Neutrophil extracellular traps drive AIM2-mediated endothelial PANoptosis and blood–brain barrier disruption after traumatic brain injury

Abstract

Background

Blood-brain barrier (BBB) disruption is a critical pathological event contributing to secondary injury and poor outcomes following traumatic brain injury (TBI). While neutrophil extracellular traps (NETs) are known to exacerbate neuroinflammation, the precise molecular mechanisms by which they compromise cerebral microvascular integrity remain elusive. This study aimed to investigate the role of the NETs-AIM2 inflammasome axis in mediating endothelial damage and to explore the therapeutic potential of targeting this pathway.

Methods

A controlled cortical impact (CCI) mouse model was utilized to simulate TBI. NET formation was modulated pharmacologically using the peptidylarginine deiminase 4 (PAD4) inhibitor Cl-amidine or genetically via adenoviral overexpression of PAD4. BBB permeability, brain edema, and cerebral blood flow (CBF) were assessed using Evans Blue extravasation, wet-to-dry weight ratio, T2-weighted magnetic resonance imaging, and laser speckle contrast imaging. Neurological function was evaluated via modified Neurological Severity Score (mNSS) and rotarod tests. Molecular mechanisms were dissected using Western blotting, immunofluorescence, and cell-type-specific co-localization analyses.

Results

TBI induced significant PAD4-dependent NET formation, peaking at day 3 post-injury, which temporally coincided with BBB breakdown and neurological deficits. Cl-amidine treatment effectively preserved tight junction proteins (zonula occludens-1, Occludin) and adherens junction protein (vascular endothelial-cadherin), reduced Evans Blue extravasation, attenuated brain edema and lesion volume, promoted CBF restoration, and improved mNSS scores and rotarod performance. Mechanistically, NETs acted as upstream danger signals that specifically triggered AIM2 inflammasome activation, which was predominantly localized in cerebral microvascular endothelial cells rather than glial cells. This endothelial-specific AIM2 activation subsequently initiated PANoptosis—a multifaceted cell death program characterized by simultaneous pyroptosis, necroptosis, and apoptosis—in the vascular endothelium. Pharmacological inhibition of AIM2 effectively suppressed all three death pathways and preserved BBB integrity. Adenoviral overexpression of PAD4 exacerbated endothelial PANoptosis, whereas AIM2 inhibition reversed these effects, establishing the causal relationship.

Conclusions

Our findings demonstrate a pathogenic axis wherein TBI-induced NETs drive endothelial PANoptosis via specific activation of the AIM2 inflammasome in cerebral microvascular endothelial cells, leading to BBB failure. Targeting this signaling cascade represents a promising therapeutic strategy to protect the neurovascular unit and improve functional recovery after TBI.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03123-4.

Plain language summary

Traumatic brain injury is a leading cause of death and disability worldwide. After severe head injury, the brain’s protective barrier—which normally prevents harmful substances from entering the brain—often breaks down, allowing fluid and toxic materials to leak into brain tissue, causing dangerous swelling and further damage.

In this study, we discovered a harmful chain reaction that destroys the brain’s protective barrier after injury. We found that immune cells called neutrophils release sticky web-like structures made of DNA into the injured brain. Although these DNA webs normally help fight infections by trapping germs, they cause serious harm in brain injury. These DNA webs trigger a molecular alarm system in the cells lining blood vessels in the brain. Once activated, this alarm system causes these cells to die through multiple pathways simultaneously, leading to the collapse of the brain’s protective barrier.

Using mice with controlled brain injuries, we tested whether blocking this harmful process could protect the brain. We found that preventing DNA web formation or stopping the molecular alarm system significantly reduced blood vessel damage, decreased brain swelling, and helped injured mice recover better movement and coordination abilities.

Our research identifies promising new targets for treating brain injury. By developing drugs that block either DNA web formation or the alarm system they trigger, we may be able to prevent barrier breakdown and improve recovery outcomes for patients. This is particularly important given that millions of people worldwide suffer traumatic brain injuries each year from vehicle accidents, falls, sports injuries, and violence.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03123-4.

Background

Traumatic Brain Injury (TBI) represents a significant global public health challenge, characterized by high morbidity, mortality, and disability rates that impose substantial economic and psychosocial burdens on society and families [1, 2]. The pathological progression of TBI is delineated into two distinct phases: the primary injury, resulting directly from the initial mechanical impact, and the secondary injury cascade, which initiates within hours to days and offers a critical therapeutic window for clinical intervention [3, 4]. This secondary phase encompasses a complex array of pathophysiological alterations, including robust neuroinflammation, cerebral edema, and dysregulated cell death [5–7], which are pivotal determinants of long-term neurological deficits and poor patient prognosis.

Within the intricate pathological network of secondary injury, the disruption of blood-brain barrier (BBB) integrity constitutes a hallmark event with profound clinical consequences. The BBB, intricately composed of brain microvascular endothelial cells (BMECs) joined by intercellular tight junctions and adherens junctions, serves as a vital physical and biochemical barrier maintaining central nervous system (CNS) homeostasis [8, 9]. Following TBI, the structural and functional integrity of the BBB is rapidly compromised, leading to a drastic increase in permeability that triggers vasogenic edema [10, 11]. This disruption facilitates the infiltration of peripheral immune cells and the extravasation of neurotoxic plasma components (e.g., thrombin, albumin) into the brain parenchyma, thereby exacerbating neuroinflammation and aggravating neuronal injury [12]. Consequently, the loss of BBB function is intrinsically linked to the injury and death of the BMECs that form its structural foundation. Despite its recognized importance, the upstream molecular triggers initiating endothelial cell death and subsequent BBB breakdown remain incompletely understood, representing a critical knowledge gap that must be addressed to develop effective therapeutic strategies for TBI.

In the context of the acute neuroinflammatory response following TBI, neutrophils are among the first immune cells recruited to infiltrate the injured brain tissue [13–15]. Beyond their conventional functions of phagocytosis and degranulation, neutrophils can undergo a distinct form of cell death termed NETosis, releasing web-like structures known as neutrophil extracellular traps (NETs). The architecture of NETs consists of a backbone of decondensed chromatin double-stranded DNA (dsDNA) decorated with cytotoxic granular proteins, including histones, myeloperoxidase (MPO), and neutrophil elastase (NE) [16–18]. While NETs exert protective antimicrobial effects by entrapping and eliminating pathogens during infection, their excessive formation under sterile inflammatory conditions, such as TBI, transforms them into potent damage-associated molecular patterns (DAMPs) that directly induce tissue damage and amplify inflammation [19–21]. Although prior studies have documented NET formation in the brains of TBI patients and animal models and correlated their presence with adverse outcomes [22, 23], the specific molecular mechanisms by which NETs drive BBB disruption and endothelial cell death post-TBI remain largely unexplored.

