Vagus Nerve Stimulation Mitigates Sepsis-Induced Lung Injury via Pulmonary proBDNF Suppression

Abstract

Background

Sepsis-associated acute respiratory distress syndrome (ARDS) remains a major contributor to mortality among critically ill patients. Vagus nerve stimulation (VNS) may attenuate inflammatory injury, but the downstream molecular mechanisms underlying its protective effects in septic lung injury remain incompletely understood. The role of proBDNF in VNS-mediated immunoregulation remains unclear. This study aimed to explore whether pulmonary proBDNF contributes to VNS-induced protection against sepsis-associated lung injury.

Methods

Sepsis-associated lung injury was induced by cecal ligation and puncture (CLP) in rats. The effects of VNS on lung injury, inflammation, and pulmonary proBDNF expression were evaluated. Vagotomy was conducted to assess the dependence of VNS effects on intact vagal signaling. Peripheral cholinergic signaling was probed using methyllycaconitine, an α7 nicotinic acetylcholine receptor (α7nAChR) antagonist. Pulmonary overexpression of proBDNF was achieved via adeno-associated virus (AAV) delivery to determine its functional significance. Histological, molecular, and functional analyses assessed lung injury, inflammation, pulmonary proBDNF, pulmonary edema, and BALF total protein levels.

Results

VNS significantly alleviated CLP-induced lung injury, reduced pulmonary inflammatory cytokine expression, decreased pulmonary edema and BALF total protein levels, and partially restored barrier-associated gene expression. Pulmonary proBDNF levels were markedly increased after CLP but suppressed by VNS, whereas vagotomy further elevated proBDNF. Brain proBDNF levels remained largely unchanged across groups. α7nAChR blockade attenuated the inhibitory effect of VNS on pulmonary proBDNF. Immunofluorescence identified macrophages as a major source of proBDNF in injured lungs. Notably, AAV-mediated pulmonary overexpression of proBDNF weakened the beneficial effects of VNS on inflammation, pulmonary edema, alveolar-capillary barrier leakage, and lung damage.

Conclusion

VNS protects against sepsis-associated lung injury through suppression of elevated pulmonary proBDNF, a process dependent on intact vagal signaling and, at least in part, mediated by α7nAChR-dependent peripheral cholinergic pathways. These findings identify proBDNF as a possible downstream effector of VNS-mediated lung protection.

Introduction

Acute respiratory distress syndrome (ARDS) is a severe form of diffuse inflammatory lung damage that can be triggered by multiple insults, including severe infection, sepsis, trauma, and aspiration. It is characterized by disruption of the alveolar-capillary barrier, pulmonary edema, inflammatory cell infiltration, and progressive impairment of gas exchange, ultimately leading to hypoxemia.^1,2 Despite advances in ventilatory support, fluid management, and critical care, mortality remains high, particularly in sepsis-associated ARDS.^3–7 This persistent burden highlights the limitations of current treatments and the need to deepen our understanding of ARDS pathogenesis.

Recently, increasing attention has been paid to the involvement of the nervous system in regulating inflammation. The concept of neuroimmune interaction provides a useful framework for understanding systemic inflammatory responses, whereby peripheral inflammatory signals are sensed by neural circuits and modulated through reflex-like feedback pathways.^8,9 The vagus nerve (VN), a key component of the parasympathetic nervous system, serves as a major communication link between the central nervous system (CNS) and peripheral organs, helping to maintain homeostasis and restrain excessive inflammation.¹⁰

A growing body of evidence shows that the VN can suppress inflammation through the cholinergic anti-inflammatory pathway. This pathway, when activated, releases acetylcholine that acts on α7 nicotinic acetylcholine receptors (α7nAChR) to suppress pro-inflammatory cytokine production.^11,12 In experimental models of sepsis and acute lung damage, VNS has been reported to attenuate pulmonary inflammation, limit edema, and alleviate histological injury.¹³ However, the downstream molecular mediators of these protective effects, as well as the specific mechanisms operating within lung tissue during ARDS, remain incompletely understood.

From a neuroimmune perspective, downstream mediators of VNS are likely to share several key features: they respond to inflammatory or injury-related signals, have functional roles in both neural and immune systems, and participate in processes such as amplification of inflammation, cellular injury, or tissue repair.¹¹ The neurotrophin family fits well within this framework and has attracted growing interest in recent years. Among its members, brain-derived neurotrophic factor (BDNF) and its precursor, proBDNF, are of particular relevance. Traditionally, proBDNF/BDNF signaling has been studied in the CNS, where it regulates neuronal development and synaptic plasticity.¹⁴ However, more recent studies have shown that proBDNF and its receptor p75NTR are also expressed in a range of immune cells, with levels that change markedly under inflammatory or stress conditions.^15–17 Compared with mature BDNF, which is generally associated with TrkB-mediated cell survival, repair, and neuroplasticity,¹⁴ proBDNF has been linked to p75NTR-dependent stress responses, inflammation, and tissue injury.^15,18 These distinct biological profiles provide a rationale for focusing on proBDNF as a candidate mediator in the context of acute septic lung inflammation. In this study, proBDNF was selected as the main molecular focus because it represents an inflammation-responsive ligand/precursor molecule whose pulmonary expression was markedly induced after CLP and suppressed by VNS in our preliminary observations, whereas p75NTR, sortilin, TrkB, and downstream signaling pathways were considered related receptor or signaling components requiring future investigation. In addition, increased proBDNF expression has been reported in several inflammatory and autoimmune diseases, suggesting a role in immune regulation.¹⁸ Despite this, its contribution to pulmonary inflammation in ARDS and its potential involvement in VN-mediated regulation remain largely unclear.

Notably, the nucleus tractus solitarius, a key relay center for vagal afferent signaling in the brainstem, has been shown to be enriched in BDNF and its receptor TrkB.¹⁹ Immunohistochemical studies also support the presence of proBDNF in brainstem-associated nuclei,²⁰ pointing to a possible connection between the BDNF/proBDNF axis and vagal pathways. In parallel, VNS has been reported to modulate a range of signaling pathways involved in inflammation and stress responses,¹¹ some of which may intersect with proBDNF/p75NTR signaling.²¹ Although direct evidence is still limited, these observations provide a reasonable basis to consider proBDNF as a potential mediator of VN-dependent immunoregulation. In this context, we hypothesized that septic lung injury may induce pulmonary proBDNF as part of the inflammatory stress response, and VNS may confer lung protection, at least in part, by limiting proBDNF-associated inflammatory amplification through peripheral cholinergic signaling.

