Potential targets for CA were predicted using the SwissTargetPrediction database (http://www.swisstargetprediction.ch). ARDS-related targets were identified using public databases, including the GeneCards database (https://www.genecards.org/) and the DisGeNET database (version 7.0; http://www.disgenet.org/). To identify relevant targets, the search terms ‘ARDS’, ‘acute lung injury’, ‘acute respiratory distress syndrome’, ‘adult respiratory distress syndrome’ and ‘acute lung injury’ were used. In the GeneCards database, target selection was carried out using the following criteria: i) Ranking targets by their relevance score; ii) retaining only those with a score >2 times the median value and iii) identifying and presenting the common targets between CA and ARDS using a Venn diagram.

Functional enrichment analysis. Metascape (http://www.metascape.org) is an online tool for functional annotation. This platform facilitates gene function annotation analysis, empowering users to leverage contemporary, prevalent bioinformatics methodologies in the examination of gene and protein batches. By doing so, it enables the elucidation of gene and protein functionalities (12). The tool was used to carry out a functional enrichment analysis of the intersected target genes using Gene Ontology (GO) (13) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses (14). The GO analysis encompassed three categories: i) Biological processes (BP); ii) molecular functions (MF); and iii) cellular components (CC). The top 10 most significantly enriched terms in both GO-BP and KEGG pathway analyses were then visualized using the SRplot bioinformatics platform (http://www.bioinformatics.com.cn/; last accessed on April 1, 2025). P<0.05 was considered to indicate a statistically significant difference.

Protein-protein interaction (PPI) network construction. Overlapping targets were analyzed using Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; https://cn.string-db.org/; version 11.0) to construct a PPI network. This network was constructed with a specific focus on Homo sapiens and a high confidence score of 0.950 to ensure the reliability of interactions. Following this, the PPI network was visually rendered using Cytoscape software (version 3.7.2) (15), facilitating a comprehensive visualization and analysis of the intricate protein interactions.

Molecular docking. Molecular docking has emerged as a fundamental technique in modern drug discovery, allowing researchers to predict binding affinities and molecular interactions by analyzing the three-dimensional conformations of protein-ligand complexes. In the present study, molecular docking simulations were carried out between CA and its key target proteins identified through the aforementioned analysis. The three-dimensional structures of CA and the pivotal proteins were sourced from PubChem (https://pubchem.ncbi.nlm.nih.gov/) and the Research Collaboratory for Structural Bioinformatics Protein Data Bank database (http://www.rcsb.org/). All docking procedures were executed using AutoDock Tools (version 1.5.6; https://autodock.scripps.edu/), with validation steps incorporated to verify the reliability of the computational predictions. Analysis of molecular docking utilized binding energy to assess protein-ligand interactions, with values ≤-5 kcal/mol indicating strong binding affinity (16). To assess structural alignment, root mean square deviation (RMSD) calculations were employed. The RMSD lower bound (RMSD l.b.) and the RMSD upper bound (RMSD u.b.) allow the diversity of conformations in the molecular docking results, and the deviation range from the target conformation to be understood. This is useful for evaluating the reliability and accuracy of the docking results. Generally, the smaller the RMSD l.b., the more similar the conformation obtained from the docking is to the target conformation, and the more reliable the docking result is. Conversely, the RMSD u.b. can assist in determining whether the docking algorithm is capable of searching a sufficiently wide enough conformational space. To facilitate a deeper understanding of the docking results, the molecular docking patterns were visually represented using PyMOL software (version 2.4; https://www.pymol.com/pymol.html), allowing for a detailed examination of the interactions at a molecular level.

MD. Groningen Machine for Chemical Simulations (GROMACS; version 2022.4; http://www.gromacs.org) was used to carry out MD simulations, which were all executed under isothermal-isobaric ensemble conditions simulating a system with a constant number of particles, temperature and pressure (NPT) with periodic boundary constraints. The linear constraint solver algorithm constrained hydrogen bonds with a 2 fsec integration time step. Electrostatic energy (ELE) interactions were calculated using the particle-mesh Ewald method, with a 1.2-nm cutoff (17), while van der Waals (VDW) interactions employed a 10 Å cutoff distance, updated every 10 steps. Temperature (298 K) and pressure (1 bar) were maintained using the V-rescale thermostat and Berendsen barostat algorithms, respectively (18). The system underwent sequential equilibration under the protocol of 100 ps (constant number of particles, volume and temperature) followed by 100 ps NPT, before a 100 ns production run, with trajectories saved every 10 ps. Post-simulation analyses included trajectory visualization using Visual Molecular Dynamics (https://www.ks.uiuc.edu/Research/vmd/) and PyMOL, along with binding free energy calculations using the GROMACS package ‘g_mmpbsa’ (19). Key evaluation metrics comprised: i) System stability parameters (RMSD of complex protein and ligand; radius of gyration of complex; root mean square fluctuation of protein residues); ii) interaction characteristics (binding site distance, buried solvent-accessible surface area, VDW/ELE energies, hydrogen bond dynamics and per-residue energy contributions); and iii) structural features (conformational superposition and interaction fingerprints). This comprehensive analysis protocol ensured rigorous assessment of the protein-ligand complex dynamics.