The identification of dsDNA as the core structural component of NETs provides a critical mechanistic link to potential downstream effector pathways. Intracellularly, absent in melanoma 2 (AIM2) functions as a cytosolic pattern recognition receptor that specifically senses dsDNA. Upon binding to exogenous or self-derived dsDNA, AIM2 rapidly oligomerizes and recruits the adaptor protein ASC and pro-caspase-1 to assemble the canonical AIM2 inflammasome [24, 25]. Activation of this complex triggers caspase-1-mediated pyroptosis—a highly pro-inflammatory form of programmed cell death—and promotes the maturation and release of pro-inflammatory cytokines. However, within the complex and severe injury microenvironment of TBI, a single mode of cell death may be insufficient to fully account for the rapid and catastrophic endothelial cell loss and the accompanying cytokine storm observed clinically. Recently, the concept of PANoptosis has emerged as an integrative inflammatory cell death pathway that may better explain the multifaceted endothelial damage in TBI. Rather than representing a novel distinct death modality, PANoptosis describes a programmed cell death network regulated by a specific molecular platform—the PANoptosome—that simultaneously integrates key components and signaling molecules from pyroptosis, apoptosis, and necroptosis [26, 27]. Emerging evidence suggests that AIM2 inflammasome activation serves not only as the canonical trigger for pyroptosis but also facilitates crosstalk with apoptotic and necroptotic pathways by activating key molecular nodes such as caspase-8, thereby functioning as a critical upstream signal driving PANoptosis [28]. This integrated cell death framework provides a more comprehensive and mechanistically precise paradigm for understanding endothelial inflammatory cell death following TBI.

Despite these advances, critical questions remain unanswered: the cell-type-specific localization of AIM2 activation within the neurovascular unit (particularly in BMECs versus glial cells), whether AIM2 triggers endothelial PANoptosis encompassing pyroptosis, apoptosis, and necroptosis, and the causal relationship between NETs and AIM2-driven endothelial death remain unclear. Therefore, this study aimed to: [1] characterize NET formation dynamics and its correlation with BBB disruption post-TBI; [2] determine the cell-type-specific localization of AIM2 activation; [3] investigate whether AIM2 initiates endothelial PANoptosis; [4] establish causality using pharmacological and genetic approaches; and [5] evaluate the therapeutic efficacy of targeting this pathway on BBB preservation and functional recovery. By elucidating the NETs-AIM2-PANoptosis axis, this study identifies novel therapeutic targets for neuroprotection following TBI.

Materials and methods

Animals

The animal study protocols were rigorously reviewed and sanctioned by the Institutional Animal Care and Use Committee at Tianjin Medical University General Hospital (IRB2024-DWFL-351). Male C57BL/6J mice (8 weeks old; body weight: 22–24 g) were procured from VitalRiver Laboratory Animal Technology Co., Ltd. (Beijing, China). We chose 8-week-old male C57BL/6J mice (22–24 g) to mimic the TBI pathology of young adults and maintain cranial biomechanical consistency during modeling, thereby precluding the confounding effects of female hormonal fluctuations on neuroinflammation and injury repair. Following acquisition, the animals were maintained in a controlled environment with a constant temperature of 20–22 °C and 50–55% humidity, adhering to a standard 12-hour light/dark cycle. Throughout the acclimatization and experimental periods, mice were provided with unrestricted access to standard chow and water. To minimize bias, the stratification of animals into experimental groups was conducted randomly by an independent coordinator who was blinded to the study design.

Experimental design

Random allocation was employed to distribute the mice into the study groups described below. To minimize potential bias, outcome assessments were conducted by independent researchers who were blinded to both the experimental interventions and group assignments. A comprehensive schematic of the experimental design is provided in Additional file: Part 1.

TBI model

A digital electromagnetic CCI instrument (eCCI6.3, Customs Design & Fabrication, Richmond, USA) was employed to establish the TBI model, adhering to previously validated protocols [29]. In brief, mice were anesthetized with 1.5–3% isoflurane mixed in 30% oxygen and 70% nitrous oxide, and placed on a thermostatic heating blanket to maintain core body temperature at 37 ± 0.5 °C. The animals were then fixed in a stereotaxic head frame, and a circular craniotomy (3.5 mm diameter) was performed over the right parietotemporal skull (2.5 mm posterior to bregma and 2.5 mm lateral to the sagittal suture) using a motorized drill. The CCI model was established with a 3-mm flat-tip impactor at the following parameters: impact velocity 5 m/s, dwelling time 200 ms, and penetration depth 2.2 mm. After surgery, mice were placed on a heating blanket for spontaneous recovery from anesthesia. Sham-operated mice underwent identical craniotomy without receiving CCI.

Drug administration

Cl-amidine (GC35706, GLPBIO, USA), acting as a PAD inhibitor, was formulated in dimethyl sulfoxide (DMSO) and further diluted with saline (5% v/v). Following prior methodologies, animals were treated intraperitoneally (i.p.) with 10 mg/kg of Cl-amidine or a vehicle control (5% DMSO/saline) daily for a three-day duration, starting 1 h post-injury [30]. Separately, the specific AIM2 inhibitor 4-sulfonic calixarene (Tokyo Chemical Industry, Cat# S0470) was prepared from DMSO stocks diluted in PBS. The dosage rationale was based on the foundational in vivo study by Green et al. [31], and further supported by recent independent applications [32], which demonstrated that an intraperitoneal (i.p.) injection of 5 mg/kg optimally and safely suppresses AIM2 inflammasome activation in mice. Accordingly, this compound was administered i.p. at this established dose of 5 mg/kg (200 µl volume) at 1 h, 24 h, and 48 h post-TBI. The control group received equivalent injections of sterile saline, strictly mirroring the regimen and volume of the experimental cohort.

Injection of adenoviruses

Adenoviral vectors, including recombinant PAD4 adenovirus (Ad-PAD4; Adeno-CMV-MCS-Padi4-3FLAG- SV40-EGFP, serotype Ad5, 4 × 10¹⁰ PFU/mL) and its control (Ad-con; Adeno-CMV-3flag-tagged SV40-EGFP), were obtained from Genechem (Shanghai, China). Viral administration was conducted one day prior to the CCI procedure. Following established protocols [33], a total of 4 µl of Ad-PAD4 or Ad-con (total dose: 1.6 × 10⁸ PFU per mouse, titer: 4 × 10¹⁰ PFU/mL) was delivered stereotactically into two distinct sites (2 µl per site) located 1.5 mm rostral and 1.5 mm caudal to the lesion epicenter, at a depth of 1.5 mm. Injections were performed using a 5 µl Hamilton syringe (Hamilton Company, USA) at a controlled rate of 0.5 µl/min. To prevent backflow, the needle was left in place for 10 minutes post-injection before being slowly withdrawn, after which the incision was sutured.

Rotarod test

Motor coordination and balance were assessed utilizing an accelerating Rotarod system (RWD Life Science, Shenzhen, China), following previously established protocols [34]. Before TBI induction, animals were subjected to a three-day acclimatization and training regimen consisting of three daily trials. The initial phase involved a 30-second acclimation at 0 rpm, followed by training sessions starting at a constant speed of 4 rpm for 1 min and progressing to an acceleration from 4 to 40 rpm over a 5-minute period. A 30-minute inter-trial interval was maintained. Post-injury assessments were conducted on days 1 and 3, where mice were tested on the Rotarod accelerating from 4 to 40 rpm over 5 min. The latency to fall was recorded for three trials per day, separated by 30-minute rest periods. Data were analyzed as the mean latency to fall, with all evaluations performed by two investigators blinded to the experimental groups.

Neurological score assessment

Neurological deficits were quantified utilizing the modified Neurological Severity Score (mNSS), adhering to previously validated methods [29]. Assessments were conducted on days 1 and 3 post-TBI. The mNSS framework encompasses a battery of tests evaluating motor function (muscle status and abnormal movements), sensory responses (visual and tactile), reflexes, and balance. The scoring system assigns one point for the inability to perform a task or the absence of a reflex. Cumulative scores range from 0 to 18, categorized as follows: 0 (normal function), 1–6 (mild impairment), 7–12 (moderate impairment), and 13–18 (severe impairment). A detailed breakdown of the scoring criteria is provided in Additional file: Part 2. To minimize bias, evaluations were performed by two independent investigators blinded to the experimental groups.

