) characterized by recurrent optic neuritis and myelitis. Blood-brain barrier (BBB) disruption is a key pathological feature of NMOSD, but the role of neutrophil extracellular traps (NETs) in this process remains poorly defined. Serum proteomic profiling in patients with acute NMOSD revealed enrichment of the NET formation pathway. Neutrophils from these patients showed an increased propensity for NETosis, accompanied by elevated myeloperoxidase-DNA complexes, neutrophil elastase-DNA complexes, and cell-free DNA in serum, as well as increased neutrophil elastase-DNA complex levels in cerebrospinal fluid. In a murine NMOSD model, neutrophil depletion, peptidylarginine deiminase 4 inhibition (Cl-amidine), and NET degradation (DNase I) each attenuated astrocyte loss, demyelination, neuroinflammation, and BBB disruption. In vitro, patient-derived NETs downregulated tight junction proteins in bEnd.3 endothelial cells through mechanisms associated with myeloperoxidase and neutrophil elastase, while c-Jun N-terminal kinase inhibition mitigated NET-induced endothelial barrier disruption. Together, these findings suggest that aberrant NETosis may contribute to BBB disruption in acute NMOSD and support further investigation of NET-associated pathways as potential therapeutic targets.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12974-026-03959-0.

Introduction

Neuromyelitis optica spectrum disorder (NMOSD) is a severe autoimmune disorder of the central nervous system (CNS) characterized by recurrent episodes of optic neuritis and myelitis [1]. Approximately 70–80% of patients with NMOSD harbor serum autoantibodies against aquaporin-4 (AQP4-IgG) [1]. Binding of AQP4-IgG to AQP4 on astrocytes triggers complement-dependent cytotoxicity and antibody-dependent cellular cytotoxicity, resulting in astrocyte injury [2–5]. This primary insult is accompanied by infiltration of monocytes/macrophages, neutrophils, eosinophils, and lymphocytes, followed by secondary oligodendrocyte injury and demyelination [6].

Blood-brain barrier (BBB) disruption is a critical early event in NMOSD pathogenesis [7–9], facilitating the entry of circulating AQP4-IgG into the CNS. Although intrathecal B cells contribute to local AQP4-IgG production [10], most AQP4-IgG-producing cells reside in peripheral compartments [11, 12]. The severity of BBB impairment, reflected by an elevated cerebrospinal fluid (CSF)-to-serum albumin ratio, correlates with clinical disease severity in NMOSD [13, 14]. Given the essential role of astrocytes in maintaining BBB integrity [15], astrocyte loss and dysfunction in NMOSD may further exacerbate BBB compromise. Several AQP4-IgG-independent serum factors have also been implicated in BBB disruption [16], including interleukin-6 [9] and autoantibodies against glucose-regulated protein 78 [17], yet the mechanisms that drive and amplify BBB disruption remain incompletely understood.

Neutrophils, the most abundant peripheral leukocytes, are rapidly recruited to sites of inflammation. Growing evidence implicates neutrophil dysfunction in NMOSD pathogenesis: elevated peripheral neutrophil counts correlate with disease activity [18, 19], and pharmacological inhibition of neutrophil elastase (NE) ameliorates brain damage in a murine model of NMOSD [20]. Prior studies have also implicated neutrophil-microglia interactions and polymorphonuclear neutrophil-dependent astrocyte loss in NMOSD pathology [21, 22]. Recent studies have further identified intrinsic neutrophil alterations in NMOSD, including impaired apoptotic responses and prolonged cell survival after stimulation in patients during clinical remission [23]. However, the extent of systemic neutrophil extracellular trap (NET) formation during the acute phase of NMOSD remains poorly characterized, and whether NETs contribute to BBB disruption and tissue injury remains unclear.

Histopathological studies have identified NET-associated markers, including citrullinated histone H3 (H3cit) and myeloperoxidase (MPO), within NMOSD lesions [24], indicating that NETosis occurs during the disease process. NETs are extracellular DNA scaffolds decorated with citrullinated histones and granular enzymes such as NE and MPO; cell-free DNA (cfDNA), MPO-DNA complexes, and NE-DNA complexes are commonly used as NET-associated biomarkers [25]. In NMOSD, NET-associated components decrease after plasma exchange in association with clinical improvement [26], while neutrophil-derived cfDNA induces a type I interferon signature [27] and correlates with neurofilament light chain levels in AQP4-IgG-seropositive patients [23]. These findings link NET-associated processes to immune dysregulation and neuroaxonal injury, but their role in BBB disruption remains insufficiently defined. Therefore, we tested the hypothesis that systemic NET formation is aberrantly enhanced during acute NMOSD attacks and that NETs may amplify BBB disruption by directly impairing endothelial tight junction integrity. We further hypothesized that pharmacological inhibition of NETosis or direct degradation of extracellular NETs would mitigate BBB disruption and tissue damage.