Most of the reagents and antibodies used in the present study are shown in Table I.

Animals and treatments. Male BALB/c mice (n=48; age, 8 weeks; weight, 21-25 g), were obtained from the Model Animal Research Center of Nanjing University. Mice were housed in a standard laboratory environment maintained at a temperature range of 20-26˚C, with a relative humidity of 40-70% and a 12-h light/dark cycle. The mice were allowed to feed and drink ad libitum throughout the present study. All animal experiments were approved by the Institutional Ethics Committee of Medical School of Nanjing University Affiliated Jinling Hospital (Nanjing, China; approval no. 2022DZGKJDWLS-00161) and complied with the Guide for the Care and Use of Laboratory Animals (https://olaw.nih.gov/policies-laws/guide-care-use-lab-animals).

The ARDS model was established through intraperitoneal (i.p.) injection of lipopolysaccharide (LPS; 10 mg/kg), a well-established method documented in previous studies (20,21). Mice were randomly allocated into the following four experimental groups (n=6/group): i) Control group (PBS; i.p.); ii) LPS group (10 mg/kg; i.p.); iii) CA group (40 mg/kg; i.p.); and iv) ML385 group (40 mg/kg CA + 30 mg/kg ML385; i.p.). All animals were anesthetized with 1% sodium pentobarbital (40 mg/kg; i.p.) before administration. A total of 30 min post-LPS injection, treatment groups received either CA (dissolved in 0.5% DMSO) or the combination of ML385 [nuclear factor erythroid 2-related factor 2 (Nrf2) inhibitor; dissolved in 0.5% DMSO; 30 mg/kg, i.p.] (22) followed by CA, while controls received equivalent volumes of PBS. Survival rates were recorded every 24 h for a 7-day observation period following the LPS challenge. Animal deaths were checked for twice daily (morning and evening) using a combination of criteria: i) The absence of spontaneous movement for >5 min of observation; ii) the lack of a response to gentle tactile stimuli (e.g., a soft touch with a clean gloved finger); and iii) the absence of visible respiration and signs of a heartbeat. For mice reaching the predefined moribund state (severe lethargy, inability to rise, dyspnea, etc.), euthanasia was performed promptly with an overdose of sodium pentobarbital (100 mg/kg, i.p.), and the time of euthanasia was recorded as the time of death for survival analysis. Upon confirmation of death, a necropsy was immediately performed. The thoracic cavity was exposed, and the lung tissues were carefully excised en bloc. Subsequently, the lungs were perfused via the trachea with ice-cold phosphate-buffered saline (PBS) to clear residual blood. After blotting the surface liquid with filter paper, the tissues were processed according to downstream experimental needs: The left lung lobe was fixed in 4% paraformaldehyde at 4˚C for 24 h for histological analysis, while the right lung lobe was snap-frozen in liquid nitrogen and then transferred to a -80˚C freezer for storage, and reserved for subsequent protein or RNA extraction. All tissue samples were maintained at -80˚C until use.