In our preliminary work, we observed a marked increase in pulmonary proBDNF expression in rats with CLP-induced sepsis-associated acute lung injury, while VNS significantly reduced its levels in lung tissues. These findings imply that proBDNF is associated with the protective effects of VNS and could play a role in shaping inflammatory responses in septic lung injury. To explore this further, we constructed a cecal ligation and puncture (CLP)-induced sepsis-associated acute lung injury model with ARDS-like pathological features and systematically examined the effects of VNS on lung injury, together with changes in proBDNF expression and their biological significance. Our aim was to provide experimental insight into the molecular mechanisms through which the vagus nerve confers protection against septic lung injury.

Materials and Methods

Animals

Male Sprague-Dawley rats (8–10 weeks old, 200–250 g) were obtained from a certified specific pathogen-free (SPF) animal facility. All animals were housed under standard conditions with controlled temperature and humidity and a 12 h light-dark cycle, with free access to water and food. All procedures were approved by the Animal Welfare and Ethics Committee of Dalian Medical University (Approval No. XL250911369) and were carried out in accordance with institutional guidelines for the care and use of laboratory animals.

Experimental Design

To explore the mechanisms underlying VNS-mediated protection in sepsis-associated acute lung injury, four separate experimental series were conducted. Animals were randomly assigned to experimental groups within each experimental series.

For all VNS-related experiments, animals in the stimulation groups underwent cervical VN electrode implantation followed by electrical stimulation. Animals in the corresponding non-stimulated groups underwent the same surgical procedures, including electrode placement, but without stimulation.

In the first series, the protective effects of VNS on CLP-stimulated inflammation and lung damage were assessed in three groups: control (CR), CLP (PR), and CLP plus VNS (PM).

In the second series, we examined the effects of VNS on pulmonary proBDNF expression and its potential central versus peripheral regulation. Five groups were included: control (CR), VNS alone (M), CLP (PR), CLP plus VNS (PM), and CLP plus left cervical vagotomy (D).

In the third series, the role of α7 nicotinic acetylcholine receptor (α7nAChR)-dependent cholinergic signaling was evaluated using four groups: control (CR), CLP (PR), CLP plus VNS (PM), and CLP plus VNS with methyllycaconitine (MLA) treatment (PM-L).

In the fourth series, the functional role of proBDNF in VNS-mediated lung protection was investigated using adeno-associated virus (AAV)-mediated pulmonary gene delivery. The groups were as follows: control plus AAV-scrproBDNF (CR-AAV-scrproBDNF), CLP plus AAV-scrproBDNF (PR-AAV-scrproBDNF), CLP plus VNS plus AAV-scrproBDNF (PM-AAV-scrproBDNF), CLP plus AAV-eoproBDNF (PR-AAV-eoproBDNF), and CLP plus VNS plus AAV-eoproBDNF (PM-AAV-eoproBDNF).

To account for any nonspecific effects of the viral vector, additional comparisons were made between untreated control animals and CR-AAV-scrproBDNF animals. These analyses were used only to assess the baseline safety and potential impact of intratracheal AAV administration and were not included in the main functional outcomes.

CLP-Induced Sepsis-Associated Acute Lung Damage Model

Sepsis-associated acute lung damage was induced using the CLP model. Rats were anesthetized with inhaled sevoflurane (induction: 5% and maintenance: 2.5–3.0%; oxygen flow rate: 2 L/min). A midline abdominal incision (~2 cm) was made to expose the cecum. The cecum was ligated at approximately one-half of the distance from the distal tip of the cecum, with care taken to preserve the ileocecal valve. The ligated cecum was then punctured once with an 18-gauge needle, creating two perforations, and a minimal volume of fecal material was gently extruded.

The cecum was carefully repositioned within the abdominal cavity, followed by layered closure of the incision. After surgery, animals received topical lidocaine cream for analgesia and subcutaneous saline (50 mL/kg, 0.9% NaCl) for fluid resuscitation.

Sham-operated animals underwent the same procedures, such as laparotomy and cecal manipulation, but without ligation or puncture. All CLP procedures were performed using the same ligation position, needle size, puncture number, fecal extrusion criterion, postoperative analgesia, and fluid resuscitation protocol to reduce procedure-related variability.

Vagus Nerve Stimulation

The left cervical VN was surgically exposed via a neck incision under sterile conditions. After careful isolation of the nerve and surrounding connective tissue, a stimulating electrode was placed directly on the surface of the left cervical VN, and a reference electrode was fixed in the ipsilateral subcutaneous cervical tissue. The stimulator body was implanted in a subcutaneous abdominal pocket, and the lead wire was tunneled subcutaneously from the abdominal region to the neck and connected to the electrode. After implantation, electrode impedance was checked, and an impedance value below 10 kΩ was considered indicative of adequate electrode connection.

VNS was initiated 4 h following CLP. The stimulation parameters were as follows: 5 V, 5 Hz, 2 ms pulse width, delivered for 10 min per session, once per hour for a total of six sessions. During stimulation, the general condition and respiratory rhythm of the animals were monitored to ensure procedural safety.

Animals in the non-stimulated groups underwent the same cervical exposure, electrode placement, stimulator implantation, and lead connection procedures but did not receive electrical stimulation.

Vagotomy

For vagotomy, the left cervical VN was exposed under anesthesia after CLP surgery. The nerve was completely transected, and a 2–3 mm segment was removed to prevent reconnection.

α7nAChR Blockade

To inhibit cholinergic anti-inflammatory signaling, the commonly used α7nAChR antagonist methyllycaconitine citrate (MLA; MedChemExpress, HY-N2332A) was administered intraperitoneally at 6 mg/kg (0.6 mg/mL, 10 mL/kg) 2 h before CLP. Control animals received an equal volume of sterile PBS.