Magnetic resonance imaging

Brain edema was assessed non-invasively three days post-CCI utilizing a 9.4T horizontal bore BioSpec MRI system (Bruker Biospec 94/30USR, Billerica, MA), in accordance with our previous methodology [35]. The imaging protocol employed the following settings: repetition time (TR) = 4000 ms; echo time (TE) = 35 ms; field of view (FOV) = 40 × 40 mm²; matrix size = 256 × 256; and slice thickness = 0.5 mm. The entire brain was imaged across 15 consecutive slices. Edema resulting from CCI was identified as hyperintense regions on T2-weighted (T2W) sequences localized to the impact site and surrounding cortex [36]. Quantification of these hyperintensities was performed using the software Radiant (version 2024), following established procedures [37]. Data analysis was conducted by investigators blinded to the study groups.

Laser speckle contrast imaging

CBF alterations were assessed at multiple time points—pre-operatively (baseline), and at 1 h, 1 day, and 3 days post-CCI—utilizing a laser speckle contrast imaging system (PeriCam PSI System; Perimed), as previously described [38]. Following deep anesthesia, mice were secured in a stereotaxic apparatus, and a midline scalp incision was made to expose the calvaria. At each specified time point, imaging was performed at a working distance of 10 cm, with continuous CBF recording for a duration of 60 s. For quantitative analysis, a standardized circular region of interest (ROI, 3.5 mm in diameter) was designated to encompass the entire hypoperfused peri-contusional cortex within the craniotomy window. The mean CBF value within this ipsilateral ROI before trauma was defined as the pre-surgery baseline (100%) for each individual animal. Subsequently, the mean CBF values obtained from the same ROI at 1 h, 1 day, and 3 days post-TBI were normalized to their respective pre-surgery baselines. The resulting percentage data were then utilized for subsequent statistical comparisons among different experimental groups. Quantification of CBF dynamics was executed using PIMsoft software (version 1.4; Perimed). All data acquisition and subsequent analyses were performed by investigators blinded to the experimental groups.

Evans blue extravasation assay

BBB integrity was evaluated via Evans blue (EB) dye extravasation, following established protocols [38]. Mice received an intravenous injection of 2% EB (4 ml/kg; Sigma-Aldrich) dissolved in sterile saline, which was allowed to circulate for 2 h prior to euthanasia. Subsequently, animals underwent transcardial perfusion with PBS to remove intravascular dye. Brain tissues were harvested, weighed, and homogenized in N, N-dimethylformamide (Sigma-Aldrich). The homogenates were incubated at 60 °C for 72 h to extract the dye. Following centrifugation, the optical density of the supernatant was measured at 620 nm using a spectrophotometer (Molecular Devices, Sunnyvale, CA). Sample processing and analysis were conducted by two investigators blinded to the experimental conditions.

Western blot analysis

Western blotting procedures were conducted adhering to previously established protocols. Protein samples were normalized, mixed with loading buffer, and denatured at 100 °C for 15 min. Subsequently, proteins were resolved via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto 0.45 μm polyvinylidene difluoride (PVDF) membranes (Millipore, Temecula, CA, USA). The membranes were blocked with 5% non-fat milk for 2 h, followed by overnight incubation at 4 °C with primary antibodies targeting proteins such as AIM2 (Cat. No. ab119791; Abcam; dilution 1:1000), ASC (Cat. No. 67824; Cell Signaling Technology; dilution 1:1000), Occludin (Cat. No. 27260-1-AP; Proteintech; dilution 1:10000), ZO-1 (Cat. No. ab276131; Abcam; dilution 1:1000), VE-cadherin (Cat. No. Ab33168; Abcam; dilution 1:1000), Claudin 5 (Cat. No. Ab131259; Abcam; dilution 1:1000), PAD4 (Cat. No. ab96758; Abcam; dilution 1:1000), H3Cit (Cat. No. ab5103; Abcam; dilution 1:1000), Histone H3 (Cat. No. 9715; Cell Signaling Technology; dilution 1:1000); GSDMD (Cat. No. A20197; Abclonal; dilution 1:5000), p-MLKL (Cat. No. 37333; Cell Signaling Technology; dilution 1:1000), p-RIPK3 (Cat. No. 91702; Cell Signaling Technology; dilution 1:1000), CASP3 (Cat. No.19677-1-AP; Proteintech; dilution 1:1000), CASP8 (Cat. No. 13423-1-AP; Proteintech; dilution 1:500). After washing, the membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies. Protein bands were detected using enhanced chemiluminescence, and the signal intensity was quantified using ImageJ software.

Immunofluorescence staining

Tissue preparation involved transcardial perfusion with 4% paraformaldehyde (PFA). Following overnight post-fixation and dehydration in graded sucrose (20% and 30%), samples were embedded in OCT compound and cut into 8 μm coronal sections. For immunostaining, sections or cells were permeabilized using 0.1% Triton X-100 and blocked with either 10% goat serum or 3% BSA. Subsequently, samples were incubated overnight at 4 °C with specific primary antibodies, including CD31 (Cat. No. Ab182981; Abcam; dilution 1:500), AIM2 (Cat. No. ab93015; Abcam; dilution 1:500), Iba1 (Cat. No. ab283346; Abcam; dilution 1:100), ZO-1 (Cat. No. 21773-1-AP; Proteintech; dilution 1:100), GFAP (Cat. No. 3670; Cell Signaling Technology; dilution 1:500), and GSDMD (Cat. No. Ab219800; Abcam; dilution 1:500). Detection was performed by incubating with Alexa Fluor 488/594-conjugated secondary antibodies for 1 h at room temperature, followed by DAPI counterstaining. To ensure consistency across all samples, images of the peri-contusional cortex were acquired strictly from regions located 0.5 to 1.5 mm lateral to the border of the primary contusion core. Image acquisition and quantification were performed by an investigator blinded to group assignments. For each animal, 3 evenly spaced coronal sections (representing the anterior, middle, and posterior aspects of the injury epicenter) were evaluated. Within each section, 3 random, non-overlapping fields of view (FOVs) were captured from the peri-contusional cortex (0.5–1.5 mm lateral to the lesion core) and averaged to represent a single biological replicate (n = 1).

Statistical analysis

Statistical analyses were performed using GraphPad Prism software (Version 8.1.2, San Diego, CA, USA) [30]. The Shapiro–Wilk test was utilized to assess data normality. Quantitative results are presented as the mean ± standard deviation (SD) with individual data points overlaid. To adhere to the 3Rs principles while ensuring adequate statistical power, sample sizes were tailored to assay variability: neurobehavioral assessments utilized n = 12 per group to account for higher biological variance, whereas biochemical, histological, and imaging analyses required n = 6 per group. For comparisons involving multiple groups, a one-way analysis of variance (ANOVA) was applied, followed by Tukey’s post hoc test. Continuous variables were analyzed using a two-way repeated-measures ANOVA with Tukey’s multiple comparisons adjustment. All statistical tests were two-tailed, and a p-value < 0.05 was considered statistically significant.

