A total of 22 prototype components were identified in the rat plasma and are referred to in the previous published literature by the research group (Liu et al., 2022). Chemical characterization analysis and quantitative analysis of QJHTD are shown in Supplementary Material. The chemical structure of 22 active components is shown in Figure 2, and their chemical information is shown in Supplementary Table S1. The SDF format file of the QJHTD components was downloaded from PubChem (https://pubchem.ncbi.nlm.nih.gov/) (Kim, 2016) and uploaded to PharmMapper databases (Liu X. et al., 2010; Wang et al., 2016; Wang et al., 2017) and SwissTargetPrediction. The targets with norm fit ≥0.6 in the output of PharmMapper and convert protein names to official gene symbols (Homo sapiens) using UniProt Knowledgebase (http://www.uniprot.org/) (UniProt Consortium, 2018). The potential drug targets predicted by the two databases were selected for further verification.

The information of the therapeutic target was searched by using “acute lung injury” as the keyword. The databases used in this study are GeneCards (https://www.genecards.org/), DisGeNet https://www.disgenet.org/home/), and TTD (http://db.idrblab.net/ttd/). Then, the components and disease overlap proteins are used as candidate targets for the treatment of ALI.

Metascape combines functional enrichment, interactome analysis, gene annotation, and membership search to leverage over 40 independent knowledge bases within one integrated portal (Zhou et al., 2019). The core target proteins of QJHTD for ALI were inputted into Metascape, after which we set p < 0.01 to analyze GO and KEGG pathway enrichment. The first 20 KEGG and GO clusters pathway information were screened and included, and the results were saved and visualized by R software.

The top 20 clusters obtained from the previous KEGG pathway enrichment analysis correspond to 22 active compounds and the core targets of QJHTD in the treatment ALI and construct the network of “compounds–targets–pathways” using CytoScape 3.8.0 to analyze the network topology parameters of the targets, including degree, betweenness, and closeness.

The virtual docking of key proteins and active components was performed using Discovery Studio 2020 software (College of Pharmacy, Henan University of Chinese Medicine) to study their interaction. For the top 20 targets with a higher degree value in the compound–target–pathway network, the corresponding 3D structure was downloaded in the RSCB PDB database (https://www.rcsb.org/) (Richardson et al., 2021), and 22 component structures were obtained from the PubChem database, and the components were prepared using the “Prepare Ligands” module to obtain the 3D structure. For protein preparation, crystallographic water molecules were removed and then used in the “Prepare Protein” module. Subsequently, CDOCKER was performed for molecular docking (Wu et al., 2003). To enable this mechanism, all default parameters were taken into account, allowing 10 poses to be generated for each ligand. Docking estimation was performed by the CDOCKER energy, which was used to assess the affinity of the proteins and ingredients. 80% of -CDOCKER ENERGY of the target protein and its corresponding prototype ligand was viewed as the threshold, and the components with higher scores were regarded as the active ingredients that interacted with the protein.

