This study employed male SPF (Specific Pathogen Free) C57BL/6 mice (Hefei Jisai Biotechnology Co., Ltd., Heifei, Anhui, China) aged 8–12 weeks, each housed individually for one week prior to experimentation. Mice were maintained under a 12-hour light-dark cycle at 25–26 °C and 50 $\pm$ 5% relative humidity with free access to chow and water. Ventilation rates were maintained at 8–15 air changes per hour, and all mice were fed standard irradiated chow.

Using a desktop animal anaesthesia ventilator system (Ruiwode Life Sciences Co., Ltd, Hefei, Anhui, China), set the isoflurane concentration to 4%, with a N₂O to O₂ ratio of 2:1 with an oxygen flow rate of 0.5–1 L/min. Place the mouse in the induction box for anaesthetic induction. Observe the mouse until loss of righting reflex is confirmed, then position it on the CCI impactor workbench (YHKJCI990313, Wuhan YiHong Sci.&Tech. Co., Ltd., Wuhan, Hubei, China) and secure it. Connect the inhalation anaesthesia mask (set to 2.5% isoflurane concentration) and maintain body temperature using a heated pad. The scalp hair was removed from the mouse’s cranial region. Following alcohol disinfection, the scalp was incised to expose the skull. A square bone window measuring 3–4 mm in diameter was created in the right cerebral hemisphere (2 mm posterior to the anterior fontanelle, 2 mm right of the sagittal line) using a cranial drill, exposing the intact dura mater. Subsequently, employ a 3mm impactor to strike the brain tissue according to pre-set parameters (velocity: 4.5 m/s, depth: 2 mm, dwell time: 200 ms), inducing CCI. Following modelling completion, suture the wound site. After disinfection with povidone-iodine, administer antibiotics via intraperitoneal injection. Upon anaesthetic recovery, transfer the animal to a mouse cage.

Following one week of adaptive feeding, mice were randomly assigned to five groups (n = 15 per group): (1) Sham group (cranial window opened without brain tissue impact), (2) TBI group, (3) TBI + taVNS group, (4) TBI + taVNS + Agonist group (receiving intraperitoneal injection of 50% glucose solution at 6 g/kg/day for 7 consecutive days starting 3 days post-surgery), (5) TBI + Inhibitor group (intraperitoneal injection of glycyrrhizic acid dissolved in 5% DMSO + physiological saline at 50 mg/kg/day for 7 consecutive days starting 3 days post-surgery). For the sake of brevity, the ‘TBI + taVNS + Agonist’ group is hereafter referred to as the ‘Agonist’ group, and the ‘TBI + Inhibitor’ group as the ‘Inhibitor’ group in the text and figures. The experimental timeline is illustrated in Fig. 1.

Fig. 1.

Experimental timeline and group design. Mice (n = 15/group) were divided into five groups: Sham, CCI, VNS, Agonist + VNS, and Inhibitor. Except for the Sham group, all animals were subjected to CCI. The experimental timeline was as follows: CCI modelling $\to$ Behavioral tests and MRI scanning (POD 2) $\to$ 7-day intervention $\to$ Repeated behavioral tests and MRI scanning $\to$ Sample collection. CCI, controlled cortical impact; VNS, vagus nerve stimulation; MRI, magnetic resonance imaging; taVNS, transcutaneous auricular vagus nerve stimulation; WB, western blot; POD 2, Post-Operative Day 2.

The taVNS stimulation parameters for this experiment were established based on prior research. The apparatus employed a vagus nerve electrical stimulator (YJT1-240710002) manufactured by Hangzhou Yijian Technology (Hangzhou, Zhejiang, China). To ensure mice could tolerate prolonged vagus nerve electrical stimulation, a bench-top animal anaesthesia ventilator (Ruiwode Life Sciences Co., Ltd.) was employed. Mice were placed in an induction chamber and anaesthetised with 2% isoflurane (0.5%–1% oxygen) to achieve and maintain anaesthesia. Both auricles were swabbed with alcohol. Electrodes were secured to both auricles, and electrical stimulation commenced. Parameters were set as follows: 0.5 mA, 30 Hz, 30 minutes daily, with 30-second on-intervals followed by 4.5-minute off-intervals. Stimulation was administered daily between 21:00 and 22:00 for a total of 7 days.