AAV-Mediated proBDNF Overexpression

Pulmonary overexpression of proBDNF was achieved using recombinant adeno-associated virus serotype 6 (AAV6). Two constructs were used: AAV-scrproBDNF as a control vector and AAV-eoproBDNF for proBDNF overexpression. Both vectors were produced by Hanbio Biotechnology (Shanghai, China) under the control of a CMV promoter, with titers ≥ 1×10¹² vg/mL.

Next, 120 μL viral suspension was administered via intratracheal instillation under anesthesia. Animals were kept in a head-up position for 1–2 min after instillation to facilitate even distribution within the lungs. CLP surgery was performed 2 weeks later to allow stable gene expression.

Tissue Collection

Animals were euthanized 12 h after CLP under deep anesthesia by exsanguination via the inferior vena cava and abdominal aorta. Lung tissues were collected and processed for different analyses. The apex of the left lung was employed for histology and transmission electron microscopy, while the remaining left lung tissue was reserved for immunohistochemistry and immunofluorescence. The right upper lobe was used for flow cytometry or RT-qPCR, and portions of the right middle and lower lobes were used for Western blotting and ELISA. For lung wet/dry weight ratio analysis, a separate portion of the right lung from non-lavaged animals was collected immediately after euthanasia and weighed to obtain the wet weight.

For BALF total protein measurement, three additional animals per group in the AAV experimental series were used. In these animals, bronchoalveolar lavage was performed before lung tissue collection, and the recovered BALF was processed for total protein analysis.

Brain tissues, including the cortex and hippocampus, were also collected for immunohistochemical analysis.

Histological Examination

Lung tissues were fixed in paraformaldehyde (4%), paraffin-embedded, and sectioned at 4 μm. One representative section per animal was analyzed. Sections underwent staining with hematoxylin and eosin (H&E). Five randomly selected, non-overlapping fields per section were examined at ×200 magnification. Lung damage was evaluated using a semiquantitative scoring system by two investigators blinded to group allocation.

Immunohistochemistry

After deparaffinization, the sections were rehydrated and subjected to antigen retrieval. Endogenous peroxidase activity and nonspecific binding were blocked before incubation. Sections were incubated overnight at 4°C with a rabbit anti-proBDNF antibody (Biosensis, R-176-20), followed by an HRP-conjugated secondary antibody. Signals were developed using DAB, and nuclei were counterstained with hematoxylin. Stained sections were scanned using a digital whole-slide scanning system. For lung tissue, one representative section per animal was analyzed, and three non-overlapping fields were selected from the corresponding tissue region under equivalent ×200 magnification. For brain tissue, one representative section per animal was analyzed, and the cortex and hippocampal regions were identified based on anatomical landmarks on whole-slide images, and three non-overlapping fields were selected from each region under equivalent ×200 magnification. ProBDNF immunoreactivity was quantified using the H-score method with SaiViewer image analysis software (Sevier, China). H-score was calculated as follows: H-score = percentage of weakly positive cells × 1 + percentage of moderately positive cells × 2 + percentage of strongly positive cells × 3, with a theoretical range of 0–300. The average value of the selected fields was used as the final value for each animal. All image analyses were performed under blinded conditions.

Immunofluorescence Staining

Double immunofluorescence staining was performed to evaluate proBDNF expression in macrophages. Sections were first exposed to an anti-CD68 antibody (Servicebio, GB11067) and visualized using TSA-Alexa Fluor 488. After antibody stripping, sections were incubated with an anti-proBDNF antibody and detected using TSA-Cy3. Nuclei underwent counterstaining with DAPI. Analyses were performed on lung tissues from the CR, PR, and PM groups. Fluorescence sections were scanned using a digital whole-slide scanning system. For each animal, one representative lung section was analyzed, and three non-overlapping fields were selected from the corresponding lung tissue region under equivalent ×200 magnification. Images were analyzed with identical image analysis parameters across groups. Quantitative analysis of CD68-positive cells, proBDNF-positive cells, and CD68/proBDNF double-positive cells was performed using SaiViewer image analysis software (Sevier, China). The average value of the three selected fields was used as the final value for each animal. All image analyses were performed under blinded conditions.

Western Blotting

Lung tissues were lysed in RIPA buffer containing protease inhibitors, and protein content was determined using a BCA assay. Equal amounts of protein (20 μg) were separated via SDS-PAGE and transferred onto PVDF membranes. Membranes were incubated with an anti-proBDNF primary antibody, followed by an HRP-labeled secondary antibody. β-actin (Abcam, ab8227) served as a loading control. Signals were detected by chemiluminescence and quantified using ImageJ.

ELISA

Pulmonary proBDNF levels were measured using a rat-specific ELISA kit (Jianglai Biological, JL21232) in accordance with the kit’s protocol.

Lung Wet/Dry Weight Ratio

Lung wet/dry weight ratio was measured to assess pulmonary edema. A separate portion of the right lung from non-lavaged animals was collected immediately after euthanasia and weighed to obtain the wet weight. The samples were then dried at 60°C for 48 h until a stable dry weight was reached. The wet/dry weight ratio was calculated by dividing the wet weight by the dry weight.

BALF Total Protein Measurement

BALF total protein concentration was measured to assess alveolar-capillary barrier leakage. For BALF collection, the trachea was exposed and cannulated after euthanasia. The lungs were gently lavaged three times with 1 mL fresh cold sterile PBS each time. The recovered BALF from the three lavages was pooled and centrifuged at 4°C to remove cells and debris. The supernatant was collected for total protein measurement. BALF total protein concentration was determined using a BCA protein assay kit (Beyotime Biotechnology, P0012, Shanghai, China) according to the manufacturer’s instructions. Each BALF sample was measured in duplicate, and the average value of the technical duplicates was used as one biological replicate for statistical analysis.

qRT-PCR

Total RNA was isolated with TRIzol reagent and reverse-transcribed into cDNA. qRT-PCR was carried out to measure the expression of inflammatory cytokines (Tnf and Il6) and barrier-related genes (Tjp1 and Cldn5). Actb was employed as the internal reference gene. Relative expression levels were measured using the 2^−ΔΔCt approach. The primer sequences were as follows: Tnf (Forward: CCAACAAGGAGGAGAAGT; Reverse: GTATGAAATGGCAAATCG), Il6 (Forward: AGTTGTGCAATGGCAATTCTGA; Reverse: CTCTGGCTTTGTCTTTCTTGTTATCTTT), Tjp1 (Forward: GATGAGCGGGCTACCTTA; Reverse: ATGGGAGCGAACTGAATG), Cldn5 (Forward: TTGGAAGGGGCTGTGGATG; Reverse: TGAGCGCCGGTCAAGGTAA), Actb (Forward: GATCATTGCTCCTCCTGAGC; Reverse: CGTCATACTCCTGCTTGCTG).