QJHTD was prepared by Tianjin Pharmaceutical Research Institute Co., Ltd. (Tianjin, China). The preparation process of QJHTD is shown in Supplementary Figure S1. Thrombin (Cat No: HY-114164), dihydrorhodamine 123 (DHR123) (Cat No: HY-101894/CS-7988), and diphenyleneiodonium chloride (DPI) (Cat No: HY-100965) were purchased from MedChemExpress LLC (Monmouth Junction, NJ, United States). Baicalin (CAS No. 21967-41-9), baicalein (CAS No. 491-67-8), and wogonin (CAS No. 632-85-9) were obtained from Shanghai Yuanye Biological Co., Ltd (Shanghai, China), and the purity of all compounds was higher than 98%. The Amino Coupling Kit (Cat No: BR-1000-50), HBS-EP buffer solution (Cat No: BR-1001-88), CM5 Sensor Chip (Cat No: BR-1003-99), and Percoll (Cat No: 17-0891-01) were purchased from GE Healthcare (Braunschweig, Germany). Hoechest 33342 (Cat No: H1399) and SYTOX™ Green Nucleic Acid Stain (Cat No: S7020) were purchased from Invitrogen (Carlsbad, CA, United States). PMA (Cat No: P1585) and LPS (from Escherichia coli O111: B4) were obtained from Sigma (St. Louis, MO, United States). Fetal bovine serum (FBS) (Cat No: BR-1003-99) was obtained from Gemini (Woodland, CA, United States). RPMI 1640 medium (Cat No: 10,491), Wright-Giemsa Stain solution (Cat No: G1020), Red Blood Cell Lysis Buffer (Cat No: R1010), Dilution Buffer (Cat No: R1017), Normal Goat Serum (Cat No: SL038), and BCA protein assay kit (Cat No: PC0020) were obtained from Solarbio Life Science (Beijing, China). The ROS detection kit (Cat No: S0033S) and Poly-L-lysine (Cat No: C0313) were obtained from Beyotime (Shanghai, China). Rabbit monoclonal to Src (ab133283), rabbit monoclonal to Src (phospho Y419) (ab185617), rabbit monoclonal to Myeloperoxidase (ab208670), and Alexa Fluor^® 488 Goat polyclonal Secondary Antibody to rabbit IgG - H&L (ab150077) were obtained from Abcam (Cambridge, MA, United States). HRP-conjugated goat anti-rabbit IgG antibody (Cat No:bs-40295G-HRP) was obtained from Bioss (Beijing, China). PE anti-rat CD11b/c Antibody (Cat No: B339537) was obtained from BioLegend (San Diego, CA, United States).

CM5 Sensor Chip was esterified with the crosslinking agents EDC and NHS. The thrombin protein at a concentration of 5 μg/ml in sodium acetate at pH 4.5 was coupled to the surface of the chip, and then the remaining reactive carboxyl on the matrix were blocked using 1 M ethanolamine, at pH 8.5. The compound of baicalin, baicalein, and wogonin were dissolved in DMSO to 10 mM, was diluted with HBS-EP buffer solution to 500 μM, and then diluted successively to 50 μM, 25 μM, 12.5 μM, 6.25 μM, 3.125 μM, 1.5625 μM, 0.7813 μM (baicalin), 12.5 μM, 6.25 μM, 3.125 μM, 1.5625 μM, and 0.7813 μM (baicalein) and 12.5 μM, 6.25 μM, 3.125 μM, 0.7813 μM, and 0.3906 μM (wogonin) using 5% DMSO HBS-EP buffer. The SPR experiment was performed using the Biacore T200 SPR instrument. The injection sample time and velocity were 120s and 20 μl/min, respectively. The protein dissociation time was 300 s.

Male Sprague–Dawley rats (220–240 g) were obtained from the Weitonglihua Experimental Animal Technology Co. (Beijing, China) [SCXK 2016–0,006], and housed at 25–28 °C and humidity of 45–55%, with free access to food and drink for 7 days before use.

Rat neutrophils were isolated from the whole blood of healthy rats by gradient centrifugation using Percoll. Briefly, blood was collected from the abdominal aorta of rats and anticoagulated with 109 mM sodium citrate (1:9 blood v/v), and the whole blood was diluted with an equal volume of dilution buffer. Diluted blood was smeared on the interface of the two layers of 82% Percoll and 69% Percoll, and centrifugation was carried out at 710 g for 30 min. Neutrophils were collected in the cell layer between the two layers and washed with PBS. Then, red blood cell lysis buffer was added, gently blown for 3–5 min, and incubated at 4°C for 15 min. Centrifugation was carried out at 290 g for 10 min. The supernatant was discarded, and the pellet was washed with PBS and centrifuged again at 250 g for 10 min. The pellet obtained at this point contains the neutrophils. The purity of neutrophil was determined by the flow cytometry and Wright-Giemsa Stain.

