Introduction

Chronic obstructive pulmonary disease (COPD) is a prevalent chronic respiratory condition characterized by airflow limitation and airway inflammation1,2. As one of the leading causes of morbidity, mortality, and health care utilization globally, COPD poses a significant threat to human health and represents a substantial economic burden3. In recent years, the prevalence of COPD has been increasing worldwide, with approximately 70–80% of affected adults remaining undiagnosed. This underdiagnosis further elevates the risk of future acute exacerbations and hospitalizations, rendering the health care challenges associated with COPD increasingly urgent and prominent.

The pathogenesis of COPD is complex and involves multiple cellular and molecular interactions4. Currently, Western medicine primarily employs palliative treatment regimens, such as oxygen therapy and pharmacologic interventions, to alleviate symptoms associated with airflow limitation. However, no specific curative treatments are available. The frequent use of antibiotics, inhaled corticosteroids, bronchodilators, and other medications can lead to adverse reactions, including osteoporosis, immunosuppression, metabolic disorders, and an increased risk of recurrent infections and a poor prognosis, which warrants careful consideration5. Traditional Chinese medicine (TCM) offers multicomponent, multitarget characteristics that provide significant advantages in alleviating exacerbated symptoms, reducing disease recurrence, and minimizing drug toxicity and side effects6. However, the mechanism of action of TCM in the treatment of COPD is still unclear. Therefore, further exploration of the pathogenesis of COPD and the identification of potential TCM treatment pathways are critical areas of modern research and clinical practice.

Guben Kechuan granules (GBKC) is composed of Dangshen (Codonopsis pilosula), Baizhu (Atractylodes macrocephala, prepared with wheat bran), Wuweizi (Schisandra chinensis, vinegar-prepared), Zhi Gancao (honey-fried liquorice), Fuling (Poria cocos), Maidong (Ophiopogon japonicus), and Buguzhi (Psoralea corylifolia, prepared with saltwater). It functions to stabilize qi, strengthen the spleen and tonify the kidney. It is suitable for the treatment of cough, phlegm, shortness of breath and wheezing caused by spleen deficiency, excessive phlegm and unstable kidney qi. Several clinical studies have demonstrated that GBKC has therapeutic effects on COPD, chronic bronchitis, and bronchial asthma, improving lung function, regulating the immune system, and reducing the inflammatory response7,8,9. However, the precise mechanisms of action remain to be fully elucidated.

In this study, ultraperformance liquid chromatography tandem mass spectrometry (UPLC‒MS/MS) was used for highly sensitive and selective detection of the active ingredients of GBKC10. As an emerging drug analysis method, network pharmacology (NP), which is based on systems biology, multidirectional pharmacology and network analysis, suggests the interaction mechanism between drugs and diseases from a high-level perspective by constructing an “active ingredient–target–disease” network, which provides an innovative research path for the field of drug research11. Molecular docking and molecular dynamics simulations can predict the stability of intermolecular contacts in protein‒ligand complexes12. GeneMANIA-based functional association network analysis (GMFA) was further integrated in this study to enrich the discovery of NPs and reveal potential multitarget synergistic mechanisms13,14. The aims of this study were to investigate the potential mechanism of GBKC in COPD using network pharmacology, molecular docking, molecular dynamics simulations and GMFA technology and to provide theoretical support for the treatment of COPD with GBKC. The research flow chart (Fig. 1) vividly depicts the entire research process, offering a clear visual summary of the crucial steps and methodologies adopted in this study.

Fig. 1
figure 1

Research Flow Chart.

Full size image

Materials and methods

Active ingredient analysis of GBKC

GBKC was produced by Hefei Cubic Pharmaceutical Co., Ltd., China (National Drug Approval Number Z20090933). GBKC is composed of Dangshen, Baizhu, Wuweizi, Zhi Gancao, Fuling, Maidong, and Buguzhi. The UPLC‒MS/MS method was used to identify its active ingredients. On the basis of a literature search, the selection of the four predicted active ingredients—liquiritigenin, isoliquiritigenin, luteolin, and psoralenol—was informed by their documented pharmacological properties and prevalence in traditional medicinal contexts15,16. A mixed standard solution containing 20 µg/mL of each of the above four active ingredients was prepared and sequentially injected and detected. The quantitative method was established by using the Analytics module in OS software, and the standard curve was generated with the concentration as the horizontal coordinate and the peak area as the vertical coordinate.