Flow Cytometry

Single-cell suspensions were prepared from lung tissue, followed by staining with APC-Cy7-conjugated anti-CD45 antibody (BD Pharmingen, 561586) and FITC-conjugated anti-CD11b antibody (BD Pharmingen, 554982). Following exclusion of dead cells by 7-AAD staining, samples were analyzed by flow cytometry using FlowJo software.

Transmission Electron Microscopy

Small lung tissue specimens (~1 mm³) were fixed in electron microscopy fixative, followed by postfixing in osmium tetroxide, dehydration, and embedding in epoxy resin. Ultrathin sections were prepared and stained with lead citrate and uranyl acetate. Samples were examined under a transmission electron microscope to assess ultrastructural changes in alveolar type II epithelial cells.

Reagents and Antibodies

Primary antibodies included anti-proBDNF (Biosensis, R-176-20), anti-CD68 (Servicebio, GB11067), and anti-β-actin (Abcam, ab8227). For flow cytometry, APC-Cy7-conjugated anti-CD45 (BD Pharmingen, 561586) and FITC-conjugated anti-CD11b (BD Pharmingen, 554982) were used. Methyllycaconitine citrate (MLA) was purchased from MedChemExpress (HY-N2332A), and the rat proBDNF ELISA kit was obtained from Jianglai Biological (JL21232).

Statistical Analysis

Data are presented as mean ± SEM. Each group included six animals for the main lung tissue-based analyses. For ELISA, all available non-lavaged samples were analyzed when applicable. For molecular and imaging-based assays, including Western blotting, RT-qPCR, immunofluorescence, immunohistochemistry, and flow cytometry, three animals per group were randomly selected. In the AAV experimental series, pulmonary proBDNF ELISA was performed using six non-lavaged animals per group, whereas other tissue-based assays and lung wet/dry weight ratio analysis were performed using three animals per group. BALF total protein was measured using three additional animals per group. Transmission electron microscopy was used for representative morphological assessment and was not quantified. No a priori sample size calculation was performed; therefore, the sample size information is reported descriptively for each assay.

Statistical tests were performed with GraphPad Prism. Differences between groups were assessed using Student’s t-test or one-way ANOVA with Tukey’s multiple comparisons test. P < 0.05 was deemed statistically significant. Exact P values, mean differences, and 95% confidence intervals for the statistically significant main comparisons in Figures 1–5 are provided in Supplementary Table S6 where applicable. For values below the reporting precision of the statistical software, P values are reported as P < 0.0001.

Figure 1.

Vagus nerve stimulation (VNS) alleviates CLP-stimulated lung damage and inflammation. (A) Experimental design. Rats underwent cecal ligation and puncture (CLP) at 0 h. VNS was initiated 4 h later using an electrode placed on the cervical vagus nerve with a subcutaneous reference electrode. Stimulation was delivered intermittently (5 V, 5 Hz, 2 ms pulse width, 10 min per session, once per hour for six sessions). Lung tissues were collected at 12 h. (B) Representative histological findings. Histological samples from control (CR), VNS alone (M), CLP (PR), and CLP plus VNS (PM) groups. CLP caused significant alveolar damage, inflammatory cell infiltration, and septal thickening, which were attenuated by VNS. Scale bar: 50 μm. (C) Lung injury scores. Quantitative assessment of lung injury in CR, M, PR, PM, and vagotomized CLP (D) groups. Each point represents an individual animal. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (D) Ultrastructural changes in alveolar type II cells. TEM images from CR, PR, and PM groups. CLP induced mitochondrial swelling and disruption of lamellar bodies, which were partially preserved by VNS. Scale bar: 2 μm. (E) Pulmonary inflammatory cytokine expression. Relative mRNA levels of Tnf and Il6 in lung tissue from CR, PR, and PM groups, measured by quantitative PCR. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3 per group). (F) Barrier-related gene expression. Relative mRNA levels of tight junction-associated genes (Tjp1 and Cldn5) in lung tissue from CR, PR, and PM groups. VNS partially restored barrier gene expression following CLP. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). *P < 0.05, **P < 0.01, ****P < 0.0001.

Figure 2.

VNS suppresses pulmonary proBDNF expression with minimal effects in the brain. (A) Pulmonary proBDNF immunohistochemistry. Representative immunohistochemical staining of proBDNF in lung tissue from CR, M, PR, PM, and vagotomized CLP (D) groups. CLP markedly increased pulmonary proBDNF expression, which was reduced by VNS and further increased after vagotomy. Scale bar: 50 μm. (B) Quantitative assessment of pulmonary proBDNF. H-score-based semi-quantitative analysis of proBDNF staining across groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (C) Western blot analysis of pulmonary proBDNF. Representative immunoblots showing proBDNF expression in lung tissue from CR, M, PR, PM, and D groups. β-actin = loading control. (D) Densitometric analysis. Quantification of proBDNF protein levels normalized to β-actin. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (E) ELISA measurement of pulmonary proBDNF. Pulmonary proBDNF levels in CR, PR, PM, and D groups. CLP significantly increased proBDNF levels, which were partially reduced by VNS and further elevated after vagotomy. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=6/group). (F) Brain proBDNF immunohistochemistry. Representative immunohistochemical staining of proBDNF in brain tissue from CR, PR, PM, and D groups. CLP induced a modest increase in proBDNF expression compared with the CR group, whereas neither VNS nor vagotomy produced noticeable changes. Scale bar: 50 μm. (G) Quantitative analysis of brain proBDNF. H-score–based analysis of proBDNF staining in brain tissue from CR, PR, PM, and D groups. Although CLP increased proBDNF expression relative to CR, no obvious differences were found among PR, PM, and D groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). *P < 0.05, ***P < 0.001, ****P < 0.0001.