Sytox Green dye was used to observe and measure the release of NETs under different conditions (Shi et al., 2019). To quantify the amount of PMA-induced formation of NETs in vitro, neutrophils isolated from rat blood (8 × 10⁴ cells/well in 200 µl of medium with 1 μM SytoxGreen) were seeded into 96-well plates. These neutrophils were activated with the media (control), QJHTD at different concentrations (0.125, 0.25, and 0.50 g/L), 50 nM PMA, and 50 nM PMA with QJHTD at different concentrations (0.125, 0.25, and 0.50 g/L), respectively. The plate was placed in a 37°C, 5% CO₂ incubator for 4 h. Fluorescence was monitored 0 and 4 h using a SpectraMax i3x plate reader (Molecular Devices, San Jose, CA) with excitation at 488 nm and emission at 525 nm. Fluorescence intensity (extracellular DNA) was calculated as (Fluorescence intensity at 4 h) - (Fluorescence intensity at 0 h). The cell status of each group was observed by Axio Observer Z1 (Carl Zeiss AG, Oberkochen, Germany) after the detection with fluorescence microplate.

Rat peripheral blood neutrophils (1.6 × 10⁵ cells) were seeded on poly-L-lysine-coated coverslips in 48-well plates and cultured for 4 h in RPMI medium containing media (control), 50 nM PMA, or 50 nM PMA with QJHTD at different concentrations (0.125, 0.25, and 0.5 g/L). Subsequently, the cells were fixed with 4% paraformaldehyde for 20 min and permeabilized with 0.2% Triton-X-100 for 20 min. After blocking with 5% normal goat serum for 1 h, it was incubated with rabbit anti-MPO overnight (1:100 dilutions) at 4°C, followed by Alexa Fluor^® 488 -conjugated goat antirabbit IgG antibody (1:200 dilutions) for 3 h in the dark. The DNA was counterstained with Hoechst 33342 for 10 min. The images were acquired using Axio Observer Z1 (Carl Zeiss AG, Oberkochen, Germany) and a C11440-42U30 digital camera (Hamamatsu Photonics, Shizuoka, Japan) and processed with BioFlux Montage software (Fluxion Biosciences, Alameda, CA, United States).

The neutrophils were preloaded with DCFH-DA and were diluted at 1:1,000 with PBS. After the extracellular DCFH-DA dye was washed, the cells were resuspended in fresh RPMI medium (8× 10⁴ cells) and were seeded in a 96-well plate. The fluorescence intensity was detected by a SpectraMax i3x plate reader (Molecular Devices, San Jose, CA) at 0, 1, 2, 3, and 4 h, and the excitation wavelength was 488 nm, and the emission wavelength was 525 nm.

Neutrophils were preloaded with 10 μM dihydrorhodamine 123 (DHR123) at 37°C for 20 min. After washing the extracellular DHR123, the cells were resuspended in fresh RPMI medium containing media (control), 50 nM PMA, or 50 nM PMA with QJHTD at different concentrations (0.125, 0.25, and 0.50 g/L) and 50 nM PMA with 20 μM diphenyleneiodonium (DPI) for 1 h. After PBS washing, ROS production was determined using a BD FACSAria II flow cytometer (BD Bioscience, New Jersey, United States).

According to the network pharmacology and molecular docking results, we selected SRC activity for Western blot verification. The neutrophils were activated using LPS (1 μg/ml), following incubation with or without QJHTD (0.125, 0.25, and 0.5 g/L) at 37°C for 1 h. After 1 h, the cells were added with ice-cold lysis buffer containing protein phosphatase inhibitor mixture, lysed on ice for 30 min, and centrifuged (12,000 rpm, 4°C, 20 min) to obtain the supernatant. Protein concentrations were determined by the BCA protein assay kit. After denaturation by boiling, the proteins were separated by SDS-PAGE and transferred to PVDF membranes. Then, the membranes were blocked with 5% BSA at room temperature for 1.5 h and then incubated with anti-Src (1:1,000 dilution) and anti-Src (phospho Y419) (1:5,000) overnight at 4°C. Subsequently, the membranes were incubated with HRP-linked secondary antibody (1:2,000) at room temperature for 1 h. The results of Western blot were visualized by using an ECL detection system (Syngene, Cambridge, United Kingdom) and analyzed by ImageJ software.