The experiment employed a Bruker Biospec 9.4T small animal magnetic resonance imaging system (Bruker Biospec MRI, Ettlingen, Germany) with a 20 mm mouse-specific brain coil. Under 4% isoflurane induction anaesthesia (maintained at 1.5%), axial T2-weighted imaging was performed on days 3 and 10 post-TBI surgery (Parameters: repetition time/echo time (TR/TE) = 3272/48 ms, rapid acquisition with relaxation enhancement (RARE) factor = 8, acquisition runs = 3, matrix = 256 $\times$ 256, slice thickness = 0.7 mm, 20 consecutive slices covering the entire brain). Throughout scanning, mouse body temperature was maintained at 37.0 $\pm$ 0.5 °C via a thermostatic system, with physiological gating synchronised to respiratory movements. Raw images were preprocessed using ITK-SNAP (v3.8.0, https://www.itksnap.org/pmwiki/pmwiki.php?n=Downloads.SNAP3). Edematous regions were semi-automatically segmented based on signal intensity thresholds ($>$2$\times$ mean of contralateral normal brain tissue), with manual refinement to exclude vascular and artefact interference. Volume changes and change rates at both time points were calculated.

Brain tissue samples were collected from mice in the taVNS group, TBI group, and control group (n = 3 per group). Sequencing analysis was performed using the DNBSEQ platform (https://www.mgi-tech.com/DNBSEQ-Technology.html). Key modules were screened for core pathways via Kyoto Encyclopedia of Genes and Genomes/Gene Ontology (KEGG/GO) enrichment analysis (ClusterProfiler). Target specificity validation was conducted using western blot and enzyme-linked immunosorbent assay (ELISA) cross-platform validation for HMGB1 and its upstream/downstream molecules (including IL-6 and IL-1$\beta$).

The experiment employed a standardised open field apparatus (40 cm $\times$ 40 cm $\times$ 40 cm, opaque black polypropylene construction, with the central area defined as a 20 cm $\times$ 20 cm square) for behavioural assessment on days 2 and 11 post-surgery. Thirty minutes prior to testing, mice were acclimatised to the behavioural laboratory environment (illuminance 50 lux, background noise 45 dB, temperature 22 $\pm$ 1 °C). During testing, individual mice were placed in the centre of the open field. An overhead camera (Noldus EthoVision XT 15.0, https://www.noldus.com.cn/ethovision-xt/) continuously recorded spontaneous activity for 5 minutes, analysing the distance travelled within the central square.

The experiment employed an ELISA to detect concentrations of HMGB1 and IL-6 in mouse serum. The mouse HMGB-1 ELISA detection kit (Lianke Bio, SEKM-0032, Shenzhen, Guangdong, China; detection range 15.6–1000 pg/mL, sensitivity 4.7 pg/mL) and the mouse IL-6 ELISA Detection Kit (EMC004, NeoBioscience Technology Co., Ltd., Shenzhen, Guangdong, China; detection range 7.8–500 pg/mL, sensitivity 2.3 pg/mL).

At the end of the experiment, all animals were deeply anesthetized with 3% isoflurane delivered via an inhalation chamber (0.5 L/min oxygen flow) until loss of the righting and pedal withdrawal reflexes. Once deep anesthesia was achieved, the animals were humanely euthanized by cervical dislocation to ensure death, in accordance with the AVMA Guidelines for the Euthanasia of Animals (2020). Whole blood was collected via ocular puncture. After standing for 30 minutes, serum was separated by centrifugation at 3000 $\times$g for 10 minutes at 4 °C. Aliquoted samples were stored at –80 °C until testing. Prior to assay, samples were thawed and diluted according to kit specifications (HMGB1: 1:20; IL-6: 1:10). Standard dilutions were incubated concurrently with samples (100 µL/well, 37 °C for 90 minutes). Following washing, the following reagents were sequentially applied: biotinylated antibody (1:1000, 37 °C for 60 minutes), streptavidin-HRP (1:2000, 37 °C for 30 minutes), and TMB chromogenic substrate (protected from light for 15 minutes). The reaction was terminated with 2M H₂SO₄ and read at dual wavelengths (450 nm/630 nm) using a microplate reader (BioTek Synergy H1, BioTek Instruments, Winooski, VT, USA).