Methods

A detailed description of the experimental procedures is provided in the Supplementary Methods.

Patient population

We retrospectively analyzed clinical data from AQP4-IgG-positive NMOSD patients admitted to Tangdu Hospital between June 1 and September 1, 2024. All patients met the 2015 diagnostic criteria for NMOSD [28]. Serum AQP4-IgG was detected using a cell-based assay. Three groups were included: 25 acute-phase patients, 21 remission-phase patients, and 50 healthy controls (HCs). The acute phase was defined as a new or worsening neurological episode consistent with NMOSD relapse, with sample collection performed within 30 days of symptom onset and before systemic corticosteroids, immunosuppressive agents, intravenous immunoglobulin, plasma exchange, or other relapse-directed therapy. Acute attacks were supported by objective neurological deterioration on examination and/or active MRI lesions. Remission was defined as the absence of new or worsening neurological symptoms for at least 2 months before sampling. HCs had no history of autoimmune disease, malignancy, active infection, or other chronic medical conditions, and none were receiving immunosuppressive or immunomodulatory medications at enrollment. Clinical assessments included Expanded Disability Status Scale (EDSS) scoring and complete blood count parameters: white blood cell count, absolute neutrophil and lymphocyte counts, neutrophil and lymphocyte percentages, and neutrophil-to-lymphocyte ratio (NLR).

Partially overlapping patient subsets were used because each assay required different biospecimen types, sample volumes, or processing conditions. Complete blood count analyses included all eligible participants with available laboratory data, whereas proteomic profiling, serum/CSF NET-associated biomarker measurements, and ex vivo NETosis assays were restricted to participants with suitable stored serum/CSF samples or fresh blood for immediate neutrophil isolation. Detailed characteristics of each assay-specific subset are provided in Supplementary Tables 1–6.

This study was approved by the Research Ethics Committee of Tangdu Hospital, Fourth Military Medical University (No. K202410-25), and written informed consent was obtained from all participants.

Induction of NMOSD pathology

All animal experiments were approved by the Animal Ethics Committee of the Fourth Military Medical University and conducted in accordance with NIH guidelines for the care and use of laboratory animals. Female C57BL/6 mice (8–12 weeks old, weight-matched) were housed under specific pathogen-free conditions prior to surgery. Mice were anesthetized with 2.5% tribromoethanol solution (Nanjing Aibei Biotechnology, China; Cat. #M2910) and immobilized in a stereotaxic frame (RWD Life Science). A midline scalp incision was made to expose the bregma, and a 1-mm burr hole was drilled 2.2 mm lateral to the bregma. A 33-gauge Hamilton needle connected to a gas-tight syringe was inserted to a depth of 3 mm, and a mixture of 4.8 µL NMOSD-IgG and 3.2 µL human complement (Innovative Research, USA; Cat. #IPLA-CSER) was infused at a rate of 0.5 µL/min, to deliver a total volume of 8 µL. The needle was then carefully withdrawn, and the incision was sutured with 4 − 0 nylon sutures.

Cell culture and treatment

bEnd.3 mouse brain-derived endothelial cells at passages 15–25 were cultured in Dulbecco’s Modified Eagle Medium containing 10% fetal bovine serum at 37 °C in a humidified 5% CO₂ atmosphere. At 80% confluence, cells were treated with NETs alone or in combination with Alvelestat (MedChemExpress, China; Cat. #HY-15651), PF-1355 (MedChemExpress, China; Cat. #HY-100873), or SP600125 (MedChemExpress, China; Cat. #HY-12041).

Statistical analysis

Statistical analyses wereperformed using GraphPad Prism 8.4.1 or R v4.2.0. Continuous data are presented as mean ± standard deviation (SD) for normally distributed data or median with interquartile range (IQR) for non-normally distributed data. Normality was assessed using the Shapiro–Wilk test. Normally distributed data were analyzed using unpaired Student’s t-test for two-group comparisons or one-way ANOVA with Tukey’s post hoc test for multiple-group comparisons. Non-normally distributed data were analyzed using the Mann–Whitney U test for two-group comparisons or the Kruskal–Wallis test followed by Dunn’s post hoc test for multiple-group comparisons. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate. Correlations were assessed using Pearson’s correlation coefficient or Spearman’s rank correlation coefficient for normally and non-normally distributed variables, respectively. Post hoc multiple-comparisons correction was applied within individual multi-group analyses, but no single global correction was applied across all experimental modules. A two-tailed P value of less than 0.05 was considered statistically significant.