All data were expressed as the mean ± SD of three independent experiments. The comparison between multiple groups was performed by one-way ANOVA followed by the LSD test when the variances were homogeneous or Dunnetts T3 test when the variances were non-homogeneous. All data were analyzed statistically using the SPSS version 23.0 (IBM, Armonk, NY, United States). p < 0.05 was considered statistically significant.

A total of 413 potential QJHTD-related targets were predicted by PharmMapper and SwissTargetPrediction (Supplementary Table S2), and 1,917 disease targets were finally summarized and obtained by removing the duplicate targets by searching the disease databases including GeneCards, DisGeNet, and TTD (Supplementary Table S3). The intersection of the QJHTD-related targets and ALI-related targets has 176 targets (Figure 3A, Supplementary Table S4 ), which include AKT1, MAPK1, F2, PPARG, and so on. The 176 targets were considered the potential therapeutic targets of QJHTD for ALI.

Metascape was used to analyze the signal pathways of QJHTD-related targets in improving ALI. The results of GO enrichment analysis are shown in Supplementary Figure S4. The biological processes in QJHTD for ALI mainly involved included protein phosphorylation, response to wounding, response to bacterium, regulation of inflammatory response, and so on. Cellular components were related to vesicle lumen, extracellular matrix, platelet alpha granule, and extrinsic component of the plasma membrane. Molecular functions analysis revealed protein serine/threonine/tyrosine kinase activity, endopeptidase activity, nitric-oxide synthase regulator activity, and phosphatase binding. The GO enrichment analysis may be related to the pathogenesis of ALI.

For each given gene list, pathway and process enrichment analysis has been carried out with the following ontology sources: KEGG pathway. All genes in the genome have been used as the enrichment background. Terms with a p-value < 0.01, a minimum count of 3, and an enrichment factor >1.5 are collected and grouped into clusters based on their membership similarities. The most statistically significant term within a cluster is chosen to represent the cluster. The first 20 clusters were selected according to their p-values to generate the bubble chart for visualization (Figure 3B). It is involved with lipid and atherosclerosis, HIF-1 signaling pathway, transcriptional misregulation in cancer, renin–angiotensin system, VEGF signaling pathway, and so on. From cluster 11, the VEGF signaling pathway expanded, and we can see that it contains VEGF signaling pathway, phospholipase D signaling pathway, platelet activation, neutrophil extracellular trap formation, and so on (Figure 3C). Neutrophils are the host’s first line of defense against microbial infection and play an important role in innate immune response (Nauseef and Borregaard, 2014; Leiding, 2017). Neutrophils survive for a short time in the blood, and they can resist pathogenic microorganisms by phagocytosis, degranulation, and formation of NETs (Kolaczkowska and Kubes, 2013). In recent years, it has been found that in ALI, neutrophil adhesion aggregation and continuous activation lead to an inflammatory cascade, during which a large number of NETs are produced (Saffarzadeh et al., 2012; Luo et al., 2014). Therefore, based on the results of network pharmacology and literature review, we adopted the PMA-induced neutrophil production NET model to verify whether QJHTD treats ALI by inhibiting NETs.

CytoScape 3.8.0 software was used to construct a component–target–pathway network of QJHTD in treating ALI and was used to calculate and sort the topological parameters (degree) of the nodes in the abovementioned network (Figure 4). The network has 218 nodes, including 22 components, 176 targets, 20 pathways, and 2,044 edges. The larger the node, the greater the degree, and the more nodes connected to it. It was surprised that the degree of MAPK14, CDK2, EGFR, F2, SRC, and AKT1 were higher in the compound–target–pathway network. It was concluded that the 22 active ingredients acted on 176 targets by relevant pathways in lipid and atherosclerosis, renin–angiotensin system, HIF-1 signaling pathway, and so on. Multiple pathways are linked to each other by common targets, indicating the synergistic action of QJHTD in treating ALI.

The 22 active components of QJHTD were used as candidate docking components. By analyzing the degree value of the targets in the component–target–pathway network, the first 20 targets (Supplementary Table S5) with higher degree values were selected for molecular docking experiments. The initial compounds of the protein were extracted from the active pockets and redocked and RMSD≤2 evaluated that the docking algorithm could reproduce the receptor-ligand binding mode (Huang et al., 2010).