Following euthanasia, 0.6-0.8 ml left ventricular blood was immediately collected for blood gas analysis testing of arterial oxygen partial pressure (PaO₂) levels and arterial carbon dioxide partial pressure (PaCO₂), FiO₂ was measured during sampling using an oxygen monitor. The PaO₂/FiO₂ ratio was then calculated as PaO₂ (mmHg) divided by FiO₂. Bronchoalveolar lavage fluid (BALF) was obtained by carrying out three cycles of instillation and withdrawal with 1-ml sterile PBS through a tracheal cannula. The BALF samples were centrifuged at 1,500 x g for 15 min at 4˚C and the resulting supernatant was aliquoted and stored at -80˚C for subsequent cytokine analysis. For histological examination, the left lung lobes were perfused with PBS and fixed in 4% paraformaldehyde at 4˚C for 24 h before paraffin embedding. Tissue sections (4-µm thick) were prepared and stained with H&E using standard protocols (hematoxylin: 0.1%, 5-8 min, room temperature; eosin: 0.5%, 1-3 min, room temperature). Lung pathology was examined under a light microscope (Olympus BX53) by an observer blinded to the groupings. Lung injury scores were graded using the following scale: 0, no injury; 2, mild injury (up to 25% injury of the field of view); 4, moderate injury (up to 50% injury of the field of view); 6, severe injury (up to 75% injury of the field of view); and 8, extremely severe injury (diffused injury) (10). The upper right lung lobes were excised and weighed to determine the wet/dry (W/D) ratio by comparing fresh weight to weight after 72 h of desiccation at 60˚C. Remaining lung tissues were quickly frozen in liquid nitrogen and maintained at -80˚C for future protein semi-quantification through ELISA and western blotting. This comprehensive sample collection and processing protocol enabled systematic evaluation of physiological parameters, histopathological changes, and molecular markers through techniques including ELISA, western blotting, RT-qPCR, and immunohistochemical (IHC) analysis using commercial assay kits, thereby providing a multi-faceted assessment of the ARDS model.

In the survival study (n=9 per group), mice received the following treatments: The ARDS model was induced by a single i.p. injection of LPS (10 mg/kg). After 30 min, the CA group received CA (40 mg/kg, i.p.), while the CA + ML385 group received ML385 (30 mg/kg, i.p.) followed by CA (40 mg/kg, i.p.), as previously described (23). Survival rates were recorded every 24 h for a 7-day observation period following LPS challenge and euthanized at the moribund stage. The humane endpoints were defined as the occurrence of moribund animal characteristics in the mice, such as severe or persistent lethargy, inability to rise from a recumbent position, convulsions, tetraplegia and dyspnea (24,25). Mortality was assessed through twice-daily monitoring (morning and evening) to ensure accurate determination of survival time points.

Cell culture. MH-S cells (murine alveolar macrophage cells) were obtained from The Cell Bank of Type Culture Collection of The Chinese Academy of Sciences. Cells were maintained in a culture medium consisting of DMEM, supplemented with 10% FBS and 1% penicillin and streptomycin. The cells were incubated at 37˚C in an environment containing 5% CO₂. Following the respective treatments, cell culture supernatants were collected for subsequent analysis by centrifugation of the culture medium at 1,000 x g for 5 min at 4˚C to remove cellular debris.

For in vitro investigations, MH-S cells were systematically allocated into the following four groups: i) Control group; ii) LPS/ATP group; iii) CA-10 group; and iv) CA-20 group. MH-S cells in the LPS/ATP, CA-10 and CA-20 groups were initially exposed to LPS (1 µg/ml) for 4 h at 37˚C, followed by ATP stimulation (5 mM) for 30 min at 37˚C (26). Subsequently, the CA-10 and CA-20 groups received either 10 or 20 µM CA and were further incubated for 48 h at 37˚C, whereas the control groups received equivalent volumes of culture medium.

Cell viability assay. A Cell Counting Kit-8 (CCK-8) assay was used to assess cell viability according to the instructions of the manufacturer. MH-S cells were plated in 96-well plates at a concentration of 5x10³ cells/well and left to adhere for 24 h under standard culture conditions. The cells were then exposed to CA (0, 0.1, 1, 10, 15, 20, 25, 30 and 35 µM) for 48 h at 37˚C. After treatment, each well received 10 µl CCK-8 solution (5 mg/ml) and the plates were incubated for 1 h at 37˚C in an environment with 5% CO₂. Absorbance was read at 450 nm with the model 680 microplate reader (Bio-Rad Laboratories, Inc.).

Reverse transcription-quantitative PCR (RT-qPCR) analysis. In accordance with a previous study (23), total RNA was extracted from lung tissues and cultured cells using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.), strictly adhering to the guidelines of the manufacturer. Subsequently, cDNA synthesis was performed using the RT Kit (Vazyme Biotech Co., Ltd.) according to the manufacturer's protocol. The qPCR amplification was then conducted on the SmartCycler® II system (Cepheid) using the following thermocycling conditions: Initial denaturation at 95˚C for 30 sec, followed by 40 cycles of denaturation at 95˚C for 10 sec and annealing/extension at 60˚C for 30 sec. A melting curve analysis was performed immediately after amplification to verify the specificity of the PCR products. GAPDH as the endogenous control. Relative gene expression levels were calculated using the 2^-ΔΔCq method (27). The specificity of the PCR amplification was confirmed by a single peak in the melting curve analysis. The primer sequences used are listed in Table II.