Experimental samples comprised mouse brain tissue perfused and fixed with 4% paraformaldehyde (BL539A, Hefei Biosharp Biotechnology Co., Ltd., Hefei, Anhui, China), rapidly frozen in liquid nitrogen and stored at –80 °C for subsequent analysis. Protein samples (approximately 20 µg per well) were denatured in loading buffer and subjected to electrophoresis on a precast polyacrylamide gel (concentration gel: 80 V for 20 minutes; separation gel: 120 V for 60 minutes). Proteins were subsequently transferred to Polyvinylidene Fluoride (PVDF) membranes (Beyotime, Shanghai, China) using a semi-dry transfer method (PVDF membranes were activated with methanol prior to transfer; both gel and membrane were equilibrated in ice-cold transfer buffer). Transfer conditions were 25 V for 30 min (Buffer: 48 mM Tris, 39 mM glycine, 0.04% SDS, 20% methanol). The membrane was blocked with 5% skimmed milk powder (or bovine serum albumin [BSA] for phosphorylated proteins) at room temperature for 1 hour or overnight at 4 °C. It was then incubated sequentially with specific primary antibodies (glyceraldehyde-3-phosphate dehydrogenase (GAPDH, 1:8000, 60004-1-Ig, Proteintech Group, Inc., Rosemont, IL, USA); HMGB1 (1:1000, ab18256, abcam, Cambridge, UK); IL-1$\beta$ (1:1000, 12242, Cell Signaling Technology, Danvers, MA, USA); p-NF-$\kappa$B p65 (1:5000, AF3387, Affinity, Cincinnati, OH, USA); p-ERK1/2 (1:5000, AP0974, Abclonal, Woburn, MA, USA), incubated at room temperature for 2 hours or overnight at 4 °C) and the corresponding HRP-labelled secondary antibody (1:5000, ZB-2305 and AB-2301, Beijing YunTai Bio‑Technology Co., Ltd. Beijing, China). Thorough washing with TBST (P1020, Solarbio, Beijing, China) was performed between each antibody incubation step. Target proteins were ultimately visualised using ECL chemiluminescent reagents (WBKLS0100, Millipore, Burlington, MA, USA), with signals captured on an imaging system (NIKON DS-U3, Nikon Corporation, Tokyo, Japan). Exposure time was optimised according to band intensity.

Mouse brain tissue was perfused and fixed with 4% paraformaldehyde, followed by gradient dehydration and embedding in paraffin. Coronal sections (5 µm thickness) were prepared using a rotary microtome (Leica RM2016; Leica Microsystems (Shanghai) Trading Co., Ltd., Shanghai, China). Deparaffinization and rehydration were performed by sequential immersion in xylene (10023418, Sinopharm, Beijing, China, three changes, 10 min each) and a graded ethanol series (100%, 100%, 95%; 5 min each), followed by a final rinse in distilled water.

Antigen retrieval was carried out by heating the sections in sodium citrate buffer (C1013, Solarbio, 10 mM sodium citrate, 0.05% Tween 20 (ST825, Beyotime, Shanghai, China), pH 6.0) using a pressurized decloaking chamber for 3 minutes. Sections were allowed to cool naturally to room temperature. After washing with PBS, sections were permeabilized with 0.1% Triton X-100 in PBS for 20 minutes at room temperature and then blocked with 3% BSA for 30 minutes at room temperature to prevent non-specific binding.

Sections were incubated overnight at 4 °C in a humidified chamber with the following primary antibodies diluted in PBS: rabbit anti-HMGB1 (1:100, 10829-1-AP, Proteintech Group) and rabbit anti-IL-6 (1:100, 83747-5-RR, Proteintech Group). After thorough washing with PBS (3 $\times$ 5 minutes), sections were incubated with an AF488-conjugated goat anti-rabbit IgG (H + L) secondary antibody (A0423, 1:500, Beyotime) for 50 minutes at room temperature in the dark. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Beyotime, C1005) for 10 minutes at room temperature, protected from light.