Result

Cl-amidine effectively inhibits TBI-induced formation of NETs

To investigate the potential involvement of NETs in the pathophysiology of TBI, we assessed NET formation in the perilesional cortex on day 3 post-TBI. Given that the citrullination of histone H3 (H3cit)—a critical step for chromatin decondensation and NET release—is primarily catalyzed by peptidylarginine deiminase 4 (PAD4), we specifically examined the expression profiles of these associated markers. Western Blot analysis revealed that, compared to the Sham group, the TBI group exhibited significantly elevated protein levels of PAD4 in brain tissues. This elevation was accompanied by a marked accumulation of MPO, a key component of NETs, and the specific marker H3cit (Fig. 1A-D), indicating a robust NETosis response in the acute phase of TBI. Notably, treatment with Cl-amidine (i.p., 10 mg/kg), a specific PAD4 inhibitor, significantly attenuated the aberrant upregulation of these NET-associated proteins (PAD4, MPO, and H3cit), confirming the efficacy of Cl-amidine in blocking in vivo NET formation. To morphologically corroborate these findings and specifically track extruded NETs rather than intact neutrophils, we performed double immunofluorescence staining for MPO and H3cit on brain sections. As the neutrophil plasma membrane ruptures during NETosis, the co-localization of these two markers is established as the gold standard for identifying and quantifying these extracellular web-like structures. In the Sham group, NET signals were negligible. Conversely, the injured cortex of the TBI group displayed extensive co-localization of MPO (green) and H3cit (red), indicative of characteristic NET-like structural release (Fig. 1E). Quantitative analysis further confirmed a sharp increase in the proportion of MPO⁺H3cit⁺ double-positive cells following TBI, whereas Cl-amidine intervention significantly reduced both the number and distribution of these cells (Fig. 1F). Collectively, these data demonstrate that TBI induces significant NET formation on day 3 post-injury, and that Cl-amidine effectively suppresses this pathological process by targeting PAD4.

Fig. 1.

PAD4 inhibitor Cl-amidine inhibited neutrophil extracellular traps formation and downregulated the expression of NETs-associated proteins after TBI. A Representative Western blot images showing the expression levels of NETs-associated proteins, including PAD4, Myeloperoxidase (MPO), and Citrullinated Histone H3 (H3cit), in the peri-contusional cortex at 3 days post-TBI. Total H3 and β-actin were used as loading controls. B–D Quantitative analysis of the protein levels for H3cit (normalized to total H3) (B), MPO (C), and PAD4 (D) (normalized to β-actin). The data show that Cl-amidine treatment (i.p., 10 mg/kg) significantly reversed the TBI-induced upregulation of these NETs markers. E Representative immunofluorescence staining images of NETs in the peri-contusional cortex. NETs formation was identified by the co-localization of MPO (green) and H3cit (red). Nuclei were counterstained with DAPI (blue). Scale bar = 20 μm. F Quantification of the percentage of MPO⁺ H3cit⁺ cells. Data are presented as mean ± SD. n = 6 per group. One-way ANOVA; ***P < 0.001, ****P < 0.0001

Inhibition of NETosis significantly attenuates BBB disruption and leakage following TBI

Given the cytotoxic potential of excessive NETs and their capacity to damage the vascular endothelium, we investigated whether inhibiting NETosis could ameliorate TBI-induced BBB dysfunction [34, 39, 40]. We first evaluated BBB permeability using the EB extravasation assay. Macroscopic inspection revealed prominent blue dye extravasation in the injured hemisphere of the TBI group, indicating severe compromise of BBB integrity. In contrast, Cl-amidine treatment significantly reduced this pathological leakage (Fig. 2A). Quantitative spectrophotometric analysis further confirmed that EB content in the brain tissue of the Cl-amidine-treated group was significantly lower than that of the Vehicle group, demonstrating that blocking NET formation effectively preserves BBB function following TBI (Fig. 2C). To delineate the molecular basis underlying the protective effects of Cl-amidine on the BBB, we examined the expression of key proteins constituting endothelial intercellular junctions, including tight junction proteins (ZO-1, Claudin5, Occludin) and the adherens junction protein (VE-cadherin). Western Blot results revealed that TBI induced a precipitous decline in the expression levels of all three junctional proteins, implying a collapse of the physical barrier between endothelial cells. However, Cl-amidine treatment significantly reversed this trend, effectively restoring the protein abundance of ZO-1, Occludin, VE-cadherin and Claudin5 (Fig. 2B, D). This suggests that inhibiting NETosis helps prevent the degradation or loss of these critical junctional proteins. Further immunofluorescence staining provided visual evidence of the structural integrity of the vascular endothelial layer. We examined the distribution of ZO-1 on CD31-labeled vascular endothelial cells. In the Sham group, ZO-1 exhibited a continuous, linear distribution along the vessel walls. Conversely, following TBI, the ZO-1 signal became discontinuous and fragmented, indicating the rupture of tight junctions. Notably, Cl-amidine administration significantly ameliorated this phenotype, maintaining the continuous coverage of ZO-1 on the vascular endothelial surface (Fig. 2E). Collectively, these results provide compelling evidence that excessive NET formation is a pivotal factor driving BBB disruption after TBI, and that inhibiting NETosis via Cl-amidine preserves the expression and distribution of tight junction proteins, thereby significantly attenuating BBB leakage.

Fig. 2.

Inhibition of NETosis attenuates BBB disruption and reduces vascular permeability following TBI. A Representative photographs of whole brains and coronal sections showing Evans Blue dye extravasation at 3 days post-TBI. The visible blue staining indicates BBB breakdown. B Representative Western blot bands of tight junction proteins (Claudin5, ZO-1, Occludin) and adherens junction protein (VE-cadherin) in the peri-contusional cortex. β-actin was used as the loading control. C Quantitative analysis of Evans Blue content in brain tissues. Cl-amidine treatment significantly reduced dye leakage compared with the Vehicle group. D Densitometric quantification of ZO-1, Occludin, VE-cadherin and Claudin5 protein levels relative to β-actin. The results demonstrate that the loss of these junctional proteins induced by TBI was significantly rescued by Cl-amidine. E Representative double immunofluorescence images of ZO-1 (red) and the endothelial marker CD31 (green) in the peri-contusional cortex. Nuclei were counterstained with DAPI (blue). The bar graph (right) shows the quantification of ZO-1 coverage on CD31-positive vessels, indicating that Cl-amidine preserved the structural continuity of tight junctions. Scale bar = 20 μm. Data are presented as mean ± SD. n = 6 per group. One-way ANOVA; **P < 0.01, ***P < 0.001, ****P < 0.0001

Inhibition of NETosis ameliorates neurological dysfunction, attenuates cerebral edema, and promotes CBF restoration following TBI

Having established the protective effects of Cl-amidine on the BBB, we further investigated whether this preservation translates into tangible improvements in neurological function. We initially assessed neurobehavioral performance using the mNSS and the Rotarod test. Following TBI, mice exhibited significant neurological deficits—characterized by elevated mNSS scores—and impaired motor coordination, evidenced by shortened latency on the Rotarod, at both 1 and 3 days post-injury. Notably, Cl-amidine treatment significantly reduced mNSS scores at both time points (Fig. 3A). Although the difference in Rotarod performance did not reach statistical significance on day 1, Cl-amidine treatment significantly prolonged the latency to fall by day 3 (Fig. 3B), indicating that inhibiting NETosis effectively facilitates the recovery of sensorimotor function. Cerebral edema is a critical pathological alteration driving elevated intracranial pressure and secondary brain injury following TBI. We quantified brain water content using the wet-to-dry ratio method, which revealed that TBI induced severe cerebral edema; however, this was significantly attenuated by Cl-amidine intervention (Fig. 3C). To visually assess the extent of brain injury in vivo, we utilized T2-weighted MRI. Mice in the TBI group displayed distinct hyperintense regions in both the cortex and subcortex, indicative of extensive edema and tissue damage. In contrast, the Cl-amidine-treated group exhibited significantly reduced lesion volumes and superior preservation of brain structural integrity (Fig. 3F, G).

