Human brain tissues were obtained in accordance with an ethically reviewed and approved protocol from Tianjin Medical University General Hospital. Four patients with severe TBI, defined as post-resuscitation Glasgow coma scale (GCS) scores less than 8, were included. Control brain specimens were obtained from three patients with glioma who underwent maximizing resection of malignant glioma to improve survival. None of the patients had any other known neurological disorder. The demographics and clinical characteristics of these patients are shown in Supplemental Table 1 .

Plasma was collected from the whole blood of humans by centrifugation at 250 × g for 20 min. Plasma DNA was quantified according to the manufacturer’s instructions using the Quant-iT PicoGreen dsDNA Assay kit (P11496, Invitrogen). We also developed a capture ELISA based on citrullinated histone H3 associated with DNA (H3cit-DNA complex), as reported previously (31). For the capture of antibodies, 5μg/ml anti-H3Cit (ab5103, Abcam, UK) was coated onto 96-well plates overnight at 4°C. The plates were then blocked with 5% BSA for 2 h at room temperature. After being washed with washing buffer (300 μL each), 50 μL of plasma was added into the wells with 80 μL incubation buffer containing a peroxidase-labeled anti-DNA mAb (Cell Death ELISA, 11774425001, Roche, Basel, Switzerland). The plates were then incubated at room temperature for 2 h. After being washed six times, the plate was developed with 100 μL ABST substrate. The absorbance (OD) at 450 nm was measured after 30 min incubation in the dark.

Male C57BL/6 mice (8-10 weeks old and 25-30g), were purchased from the Experimental Animal Laboratories of the Academy of Military Medical Sciences (Beijing, China). All mice were housed in a temperature (20°C–24°C) and humidity (50%–60%) controlled environment with food and water available ad libitum under a standard 12-h light/dark cycle. All experimental procedures were conducted in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by Tianjin Medical University Animal Care and Use committee.

All experimental mice were randomly assigned to five experimental groups. Surgeries, histological, and neurological function assessments were carried out in a blinded manner.

TBI was induced by a digital electromagnetically controlled cortical impact (CCI) device (CCI-6.3 device, Custom Design & Fabrication, USA) as previously described (32). Briefly, each mouse was anesthetized with 1–1.5% isoflurane in 30% oxygen and 70% nitrous oxide during the surgical procedure, and the body temperature was maintained at 37 ± 0.5°C using a heating blanket. After fixation on a custom head-fixing apparatus, a craniotomy was performed using a high-speed micro drill on the right parietal cortex (2.5 mm posterior from the bregma and 2.5 mm lateral to the sagittal suture) (33). A 4-mm-flat impactor tip was used to conduct a unilateral 2.2 mm depth impact on the mice at 5 m/s with a 200 ms dwell time. The sham mice were subjected to the same surgical procedure without the infliction of CCI. All procedures were performed with a strict aseptic technique.

Recombinant PAD4 adenovirus (Adeno-PAD4; Adeno-CMV-Padi4-3*flag-tagged SV40-EGFP, 4 × 10¹⁰ plaque-forming-unit/mL) and empty adenovirus (Adeno-CMV-3*flag-tagged SV40-EGFP) were purchased from Genechem (Shanghai, China). A total of 4 μL of Adeno-PAD4 (Ad-PAD4) or Adeno-EGFP (Ad-con) was injected stereotactically into two sites of the right cortex (1.5 mm caudal and 1.5 mm rostral from lesion epicenter) at a dose of 2 μL, respectively, 24 h before mice subjected to CCI. The injections were performed with a 5 μL Hamilton syringe (Hamilton Company, USA) at a flow rate of 0.5 μL/min through the hole at a depth of 1.0 mm (33, 34). After the injection, the needle was held for 10 min before retraction, and the scalp was stitched.

As previously described (35), to deplete neutrophils in vivo, each mouse received an intraperitoneal (i.p.) injection of 100 μg monoclonal anti-mouse Ly6G antibody (1A8 clone; specific for neutrophils, BE0075-1, BioXCell, NH) 24 h before TBI and 24 h after TBI. Each control mouse received 100 μg of the isotype antibody rat IgG2a in the same manner.