For sample pretreatment, 0.02 g of sample was weighed precisely, 10 mL of 70% methanol water was added, the mixture was vortexed, mixed well, and extracted by ultrasonic extraction at room temperature for 30 min (300 W, 40 kHz), and 70% methanol water was added to compensate for the lost weight of the extraction reagent. After centrifugation at 4200 rpm for 10 min at room temperature, the supernatant was carefully collected. Then, the supernatant was passed through a 0.22 μm organic nylon filter membrane and diluted 4-fold (with 70% methanol water) for UPLC‒MS/MS.

Six parallel experiments were carried out. The separation was performed on a WatersACQUITYHSST3 column (100 mm×2.1 mm, 1.8 μm) with 0.1% formic acid in water (A) and acetonitrile (B) as the mobile phase, and the parameters were set as follows: flow rate, 0.3 mL/min; injection volume, 5 µL; and column temperature, 35 °C. The elution time program is shown in Table 1. An AB SCIEX 4500 mass spectrometer was used for the analysis, with an electrospray ionization (ESI) source. The parameters were set as follows: the ionization voltage was set to −4500 V, the temperature of the ion source was controlled at 550 °C, the pressure of the nebulizing gas (GS1) was 50 psi, the pressure of the auxiliary heating gas (GS2) was 55 psi, and the pressure of the curtain gas (CUR) was 35 psi. In addition, Medium-9 was used for collision gas/collision-activated dissociation (CAD), negative ion scanning was used in scanning mode, and multiple reaction monitoring (MRM) was used in detection mode.

Table 1 Liquid phase gradient elution ratios.
Full size table

Screening of the active ingredients and target prediction of GBKC

We conducted a systematic search for the active components of GBKC using the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) (https://tcmspw.com/tcmsp.php) and the Encyclopedia of Traditional Chinese Medicine (ETCM) (http://www.tcmip.cn/ETCM2/front/#/). The keywords “Codonopsis,” “Atractylodes,” “Poria,” “Psoralea,” “Honey-fried Licorice,” “Schisandra,” and “Ophiopogon” were employed to identify the effective constituents of the formulation. The parameters of oral bioavailability (OB) and drug-like properties (DL) were set, and the drug components with OB ≥ 30% and DL ≥ 0.18 and their corresponding targets were screened. We subsequently utilized the Swiss Target Prediction platform (http://www.swisstargetprediction.ch/) to predict the potential action targets of the identified chemical constituents. Targets with a confidence level ≥ 0.1 were selected, and duplicate targets were removed from the analysis.

COPD-related targets

The keyword “chronic obstructive pulmonary disease” was used to search for relevant targets in the Online Mendelian Inheritance in Man (OMIM) database (https://omim.org), GeneCards (https://www.genecards.org/), the DisGeNET database (https://www.disgenet.org/), and DrugBank (https://go.drugbank.com/). After merging, duplicate targets were deleted to obtain targets corresponding to COPD.

Screening for drug‒disease intersection targets

The targets of Guben Kechuan granules and the targets of COPD were imported into the online Venn diagram software VENNY 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/index.html)17 to obtain the drug‒disease intersection targets.

Constructing protein‒protein interaction (PPI) networks and screening key targets

The identified drug‒disease common targets were analysed using the STRING database (https://cn.string-db.org/). The species was set to “Homo sapiens,” with a minimum interaction score of “medium confidence (0.400).” At the same time, we set up a hidden network without joints. We output TSV files and imported them into Cytoscape 3.10.2 for visual analysis. Using the Analyze Network plugin, the top five targets were selected as core proteins on the basis of their degree value.