Western blotting. Protein lysates were prepared from both treated MH-S cells and mouse lung tissues using RIPA buffer supplemented with protease and phosphatase inhibitors. Protein concentrations were evaluated through the BCA assay kit, according to the manufacturer's protocol. Prior to electrophoresis, protein samples were mixed with 5X Laemmli sample buffer containing 5% β-mercaptoethanol (or 100 mM DTT) as a reducing agent and denatured by boiling at 95-100˚C for 5-10 min. For western blotting, protein lysates were adjusted to a concentration of 0.75 µg/µl for lung tissues or 1.0 µg/µl for MH-S cells using RIPA buffer. A total of 15 µg protein/lane was loaded for all samples, corresponding to 20 µl for lung tissues and 15 µl for cell lysates. Proteins were then separated by SDS-PAGE on 8-12% gels (depending on target protein size) under reducing conditions. The resolved proteins were then electrophoretically transferred onto PVDF membranes (0.45-µm pore size; MilliporeSigma) using a semi-dry transfer system.

Each target protein and its loading control, GAPDH, were ultimately detected on the same physical membrane to ensure direct comparability. This was achieved through sequential detection (stripping and re-probing) on a single membrane: For analysis of the inflammasome pathway, the membrane was first probed for NLRP3, ASC, and caspase-1 (p20). After imaging, the antibodies were stripped, and the same membrane was then re-probed for GAPDH. For analysis of GSDMD cleavage, the membrane was first probed for GSDMD-FL and GSDMD-N. After imaging, the antibodies were stripped, and the same membrane was then re-probed for GAPDH. Electrophoresis conditions were optimized to ensure sufficient separation between the GAPDH (~36 kDa) and GSDMD-N (30-35 kDa) bands prior to stripping and re-probing, allowing for accurate post hoc alignment and densitometric analysis.

Following transfer, all membranes were stained with Ponceau S to confirm uniform protein loading and transfer efficiency. The membranes were then blocked with 5% skimmed milk in Tris-buffered saline containing 0.1% Tween-20 (TBST) at 4˚C overnight (or for 1 h at room temperature if applicable). After washing three times with TBST (10 min each), the membranes were incubated with corresponding HRP-conjugated secondary antibodies diluted in 5% skimmed milk/TBST for 1 h at room temperature. Immunoreactive bands were detected using the Enhanced ECL Kit (cat. no. P0018; Beyotime Biotechnology) according to the manufacturer's instructions, and visualized using a Tanon-5200 Imaging System (Tanon Science & Technology Co., Ltd.). Between probing cycles, the membranes were stripped using a mild stripping buffer. Complete removal of previous antibodies was verified by incubating the membrane with ECL substrate and confirming the absence of signal prior to the next round of antibody incubation. Densitometric analysis of the bands was performed using ImageJ software (V1.37; National Institutes of Health). The signal intensity of each target protein was normalized to that of GAPDH obtained from the same membrane lane.

Immunohistochemical (IHC) analysis. Firstly, lung tissue samples were preserved in 4% paraformaldehyde at 4˚C for 1 day, processed through a graded ethanol series and embedded in paraffin blocks. Serial sections (5 µm) were then cut using the RM2235 rotary microtome (Leica Biosystems). The staining procedure involved: i) Deparaffinization in xylene (three changes, 5 min each); ii) rehydration through a graded ethanol series (100-70%); iii) antigen retrieval in citrate buffer (pH 6.0) at 95˚C for 20 min; iv) endogenous peroxidase blocking with 3% H₂O₂ for 10 min at room temperature; and v) non-specific site blocking with 5% goat serum for 30 min at room temperature.

Sections were incubated with a primary antibody against gasdermin D (GSDMD; 1:1,000) overnight at 4˚C in a humidified environment. After three washes with PBS (5 min each), the sections were exposed to an HRP-conjugated secondary antibody (1:200) for 20 min at room temperature, followed by 3,3'-diaminobenzidine chromogen development (6 min) with microscopic monitoring. Counterstaining was performed with hematoxylin at room temperature for 30 sec, followed by dehydration through graded ethanol and xylene clearing. Slides were examined under the LSM700 confocal microscope (Zeiss GmbH). Semi-quantitative analysis of IHC staining was performed using ImageJ software.