Neurological function was evaluated at 1, 3, 5, and 7 d after TBI using the well-established mNSS, as described previously (39). The mNSS consists of a motor (muscle state and abnormal action), sensory (visual, tactile), reflex, and balance tests. The higher score represents the more serious neurological impairment (normal score, 0; maximal score, 18).

As described previously (40), the rotarod test was performed to assess the sensorimotor function, coordination, and balance at 1, 3, 5, and 7 d post-injury with an accelerating Rota-rod apparatus (RWD Life Science, Shenzhen, China). Before induction of CCI, mice in each group were trained for three consecutive days. Each mouse was placed in each lane on the rotating cylinder, which accelerated from 4 to 40 rpm/min within 5 min. The latency to fall for each mouse was recorded. Each mouse was tested three times with an interval of 30 min between trials, and the average latency to falling was used for analysis.

H&E staining was performed to evaluate damaged cells as previously described (41). The paraffin-embedded 4% paraformaldehyde-fixed brains were sliced into 10um thick sections. The brain sections were then deparaffinized with xylene and dehydrated in a graded series of alcohols. Sections were incubated with hematoxylin (Sigma Aldrich) for 3 mins and washed three times with PBS. Then, sections were stained with eosin for 4 min. The average number of normal neurons and damaged neurons were calculated by ImageJ software (Version 1.46r, Wayne Raband, USA).

Nissl staining was performed to evaluate neuronal damage as previously described (32). The paraffin-embedded 4% paraformaldehyde-fixed brains were sliced into 10um thick sections. The brain sections were then deparaffinized with xylene and dehydrated in a graded series of alcohols. Sections were incubated with Nissl staining solution (Sigma Aldrich) for 20 mins at 60°C in the oven. The damaged neurons were characterized by a shrunken cytoplasm and condensed staining, while normal neurons were characterized by a large and full soma. The average number of normal neurons and damaged neurons were calculated by ImageJ software (Version 1.46r, Wayne Raband, USA).

After the sacrifice of mice, the brains were removed and immersed in 4% paraformaldehyde, 15% sucrose, and 30% sucrose to complete fixation and dehydration. Brains were then sliced into 8μm-thick coronal sections using a cryostat (Leica, Model CM1950, Germany). After being washed in PBS, the sections were permeabilized with 0.1% Triton X-100 (Sigma Aldrich) for 30 mins and incubated with 3% BSA for 1h at room temperature. Thereafter, sections were incubated overnight at 4°C with primary antibodies, including mouse anti-NeuN (1:1000, ab104224), rabbit anti-NeuN (1:1000, ab177487), rabbit anti-H3Cit (1:1000, ab5103, all from Abcam, UK), rabbit anti-STING (1:500, 13647, Cell Signaling Technology), goat anti-MPO (1:500, AF3667, R&D Systems), goat anti-CD31(1:200, AF3628, R&D Systems), rat anti-mouse Ly6G (1:200, 551459, BD Biosciences), mouse anti-Caspase1 (1:500, sc-56036, Santa Cruz Biotechnology), mouse anti-NLRP1 (1:500, sc-390133, Santa Cruz Biotechnology). The sections were then incubated with the species-appropriate Alexa Fluor-conjugated IgG (1:500, Invitrogen, USA) for 1h at room temperature. DNA was stained with 4’,6-diamidino-2-phenylidole (DAPI, Abcam). Staining was visualized and captured by an inverted fluorescence microscope (Olympus, Japan). Images were then analyzed using ImageJ software (Version 1.46r, Wayne Raband, USA).

For quantification of neuronal apoptosis at 3 days after TBI, double staining of neuron marker NeuN (red) and TUNEL (green) was conducted using the In Situ Cell Death Detection kit (Roche, South San Francisco, CA, USA), according to the manufacturer’s instructions. Coronal sections were counterstained with mouse anti-NeuN (1:1000, ab104224, Abcam, UK), at 4°C overnight and subsequently incubated with the In Situ Cell Death Detection kit and a secondary donkey anti-mouse Alexa 594 antibody for 1h at 37°C in the dark. The average number of TUNEL-positive neurons in the contused cortex was calculated by ImageJ software (Version 1.46r, Wayne Raband, USA).