Results

Systemic and intrathecal NET activation in NMOSD patients

Serum proteomic analysis was performed using data-independent acquisition (DIA)-based label-free quantitative proteomics on samples from 7 acute NMOSD patients and 5 age- and sex-matched HCs (Supplementary Table 1). Principal component analysis (PCA) showed distinct clustering between the two groups (Fig. 1A). Volcano plot analysis identified significantly upregulated and downregulated proteins using thresholds of fold change ≥ 1.5 or ≤ 0.67 and Benjamini-Hochberg adjusted P < 0.05 (Fig. 1B). Several NET formation pathway-related proteins showed an overall upward trend and are highlighted in Fig. 1B, including high‑mobility group protein B1 (HMGB1) [29], cathepsin G (CTSG) [30], phosphoinositide-3-kinase regulatory subunit 1 (PIK3R1) [31], complement 5 (C5) [32], peptidylarginine deiminase 4 (PAD4) [33], and Rac2 [34]. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis ranked the NET formation pathway as the second most significantly enriched pathway (Fig. 1C), and hierarchical clustering showed increased abundance of NET formation-associated proteins in NMOSD patients (Fig. 1D).

Fig. 1.

Serum proteomic profiling and peripheral neutrophil expansion in acute NMOSD. A PCA of serum proteomes from 7 acute NMOSD patients and 5 HCs using DIA quantitative proteomics. B Volcano plot of differentially expressed proteins using thresholds of fold change ≥ 1.5 or ≤ 0.67 and Benjamini–Hochberg adjusted P < 0.05. Representative NET formation pathway-related proteins are labeled. C Top 10 enriched KEGG pathways among differentially expressed proteins in NMOSD patients. The NET formation pathway is highlighted. D Hierarchical clustering heatmap of NET formation pathway-related proteins. The color scale represents z-score-normalized abundance (blue, low; pink, high). E Neutrophil counts in HCs (n = 50), acute NMOSD patients (n = 25), and remission-phase NMOSD patients (n = 21). F Neutrophil percentages in HCs (n = 50), acute NMOSD patients (n = 25), and remission-phase NMOSD patients (n = 21). G NLR in HCs (n = 50), acute NMOSD patients (n = 25), and remission-phase NMOSD patients (n = 21). H Correlation between EDSS scores and neutrophil counts in acute NMOSD patients (n = 25). I Correlation between EDSS scores and NLR in acute NMOSD patients (n = 25). Data are presented as median and IQR. Statistics: Kruskal–Wallis test followed by Dunn’s post hoc test (E-G) or Spearman’s rank correlation analysis (H, I). *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001

Peripheral blood analysis revealed significantly elevated absolute neutrophil counts in acute NMOSD patients compared with HCs and remission-phase patients, whereas neutrophil percentages were increased in acute NMOSD patients compared with HCs (Fig. 1E and F). NLR was also higher in acute NMOSD patients than in HCs (Fig. 1G). Total leukocyte counts were increased in acute NMOSD patients relative to patients in remission (Supplementary Fig. 1A), whereas lymphocyte counts and percentages were lower in acute NMOSD patients than in HCs (Supplementary Figs. 1B, 1 C). Both neutrophil counts and NLR positively correlated with EDSS scores in acute NMOSD patients (Fig. 1H and I). Demographic and clinical characteristics of HCs and NMOSD patients included in the complete blood count analysis are detailed in Supplementary Table 2.

We next quantified NET-associated biomarkers in serum and CSF. Serum MPO-DNA and NE-DNA complexes were elevated in acute NMOSD patients compared with HCs (Fig. 2A and B). Serum cfDNA was significantly elevated in acute NMOSD patients compared with HCs, with a similar upward trend in remission-phase patients (Fig. 2C). In CSF, NE-DNA complex levels were significantly higher in acute NMOSD patients than in HCs (Fig. 2E), whereas MPO-DNA and cfDNA levels did not differ significantly between groups (Fig. 2D and F). Clinical characteristics of patients and HCs included in the NET-associated biomarker analysis are provided in Supplementary Table 3.

Fig. 2.