Fig. 3.

Cl-amidine treatment ameliorates neurological deficits and alleviates brain edema and cerebral hypoperfusion following TBI. A Assessment of neurological deficits using the modified Neurological Severity Score (mNSS) at 1 and 3 days post-TBI. Cl-amidine treatment significantly lowered the mNSS scores compared to the Vehicle group. n = 12 per group. B Motor function evaluation using the Rotarod test. While no significant difference was observed at 1 day post-TBI (ns), Cl-amidine significantly improved the latency to fall at 3 days post-TBI. n = 12 per group. C Quantification of brain edema using the wet-to-dry weight ratio method. The TBI-induced increase in brain water content was significantly attenuated by Cl-amidine. n = 6 per group. D Representative laser speckle contrast images showing CBF changes in the cortical surface at pre-surgery, 1 h, 1 day, and 3 days post-TBI. Warmer colors (red/yellow) indicate higher perfusion, while cooler colors (blue) indicate lower perfusion. E Quantitative analysis of CBF within the 3.5-mm ipsilateral injury ROI, expressed as a percentage of the pre-surgery baseline. Cl-amidine treatment mitigated the post-traumatic reduction in CBF over time. Two-way ANOVA (F) Representative T2-weighted MRI images of mouse brains at 3 days post-TBI. The hyperintense (bright) regions outlined in green represent the edema/lesion volume. G Quantification of brain edema volume based on MRI scans. Data are presented as mean ± SD. n = 6 per group.One-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant

Furthermore, the restoration of CBF is vital for the survival of compromised brain tissue. Monitoring via Laser Speckle Contrast Imaging (LSCI) revealed a precipitous decline in CBF within the impact core and penumbra following TBI, manifested as dark blue/cool tones. While blood flow recovery was slow and incomplete in the Vehicle group, the Cl-amidine group demonstrated more robust reperfusion capabilities at days 1 and 3 post-injury. This was evidenced by a significant resurgence of blood flow signals (yellow/red warm tones), with quantitative analysis of the ipsilateral 3.5-mm ROI confirming that CBF restoration levels were significantly superior to those of the control group (Fig. 3D, E). Taken together, these data indicate that targeted inhibition of NETs via Cl-amidine not only mitigates cerebral edema and structural damage but also enhances cerebral perfusion, ultimately facilitating neurological functional recovery following TBI.

NETs drive AIM2 inflammasome activation with specific enrichment in vascular endothelial cells

To dissect the downstream molecular mechanisms through which NETs mediate brain injury, we focused on the AIM2 inflammasome—a pivotal intracellular receptor capable of sensing cytosolic dsDNA [32]. Given that the structural backbone of NETs consists of extracellularly released DNA, we hypothesized that NETs function as upstream agonists for AIM2 [41].

First, we delineated the temporal dynamics of AIM2 expression following TBI. Western Blot analysis indicated that AIM2 protein levels escalated rapidly post-injury, peaking at day 3 before gradually subsiding (Fig. 4A, E). This temporal window coincides precisely with the aforementioned peak of NET formation [30, 42], suggesting a potential spatiotemporal correlation. To corroborate this regulatory relationship, we examined the impact of NETosis inhibition. Cl-amidine treatment not only significantly attenuated AIM2 expression but also effectively suppressed levels of its downstream adaptor protein, ASC (Fig. 4C, F). Intriguingly, double immunofluorescence staining revealed a distinct spatial distribution: within the perilesional cortex, robust AIM2 signals (red) were closely clustered around, and even appeared to envelop, H3cit-marked NET structures (green). Conversely, Cl-amidine treatment, while clearing NETs, concurrently diminished these AIM2 aggregates (Fig. 4I). This strongly suggests that extracellular DNA released by NETs may be internalized or recognized by neighboring cells, thereby “igniting” the AIM2 inflammasome signaling pathway. We next verified the specificity of our pharmacological intervention and explored the functional consequences of AIM2 activation. Administration of a specific AIM2 inhibitor (4-sulfonic calixarene) successfully downregulated the TBI-induced upregulation of AIM2 and ASC (Fig. 4B, G). More importantly, to determine whether AIM2 activation directly contributes to BBB breakdown, we assessed the expression of key junctional proteins. TBI resulted in a severe loss of tight junction proteins (ZO-1, Occludin) and the adherens junction protein (VE-cadherin). However, this degradation was significantly reversed by AIM2 inhibition (Fig. 4D, H). These results provide compelling evidence that the AIM2 inflammasome acts as a critical downstream effector of NETs, directly mediating the structural dismantling of the BBB.

Fig. 4.

Spatiotemporal correlation between NETs and AIM2 inflammasome, and the protective effect of AIM2 inhibition on BBB integrity. A, E Temporal expression profile of AIM2 in the peri-contusional cortex following TBI. Western blot analysis (A) and quantification (E) showed that AIM2 protein levels gradually increased, peaking at 3 days post-injury, consistent with the time course of NETs formation. Two-way ANOVA. C, F The regulatory effect of PAD4 inhibition on the AIM2 inflammasome. Representative Western blots (C) and quantitative analysis (F) demonstrated that treatment with the PAD4 inhibitor Cl-amidine significantly reduced the protein levels of AIM2 and ASC compared to the Vehicle group, indicating that AIM2 activation is downstream of NETosis. n = 6 per group. B, G Validation of the specific AIM2 inhibitor. Western blots (B) and densitometric quantification (G) confirmed that the administration of the AIM2 inhibitor effectively suppressed the TBI-induced upregulation of AIM2 and ASC. n = 6 per group. D, H The protective effect of AIM2 inhibition on BBB structural proteins. Representative Western blots (D) and quantification (H) of tight junction proteins (Claudin5, ZO-1, Occludin) and the adherens junction protein (VE-cadherin). TBI caused a severe loss of these proteins, which was significantly restored by the AIM2 inhibitor treatment. n = 6 per group. I Representative double immunofluorescence images of H3cit (green, NETs marker) and AIM2 (red). The merged images and quantification (right) reveal a high degree of co-localization between NETs and AIM2. Importantly, Cl-amidine treatment significantly reduced the number of H3cit⁺AIM2⁺ double-positive cells. Scale bar = 20 μm. Data are presented as mean ± SD. n = 6 per group.One-way ANOVA; *P < 0.05, ***P < 0.001, ****P < 0.0001

A critical ensuing question was determining which cell type primarily responds to NET signals to activate AIM2. We utilized cell-specific markers for detailed co-localization analysis. Unexpectedly, although microglia (IBA1⁺) and astrocytes (GFAP⁺) are the primary immune effector cells in the brain, they exhibited relatively low co-localization with AIM2 (Fig. 5C-F). In stark contrast, we observed profound AIM2 overexpression within vascular endothelial cells (CD31⁺) (Fig. 5A, B). Quantitative analysis confirmed that the vast majority of activated AIM2 signals post-TBI were localized to the vascular endothelial layer, a phenomenon specifically blocked by the AIM2 inhibitor. Collectively, these findings unveil a distinct pathological axis: TBI-induced NET release acts as a danger signal that preferentially targets and activates the AIM2 inflammasome in brain microvascular endothelial cells rather than in conventional glial cells, providing a precise cellular and molecular explanation for the observed BBB disruption.

Fig. 5.