For the co-detection of GSDMD-FL and GSDMD-N, the same membrane was used. Due to the close molecular weights of GSDMD-N and GAPDH, which could lead to overlapping bands and imprecise semi-quantification, GAPDH was probed on a separate membrane. Similarly, other target proteins (NLRP3, caspase-1 and ASC) were detected on dedicated membranes to optimize antibody conditions and avoid excessive stripping/re-probing cycles. To ensure perfect comparability between different membranes, total protein lysates for all samples were run on two identical SDS-PAGE gels in parallel under the same electrophoretic conditions. Proteins were then transferred in parallel onto separate PVDF membranes under identical transfer conditions. To control for loading and transfer variations, the membranes were stained with Ponceau S (Sigma-Aldrich) after transfer. Densitometric analysis of the total protein stain (Ponceau S) for each lane was performed using ImageJ software, and the signal intensity of each target protein was normalized to the total protein signal of its corresponding lane.

Immunofluorescence analysis. Cells were rinsed with PBS, fixed with 4% paraformaldehyde for 10 min at room temperature and then washed three times with TBST (5 min each). Permeabilization with 0.1% Triton X for 20 min at room temperature was carried out after another three TBST washes (5 min each). Primary antibodies of Nrf2 (as aforementioned) diluted in blocking buffer were then applied and incubated overnight at 4˚C in a humidified chamber. Subsequently, the cells were washed, incubated with secondary antibody for 1 h in the dark at room temperature and rinsed three times with TBST. Each sample received 200 µl antifade mounting medium with DAPI and visualization was performed using the LSM700 confocal microscope. Semi-quantitative analysis of IF staining was performed using ImageJ software.

ELISA. Concentrations of IL-1β and IL-18 in MH-S cell culture supernatants and BALF were quantified using ELISA kits following the manufacturer's protocols. Standards and samples (100 µl/well) were loaded in duplicate onto 96-well plates and incubated at 37˚C for 90 min. After washing, 100 µl biotinylated detection antibody was added to each well and incubated for 60 min at 37˚C. After three washes, 100 µl HRP-conjugated streptavidin was added and incubated for 30 min at 37˚C in the dark. The enzymatic reaction was terminated by adding 50 µl stop solution and the absorbance was immediately measured at 450 nm (with 570 nm correction) using the Bio-Rad Model 680 microplate reader (Bio-Rad Laboratories, Inc.). Cytokine concentrations were interpolated from a four-parameter logistic standard curve generated with reference standards.

Annexin V-FITC staining. MH-S cells were cultured in a 24-well plate. After treatment, the plates were centrifuged at 1,000 x g for 5 min at room temperature, the supernatant was removed and the cells were washed twice with PBS. Subsequently, the cells were stained with 250 µl Annexin V-FITC/PI solution for 15 min at room temperature in the dark. Images were then acquired using an IX73 inverted fluorescence microscope (Olympus Corporation).

Lactate dehydrogenase (LDH) release assay. Cytotoxicity was assessed by quantifying LDH activity in cell culture supernatants and BALF using a commercial LDH detection kit assay, following the product description. A total of 100 µl cell-free supernatant from each sample was transferred to a 96-well plate and incubated with the reaction mixture for 30 min at room temperature in the dark. The enzymatic reaction was terminated by adding stop solution and absorbance was immediately measured at 450 nm using the model 680 microplate reader (Bio-Rad Laboratories, Inc.).

Assessment of intracellular reactive oxygen species (ROS) levels. Intracellular ROS levels were quantified using 2',7'-dichlorodihydrofluorescein diacetate (DCFH-DA) as a fluorescent probe. Cells were seeded in 24-well plates (5x10⁴ cells/well) and treated as indicated for 48 h at 37˚C. Subsequently, the cells were incubated with 10 µM DCFH-DA in serum-free medium at 37˚C for 20 min in the dark. After three washes with warm DMEM to remove excess probe, ROS detection was carried out using the IX73 inverted fluorescence microscope (Olympus Corporation).