Elevated NET-associated biomarkers and enhanced spontaneous NET formation in NMOSD patient samples. A Serum MPO-DNA complex levels in HCs (n = 10), acute NMOSD patients (n = 18), and remission-phase NMOSD patients (n = 15). B Serum NE-DNA complex levels in HCs (n = 10), acute NMOSD patients (n = 18), and remission-phase NMOSD patients (n = 15). C Serum cfDNA levels in HCs (n = 10), acute NMOSD patients (n = 18), and remission-phase NMOSD patients (n = 15). D CSF MPO-DNA complex levels in HCs (n = 10) and acute NMOSD patients (n = 18). E CSF NE-DNA complex levels in HCs (n = 10) and acute NMOSD patients (n = 18). F CSF cfDNA levels in HCs (n = 10) and acute NMOSD patients (n = 18). G Representative immunofluorescence images showing H3cit (green), MPO (magenta), and DAPI (blue) staining in neutrophils from HCs and acute NMOSD patients at 2 h and 6 h. H Quantification of MPO and H3cit fluorescence intensities (n = 4 per group). Data are presented as median and IQR (A–F) or mean ± SD (H). Statistics: Kruskal–Wallis test followed by Dunn’s post hoc test (A–C), Mann–Whitney U test (D–F), or unpaired Student’s t-test for comparisons between HCs and acute NMOSD patients at each time point (H). *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001. Scale bar: 20 μm (G)

To directly assess NETosis capacity, peripheral neutrophils isolated from acute NMOSD patients and HCs were cultured ex vivo and stained for MPO and H3cit. Neutrophils from NMOSD patients showed significantly increased MPO and H3cit fluorescence intensities at 2 h, with H3cit remaining elevated at 6 h (Fig. 2G and H), supporting an enhanced propensity for spontaneous NET formation in neutrophils from acute NMOSD patients.

Neutrophil depletion, NETosis suppression, or NET degradation attenuates neuropathology and improves functional outcomes in NMOSD

Consistent with observations in human NMOSD specimens [24], MPO/H3cit/DAPI-colocalized NET-like structures were detected in the corpus callosum lesions of NMOSD mice at 4 days post-induction, whereas sham-treated mice showed minimal MPO/H3cit positivity (Figs. 3A-C). High-magnification confocal analysis, including orthogonal views and 3D reconstruction, showed H3cit-positive DNA extrusion from neutrophil-associated structures (Fig. 3C). Dual immunofluorescence staining revealed MPO or H3cit immunoreactivity in Ly-6G⁺ neutrophils within NMOSD lesions, consistent with lesion-associated neutrophil NET formation (Fig. 3D). This was further supported by elevated MPO and H3cit levels in lesion tissue homogenates by Western blot (Fig. 3E and F).

Fig. 3.

Neutrophil extracellular traps accumulate in NMOSD mouse lesions. A Representative confocal images of MPO (magenta), H3cit (green), and DAPI (blue) staining in sham and NMOSD mice at 4 days post-NMOSD induction. NET-like structures were identified by colocalization of MPO, H3cit, and DAPI within corpus callosum lesions in NMOSD mice, whereas minimal MPO/H3cit colocalization was observed in sham controls. B Higher-magnification view of the boxed region in (A), showing neutrophils at different stages of NET formation, including early nuclear decondensation with intracellular H3cit (open arrowheads) and extracellular NET release (open arrows). C NET structural analysis of the boxed region in (B). Orthogonal views and single-channel images show colocalization of MPO and H3cit with DAPI-positive DNA, while the 3D reconstruction illustrates H3cit-positive DNA extrusion from a neutrophil-associated structure. D Representative dual immunofluorescence images of Ly-6G (green) with MPO (magenta) or H3cit (magenta) in sham and NMOSD mice. Nuclei were counterstained with DAPI (blue). E Representative immunoblots of MPO and H3cit in lesion lysates. β-actin served as a loading control. F Quantification of relative MPO and H3cit protein expression (n = 5 per group). Data are presented as mean ± SD. Statistics: unpaired Student’s t-test. *P < 0.05 and **P < 0.01. Scale bars: 50 μm (A); 10 μm (B, D)

To determine whether NETs contribute to NMOSD-like neuropathology, we evaluated three interventions: neutrophil depletion via anti-Ly-6G antibody, inhibition of NETosis with the PAD4 inhibitor Cl-amidine, and NET degradation with DNase I. Circulating neutrophils were depleted via anti-Ly-6G administration (Supplementary Figs. 2 A, 2B). Compared with untreated NMOSD mice, anti-Ly-6G, Cl-amidine, and DNase I reduced H3cit⁺ signals by approximately 61.8%, 48.5%, and 21.7%, respectively, and MPO⁺ signals by approximately 89.6%, 89.6%, and 83.7%, respectively (Supplementary Figs. 3 A, 3B). These interventions also improved neurological outcomes, including motor performance in the beam walking tests (Figs. 4A-D) and recognition memory in the novel object recognition test (Fig. 4E). Histological analysis showed reduced demyelination (Fig. 4F and G) and attenuated loss of AQP4 and glial fibrillary acidic protein (GFAP), with reduced degraded myelin basic protein (dMBP)-associated pathology most evident after anti-Ly-6G and Cl-amidine treatment (Fig. 4H and I), compared with untreated NMOSD mice.