AIM2 inflammasome is predominantly activated in vascular endothelial cells rather than glial cells in the peri-contusional cortex. A, C, E Representative double immunofluorescence images showing the co-localization of AIM2 (red) with the endothelial cell marker CD31 (A), the astrocyte marker GFAP (C), and the microglial marker Iba1 (E) in the peri-contusional cortex at 3 days post-TBI. Nuclei were counterstained with DAPI (blue). B, D, F Quantitative analysis of the density of double-positive cells. B The number of AIM2 ⁺ CD31⁺ cells was significantly increased after TBI and markedly reduced by the AIM2 inhibitor treatment. D, F Although TBI induced some AIM2 expression in astrocytes (D) and microglia (F), the AIM2 inhibitor treatment did not result in a statistically significant reduction in AIM2 ⁺ GFAP ⁺ or AIM2 ⁺ Iba1⁺ cells (ns). Conclusion: The quantitative data demonstrate that the expression of AIM2 is significantly higher in endothelial cells compared to glial cells, suggesting that endothelial cells are the primary cellular source of AIM2 inflammasome activation in this model. Scale bar = 20 μm. Data are presented as mean ± SD. n = 6 per group. One-way ANOVA; ***P < 0.001, ****P < 0.0001; ns, not significant

AIM2 mediates PANoptosis activation in endothelial cells following TBI

Given the specific accumulation of the AIM2 inflammasome in vascular endothelial cells following TBI, we further investigated the downstream execution mechanisms of cell death. PANoptosis is a identified mode of cell death characterized by the synchronous activation of pyroptosis, apoptosis, and necroptosis pathways [26, 27]. To verify whether TBI triggers this complex death program, we systematically assessed the activation status of core molecules. Regarding the pyroptotic pathway, Western Blot analysis showed that TBI significantly promoted GSDMD cleavage (generating the active GSDMD-N fragment) and Caspase-1 activation (Fig. 6A, G), hallmarks of inflammatory cell death. In the necroptotic pathway, we observed substantially elevated phosphorylation levels of the key kinase RIPK3 and its downstream effector MLKL (p-RIPK3, p-MLKL) post-TBI (Fig. 6C, H), indicating compromised cell membrane integrity. Concurrently, the apoptotic pathway was significantly activated, evidenced by the upregulation of pro-apoptotic proteins BAX, Cleaved-Caspase-8, and Cleaved-Caspase-3, alongside the downregulation of the anti-apoptotic protein BCL-2 (Fig. 6E, I, J).

Fig. 6.

AIM2 inflammasome activation triggers PANoptosis in endothelial cells following TBI. A Representative Western blots showing the expression levels of pyroptosis-related proteins (Cleaved-Caspase−1, GSDMD-N) in the peri-contusional cortex at 3 days post-TBI. B Representative double immunofluorescence staining of GSDMD-N (red) and CD31 (green). The quantification (right) shows that the percentage of CD31 ⁺ vessels expressing GSDMD-N was significantly reduced by the AIM2 inhibitor. Scale bar = 20 μm. C Representative Western blots of necroptosis markers (p-MLKL, p-RIPK3). D Double immunofluorescence images of p-MLKL (red) and CD31 (green), demonstrating the localization of necroptosis signaling in endothelial cells and its suppression by AIM2 inhibition. Scale bar = 20 μm. E Representative Western blots of apoptosis-related proteins (Cleaved-Caspase−3, Cleaved-Caspase−8, BCL−2, BAX). F Double immunofluorescence staining of Cleaved-Caspase−3 (red) and CD31 (green), confirming endothelial apoptosis. Scale bar = 20 μm. G–J Densitometric quantification of Western blot data. G Quantification of pyroptosis markers (GSDMD-N, Cleaved-Caspase−1). H Quantification of necroptosis markers (p-MLKL/MLKL, p-RIPK3/RIPK3). I Quantification of apoptosis executioners (Cleaved-Caspase−3, Cleaved-Caspase−8). J Quantification of apoptosis regulators (BCL−2, BAX). Conclusion: The concurrent upregulation of pyroptotic, necroptotic, and apoptotic markers and their co-localization with endothelial cells confirm the activation of PANoptosis, which was effectively reversed by the specific AIM2 inhibitor. Data are presented as mean ± SD. n = 6 per group. One-way ANOVA; **P < 0.01, ***P < 0.001, ****P < 0.0001

Crucially, blocking upstream signaling with an AIM2 inhibitor synchronously and significantly suppressed the aberrant activation of all key markers across these three death pathways (GSDMD-N, p-MLKL, Cleaved-Caspase-3, etc.). This identifies AIM2 as the critical conductor orchestrating this “death ensemble.” To confirm the specific localization of PANoptosis in situ, we performed double immunofluorescence staining. The results clearly demonstrated that the key executioners of PANoptosis—GSDMD-N (pyroptosis marker), p-MLKL (necroptosis marker), and Cleaved-Caspase-3 (apoptosis marker)—specifically co-localized with the vascular endothelial marker CD31 following TBI (Fig. 6B, D, F). This high degree of overlap visually confirms endothelial cells as the primary victims of PANoptosis. Moreover, treatment with the AIM2 inhibitor significantly reduced the coverage of these death markers on blood vessels, further corroborating the central role of the AIM2-PANoptosis axis in mediating post-TBI endothelial injury and microvascular collapse. In summary, these data depict a clear molecular landscape: activated AIM2 post-TBI functions as a molecular switch, initiating pyroptotic, necroptotic, and apoptotic programs, thereby driving endothelial PANoptosis and precipitating severe microvascular damage.

Overexpression of PAD4 exacerbates endothelial PANoptosis, whereas AIM2 inhibition reverses this pathological process

To definitively establish the causality of the “NETs-AIM2-PANoptosis” axis, we designed a rigorous rescue experiment. We intracerebrally injected an adenovirus (Ad-PAD4, total dose: 1.6 × 10⁸ PFU per mouse; titer: 4 × 10¹⁰ PFU/mL) to artificially overexpress PAD4 (administered one day prior to TBI induction), thereby simulating an environment with intensified NET formation, and investigated whether an AIM2 inhibitor could interrupt the resultant cascade. First, Western Blot analysis verified the validity of the model: compared to the control virus group (Ad-Con), Ad-PAD4 injection significantly upregulated PAD4 protein levels in the TBI mouse brain, successfully constructing a microenvironment with a high NET burden (Fig. 7A, G). Subsequent assays revealed that PAD4 overexpression exerted a series of deleterious effects: it not only further potentiated AIM2 expression (Fig. 7B, H) but also significantly exacerbated the activation of PANoptosis-associated markers (GSDMD-N, p-MLKL, and Cleaved-Caspase-3) (Fig. 7C, I). This confirms that excessive NETs indeed amplify cell death signals by upregulating AIM2.

Fig. 7.