Measurement of oxidative stress markers. Myeloperoxidase (MPO), malondialdehyde (MDA) and superoxide dismutase (SOD) oxidative stress marker activity was quantitatively analyzed in both cell culture supernatants and BALF using commercial assay kits according to the manufacturer's protocols. All spectrophotometric measurements were carried out in triplicate using the model 680 microplate reader (Bio-Rad Laboratories, Inc.), with appropriate blank and standard controls included in each assay plate.

Statistical analysis. All quantitative data, except for survival data, are presented as the mean ± standard deviation (SD) from at least three independent experiments, with each experiment carried out in triplicate. Statistical analyses were carried out using GraphPad Prism (version 6.0; Dotmatics). For comparisons between multiple groups, one-way analysis of variance (ANOVA) was performed: for data with homogeneous variances, Tukey's post hoc test was used; for data with heterogeneous variances, Welch's ANOVA followed by Games-Howell's post hoc test was employed. For the survival study, Kaplan-Meier survival curves were generated, and differences between groups were compared for statistical significance using the log-rank (Mantel-Cox) test. In all analyses, a P-value of less than 0.05 (P<0.05) was considered to indicate a statistically significant difference.

The two-dimensional structure of CA was retrieved from the PubChem database (Fig. 1A). Target prediction analysis using the SwissTargetPrediction platform identified 106 potential CA target genes (probability score, >0.1). In addition, 496 ARDS-associated genes were compiled from comprehensive disease databases (GeneCards, 387 targets; DisGeNET, 109 targets) after removing duplicates. Comparative analysis through a Venn diagram revealed 19 overlapping genes between CA targets and ARDS-related genes (Fig. 1B), representing potential therapeutic targets through which CA may exert its pharmacological effects against ARDS. These 19 candidate targets were subsequently prioritized for further network pharmacology analysis.

Functional enrichment analysis. Bioinformatics analysis using the Metascape platform revealed significant functional enrichment among the 19 common targets, identifying 466 GO BP, 22 GO CC, 56 GO MF and 37 KEGG pathways. BP analysis demonstrated that these targets were predominantly involved in ‘regulation of reactive oxygen species metabolic process’ (GO: 2000377), ‘positive regulation of cell death’, ‘regulation of apoptotic signaling pathway’ and ‘regulation of DNA-binding transcription factor activity’, suggesting these BP may mediate the therapeutic effects of CA in ARDS (Fig. 1C). MF analysis results demonstrated that the common target genes were most involved in ‘metallopeptidase activity’, ‘serine-type peptidase activity’ and ‘serine hydrolase activity’. CC analysis results indicated that the common target genes were the most predicated on the ‘extracellular matrix’, ‘external encapsulating structure’ and ‘tertiary granule’. Results of the KEGG analysis revealed that the 19 genes were mainly enriched in ‘Fluid shear stress and atherosclerosis’, ‘Pathways in cancer’, ‘Lipid and atherosclerosis’ and the ‘IL-17 signaling pathway’ (Fig. 1D).

PPI network construction. Among the 19 genes, ADCY10 was excluded from PPI network construction as it exhibited no association with the other genes. The finalized network comprised 18 nodes and 80 edges, illustrating the complex web of interactions among the remaining targets (Fig. 1E). Hub targets were subsequently identified to uncover the core characteristics of the network. These key nodes were pinpointed based on their degree values, which effectively captured their relative importance and connectivity within the network, providing insight into the functional architecture of the network.

Molecular docking and MD. Following comprehensive analysis of the PPI network, NFE2L2 (encoding the Nrf2 protein) emerged as the pivotal target. Subsequently, to explore the potential binding mode and interaction mechanism, molecular docking experiments were conducted between NFE2L2 and CA. A total of 10 different docking sites were assessed, with findings indicating that the docking affinity was <-5 kcal/mol (Table III). Subsequently, the top-ranked conformation from the molecular docking results, which demonstrated the highest binding affinity, was subjected to MD simulations for further validation. All results suggested that NFE2L2 and CA are capable of stable binding (Fig. 2). As shown in Fig. 2A-C, the RMSD of the complex, protein and ligand CA, the radius of gyration, and the root-mean-square fluctuation of amino acid residues all stabilized over the 100 nsec molecular dynamics simulation, indicating that the complex system reached equilibrium and the binding conformation remained stable. Further binding analysis revealed that the distance between CA and the Nrf2 binding site, as well as the buried solvent-accessible surface area, remained stable during the later stages of the simulation (Fig. 2D-E). Moreover, superimposed simulation conformations demonstrated that CA stably bound the initial docking site (Fig. 2F). Binding free energy calculations (molecular mechanics Poisson-Boltzmann surface area) revealed that van der Waals interactions served as the primary driving force for binding. Additionally, residual energy decomposition analysis identified key amino acids, such as GLY-464 and VAL-463, that contributed significantly to the binding energy (Fig. 2H). These results confirm at the atomic level that CA and Nrf2 can form a stable complex, providing a theoretical foundation for subsequent experimental validation.