Total proteins in the brain tissues of TBI mice were extracted using RIPA lysis buffer (Sigma-Aldrich, MO, USA). Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a 0.45um pore size polyvinylidene difluoride (PVDF) membranes (Millipore, Temecula, CA, USA). The membranes were blocked with 5% skimmed milk for 2 h and then incubated primary antibodies: including rabbit anti-IREα (1:1000, 3294), rabbit anti-ASC (1:1000, 67824), rabbit anti-Histone H3 (anti-H3; 1:1000, 9715), rabbit anti-STING (1:1000, 13647), rabbit anti-β-actin (1:2000, 4970, all from Cell Signaling Technology, MA), rabbit anti-H3Cit (1:1000, ab5103), rabbit anti-p-IREα (1:1000, ab124945), rabbit anti-IL-18 (1:1000, ab207323), rabbit anti-MPO (1:1000, ab208670), rabbit anti-PAD4 (1:1000, ab96758, all from Abcam, UK), anti-Caspase1 (22915–1-AP, 1:2000, Proteintech Group, Wuhan, China), anti-GSDMD (AF4012, 1:2000, Affinity Biosciences), mouse anti-NLRP1 (1:500, sc-390133, Santa Cruz Biotechnology). The membranes were then washed three times and incubated with species-appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (1:5000, Cell Signaling Technology, USA) for 1h at room temperature. Finally, the immunoblot bands were visualized using ECL under an imaging system (Bio-Rad, Hercules, CA, USA). Protein quantification was measured in optical density units using Image Lab-5.2.1 software (Bio-Rad, CA, USA) and was normalized to the corresponding sample expression of β-actin.

All statistical analysis was performed with Graph-Pad Prism software (Graph Pad Software, Version8.1.2 San Diego, CA, USA). Multiple comparisons were analyzed by one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison post hoc test. The results were expressed as means ± (SD). Comparison between 2 groups was performed by unpaired Student’s t-test or Mann-Whitney test. A value of P < 0.05 was considered statistically significant.

To explore whether neutrophils are present in the brain tissues of TBI patients, we obtained brain specimens from four severe TBI patients who underwent a decompressive craniotomy to remove the hematoma and/or to reduce life-threatening high intracranial pressure. We found that infiltration of both neutrophils and neutrophils-derived NETs was significantly increased in the brain tissues of TBI patients ( Figures 1A, B ). A typical head CT image of a patient with TBI observed a brain contusion in the right frontal lobe with intracranial hematoma ( Figure 1C ). Plasma cell-free DNA (cf-DNA) ( Figure 1D ), a rough biomarker of NETs, and plasma H3Cit-DNA complex ( Figure 1E ), a specific biomarker of NETs, were significantly elevated in TBI patients compared with matched healthy donors. Notably, a significant inverse correlation was observed between plasma DNA levels and GCS score, and a significant positive correlation was detected between plasma DNA levels and ICP ( Figures 1F, G ). Similarly, plasma H3Cit-DNA complex levels positively correlated with ICP, whereas inversely correlated with the GCS score. ( Figures 1H, I ). These results raise the unexplored possibility that NETs contribute to neurological deficits and cerebral edema after TBI.

To verify the presence of neutrophil infiltration and NETs formation in the brain tissue of 3d mice after TBI. Double immunofluorescence staining of Ly6G and NETs-specific markers (citrullinated histone H3, H3cit), neutrophil, and neutrophil-derived NETs could be simply observed in the mouse brain sections after TBI ( Figure 2A ). It has been confirmed that the peak of neutrophil infiltration in mouse brain tissue is 3 days after TBI (42). Western Blot results that the protein expression of MPO, PAD4, and H3Cit was significantly increased in the cortex around the trauma area 3d post-TBI ( Figures 2B–E ).