Abbreviation: ns, not significant.

Figure 3.

Vagal integrity is required for VNS-mediated lung protection and suppression of pulmonary proBDNF. (A) Lung histopathology. Representative histological samples from CR, PR, PM, and D groups. CLP caused significant alveolar septal thickening and inflammatory cell infiltration. These changes were attenuated by VNS but reappeared following vagotomy. Scale bar: 50 μm. (B) Lung injury scores. Quantitative assessment of lung injury in CR, PR, PM, and D groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (C) Pulmonary proBDNF expression. Semi-quantitative H-score analysis of pulmonary proBDNF levels in PR, PM, and D groups. VNS reduced proBDNF expression, whereas vagotomy restored the elevated levels induced by CLP. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (D) Pulmonary inflammatory cytokine expression. Relative mRNA levels of Tnf and Il6 in lung tissue from PR, PM, and D groups, determined by quantitative PCR. VNS decreased inflammatory cytokine expression, whereas cytokine expression increased after vagotomy. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4.

α7nAChR-mediated cholinergic signaling contributes to VNS-induced suppression of pulmonary proBDNF. (A) Pulmonary proBDNF immunohistochemistry. Representative immunohistochemical staining of proBDNF in lung tissue from CR, PR, PM, and MLA-treated (PM-L) groups. VNS reduced proBDNF expression compared with the PR group, whereas α7nAChR blockade attenuated this VNS-associated reduction. Scale bar: 50 μm. (B) Quantitative analysis of pulmonary proBDNF. H-score–based quantification of proBDNF staining in CR, PR, PM, and PM-L groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (C) ELISA measurement of pulmonary proBDNF. Pulmonary proBDNF levels in CR, PR, PM, and PM-L groups, determined by ELISA. VNS decreased proBDNF expression relative to the PR group, while MLA treatment attenuated this VNS-associated reduction. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=6/group). **P < 0.01, ****P < 0.0001.

Figure 5.

AAV-mediated proBDNF overexpression attenuates the protective effect of VNS against CLP-stimulated lung damage. (A) Pulmonary proBDNF immunohistochemistry. Representative immunohistochemical staining of proBDNF in lung tissue from CR-AAV-scrproBDNF, PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups. AAV-mediated overexpression markedly increased pulmonary proBDNF levels. Scale bar: 50 μm. (B) Western blot analysis of pulmonary proBDNF. Representative immunoblots showing proBDNF protein expression in lung tissue from CR-AAV-scrproBDNF, PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups. β-actin = loading control. (C) Densitometric analysis. Quantification of proBDNF protein levels normalized to β-actin across CR-AAV-scrproBDNF, PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (D) Lung histopathology. Representative histological samples from CR-AAV-scrproBDNF, PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups. Under control vector conditions, VNS attenuated CLP-induced lung injury, whereas proBDNF overexpression weakened this protective effect. Scale bar: 50 μm. (E) Lung injury scores. Quantitative assessment of lung injury across CR-AAV-scrproBDNF, PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (F) Pulmonary inflammatory cytokine expression. Relative mRNA levels of Tnf and Il6 in lung tissue from PR-AAV-scrproBDNF, PM-AAV-scrproBDNF, PR-AAV-eoproBDNF, and PM-AAV-eoproBDNF groups, determined by quantitative PCR. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n=3/group). (G) Lung wet/dry weight ratio in the AAV experimental groups. CLP increased the lung wet/dry weight ratio, whereas VNS reduced this increase under control vector conditions. Pulmonary proBDNF overexpression attenuated the VNS-associated reduction in the wet/dry weight ratio. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n = 3/group). (H) BALF total protein concentration in the AAV experimental groups. CLP increased BALF total protein levels, whereas VNS reduced this increase under control vector conditions. Pulmonary proBDNF overexpression attenuated the VNS-associated reduction in BALF total protein levels. Mean ± SEM. One-way ANOVA with Tukey’s multiple comparisons test (n = 3/group using additional animals). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Abbreviation: ns, not significant.

Figure 5.

Continued.

Clinical Trial Number

Not applicable.

Results

VNS Alleviates CLP-Stimulated Lung Damage and Inflammatory Responses

To examine the protective role of VNS in sepsis-associated acute lung damage, a CLP-induced rat model was established following the experimental timeline shown in Figure 1A.

H&E analysis showed that lung structure was largely preserved in the CR and M groups, with thin alveolar septa and minimal inflammatory infiltration. Conversely, the PR group displayed pronounced pathological alterations, including disruption of alveolar architecture, thickened septa, and extensive inflammatory cell infiltration. These alterations were remarkably attenuated in the PM group (Figure 1B).

Consistent with these observations, lung injury scores were dramatically higher in the PR group than in the CR group, whereas VNS markedly reduced injury severity in the PM group (Figure 1C).

TEM further supported these findings. In the CR group, alveolar type II epithelial cells exhibited intact ultrastructure. In the PR group, however, clear signs of subcellular damage were observed, including mitochondrial swelling and disorganization of lamellar bodies. These abnormalities were partially reversed by VNS in the PM group, suggesting preservation of epithelial integrity under septic conditions (Figure 1D). Additional representative images are provided in Figure S2A.

To further assess inflammation, the expression of pro-inflammatory cytokines was determined by quantitative PCR. Compared to the CR group, the PR group exhibited a marked upregulation of Tnf and Il6 levels, whereas VNS significantly reduced their levels in the PM group (Figure 1E).

We next examined genes associated with barrier function. The levels of the tight junction-related genes Tjp1 and Cldn5 were markedly downregulated in the PR group but partially restored following VNS treatment (Figure 1F). Similar trends were observed across additional groups (Figure S3A).

Flow cytometric analysis revealed that CLP markedly upregulated the proportion of CD45⁺CD11b⁺ myeloid cells in the lung, whereas VNS significantly reduced their accumulation (Figure S1).