A CCK-8 assay was used to assess the cytotoxic effects of CA on MH-S cells. The results demonstrated that low CA concentrations (1-20 µM) did not significantly affect cell viability. Based on these findings, 10 and 20 µM CA concentrations were selected for subsequent experimental investigation (Fig. 3A).

CA inhibits pyroptosis in LPS/ATP-stimulated MH-S cells. To assess the impact of CA on pyroptosis, MH-S cells were treated with LPS and ATP. Pyroptosis was then quantified by measuring the proportion of PI-positive cells (Annexin V-FITC and PI double staining, a combination of methods evaluating membrane integrity and inflammasome activation) and LDH levels in the cell supernatant. The findings indicated a notable reduction in both LDH release (Fig. 3B) and PI-positive cells (Fig. 3C and D) following CA treatment. During the early stages of apoptosis and pyroptosis, phosphatidylserine becomes externalized on the cell membrane. However, pyroptosis ultimately leads to complete membrane rupture. Consequently, a cell population positive for Annexin V-FITC (green) and negative for PI (red) indicates an early stage of cell death, while cells positive for both Annexin V and PI represent a later stage characterized by loss of membrane integrity. The results demonstrated that LPS/ATP stimulation significantly increased the proportion of double-positive cells, an effect that was effectively reversed by CA treatment. To assess NLR family pyrin domain containing 3 (NLRP3) inflammasome activation, key markers were examined, including apoptosis-associated speck-like protein containing a CARD (ASC), caspase-1 (CASP1), the N-terminal fragment of GSDMD and the downstream cytokines IL-18 and IL-1β. CA treatment significantly inhibited the protein expression of these key markers induced by LPS/ATP and their corresponding mRNA upregulation (Fig. 3E-G). Furthermore, CA markedly attenuated the release of IL-18 and IL-1β in the supernatant compared with the LPS/ATP-stimulated group (Fig. 3H). Collectively, these outcomes indicated that CA may effectively hinder LPS/ATP-induced pyroptosis in MH-S cells by inhibiting NLRP3 activation and averting GSDMD cleavage.

CA activates Nrf2 and alleviates oxidative stress in LPS/ATP-stimulated MH-S cells. CA has previously been documented to exhibit potent antioxidative properties (28); however, its protective effects against oxidative stress in MH-S cells remain unclear. Thus, the antioxidant effects and underlying mechanisms of CA in LPS/ATP-challenged MH-S cells were investigated.

To assess the antioxidant properties of CA, the key oxidative stress markers MDA, MPO and SOD were analyzed. CA treatment significantly reduced MDA and MPO levels while enhancing SOD activity in MH-S cells compared with the LPS/ATP group (Fig. 4A-C). In addition, LPS/ATP stimulation led to a notable increase in intracellular ROS production in MH-S cells compared with that in the control group, with CA treatment effectively mitigating this increase (Fig. 4D and E).

Subsequently, the molecular mechanisms responsible for the antioxidant properties of CA were explored. Nrf2, a regulator of cellular antioxidant responses, was focused on to determine its involvement in CA-mediated oxidative stress protection (29). The findings demonstrated that, compared to the LPS/ATP-stimulated group, CA treatment significantly augmented the protein expression of both total Nrf2 and heme oxygenase-1 (Fig. 4F and G) and facilitated the nuclear translocation of Nrf2 (Fig. 4H). Collectively, these findings suggested that CA may alleviate oxidative stress and activate Nrf2 in MH-S cells.

CA attenuates LPS-induced ARDS in mice. Arterial blood gas analysis showed that, in comparison with the control group, the LPS group exhibited a notable decrease in PaO₂ and PO₂/FiO₂ levels, and a marked increase in PaCO₂ levels (Fig. 5A-C). These characteristic alterations in blood gas parameters confirm the successful establishment of the ARDS mouse model (23). The mice that received treatment with CA demonstrated an elevation in PaO₂ levels and a significant decrease in PaCO₂ levels, indicating a marked improvement in pulmonary gas exchange, thus supporting its potential for clinical application.