Inhibition of AIM2 reverses PAD4-overexpression-induced endothelial PANoptosis and vascular damage. A Representative Western blots validating the efficiency of adenovirus-mediated PAD4 overexpression (Ad-PAD4, total dose: 1.6 × 10⁸ PFU per mouse; titer: 4 × 10¹⁰ PFU/mL, administered one day prior to TBI induction) in the peri-contusional cortex. Ad-PAD4 significantly increased PAD4 protein levels compared to the Ad-Con group. B Western blot analysis of AIM2 expression. Overexpression of PAD4 led to a marked upregulation of AIM2, which was effectively suppressed by the administration of the AIM2 inhibitor. C Representative Western blots showing the expression of PANoptosis markers (GSDMD-N, p-MLKL, Cleaved-Caspase−3). PAD4 overexpression aggravated the activation of these pathways, whereas treatment with the AIM2 inhibitor reversed these effects. D–F Representative double immunofluorescence images of CD31 (green) with GSDMD-N (D), Cleaved-Caspase−3 (E), and p-MLKL (F) (red). Scale bar = 20 μm. G–J Quantitative analysis of the Western blot and immunofluorescence data. G Quantification of PAD4 levels confirming successful overexpression. H Quantification of AIM2 levels showing the regulatory link between PAD4 and AIM2. I Densitometric analysis of PANoptosis-related proteins. J Quantification of the coverage of PANoptosis markers on CD31 ⁺ vessels. Conclusion: The results indicate that PAD4 acts as an upstream regulator of AIM2, and the exacerbation of endothelial PANoptosis caused by excess PAD4 can be mitigated by blocking AIM2 function. Data are presented as mean ± SD. n = 6 per group. One-way ANOVA; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Crucially, the administration of the AIM2 inhibitor served as a pivotal intervention. When mice overexpressing PAD4 were concurrently treated with the AIM2 inhibitor, the downstream PANoptosis cascade was effectively abrogated, despite the persistence of high upstream PAD4 levels. Western Blot data demonstrated that the AIM2 inhibitor significantly reversed the Ad-PAD4-induced hyperactivation of GSDMD, MLKL, and Caspase-3 (Fig. 7C, I). These biochemical findings were corroborated by histomorphological evidence. Immunofluorescence staining demonstrated that in the Ad-PAD4 group, vascular endothelial cells (CD31⁺) exhibited an extremely high density of PANoptosis signals (characterized by a surge in GSDMD-N/p-MLKL/Caspase-3 coverage), indicating severe microvascular compromise. However, intervention with the AIM2 inhibitor effectively protected the vascular endothelium, significantly reducing the accumulation of these death markers on the vessel walls (Fig. 7D-F, J).

In conclusion, our results construct a comprehensive pathological mechanism: TBI induces neutrophil NET release (catalyzed by PAD4), which acts as an upstream danger signal to specifically activate the AIM2 inflammasome in brain microvascular endothelial cells; activated AIM2 subsequently initiates a destructive PANoptosis program, leading to endothelial cell death, BBB disintegration, and ultimately aggravating neurological deficits. Our data compellingly suggest that blocking any key node in this axis—whether by inhibiting PAD4 or AIM2—offers a promising therapeutic strategy for neuroprotection following TBI.

Discussion

Through multidimensional in vivo experiments, this study unveils a novel pathological mechanism underlying BBB disruption following TBI: the release of NETs acts as an upstream initiating factor that specifically activates the AIM2 inflammasome in cerebral vascular endothelial cells, thereby inducing endothelial PANoptosis. Our data establish a causal link between PAD4-mediated NETosis and AIM2-driven PANoptosis, demonstrating that blocking this axis—whether by inhibiting PAD4 with Cl-amidine or directly inhibiting AIM2—significantly attenuates cerebral edema, improves microcirculatory perfusion, and ultimately promotes neurological recovery.

Neutrophil infiltration is a hallmark event of sterile inflammation in the acute phase of TBI. Although the deleterious role of neutrophils in secondary brain injury is well-established, their specific effector mechanisms have long been simplistically attributed to the release of proteases or ROS. Our findings advance this understanding by confirming that PAD4-catalyzed histone citrullination and subsequent NET release represent a critical form of tissue injury. Consistent with findings by Vaibhav et al. [43] in an ischemic stroke model and recent observations in neurotrauma [44], we observed significant co-localization of H3cit and MPO in the TBI mouse brain. While previous studies have largely focused on clearing extracellular NETs using nucleases like DNase I [39], our rescue experiments demonstrated that artificial overexpression of PAD4 was sufficient to exacerbate neuronal injury, whereas pharmacological inhibition of PAD4 effectively reversed this process. This highlights the translational potential of targeting chromatin decondensation (the rate-limiting step of NET formation) in treating acute brain injury, suggesting that Cl-amidine or its analogs may offer specific therapeutic advantages over broad-spectrum anti-inflammatory drugs by arresting the cascade at its intracellular origin. Furthermore, although H3cit extensively co-localized with the neutrophil marker MPO, PAD4 is also expressed in macrophages and neurons. Therefore, systemic Cl-amidine administration cannot entirely rule out the protective effects of PAD4 inhibition in non-neutrophil populations. Future studies employing neutrophil-specific PAD4 knockout mice or extracellular NET degraders (e.g., DNase I) are required to definitively isolate these cell-specific contributions.

The most significant innovation of this study lies in elucidating the molecular receptor mechanism linking NETs to BBB disintegration. Although extracellular DNA is recognized as a potent DAMP, its specific sensor within the TBI microenvironment has remained elusive. While other nucleic acid sensors, such as the cGAS-STING and TLR9 pathways, have been predominantly implicated in driving neuroinflammation post-TBI [45], our data indicate that the AIM2 inflammasome serves as the critical “bridge” specifically for NET-induced damage. Unlike the widely studied NLRP3 inflammasome, which primarily responds to ionic flux disturbances or lysosomal rupture [46], AIM2 specifically recognizes dsDNA. Specifically, the overexpression of PAD4 to mimic a high-NETs microenvironment directly precipitated a surge in AIM2 expression, whereas the pharmacological inhibition of NETosis significantly attenuated AIM2 activation. Furthermore, the observation that AIM2 inhibitors retain their protective effect even under conditions of PAD4 overexpression logically identifies AIM2 as a key downstream effector node of NETs, refining our understanding of sterile inflammatory signal transduction.

At the cellular level, our study challenges the traditional view that neuroinflammation is primarily mediated by glial cells. Previous research on AIM2 in CNS diseases has largely focused on microglia or neurons [47]. However, we postulate that since NETs are predominantly extruded into the vascular lumen or perivascular space, the adjacent endothelial cells are exposed to the highest local concentrations of extracellular DNA. Consequently, using double immunofluorescence, we unexpectedly found that activated AIM2 post-TBI is predominantly enriched in CD31-positive vascular endothelial cells, rather than IBA1-positive microglia or GFAP-positive astrocytes. This unique cellular localization explains why NET formation directly leads to severe vasogenic edema and microcirculatory dysfunction. As endothelial cells constitute the physical core of the BBB, activation of the AIM2 pathway within them directly results in the loss of critical tight and adherens junction proteins (including Claudin-5, ZO-1, VE-cadherin, and Occludin), which existing literature consistently identifies as the fundamental structural basis of TBI-induced BBB collapse [48]. This suggests that immunomodulation targeting the cerebral endothelium may be a breakthrough point for preventing secondary injury after TBI. While our quantitative analysis demonstrated a significant enrichment of AIM2 within CD31-positive endothelial cells compared to macroglial cells, it is noteworthy that prominent inflammasome activation signals were also visibly apparent in cells neighboring or inside the CD31 + brain vasculature (Figs. 5A and 6B). In the highly complex microenvironment of acute TBI, we speculate that neighboring positive signals may originate from other cellular components of the neurovascular unit, such as pericytes, which closely interact with endothelial cells. Additionally, infiltrating peripheral immune cells (e.g., blood-derived macrophages) or severely damaged neurons may also upregulate AIM2 in response to the massive release of DAMPs. Future studies employing single-cell RNA sequencing or additional multi-labeling approaches are warranted to fully map the cell-type-specific landscape of AIM2 activation post-TBI. An important conceptual question arising from our findings is how extracellular NET-DNA is sensed by the intracellular cytosolic receptor AIM2 in endothelial cells. While not directly investigated in our current study, previous research provides a plausible mechanistic framework. Extracellular DNA and associated damage-associated molecular patterns (DAMPs) can be internalized via endocytosis or through specific cell surface receptors, such as the receptor for advanced glycation endproducts (RAGE) [49]. Crucially, NETs are decorated with highly cytotoxic components, including histones and neutrophil elastase. Once internalized into endosomes, these toxic components have been shown to induce endosomal membrane destabilization and rupture [50]. This endosomal leakage would subsequently allow the NET-derived DNA to escape into the cytosol, where it can directly engage and activate AIM2. Future studies utilizing specific receptor antagonists or genetic ablation of surface transporters are warranted to precisely delineate the upstream mechanisms mediating NET internalization in the brain endothelium.