Finally, the sections were washed with PBS, briefly air-dried, and mounted with an anti-fade mounting medium (P0126, Beyotime). Images were captured using a fluorescence microscope (Nikon Eclipse E100, Shanghai, China) at 400$\times$ magnification. Nuclei stained with DAPI appeared blue, while positive signals for HMGB1 and IL-6 were visualized as green fluorescence.

The modified neurological sign score (mNSS) was assessed blindly by two researchers, who were completely unaware of the experimental groups and treatments, on postoperative days 2 and 11, respectively. The scoring system encompasses four dimensions: motor function (limb symmetry, crawling ability), sensation (tactile/pain reflexes), balance (balance beam walking, righting reflex), and abnormal behaviour (tremors, circling). The total score ranges from 0 to 18 points (0 = no deficit, 18 = severe deficit). Prior to assessment, mice were acclimatised for 30 minutes in a quiet environment (25 $\pm$ 1 °C, light intensity 30 lux). Each test was repeated three times, with the mean value recorded. Non-specific behaviours resulting from anaesthesia or surgical trauma (e.g., incision pain reflex) were excluded.

Transcriptomic analysis revealed 1170 differentially expressed genes in the CCI group, comprising 511 upregulated and 659 downregulated genes. taVNS, the number of differentially expressed genes decreased to 929, including 455 upregulated and 474 downregulated genes. KEGG enrichment analysis confirmed that HMGB1-related pathway genes, such as Rat sarcoma (Ras) and mitogen-activated protein kinase (MAPK), were significantly enriched among the upregulated genes in the CCI group. Moreover, VNS, HMGB1-related pathway genes, including Ras-related protein 1 (Rap1), Ras, and MAPK, were markedly enriched among the downregulated genes in the VNS group compared to the CCI group (Fig. 2). Preliminary transcriptomic findings suggest that taVNS may regulate HMGB1 levels and inflammation by modulating HMGB1-associated pathways.

Fig. 2.

Transcriptome analysis image. (A) DEGs in CCI versus Sham (volcano plot). (B) DEGs in VNS versus CCI (volcano plot). (C) Pathway enrichment of CCI-upregulated genes identifies HMGB1 signaling as a key upregulated node. (D) Pathway enrichment of VNS-downregulated genes confirms HMGB1 signaling as a primary target for reversion. DEGs, differentially expressed genes; KEGG, Kyoto Encyclopedia of Genes and Genomes; cGMP, cyclic guanosine monophosphate; PKG, protein kinase G; PI3K-Akt, phosphoinositide 3-kinase–protein kinase B signaling pathway; cAMP, cyclic adenosine monophosphate; MAPK, mitogen-activated protein kinase; HMGB1, high-mobility group box 1.

Magnetic resonance imaging revealed marked cerebral edema in the group compared to the control group at days (Fig. 3A). Combined with analysis, this indicated significant neurological deficits in the group at day 3 distance travelled in the central zone of the open field test decreased by 60% compared to controls (915.20 $\pm$ 318.59 cm vs 2296.87 $\pm$ 505.15 cm, p 0.0001, Fig. 3B), and the mNSS increased to 9.10 $\pm$ 2.64 points (control group 0.50 $\pm$ 0.69 points, p 0.0001, Fig. 3C). At the molecular level, the traumatic group exhibited significantly higher serum concentrations of HMGB1 (1325.40 $\pm$ 202.64 vs 795.78 $\pm$ 433.77 pg/mL; p = 0.0409) and IL-6 (76.16 $\pm$ 10.35 vs 46.74 $\pm$ 19.29 pg/mL; p = 0.0088) compared to the control group. Western blot analysis revealed that the expression of HMGB1 protein was 3.7-fold higher, and that of IL-1$\beta$ was 2.8-fold higher, compared to the control group (both p 0.0001). Using immunofluorescence, we further observed that HMGB1-positive signals (green fluorescence) were significantly increased in the traumatic group compared to the control group (Fig. 3D). The aforementioned multidimensional data consistently confirmed the successful establishment of the model.

Fig. 3.