The rotary test and mNSS were used to assess neurological function in the acute phase after TBI. we found that anti-Ly6G treatment effectively improved neurological deficits in mice at 3 d after TBI ( Figures 3A, B ). Hematoxylin and eosin staining and Nissl staining showed massive degeneration and necrosis of neuronal cells 3 days after TBI. However, the number of neuronal deaths was significantly reduced in anti-Ly6G-treated mice compared with the vehicle group ( Figures 3C–E ). Additionally, we observed neuronal death by immunofluorescence staining of NEUN (red) and TUNEL (green) in the different groups ( Figures 3F, G ). The results showed that neuronal death was significantly reduced in the anti-Ly6G treatment group compared with the TBI+ vehicle group. NLRP1 (nucleotide-binding domain leucine-rich repeat pyrin domain containing 1) inflammasomes are primarily expressed in neurons and is closely related to neuronal pyroptosis. The results observed that the number of NLRP1 positive neurons in the anti-Ly6G treatment group was significantly reduced compared with the vehicle group ( Figures 3H, I ). Therefore, neutrophil infiltration was closely related to NLRP1-mediated neuronal pyroptosis after TBI by double immunofluorescence staining of NEUN (red) and NLRP1(green).

We next confirm whether the degradation of NET-associated DNA with DNase 1 could ameliorate neuronal pyroptosis by inhibiting the STING-IRE1α-NLRP1 pathway. Immunofluorescence staining showed the number of TUNEL-positive neurons remarkably increased in the cortex around the trauma area of TBI mice compared with the sham group ( Figures 4A, B ). In TBI mice treated with Dnase 1 the number of TUNEL-positive neurons significantly reduced in the peripheral cortex of the trauma area ( Figures 4A, B ). Phosphorylated STING is an activated form of STING. In order to verify the activation of STING in neurons after TBI, immunofluorescence staining was used to observe the expression of P-STING in neurons ( Supplementary Figure 1 ). The results showed that the expression of P-STING increased significantly in neurons after TBI, and DNase1 treatment could reverse this result. This effect of DNase 1 was abolished in mice that also received the second messenger cGAMP 3 d after TBI ( Figures 4A, B ). Consistent with this result, compared with TBI +vehicle group, DNase 1 treatment could significantly reduce the number of STING-positive neurons. However, cGAMP treatment can reverse the treatment advantage of DNase 1 ( Figures 4A, C ). Meanwhile, the number of NLRP1-positive neurons was significantly reduced in DNase 1 treatment mice with TBI. Interestingly, cGAMP treatment also reversed the therapeutic effects of DNase 1 ( Figures 4A, D ). Therefore, STING can activate the pyroptosis pathway of neurons. We subsequently found that the protein expression of P- IRE1α, NLRP1, ASC, Caspase-1 p20, N-GSDMD, IL-18, and IL-1β were remarkably upregulated in the cortex of mice subjected to TBI ( Figures 4E–L ). DNase 1 treatment significantly rescued the activation of the STING-IRE1α-NLRP1 pathway and reduced protein products associated with neuronal pyroptosis. However, treatment with the cGAMP effectively reversed the effect of DNase 1 on the STING-IRE1α-NLRP1 pathway ( Figures 4E–L ). These results showed that degradation of neutrophil-released NETs associated DNA with DNase 1 could recuse neuronal pyroptosis by inhibiting the STING-IRE1α-NLRP1 pathway.