Further mechanistic exploration revealed that AIM2 does not execute endothelial cells via a single death mode but rather initiates a more robust PANoptosis program. We detected high activation of markers for pyroptosis (GSDMD-N), necroptosis (p-MLKL), and apoptosis (Cleaved-Caspase-3) simultaneously on damaged vessels. Previous TBI studies have extensively investigated these programmed cell death pathways in isolation; however, increasing evidence suggests profound pathological crosstalk among them. Our observation aligns well with the concept of the PANoptosome proposed by Samir et al. [27], wherein innate immune sensors—such as the AIM2 identified herein—function as molecular scaffolds. By orchestrating the assembly of a multi-protein complex comprising RIPK3, Caspase-8, and ASC, this machinery dismantles the conventional boundaries between death pathways. Under the severe impact of TBI, the synchronous activation of these multiple death pathways leads to a catastrophic collapse of the endothelial barrier. This extensive molecular redundancy explains why treatments targeting single cell-death pathways (e.g., pan-caspase inhibitors) have often shown limited efficacy in clinical trials [51]. Our findings suggest that only by blocking an upstream regulator like AIM2 can the downstream divergent death signals be simultaneously curbed, thereby achieving effective BBB protection.

Despite compelling evidence, this study has several limitations to be addressed in future work. First, we used only male mice; given the potential influence of sex hormones on immune responses and BBB permeability, future studies must include females to evaluate sex differences. Second, we utilized 8-week-old young adult mice to ensure a mature, stable neurovascular baseline devoid of adolescent developmental confounders; however, because cerebrovascular architecture and neuroimmune responses change significantly with age, future studies incorporating aged cohorts are needed to comprehensively represent TBI pathophysiology in the elderly population. Third, we focused primarily on the acute phase (within 3 days) of TBI; the long-term role of the NETs-AIM2 axis in chronic neurodegeneration remains to be elucidated. Fourth, our study focused on the local microenvironment of the injured brain; whether systemic NETs in the peripheral circulation also contribute to BBB disruption warrants further investigation. Furthermore, while closed-head injury (CHI) models better represent human TBI biomechanics, we utilized an open-skull controlled cortical impact (CCI) model to ensure the high spatial reproducibility required for microvascular research. Unpredictable macro-hemorrhages secondary to skull fractures in CHI models would severely confound the specific analyses of endothelial PANoptosis and BBB integrity, though future CHI-based studies are necessary to validate this axis in a clinically relevant context. Additionally, while we validated the pathway using pharmacological inhibitors, adenoviral vectors, and targeted in vitro models, the use of cell-type-specific conditional knockout mice (such as endothelial- or neutrophil-specific AIM2, ASC, and GSDMD deficient lines) would provide more definitive genetic evidence and remains a critical avenue for our future investigations. Finally, although we confirmed that NET-derived DNA activates AIM2, we cannot entirely rule out the involvement of self-DNA from other sources (e.g., mitochondrial DNA released by damaged neurons) in this process.

Conclusion

In summary, the present study elucidates a novel pathogenic axis underlying TBI-induced BBB disruption and neurological dysfunction. We demonstrate that following TBI, PAD4-mediated formation of NETs serves as a critical upstream danger signal that specifically targets cerebral vascular endothelial cells. This interaction triggers the activation of the AIM2 inflammasome, which subsequently orchestrates a robust PANoptosis program—encompassing pyroptosis, apoptosis, and necroptosis—thereby leading to the collapse of endothelial tight junctions and severe BBB leakage. Importantly, our findings confirm that interrupting this cascade, either by inhibiting PAD4-dependent NETosis or by blocking downstream AIM2 signaling, effectively preserves endothelial integrity, restores cerebral perfusion, and ameliorates neurological outcomes. These data not only uncover the molecular mechanism by which innate immune responses compromise the neurovascular unit but also highlight the NETs-AIM2-PANoptosis axis as a promising therapeutic target for mitigating secondary injury following TBI.

Abbreviations

AIM2

Absent in melanoma 2

ASC

Apoptosis-associated speck-like protein containing a CARD

BAX

BCL2-associated X protein

BBB

Blood-brain barrier

BCL-2

B-cell lymphoma 2

BMECs

Brain microvascular endothelial cells

CBF

Cerebral blood flow

CCI

Controlled cortical impact

CD31

Cluster of differentiation 31 (PECAM-1)

CNS

Central nervous system

DAMPs

Damage-associated molecular patterns

DMSO

Dimethyl sulfoxide

dsDNA

Double-stranded DNA

EB

Evans blue

GFAP

Glial fibrillary acidic protein

GSDMD

Gasdermin D

GSDMD-N

N-terminal fragment of gasdermin D

H3cit

Citrullinated histone H3

IBA1

Ionized calcium-binding adapter molecule 1

i.p.

Intraperitoneal

LSCI

Laser speckle contrast imaging

MLKL

Mixed lineage kinase domain-like protein

mNSS

modified neurological severity score

MPO

Myeloperoxidase

MRI

Magnetic resonance imaging

NE

Neutrophil elastase

NETs

Neutrophil extracellular traps

PAD4

Peptidylarginine deiminase 4

PANoptosis

Pyroptosis, apoptosis, and necroptosis

PFU

Plaque-forming units

p-MLKL

Phosphorylated MLKL

p-RIPK3

Phosphorylated RIPK3

RIPK3

Receptor-interacting protein kinase 3

TBI

Traumatic brain injury

VE-cadherin

Vascular endothelial-cadherin

ZO-1

Zonula occludens-1

Authors’ contributions

X.C., L.L., and H.F.Z. designed the experiment; H.F.Z., L.L., and Y.B.Y. performed most of the experiments; H.C. and P.Y.L. assisted with the experiments. H.F.Z. performed the data analysis; H.F.Z. and L.L. wrote the manuscript; X.C., S.Z. and Y.C. provided the overall guidance.

Funding

This work was supported by grants from the National Natural Science Foundation of China (No. 82471397, No. 82271399, No. 82401605), Joint Funds of the Natural Science Foundation of Tianjin (No. 25JCLZJC00280), the Tianjin Health Care Elite Prominent Young Doctor Development Program, the Tianjin Health Research Project (TJWJ2024QN004), the Young and Middle-aged Backbone Innovative Talent Program, Clinical Talent Training “123” Climbing Plan of Tianjin Medical University, and Tianjin Key Medical Discipline (Neurosurgery) Construction Project (TJYXZDXK-002A).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

All aspects of the study were approved by the Experimental Animal Welfare Ethics Committee of Medical University General Hospital (IRB2024-DWFL-351).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.