The effects of CA against LPS-induced ARDS were subsequently evaluated by performing histopathological analysis of lung tissues using H&E staining (Fig. 5D). Compared with in the control group, LPS-challenged mice exhibited significantly elevated histological scores, characterized by extensive inflammatory cell infiltration, alveolar hemorrhage and interstitial edema (Fig. 5E). Notably, CA treatment markedly attenuated these pathological alterations and significantly improved histological scores. These findings provide histological evidence for the protective role of CA in mitigating LPS-induced acute lung injury.

To further assess pulmonary edema formation, lung tissue water content was quantified by measuring the W/D weight ratio. As shown in Fig. 5F, ARDS mice exhibited a significantly elevated lung W/D ratio compared with that in the control group, indicating severe pulmonary edema. Notably, CA treatment effectively normalized this parameter. Together with the histological findings, these results demonstrated that CA could confer marked protection against ARDS by attenuating pathological vascular permeability and edema formation.

Consistent with its reversal of CA-mediated improvements in lung injury parameters (Fig. 5A-F), pharmacological inhibition of Nrf2 with ML385 also significantly attenuated the survival benefit conferred by CA. Survival analysis via the Kaplan-Meier method revealed marked differences between groups (log-rank test, Fig. 5G). LPS-challenged mice only exhibited a 20% survival rate, treatment with CA markedly improved survival rates to >50%. Pharmacological inhibition of Nrf2 with ML385 attenuated the protective effects of CA, thus reducing survival rates. These findings suggested that CA may provide a notable survival benefit in LPS-induced ARDS, mediated through Nrf2 pathway activation.

CA inhibits NLRP3 inflammasome-mediated pyroptosis in mice with LPS-induced ARDS through activation of Nrf2. To further explore the anti-pyroptotic effect of CA and its protective role in LPS-induced ARDS, the impact of CA on NLRP3 inflammasome-mediated pyroptosis was investigated in vivo. Quantitative analysis revealed that LPS challenge induced pyroptotic activation, as evidenced by significant upregulation of NLRP3 inflammasome components (NLRP3, ASC, CASP1 and GSDMD) at both transcriptional and translational levels (Fig. 6A-E). This was associated with elevated BALF concentrations of pro-inflammatory cytokines IL-1β (Fig. 6F) and IL-18 (Fig. 6G), and increased LDH release (Fig. 6F). Notably, CA exerts its effects by selectively inhibiting the activation of the NLRP3 inflammasome, reducing LDH release and decreasing IL-18 and IL-1β secretion, thereby demonstrating the anti-pyroptotic effects observed in vitro through suppression of the NLRP3 inflammasome pathway.

To explore whether the effects of CA were Nrf2-dependent, mice were pretreated with the Nrf2 inhibitor ML385. This pretreatment eliminated the ability of CA to downregulate NLRP3 inflammasome components and reduce inflammatory mediator release (Fig. 6A-H). Therefore, these findings suggested that CA could attenuate LPS-induced ARDS by suppressing NLRP3 inflammasome-mediated pyroptosis, a process partially dependent on Nrf2 activation.

CA alleviates LPS-induced ARDS through Nrf2-dependent antioxidant mechanisms. MPO, MDA and SOD activity was assessed to evaluate whether CA exerted a protective effect against LPS-induced ARDS through its antioxidant properties. In addition, the protein and mRNA levels of Nrf2 and HO-1 were measured. The findings revealed that LPS administration notably elevated MPO and MDA activities, while suppressing SOD levels in the BALF (Fig. 7A-C). However, these changes were effectively prevented by pretreatment with CA. Additionally, treatment with CA resulted in a notable upregulation of both Nrf2 and HO-1 at both transcriptional and translational levels in the lung tissue of mice (Fig. 7D-F).

To further substantiate whether the antioxidant protection offered by CA against LPS-induced ARDS was dependent upon Nrf2 activation, mice were pretreated with the aforementioned ML385. ML385 not only abolished the CA-mediated reduction in MPO and MDA levels and the restoration of SOD activity (Fig. 7A-C), but also attenuated the CA-induced upregulation of both Nrf2 and HO-1 (Fig. 7D-F). These findings indicated that Nrf2 activation constitutes an essential mechanism underlying the antioxidant protection of CA in LPS-induced ARDS.