Fig. 4.

Neutrophil depletion, NET formation inhibition, or NET degradation attenuates NMOSD lesion progression. A, B Beam walking test on a 1.2 cm × 80 cm beam: (A) number of paw slips and (B) crossing time (n = 8–10 per group). C, D Beam walking test on a 0.6 cm × 80 cm beam: (C) number of paw slips and (D) crossing time (n = 7–10 per group). E Discrimination index in the novel object recognition test (n = 4–5 per group). F Representative Luxol fast blue (LFB) staining images showing demyelination. G Quantification of LFB staining loss area (n = 4–5 per group). H Representative immunofluorescence images of AQP4 (magenta), GFAP (green), and dMBP (magenta). I Quantification of the loss of AQP4, GFAP, and dMBP immunoreactivity (n = 4 per group). Higher values indicate greater pathological loss. J Representative immunofluorescence images of Iba1 (green). K Quantification of Iba1-positive area (n = 4 per group). L Representative immunofluorescence images of CD3 (green). Nuclei were counterstained with DAPI (blue). Lower panels show higher-magnification views of the boxed regions. M Quantification of CD3⁺ T cells (n = 5 per group). Data are presented as mean ± SD. Statistics: Kruskal–Wallis test followed by Dunn’s post hoc test (A) or one-way ANOVA with Tukey’s post hoc test (B-E, G, I, K, M). *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001. Scale bars: 500 μm (F, H, J); 50 μm (L, upper panels); 10 μm (L, lower panels)

All three interventions also reduced neuroinflammation, as evidenced by decreased ionized calcium binding adaptor molecule 1⁺ (Iba1⁺) macrophage/microglia immunoreactive area (Fig. 4J and K) and fewer CD3⁺ T cells (Fig. 4L and M). Representative hematoxylin and eosin (H&E)-stained sections also showed less inflammatory infiltration after these interventions (Supplementary Fig. 4). The parallel protective effects of neutrophil depletion, NETosis inhibition, and NET degradation indicate that neutrophil- and NET-associated mechanisms contribute to astrocyte loss, demyelination, and neuroinflammation in NMOSD.

Neutrophil depletion, NETosis inhibition, or NET degradation mitigates BBB damage following NMOSD induction

NETs have been implicated in BBB disruption in multiple sclerosis [35]. We therefore assessed the effects of neutrophil depletion, NETosis inhibition, and NET degradation on BBB integrity in the NMOSD mouse model. Immunofluorescence staining revealed significant downregulation of tight junction proteins, including occludin, claudin-3, and claudin-5, in NMOSD mice compared with sham controls (Fig. 5A and B). All three interventions mitigated BBB impairment, with significant upregulation of claudin-5 and occludin; claudin-3 showed an upward trend after anti-Ly-6G, Cl-amidine, and DNase I treatment, although these differences did not reach statistical significance (NMOSD vs. NMOSD + Ly-6G, P = 0.1271; NMOSD vs. NMOSD + Cl-amidine, P = 0.0933; NMOSD vs. NMOSD+DNase I, P = 0.2801) (Fig. 5A and B). Consistent with these findings, increased IgG extravasation in NMOSD mice, reflecting impaired BBB integrity, was attenuated by each treatment strategy (Fig. 5C and D). Together, these results support NETs as important mediators of BBB disruption in NMOSD.

Fig. 5.

Neutrophil depletion, NET formation inhibition, or NET degradation attenuates blood–brain barrier disruption in NMOSD mice. A Representative immunofluorescence images of tight junction proteins, including claudin-5 (magenta), claudin-3 (green), and occludin (magenta), in brain sections. B Quantification of fluorescence intensity for claudin-5 (n = 4 per group), claudin-3 (n = 4 per group), and occludin (n = 4–5 per group). C Representative immunofluorescence images of IgG extravasation (magenta). D Quantification of IgG fluorescence intensity (n = 4 per group). Data are presented as mean ± SD. Statistics: one-way ANOVA with Tukey’s post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001. Scale bars: 50 μm (A, C)

NMOSD patient-derived NETs compromise endothelial barrier integrity in vitro

To investigate the effects of NETs on endothelial barrier function, NETs were isolated from neutrophils of acute NMOSD patients after stimulation with phorbol 12-myristate 13-acetate (PMA), a well-established inducer of NETosis [36]. Compared with unstimulated NMOSD neutrophils, PMA-stimulated neutrophils exhibited increased MPO and H3cit signals, confirming successful NETosis induction (Fig. 6A and B).