The original ligands of the selected crystal structures of 20 targets were used to define the active pockets. The original ligands were taken out, and the CDOCKER method was used. Re-dock to the set active pocket, calculate its RMSD and scoring value and its active pocket information, and calculation results are shown in Table 1. It can be seen from the table that the RMSD of the 20 targets are all less than 2 Å, indicating that the selected docking method and parameter settings are reasonable and can be used for further molecular docking research. Molecular docking results in the top 20 target proteins with 22 chemicals and are shown in Supplementary Table S6 and Table 1. The higher -CDOCKER ENERGY is, the more likely the chemical and target are to interact with each other. The results showed that most chemical components had good interaction and binding activities with the targets. Components such as baicalin, oroxylin A-7-glucuronide, hispidulin-7-O-β-D-glucuronide, wogonoside, baicalein, wogonin, tianshic acid, and mangiferin are bound to most of the target proteins, which may be the core components of QJHTD for ALI.

Studies have shown that the inflammation of ALI depends on tissue factors and thrombin (Lou et al., 2019). Plasma and lavage fluid thrombin elevated evidently in the animal models of ALI, and thrombin has been found as a key molecule linking coagulation and inflammation (Lou et al., 2019; Arroyo et al., 2021; Kaspi et al., 2021). The results of molecular docking experiments showed that the docking scores of baicalein, wogonin, tianshic acid, and baicalin to F2 (thrombin) reached the effective binding scores. Two- and three-dimensional ligand–protein interactions of baicalein, tianshic acid, wogonin, and baicalin with F2 (PDB ID: 3QWC) are shown in Figure 5A. Both baicalin and wogonin inhibited thrombin-catalyzed fibrin polymerization and platelet functions and inhibited the activities of thrombin (Ku and Bae, 2014; Lee, et al., 2015b). Affinity capillary electrophoresis (ACE) is one of the predominant methods for interaction studies, and baicalein had the greatest affinity with thrombin in comparison with the Kb values of other flavonoid compounds (Li et al., 2018), and it has indicated more OH groups in the A-ring and more thrombin inhibitory activity (Liu L. et al., 2010). Taking into account all these, we thus selected baicalin, wogonin, and baicalein for the experiment of SPR.

The results of molecular docking experiments showed that the docking scores of 13 compounds including baicalein and wogonin to SRC reached the effective binding scores (Supplementary Table S6). Two- and three-dimensional ligand–protein interactions of baicalein, tianshic acid, wogonin, and baicalin with SRC (PDB ID: 2H8H) are shown in Figure 5B. The Src kinase family plays an important role in LPS-induced ALI, and studies have shown that bletinib and resveratrol ameliorates neutrophil inflammation and lung injury via inhibition of Src family kinases (Tsai et al., 2019; Kao et al., 2021).

In the SPR experiment, baicalein, wogonin, and baicalin were screened out as in vitro validation molecular models, and small-molecule and macromolecular interaction experiments with thrombin protein were performed. The binding affinity (K_D) describes the strength of the binding between the ligand and the analyzing molecule. K_D can be obtained by either “steady-state” or “kinetic” methods. The steady-state method was used in the “fast-up and fast-down” binding mode to obtain affinity, such as baicalein (Figure 6A) and wogonin (Figure 6B). Kinetic analysis K_D was obtained from the association rate constant (Ka) and dissociation rate constant (Kd), which was the combined result of the two processes of association and dissociation, such as baicalin (Figure 6C). The results showed that the direct binding ability of baicalein, wogonin, and baicalin to thrombin protein was all micromolar, and their K_D values were 11.92 μM , 1.303 μM , and 1.146 μM, respectively (Figure 6). Therefore, it was speculated that baicalein, wogonin, and baicalin bind to the thrombin protein and inhibit its activities and might exert thrombosis prevention in ALI to some extent.