Taken together, these results indicate that VNS alleviates CLP-stimulated lung damage, suppresses pulmonary inflammation, and partially restores lung barrier integrity.

VNS Suppresses Aberrant Pulmonary proBDNF Expression While Having Minimal Effects on Brain proBDNF

To examine how VN activity influences proBDNF expression in sepsis-associated lung injury, we assessed proBDNF levels in both lung and brain tissues and explored its cellular localization in the injured lung.

Immunohistochemical staining showed low basal proBDNF expression in lung tissue from the CR and M groups. In contrast, CLP markedly increased proBDNF immunoreactivity in the PR group. This increase was significantly attenuated by VNS in the PM group, whereas vagotomy led to a further rise in proBDNF staining in the D group (Figure 2A).

Semi-quantitative analysis was consistent with these observations. Pulmonary proBDNF levels were remarkably higher in the PR group than in the CR group, reduced in the PM group, and further elevated following vagotomy (Figure 2B).

Western blotting confirmed these results at the protein level. ProBDNF expression was significantly upregulated after CLP, downregulated with VNS treatment, and markedly elevated after vagotomy (Figure 2C and D). ELISA measurements showed a similar pattern, with proBDNF levels rising after CLP, partially decreasing with VNS, and increasing further in the D group (Figure 2E).

To identify the cellular source of proBDNF in the lung, double immunofluorescence staining was performed. A substantial proportion of proBDNF-positive cells co-localized with the macrophage marker CD68, indicating that macrophages are a major source of proBDNF in injured lung tissue. Both the proportion of CD68⁺ macrophages among total cells and the fraction of proBDNF-positive macrophages within the CD68⁺ population were dramatically increased in the PR group and partially reduced by VNS (Figure S4).

We next examined whether VNS affected proBDNF expression in the brain. Immunohistochemical analysis showed a slight increase in brain proBDNF levels following CLP compared with the CR group. However, no marked differences were found among the PR, PM, and D groups (Figure 2F and G).

Taken together, these findings suggest that VNS primarily regulates proBDNF expression in peripheral lung tissue, with minimal impact on the CNS in this model.

Vagal Integrity is Required for VNS-Associated Lung Protection and Suppression of Pulmonary proBDNF

To further assess the role of vagal signaling in the protective effects of VNS, vagotomized animals (D group) were compared with the PR and PM groups.

Histological analysis showed that CLP-induced lung injury was markedly alleviated by VNS but worsened following vagotomy. In the PR group, lung tissue displayed severe pathological changes, including thickened alveolar septa and prominent inflammatory cell infiltration. These changes were markedly reduced in the PM group but became more pronounced in the D group (Figure 3A).

Consistent with these observations, lung injury scores were remarkably lower in the PM group than in the PR group, whereas vagotomy led to a significant increase in injury severity (Figure 3B).

We next examined pulmonary proBDNF expression. Compared with the PR group, proBDNF levels were markedly reduced in the PM group but remarkably increased after vagotomy (Figure 3C), indicating that the suppressive effect of VNS on proBDNF depends on intact vagal signaling.

Inflammatory responses were further evaluated by measuring Tnf and Il6 expression. VNS remarkably downregulated the expression of these cytokines in the PM group, whereas vagotomy markedly increased their levels (Figure 3D).

Together, these results show that the protective role of VNS in lung damage, as well as its ability to suppress pulmonary proBDNF and inflammation, depends on preserved vagal signaling.

α7nAChR-Mediated Cholinergic Signaling Participates in VNS-Induced Suppression of Pulmonary proBDNF

To determine whether cholinergic anti-inflammatory signaling mediates the effect of VNS on proBDNF expression, the α7nAChR antagonist methyllycaconitine (MLA) was used to block peripheral cholinergic signaling.

Immunohistochemical analysis showed that pulmonary proBDNF expression was markedly higher in the PR group than in the CR group. This increase was significantly reduced by VNS in the PM group. However, MLA treatment largely attenuated this effect, resulting in elevated proBDNF expression in the PM-L group (Figure 4A).

Quantitative H-score analysis confirmed these findings, showing increased proBDNF levels after CLP, a reduction with VNS, and a significant rebound following MLA administration (Figure 4B). ELISA measurements showed a similar pattern, with proBDNF levels decreased by VNS but rising again after α7nAChR blockade (Figure 4C).

Collectively, these results indicate that the suppressive effect of VNS on pulmonary proBDNF expression is mediated, at least in part, through α7nAChR-dependent cholinergic signaling.

AAV-Mediated Overexpression of proBDNF Attenuates the Protective Effect of VNS Against CLP-Stimulated Lung Damage

To determine whether proBDNF functionally contributes to VNS-mediated lung protection, we established AAV-mediated pulmonary overexpression of proBDNF prior to CLP induction.

Immunohistochemistry confirmed successful proBDNF overexpression. Compared with the CR-AAV-scrproBDNF group, proBDNF levels were markedly elevated in PR-AAV-scrproBDNF animals. VNS reduced pulmonary proBDNF in PM-AAV-scrproBDNF animals, whereas animals receiving AAV-eoproBDNF showed persistently high proBDNF levels in both PR-AAV-eoproBDNF and PM-AAV-eoproBDNF groups (Figure 5A).

Western blotting and ELISA analyses further corroborated these findings, confirming elevated proBDNF protein in the overexpression groups (Figure 5B and C; Figure S5). Histological examination revealed that VNS markedly alleviated CLP-induced lung injury in control virus animals. In contrast, under conditions of proBDNF overexpression, severe lung injury persisted despite VNS treatment (Figure 5D). Consistently, lung injury scores were significantly reduced by VNS in control virus animals but remained elevated in both PR-AAV-eoproBDNF and PM-AAV-eoproBDNF groups (Figure 5E).