Peptidylarginine deiminase 4 (PAD4) catalyzes histone citrullination that plays a key role in NETs formation (43, 44). We examined a hypothesis on whether inhibition of PAD4(CL-amidine) to reduce NETs-formation improved neural function and reduced neuronal pyroptosis. First, Nissl staining showed that the number of neuronal deaths was significantly reduced in CL-amidine treatment mice compared with the vehicle group ( Figures 5A, B ). However, treatment with cGAMP could revere the effectiveness of CL-amidine after TBI ( Figures 5A, B ). First, confocal microscopy was used to observe the brain tissue stained by immunofluorescence, and it was found that STING/NLRP1 and STING/Caspase-1 could be expressed simultaneously in the same neuron ( Supplementary Figure 2 ). In addition, immunofluorescence staining showed that treatment with Cl-amidine significantly reduced the number of STING-positive, NLRP1-positive, and Caspase-1-positive neurons in the cortex surrounding the trauma area ( Figures 5C–F ). On the contrary, cGAMP treatment could block the effectiveness of CL-amidine ( Figures 5C–F ). The protein levels of P- IRE1α, NLRP1, ASC, Caspase-1 p20, N-GSDMD, IL-18, and IL-1β were detected by western blot and it was found that the protein expression level of STING/P- IRE1α-NLRP1 pathway and neuronal pyroptosis-related protein products in the CL-amidine treatment group were significantly down-regulated compared with TBI+ vehicle group ( Figures 5G–N ). However, the protein level of the STING-IRE1α-NLRP1 pathway and neuronal pyroptosis-related protein products were still up-regulated in the mice that interfered with both CL-amidine and cGAMP ( Figures 5G–N ). Meanwhile, the results of the rotary test and mNSS scores showed that Inhibition of PAD4 could remarkably ameliorate neurological deficits of mice on day 3 after TBI ( Figures 5O, P ). These results showed that Inhibition of PAD4 could ameliorate neurological deficits and recuse neuronal pyroptosis by inhibiting the STING-IRE1α-NLRP1 pathway.

To study whether that increased formation of NET, orchestrated by PAD4, contributed to neuronal pyroptosis, we first studied the role of overexpression of PAD4 on neuronal pyroptosis in mice with TBI. Immunofluorescence staining showed extensive expression of recombinant adeno-PAD4-EGFP-infected cells in the cortex 3 d after injection ( Supplementary Figure 3 ). It can be seen from the figure that the expression of PAD4 is mainly concentrated in neutrophils. Immunofluorescence staining showed that administration of PAD4 adenovirus into the cortex led to a remarkable increase in the number of neuronal death (TUNEL-positive neurons), STING-positive neurons, and NLRP1-positive neurons ( Figures 6A–D ). Interestingly, STING inhibitor C176 could block the function of PAD4 adenovirus ( Figures 6A–D ). Meanwhile, the protein expression level of the STING- P- IRE1α-NLRP1 pathway and neuronal pyroptosis-related protein products were also detected by Western Blot. The results indicated that the protein levels of P- IRE1α, NLRP1, ASC, Caspase-1 p20, N-GSDMD, and IL-18 were significantly increased after the administration of PAD4 adenovirus into the cortex ( Figures 6E–K ). However, STING inhibitor C-176 could decrease the expression level of related proteins after TBI and reverse the effect of PAD4 adenovirus ( Figures 6E–K ). Finally, we evaluated neurological function during the acute phase post-TBI by rotary test and mNSS scores after administration of PAD4 adenovirus and PAD4 adenovirus + C176. The results showed that the PAD4 adenovirus group aggravated nerve damage after TBI, and C176 improved the adverse effects caused by PAD4 adenovirus ( Figures 6L, M ). These results showed PAD4 played a key role in NLPR1-induced neuronal pyroptosis via STING/IRE1α pathway.

To explore whether IRE1α is a key part of the pathway that NETs cause neuronal pyroptosis. We used IRE inhibitor Kira6 after the administration of PAD4 adenovirus into the cortex to observe the expression of pathway proteins. The results found that the protein expression of P-IRE1α was significantly increased in the PAD4 adenovirus group and IRE inhibitor Kira6 could block the effect of PAD4 adenovirus ( Figures 7A, B ). Meanwhile, Immunoblotting showed the protein expression of NLRP1, ASC, Caspase-1 p20, N-GSDMD, and IL-18 were significantly reduced after the administration of Kira6 compared with the DMSO group and the PAD4 adenovirus group. ( Figures 7A, C–G ). These data suggest that IRE1α plays a key role in NLRP1-induced neuronal pyroptosis by promoting NETs formation.