Enrichment analysis

Using the DAVID database (https://david.ncifcrf.gov/home.jsp), we conducted Gene Ontology (GO) biological process and Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway enrichment analyses on the intersection of component–disease target points. KEGG pathway identifiers and names were retrieved from the KEGG database (https://www.kegg.jp/kegg/kegg1.html)18. The GO enrichment analysis results were categorized into biological process, cellular component, and molecular function categories, with statistical significance determined at P < 0.05. The results were visualized using the Microbioinformatics platform19.

Molecular docking

To explore protein‒ligand interactions through molecular docking, we first obtained the 2D structure of the small-molecule ligands from the PubChem database (http://pubchem.ncbi.nlm.nih.gov/). Using ChemOffice software, these structures were converted into 3D conformations and saved in.mol2 format. High-resolution protein target crystal structures were retrieved from the RCSB PDB database (http://www.rcsb.org/) as molecular receptors. PyMOL was employed to prepare the protein structures by removing water molecules and phosphate groups, which were then saved as.pdb files. Molecular docking was conducted using AutoDock Vina 1.5.6. All the rotatable bonds of the ligands were treated as flexible during molecular docking. The “exhaustiveness” parameter was set to 9. Preprocessing with AutoDock Tools involved protonating the protein (with hydrogen atoms added) and dehydrating it, whereas small-molecule ligands were protonated and had their rotatable bonds assigned (torsional bonds defined). Docking box coordinates were set, and optimal binding conformations were identified by comparing the affinity scores of the docked poses. Finally, PyMOL and Discovery Studio 2019 were used to generate 2D schematics and 3D visualizations, enabling analysis of the interactions between test compounds and key amino acid residues in the binding site.

Molecular dynamics simulations

A 100 ns molecular dynamics simulation of the complex structure was performed by applying Gromacs 2023. In this simulation, the force field parameters of the protein were chosen to be CHARMM 36, while the topology of the ligand was constructed on the basis of the GAFF2 force field parameters20. Periodic boundary conditions were used, and the protein‒ligand complex was placed in a cube box. The box was filled with water molecules using the TIP3P water model21. Particle Mesh Ewald (PME) is used to treat “long-range” electrostatic interactions. Electrostatic interactions beyond 1.0 nm were calculated using the PME method, whereas short-range interactions within the cut-off were computed in real space. The Verlet al.gorithm is used to integrate the equation of motion. Before the simulation, 100,000 steps of NVT ensemble (isothermal isovolume) equilibration and NPT ensemble (isothermal isopressure) equilibration were performed, in which a coupling constant of 0.1 ps was used and the equilibration process lasted for 100 ps22. During equilibration, both van der Waals and Coulomb forces were calculated using a cut-off value of 1.0 nm. After equilibration, the system was subjected to a constant temperature of 300 K and a constant pressure of 1 bar using Gromacs 2023 for a total duration of 100 ns of molecular dynamics simulations. As key metrics, we monitored the radius of gyration (Rg), root mean square deviation (RMSD), solvent accessible surface area (SASA), root mean square fluctuation (RMSF), and protein‒ligand contact23. RMSD and RMSF analyses were conducted using all the atoms of the system, including both heavy atoms and hydrogens. This comprehensive approach was applied to assess the global conformational stability of the protein‒ligand complex over the course of the molecular dynamics simulation. The reference structure for the RMSD calculations was the initial minimized structure (docked pose), and alignment was performed on the entire complex to avoid bias from flexible regions. The SASA plots presented in the manuscript represent the protein alone, excluding the ligand.

GeneMANIA-based functional association network analysis

To comprehensively identify the genes functionally related to the core targets of GBKC in COPD treatment, we used a novel GMFA. This method screens the 20 most strongly associated additional genes for each core target gene by a gene interaction network. All the newly screened genes were subsequently merged with the initial core genes to create a GMFA-based extended database (GMFA-ED). The GMFA-ED dataset was analysed for KEGG pathway enrichment.