Fig. 6.

NETs regulate endothelial function in bEnd.3 cells. A Representative immunofluorescence images of MPO (green) and H3cit (magenta) in neutrophils from NMOSD patients under unstimulated conditions or after PMA stimulation. Nuclei were counterstained with DAPI (blue). B Quantification of MPO and H3cit fluorescence intensities (n = 3–4 per group). C Representative images of EdU-labeled cells (green) and DAPI (blue). D Percentage of EdU-positive cells (n = 5 per group). E Representative immunofluorescence images of cleaved caspase-3 (green) and DAPI (blue). F Quantification of cleaved caspase-3 fluorescence intensity (n = 5 per group). G Representative images of endothelial cell migration in the transwell assay. H Quantification of migrated endothelial cells in the transwell assay (n = 6 per group). I Representative immunofluorescence images of claudin-5 (green), occludin (magenta), and ZO1 (green). Nuclei were counterstained with DAPI (blue). J Quantification of claudin-5, occludin, and ZO1 fluorescence intensities (n = 5 per group). K Representative immunoblots of claudin-5, occludin, and ZO1. β-actin served as a loading control. L Quantification of relative claudin-5, occludin, and ZO1 expression (n = 5 per group). M TEER measurements in bEnd.3 cells (n = 5–6 per group). Data are presented as mean ± SD. Statistics: unpaired Student’s t-test (B) or one-way ANOVA with Tukey’s post hoc test (D, F, H, J, L, M). *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001. Scale bars: 20 μm (A, C, E); 50 μm (G, I)

NET treatment significantly inhibited bEnd.3 endothelial cell proliferation and promoted apoptosis; both effects were partially rescued by co-treatment with the MPO inhibitor PF-1355 or the NE inhibitor Alvelestat (Figs. 6C-F). NET exposure also suppressed endothelial cell migration, an effect attenuated by PF-1355 but not by Alvelestat (Fig. 6G and H), suggesting partially distinct roles for MPO and NE in regulating endothelial motility.

Regarding barrier integrity, NET treatment reduced the expression of tight junction proteins, including claudin-5, occludin, and zonula occludens-1 (ZO1), and these changes were attenuated by Alvelestat and PF-1355 (Figs. 6I-L). Consistent with these findings, either inhibitor mitigated the NET-induced reduction in transendothelial electrical resistance (TEER) (Fig. 6M). Collectively, these data indicate that NMOSD patient-derived NETs impair endothelial barrier integrity through MPO- and NE-associated mechanisms, providing in vitro cellular evidence for NET-mediated BBB disruption in NMOSD.

NETs disrupt endothelial barrier integrity through JNK signaling

We then performed RNA sequencing of bEnd.3 cells treated with NETs or control culture medium. PCA showed clear separation between the two groups (Fig. 7A). Comparative analysis identified 5,107 upregulated and 3,447 downregulated genes in NET-treated cells, with representative mitogen-activated protein kinase (MAPK)/c-Jun N-terminal kinase (JNK) signaling-related genes labeled in the volcano plot (Fig. 7B). KEGG enrichment analysis showed significant enrichment of the MAPK signaling pathway among upregulated genes (Fig. 7C), and the upregulated genes within this pathway were further visualized (Fig. 7D).

Fig. 7.

NETs impair endothelial barrier integrity through JNK signaling. A PCA of transcriptomes from control and NET-treated bEnd.3 cells. B Volcano plot of differentially expressed genes, with representative MAPK/JNK signaling-related genes labeled. C Top 10 enriched KEGG pathways among upregulated genes in NET-treated endothelial cells. The MAPK signaling pathway is highlighted. D Upregulated genes involved in the MAPK signaling pathway in NET-treated endothelial cells (expression z-score: blue = -2; pink = 2). E RT-qPCR analysis of MAPK pathway components (Jnk1/2, p38Mapk, and Erk) mRNA expression (n = 10 per group). F Representative immunoblots of JNK, p-JNK, and β-actin. G Quantification of JNK/β-actin, p-JNK/β-actin, and p-JNK/JNK expression ratios (n = 5 per group). H Representative immunoblots of tight junction proteins (claudin-5, occludin, and ZO1). I Quantification of relative claudin-5, occludin, and ZO1 protein expression (n = 5 per group). J Representative immunofluorescence images of tight junction proteins, including claudin-3 (magenta), claudin-5 (green), occludin (magenta), and ZO1 (green). Nuclei were counterstained with DAPI (blue). K Quantification of tight junction fluorescence intensities (n = 4 per group). L TEER measurements in bEnd.3 cells (n = 6 per group). Data are presented as mean ± SD. Statistics: unpaired Student’s t-test was used for Jnk1/2 comparisons in panel E and for JNK/β-actin and p-JNK/β-actin comparisons in panel G; the Mann–Whitney U test was used for p38Mapk and Erk comparisons in panel E and for p-JNK/JNK comparisons in panel G; one-way ANOVA with Tukey’s post hoc test was used for panels I, K, and L. *P < 0.05, **P < 0.01, ***P < 0.001, and #P < 0.0001. Scale bar: 20 μm (J)