Inflammatory cytokine analysis further supported these findings. Notably, VNS significantly suppressed Tnf and Il6 expression in PM-AAV-scrproBDNF animals, whereas proBDNF overexpression weakened this anti-inflammatory effect (Figure 5F). Similarly, overexpression of proBDNF impaired the restoration of lung barrier-related genes (Figure S3B), and ultrastructural analysis further supported these observations (Figure S2B). To further evaluate pulmonary edema and alveolar-capillary barrier leakage, lung wet/dry weight ratio and BALF total protein concentration were measured in the AAV experimental series. CLP significantly increased the lung wet/dry weight ratio and BALF total protein levels in PR-AAV-scrproBDNF animals compared with CR-AAV-scrproBDNF animals. VNS significantly reduced both indices in PM-AAV-scrproBDNF animals. However, pulmonary proBDNF overexpression further increased the wet/dry weight ratio and BALF total protein levels, and VNS failed to significantly reduce these changes under proBDNF-overexpressing conditions (Figure 5G and H). These results provide additional functional evidence that sustained pulmonary proBDNF elevation weakens VNS-mediated protection against CLP-induced pulmonary edema and alveolar-capillary barrier leakage.

Together, these findings indicate that sustained elevation of pulmonary proBDNF weakens the protective effects of VNS against CLP-induced lung injury, suggesting that proBDNF may act as an important downstream contributor to VNS-regulated lung protection.

Discussion

Herein, we investigated the immunomodulatory effects of VNS in a CLP-induced rat model of sepsis-associated acute lung injury with ARDS-like pathological features. By integrating observations of lung injury phenotypes, inflammatory mediators, and proBDNF expression, we systematically evaluated the impact of VNS on CLP-stimulated lung injury. Our results show that VNS provides marked protection against septic lung injury, accompanied by a marked attenuation of elevated pulmonary proBDNF levels.

The role of VN in systemic inflammatory regulation is well established. The cholinergic anti-inflammatory pathway, when activated, inhibits pro-inflammatory cytokine release and confers protection in sepsis and acute lung damage models.^11,22 Prior studies have shown that VNS can attenuate pulmonary inflammation and enhance histological outcomes in experimental ARDS.^13,23 However, most reports have focused on cytokine profiles, canonical signaling pathways, or organ-level outcomes, with limited exploration of molecular changes within the injured lung tissue itself.^22,23 Here, in addition to confirming the protective effects of VNS, we identified pulmonary proBDNF as a molecular feature closely linked to these effects, providing mechanistic insight at the tissue level into how VNS mediates lung protection.

ProBDNF, the precursor of mature BDNF, is increasingly recognized as having biological functions distinct from its mature form. Although BDNF-TrkB signaling is generally associated with cell survival, repair, and plasticity, proBDNF preferentially engages p75NTR-dependent pathways, which are associated with stress responses, inflammation, and tissue injury.^24–26 Prior reports have shown that proBDNF levels rise in various inflammatory and injury-related conditions, where it may contribute to immune dysregulation, cellular damage, and disruption of tissue homeostasis.^15,18,27 However, its role in acute lung injury and ARDS remains poorly defined. In our model, pulmonary proBDNF levels were markedly elevated following CLP, correlating with worsened lung injury, increased pro-inflammatory cytokine expression, and impaired barrier-related gene expression. Importantly, VNS significantly mitigated this abnormal increase, suggesting that proBDNF is not merely a passive marker of tissue injury but may actively participate in shaping the inflammatory microenvironment in septic lung injury.

A key finding of this study is that VNS-mediated regulation of proBDNF occurs predominantly in peripheral tissues rather than in the CNS. Although CLP induced a modest increase in brain proBDNF, neither VNS nor vagotomy significantly altered its levels. In contrast, pulmonary proBDNF responded dynamically to both vagal stimulation and vagotomy. Moreover, pharmacological blockade of α7nAChR largely attenuated the suppressive effect of VNS on pulmonary proBDNF, indicating that this regulation is at least partly mediated via the peripheral cholinergic anti-inflammatory pathway rather than central mechanisms. However, because MLA was used as a pharmacological antagonist, potential off-target effects cannot be fully excluded. Therefore, the MLA-based findings should be interpreted as supportive evidence for the involvement of α7nAChR-dependent cholinergic signaling rather than definitive proof of pathway specificity. These results support the notion that proBDNF functions as a peripheral, inflammation-associated mediator in septic lung injury, rather than being restricted to a central neurotrophic role. The findings from vagotomy experiments further emphasize the importance of vagal signaling. Unilateral left cervical vagotomy aggravated CLP-induced lung injury and further increased pulmonary proBDNF and inflammatory cytokine levels, in contrast to the protective effects observed with left-sided VNS. Although vagotomy was performed without concurrent stimulation, the loss of protection along the same anatomical pathway targeted by VNS supports an important role of intact vagal signaling in modulating pulmonary inflammation and proBDNF expression.

To better define the cellular context of proBDNF expression in the lung, we combined immunofluorescence staining with flow cytometric analysis. Macrophages are well-established contributors to the initiation and progression of inflammation in ARDS.^28,29 In our study, double immunofluorescence revealed that CD68-positive macrophages accounted for a substantial portion of proBDNF-expressing cells in lung tissues. Both the number of pulmonary macrophages and the fraction of proBDNF-positive macrophages were increased following CLP and were partially reduced by VNS. These findings indicate that pulmonary proBDNF expression is closely associated with inflammatory immune cell populations, with macrophages serving as a major cellular source. This observation aligns with prior reports demonstrating that proBDNF and p75NTR can be induced in monocyte/macrophage lineage cells under inflammatory or stress conditions, where they contribute to immune regulation.¹⁷ In our model, proBDNF expression was not randomly distributed but appeared tightly linked to immune cell infiltration and activation. Consistently, flow cytometric analysis demonstrated a marked increase in pulmonary CD45⁺CD11b⁺ myeloid cells after CLP, which was attenuated by VNS. Collectively, these data suggest that VNS may modulate the inflammatory microenvironment of the lung, influencing both immune cell recruitment and the molecular characteristics of these cells, including proBDNF expression.

Despite these findings, the relationship between proBDNF and pulmonary inflammation requires careful interpretation. The CLP model represents a complex inflammatory environment involving multiple interacting cell types and signaling pathways. Therefore, although our results demonstrate a clear association between proBDNF expression, inflammatory severity, and immune cell infiltration, they do not establish a direct causal relationship. It remains unclear whether proBDNF acts as a driver of inflammation, a consequence of immune activation, or both. Tissue-level observations alone are insufficient to fully define its cellular function under inflammatory conditions.