Successful establishment of the CCI model in mice. (A) Representative MRI images of the brain from the Sham and CCI groups at 3 days post-surgery, showing the distinct lesion area (The area outlined in red dashed lines). Scale bar = 1 mm. (B) Quantitative analysis of the distance traveled in the center of the open field, demonstrating reduced exploratory behavior in the CCI group compared to the Sham group. (C) mNSS scores assessed at 3 days post-surgery, indicating significant neurological deficits in the CCI group. (D) Representative immunofluorescence images showing enhanced expression of HMGB1 (green) in the CCI group compared to the Sham group. Cell nuclei were counterstained with DAPI. Scale bar = 25 µm. mNSS, modified neurological sign score; DAPI, 4′,6-diamidino-2-phenylindole.

ITK-SNAP three-dimensional reconstruction revealed that the taVNS group exhibited a significant reduction in cerebral volume of 74.7 $\pm$ 12.1% at 10 days compared to 3 days the trauma group demonstrated only natural absorption of 53.5 $\pm$ 16.2% (p = 0.024, Fig. 4E), the central area traversed by the taVNS group at 10 days recovered to 76.3% of the control group level (1753.97 $\pm$ 588.97 cm vs 915.20 $\pm$ 318.59 cm in the trauma group, p = 0.0031, Fig. 4A), while the mNSS score decreased to 2.58 $\pm$ 1.31 points (group: 9.10 $\pm$ 2.64 points, p 0.0001, Fig. 4B). Molecular analysis showed that the taVNS group resulted in a significant reduction in serum HMGB1 levels (853.39 $\pm$ 145.94 pg/mL) compared to the traumatic group (1325.40 $\pm$ 202.64 pg/mL; p = 0.0427). Similarly, serum IL-6 levels were also significantly lower in the taVNS group (47.83 $\pm$ 12.64 pg/mL) than in the traumatic group (76.16 $\pm$ 10.35 pg/mL; p = 0.0052). The protein expression of HMGB1 in brain tissue was reduced to a level 1.5 times that of the control group (compared to 3.7-fold in the trauma group, p = 0.0003), while IL-1$\beta$ levels were downregulated to 1.4-fold (compared to 2.82-fold in the trauma group, p = 0.0004) (Fig. 4D, the original WB images can be found in the Supplementary Materials). Immunofluorescence revealed that taVNS reversed - induced HMGB1 levels (Fig. 5). These findings confirm that taVNS accelerates resolution and neural functional by suppressing HMGB1-driven inflammation.

Fig. 4.

taVNS improves functional recovery, suppresses neuroinflammation, and promotes edema resolution post-TBI. (A) Analysis of the distance traveled in the center of the open field for all five experimental groups. (B) mNSS neurological scores across all five groups. (C) Serum levels of HMGB1 and IL-6 were quantified by ELISA in the five groups. (D) Protein expression levels of HMGB1 and IL-1$\beta$ in brain tissue were detected by western blot. The blots for HMGB1, IL-1$\beta$, and the loading control GAPDH are displayed. Representative images from three independent experiments are shown. (E) Left: Representative MRI images comparing the CCI and VNS groups after the 7-day intervention, demonstrating enhanced edema resolution with VNS. Scale bar = 1 mm. Right: quantitative analysis of cerebral edema from the four TBI groups, showing the edema volume reduction rate (left graph) and the absolute volume of edema resolved over 7 days (right graph). Data are presented as mean $\pm$ standard deviation (SD). *p 0.05, **p 0.01, ***p 0.001, ****p 0.0001 (by one-way ANOVA with Tukey’s post-hoc test). TBI, traumatic brain injury; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IL-1$\beta$, interleukin-1$\beta$.

Fig. 5.

taVNS suppresses HMGB1 and IL-6 expression in brain tissue post-TBI. (A) Representative immunofluorescence images from the five experimental groups. Cell nuclei were counterstained with 4^′,6-diamidino-2-phenylindole (DAPI) (blue). Positive signals for HMGB1 and IL-6 are shown in green. Scale bar = 25 μm. (B,C) Quantitative analysis of the mean fluorescence intensity for (B) HMGB1 and (C) IL-6 in the peri-lesion cortex. Data are presented as mean $\pm$ SD. *p 0.05, **p 0.01 (by one-way ANOVA with Tukey’s post-hoc test).