Because KEGG enrichment of the MAPK pathway does not necessarily indicate uniform activation of all MAPK branches, we next examined JNK, p38 MAPK, and extracellular signal-regulated kinase (ERK) signaling. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) revealed selective upregulation of Jnk1/2 mRNA, whereas p38Mapk mRNA was downregulated and Erk mRNA remained unchanged (Fig. 7E). We next assessed JNK activation at the protein level. Western blot analysis showed a marked increase in phosphorylated JNK (p-JNK), with no significant change in total JNK protein (Fig. 7F and G), supporting activation of the JNK branch in NET-treated endothelial cells. The JNK inhibitor SP600125 suppressed NET-induced p-JNK activation (Supplementary Figs. 5 A, 5B). SP600125 also partially rescued the expression of tight junction proteins claudin-5, occludin, and ZO1 that had been downregulated by NETs, with the clearest effect observed at 40 µM (Fig. 7H and I). Immunofluorescence staining further showed partial recovery in claudin-3, claudin-5, and ZO1 signals after SP600125 treatment (Fig. 7J and K). In TEER assays, SP600125 significantly mitigated the NET-induced reduction in barrier resistance (Fig. 7L). Collectively, these results support JNK signaling as an important downstream pathway involved in NET-induced endothelial barrier disruption in our in vitro model.

Discussion

In this study, serum proteomic analysis identified upregulation of the NET formation pathway in acute NMOSD, accompanied by elevated NET-associated components in both serum and CSF. In a murine NMOSD model, neutrophil depletion, NETosis inhibition, and NET degradation each attenuated astrocyte loss, demyelination, neuroinflammation, and BBB disruption. Mechanistically, NMOSD patient-derived NETs downregulated tight junction proteins in endothelial cells through mechanisms involving MPO and NE, while JNK inhibition mitigated NET-induced tight junction loss and barrier dysfunction in vitro. Collectively, the convergent protection achieved by three independent NET-targeting strategies in vivo, together with the direct barrier-disrupting effects of isolated NETs in vitro, supports aberrant NETosis as an important amplifier of BBB disruption in NMOSD.

Our findings should be interpreted in relation to Schroeder-Castagno et al. [23], who reported impaired neutrophil cell-death responses in AQP4-IgG-seropositive NMOSD. Critically, that study examined treated patients in stable remission after PMA stimulation, and the impaired response primarily reflected reduced apoptotic rather than reduced NET formation. Our study, by contrast, assessed spontaneous NET formation in treatment-naïve patients during acute attacks. This distinction may also explain why serum MPO-DNA and NE-DNA complexes were both elevated in our acute NMOSD cohort, whereas that study reported no elevation of serum MPO-DNA levels in treated remission-phase patients [23]. Thus, MPO-DNA elevation may be more apparent during acute, untreated disease. In CSF, only NE-DNA complexes were significantly increased. This may reflect the limited CSF sample size, differences in the stability or clearance of MPO-DNA and NE-DNA complexes, or preferential enrichment of NE-associated NET components within the CNS compartment. Given the established capacity of NE to degrade extracellular matrix and tight junction-related proteins [37], elevated CSF NE-DNA may be relevant to BBB disruption. Larger paired serum-CSF cohorts are needed to validate this pattern and define its mechanistic significance.

The triggers of BBB damage in NMOSD remain incompletely characterized. Infiltrating polymorphonuclear leukocytes, predominantly neutrophils, can directly induce BBB disruption [22], yet the mechanisms linking neutrophil activation to endothelial barrier failure have not been established in NMOSD. Given the central role of complement in AQP4-IgG-positive NMOSD and the trend toward of C5 in our proteomic screen, our findings support a model in which NETs may act downstream of, and potentially in parallel with, complement-mediated injury rather than independently of it. Initial access of AQP4-IgG and complement to the CNS may promote local complement activation and neutrophil recruitment, consistent with previous evidence that C5a-preactivated neutrophils exacerbate astrocyte injury [38]. In this setting, NET release by circulating and lesion-recruited neutrophils may contribute to endothelial barrier disruption, with downstream pathological associations summarized in Fig. 8.

Fig. 8.