To address this question, we conducted functional intervention experiments using AAV-mediated pulmonary overexpression of proBDNF. Under control conditions, VNS significantly improved lung histopathology, decreased inflammatory cytokine levels, and lowered lung injury scores. However, sustained overexpression of proBDNF markedly diminished the protective effect of VNS. Functional assessments further supported these findings: VNS reduced the lung wet/dry weight ratio and BALF total protein concentration under control vector conditions, whereas sustained pulmonary proBDNF overexpression weakened these protective effects. These results imply that, although proBDNF is not the sole determinant of lung injury severity, its expression level critically influences the efficacy of VNS-mediated protection. Therefore, proBDNF likely functions as an important modulatory factor within the inflammatory network, limiting the beneficial effects of VNS when excessively elevated, rather than serving solely as a passive marker of tissue injury.

Nevertheless, several limitations should be acknowledged. First, all analyses were conducted at a single early time point following CLP, which does not capture dynamic alterations in proBDNF expression, immune cell recruitment, lung injury progression, long-term recovery, survival outcomes, progression of ARDS-like pathology, or sustained responses to VNS and proBDNF modulation over time. Second, although each experimental group included six animals for the main lung tissue-based analyses, several molecular, imaging-based, and functional assays were performed using three animals per group, and no a priori sample size calculation was performed. This may limit the statistical power of some comparisons, particularly multi-group comparisons and semi-quantitative analyses. Third, only male rats were used in this study; therefore, potential sex-related differences in VNS responsiveness, inflammatory regulation, and proBDNF expression require further investigation. Fourth, although lung wet/dry weight ratio and BALF total protein measurement were added to strengthen the functional assessment of pulmonary edema and alveolar-capillary barrier leakage, more direct permeability assays, such as Evans blue extravasation or tracer-based permeability assays, were not performed. Therefore, the conclusions regarding barrier protection should be interpreted as barrier-related functional evidence rather than definitive proof of complete barrier restoration. Fifth, although macrophages were identified as a major source of pulmonary proBDNF, the current in vivo findings do not fully clarify whether proBDNF directly regulates macrophage function or contributes to lung injury through other cell types and signaling interactions. Moreover, the immune analyses were limited to inflammatory cytokine expression, CD45⁺CD11b⁺ myeloid cell assessment, and CD68/proBDNF double immunofluorescence, without comprehensive macrophage subtype or cell-specific functional analyses. In addition, mature BDNF, p75NTR, sortilin, TrkB, and downstream signaling pathways were not directly examined in this study; therefore, the proposed involvement of proBDNF-related receptor signaling remains inferential. Sixth, several intervention-related methodological issues should be considered when interpreting the mechanistic findings. Although MLA was used to probe α7nAChR-dependent cholinergic signaling, its pharmacological specificity remains a limitation. In addition, although electrode impedance and respiratory rhythm were monitored during VNS, heart rate, autonomic parameters, and fiber-specific activation were not directly recorded. The vagotomy experiments should also be interpreted cautiously, because unilateral cervical vagotomy may influence autonomic balance, respiration, gastrointestinal function, and systemic stress responses. Finally, because left-sided cervical VNS was used consistently in the stimulation experiments, the present findings should be interpreted within the context of left-sided VNS and should not be directly extrapolated to right-sided VNS. Future studies directly comparing left- and right-sided VNS will be needed to determine whether laterality-dependent autonomic or pulmonary effects exist in this model.

Despite these limitations, the current findings consistently show that VNS attenuates CLP-induced lung injury and pulmonary inflammation and suppresses pulmonary proBDNF elevation, whereas sustained pulmonary proBDNF overexpression weakens VNS-mediated protection. Future studies using complementary models, multiple time points, cell-specific approaches, and human samples will be needed to further clarify the cellular mechanisms of proBDNF-mediated lung injury and validate the translational relevance of these findings.

In summary, our study provides integrative evidence that proBDNF is an important downstream component of vagus nerve-mediated protection in septic lung injury. Future studies employing cell-specific approaches, particularly focusing on alveolar macrophages and targeted modulation of proBDNF signaling, will be necessary to further clarify the underlying mechanisms. Such research may help determine whether targeting proBDNF has therapeutic relevance for neuroimmune regulation in sepsis-associated lung injury.

Conclusion

VNS exerts protective effects in a CLP-induced rat model of sepsis-associated acute lung injury with ARDS-like pathological features by attenuating lung injury and suppressing inflammatory responses. VNS reduces the abnormal elevation of pulmonary proBDNF at least in part through the peripheral cholinergic anti-inflammatory pathway, without corresponding effects in the CNS. The level of pulmonary proBDNF is closely associated with the efficacy of VNS-mediated lung protection, and sustained pulmonary overexpression of proBDNF markedly diminishes this protective effect. Additionally, proBDNF expression in septic lung injury is closely associated with changes in immune cell composition, with pulmonary macrophages representing a major proBDNF-expressing population. These findings suggest that proBDNF is an important molecular component of the inflammatory microenvironment in sepsis-associated lung injury and may contribute to VNS-mediated lung protection.

Funding Statement

This research was funded by the Beijing Life Oasis Public Welfare Service Center (Grant No. HXLX20240041). The funder was not involved in the study design, data handling, analysis, or publication process.

Data Sharing Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Ethics Approval

All animal procedures complied with the ARRIVE guidelines and the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Welfare and Ethics Committee of Dalian Medical University (Approval No. XL250911369).

Author Contributions

YG: Conceptualization, Methodology, Investigation, Formal analysis, Data curation, Visualization, Writing – original draft. NW: Investigation, Formal analysis, Data curation, Validation, Visualization, Writing – review and editing. ZX: Conceptualization, Methodology, Supervision, Project administration, Funding acquisition, Resources, Writing – review and editing. YG and NW contributed equally to this work. All authors made significant contributions to the conception and design of the study, acquisition of data, or analysis and interpretation of data. All authors participated in drafting, revising, or critically reviewing the article; reviewed and approved all versions of the article before submission and during revision; agreed on the journal to which the article has been submitted; and agree to take responsibility and be accountable for all aspects of the work.

Disclosure

The authors declare no competing interests in this work.