The purity of neutrophils is >80% with PE Mouse anti-rat CD11b/c antibody was used to label neutrophils (Supplementary Figure S5A). By Wright-Giemsa staining, it was further confirmed that the purity of neutrophils was >80%. At the same time, it could be observed that the neutrophils had complete morphology, the cytoplasm was light red, and the nucleus was purple and lobulated, divided into two–five leaves (Supplementary Figure S5B).

Phorbol 12-myristate 13-acetate (PMA) is most widely used as an inducer of NETosis. Sytox Green dye is a high-affinity nucleic acid stain that readily cross-damaged cell membranes but does not penetrate the membranes of living cells. At the same time, it could be used to observe and measure the release of NETs (Shi et al., 2019). QJHTD (0.125, 0.25 g/L) showed no significant difference compared with that of the control group (p > 0.05), and the immunofluorescence intensity of the QJHTD (0.50 g/L) group was significantly lower than that of the control group (p < 0.05), suggesting that the QJHTD group has no cytotoxic effect on neutrophils, and the QJHTD (0.50 g/L) group may play a protective role (Figure 7A). Meanwhile, as shown in Figure 7B, QJHTD (0.125, 0.25, and 0.50 g/L) did not change the morphology of neutrophils.

The immunofluorescence intensity values of the PMA + QJHTD group with different concentrations (0.25 and 0.50 g/L) were significantly decreased compared with those of the PMA group (p < 0.01), indicating that QJHTD could inhibit PMA-induced neutrophil releasing NETs (Figure 7C). From Figure 7D, the result was consistent with the fluorescence intensity of each well detected by the fluorescence microplate. It was found that 4 h after PMA stimulation of neutrophils, compared with the control group, the morphology of neutrophils in the PMA group had changed, the cells were flattened, and neutrophils released a large number of green fluorescent filament-like and reticular NETs. Compared with the PMA group, the PMA + QJHTD group with different concentrations (0.25 and 0.50 g/L) did not change the flattened morphology of neutrophils, but the green fluorescent filamentous structures and NETs were significantly reduced, suggesting that QJHTD inhibits the release of NETs from the PMA-stimulated neutrophils.

We measured the release of NETs by detecting the release of MPO-DNA complexes. Immunofluorescent staining of MPO (neutrophil marker) and Hoechst-33342 (nucleic staining) further confirmed the inhibitory effect of QJHTD (Figure 7E). Compared with the control group, neutrophils lost their original structure after PMA stimulation, and their nuclear morphology also showed depolymerization and expansion. There was no significant difference between the PMA + QJHTD group (0.125 g/L) and the PMA group, but the PMA + QJHTD group (0.25 and 0.5 g/L) released significantly less NETs to the extracellular network and filamentous structures (p < 0.01) (Figure 7E, Figure 7F). These experiments documented that QJHTD inhibited the formation of PMA-stimulated NETs.

NETosis is a multifactorial process, but the detailed molecular mechanisms are not fully understood. The formation of PMA-induced NETs is closely related to ROS generation by nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Fuchs et al., 2007; De Bont et al., 2018; Fousert et al., 2020). DPI, an NADPH inhibitor, was used as a positive control for the inhibition of ROS production. In the PMA-stimulated neutrophils, the ROS production increased with longer time and the neutrophils were stimulated by PMA (Supplementary Figure S6). As can be seen from Figure 8 and Supplementary Figure S6, QJHTD treatment at concentrations of 0.125, 0.25, and 0.5 g/L significantly inhibited PMA induced ROS production (p < 0.01), suggesting that the inhibitory effect of QJHTD on NET formation is mediated by ROS inhibition.

The Src kinase family plays an important role in the systemic inflammatory response induced by LPS (Lee et al., 2007; Toumpanakis et al., 2017). Neutrophils have a key role in innate immunity and the development of infections and inflammation, and activation of the Src-dependent Smad3 signaling pathway mediates neutrophil inflammation and oxidative stress (Li et al., 2015). In this study, QJHTD (0.5 g/L) was shown to inhibit the phosphorylation of Src in the LPS-activated neutrophils (Figure 9).