Proposed mechanism of NET-mediated pathology in NMOSD. Schematic showing the proposed role of NETs in NMOSD-associated BBB disruption. Solid arrows indicate relationships directly supported by this study, including NET-induced endothelial p-JNK activation, tight junction loss, and barrier dysfunction in vitro. Dashed arrows indicate inferred relationships based on the present findings and prior literature, including neutrophil recruitment and the association of BBB disruption with AQP4-IgG entry, inflammatory infiltration, astrocyte damage, demyelination, and neuroinflammation

Building on previous evidence that NE contributes to NMOSD-like injury [20], our in vitro findings link MPO and NE to NET-associated endothelial barrier injury. Inhibiting either enzyme partially restored tight junction protein expression and improved TEER after NET exposure. Their effects were not identical: PF-1355, but not Alvelestat, improved NET-impaired endothelial migration, suggesting that MPO may have a more prominent role in endothelial motility. Transcriptomic profiling and protein analyses further showed selective activation of JNK signaling in NET-treated endothelial cells. Consistently, JNK inhibition partially restored tight junction proteins and mitigated NET-induced barrier dysfunction. Together with previous studies linking JNK signaling to tight junction degradation and BBB permeability [39, 40], these findings suggest that MPO/NE-associated endothelial injury and JNK-dependent tight junction disruption are complementary mechanisms of NET-associated barrier dysfunction. Future in vivo studies should determine whether MPO, NE, or JNK inhibition mitigates BBB disruption within NMOSD lesions.

This study has several limitations. First, the single-center design, short recruitment window from June to September 2024, predominantly Han Chinese cohort, and modest, partially overlapping patient subsets may limit generalizability and statistical power. Although clinically evident infection was excluded, seasonal or subclinical inflammatory effects and tertiary referral bias cannot be fully ruled out. Second, the intracerebral AQP4-IgG/complement model does not recapitulate systemic immune priming in NMOSD. However, the presence of Ly-6G⁺ neutrophils and NET-like structures within lesions, together with protection after systemic anti-Ly-6G, Cl-amidine, or DNase I treatment, suggests that circulating neutrophils/NETs contribute to lesion amplification after local AQP4-IgG/complement-mediated injury. Future systemic models are needed to define how peripheral NETosis is initiated and coupled to CNS lesion development. Third, the role of the JNK signaling pathway in NET-mediated BBB disruption was examined only in vitro and was not directly validated in the NMOSD mouse model. Fourth, the intervention data require cautious interpretation because NETs may not be the only neutrophil-derived mediators. Cl-amidine is a pan-PAD inhibitor rather than a neutrophil-specific agent [41], and PAD enzymes are expressed in multiple immune and non-immune cell types [42]. Translational targeting of NETs also requires caution, given the lack of clear efficacy of recombinant human DNase I in lupus nephritis [43] and the possibility that broad NET inhibition may compromise antimicrobial host defense [44]. Future gain-of-function studies using purified NETs are needed to determine whether NETs are sufficient to exacerbate NMOSD-like injury in vivo. Fifth, the in vitro endothelial assays relied on PMA-induced NETs and bEnd.3 cells, which may not fully reproduce in vivo NET composition or primary human BBB properties. These findings should therefore be interpreted as mechanistic rather than fully physiological BBB evidence.

In conclusion, this study links aberrant NET formation to BBB disruption in acute NMOSD and shows that neutrophil/NET-targeting interventions attenuate BBB damage and lesion pathology in vivo. It also identifies endothelial JNK signaling as a relevant in vitro pathway in NET-induced barrier dysfunction. These findings support further investigation of NET-associated pathways as potential therapeutic targets.

Authors’ contributions

All authors have read and approved the article. Xue Ma was responsible for funding acquisition, methodology, investigation, validation, and writing – original draft. Yao Wang contributed to data curation, investigation, supervision, and writing – original draft. Xin Chen performed data curation and investigation. Zonghao Pan conducted investigation, Methodology, and Writing – original draft. Yi Li was involved in supervision. Kaixi Ren participated in validation. Daidi Zhao assisted in methodology and supervision. Hongzeng Li contributed to conceptualization and project administration. Jun Guo contributed to conceptualization and project administration, and carried out formal analysis, funding acquisition, and writing – review & editing.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82301531, No. 82171339, and No. 82571536), Cross Integration Special Project of AFMU (No. 2024JC029), the Youth Innovation Team of Shaanxi Universities (No. 2023-997-71), and the Shaanxi Province Natural Science Foundation (No. 2025JC-YBQN-1141).

Data availability

The datasets generated and/or analysed during the current study are not publicly available due to ethical restrictions and ongoing further research, but are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.