Results

Analysis of the active ingredients of GBKC

UPLC‒MS/MS analysis was performed on the mixed standard solution and the sample. A comparison of the four predicted active compounds, namely, luteolin, psoralenol, liquiritigenin, and isoliquiritigenin (Fig. 2A), revealed that the peak elution times of the four compounds in the sample and the standard were consistent and that the ion-pair peak areas of the four compounds could be utilized for quantification (Fig. 2B and C). For method validation, the parameters included specificity, linearity, precision, stability, repeatability and recovery, and the results revealed that the correlation coefficients (R²) of the linear relationships of the four active ingredients were greater than 0.99 (Table 2), which indicated that the four compounds to be measured had good linearity in the corresponding range and could be used for accurate quantification. These findings verified that this method can be used for the analysis of the active ingredients of GBKC.

Fig. 2
figure 2

The UPLC-MS/MS Analysis of Guben Kechuan granules. (A) The chemical compositions of Guben Kechuan granules were analyzed and identified with UPLC-MS/MS. (B) Spectra of the peaks of the mixed standard solution. (C) Sample peak spectrum.

Full size image

According to the linear calculation results, the average content of liquiritigenin in the six samples was 45.64 µg/g, the average content of isoliquiritigenin was 6.73 µg/g, the average content of luteolin was 0.25 µg/g, and the average content of psoralenol was 98.55 µg/g, with the relative standard deviations (RSDs) of the six samples being less than 15% (Table 3).

Table 2 Identification parameters of the 4 predicted active components.
Full size table
Table 3 Contents of the 4 predicted active components in the sample.
Full size table

Screening of the active ingredients and targets of GBKC

A total of 67 active ingredients that met the screening conditions were identified in GBKC, including 7 from Atractylodes macrocephala, 22 from Radix et Rhizoma Ginseng, 15 from Poria, 1 from Psoralea, 17 from Radix et Rhizoma Glycyrrhizae, and 5 from Maitake, and the top 10 active ingredients were obtained after screening in descending order of OB values (Table 4).

Table 4 Top ten active components in GBKC.
Full size table

Intersection of the GBKC and COPD targets

A total of 173 relevant drug targets were identified by converting target names into standardized gene names, followed by the removal of duplicates and invalid targets. Additionally, 1,952 relevant disease targets were retrieved from the OMIM and GeneCards databases. By intersecting the drug targets with the disease targets using the Venn diagram tool, we identified 72 potentially relevant targets (Fig. 3A).

Fig. 3
figure 3

PPI network and enrichment analysis of Guben Kechuan granules and COPD intersection targets. (A) Wayne diagram of intersecting targets. (B) PPI network of intersecting targets. (C) The PPI network is represented by nodes, comprising 70 nodes and 1036 edges. (D) Bar graph of GO enrichment analysis results showing the top 10 entries for BP, CC, and MF. (E) Bubble chart of KEGG pathway enrichment analysis, displaying the top 20 pathways. Larger bubbles represent a higher number of associated genes, while greener colors indicate a lower p-value. The KEGG pathway annotations in this figure were retrieved from the KEGG database (www.kegg.jp/kegg/kegg1.html). (F) Thermogram of molecular docking binding energy visualization.

Full size image

PPI network construction

PPI network analysis was conducted on the 72 overlapping targets using the STRING database, with the results imported into Cytoscape 3.10.2 for the construction of a PPI network comprising 70 nodes and 1,036 edges. In this network, nodes with darker colours and larger sizes are more important. The top five core targets of GBKC in COPD, based on the degree, were AKT1, TNF, IL6, ACTB, and TP53 (Fig. 3B and C).

Enrichment analysis

The 72 common targets were entered into the DAVID database and screened with a significance threshold of P < 0.05. GO analysis yielded a total of 503 entries, comprising 375 biological process (BP) terms, 46 cellular component (CC) terms, and 82 molecular function (MF) terms. On the basis of the P value ranking, the top 10 entries for BP, CC, and MF were selected and visualized to generate a histogram (Fig. 3D). The results indicated that BP was predominantly associated with the negative regulation of the apoptotic process, response to xenobiotics, and positive regulation of miRNA transcription. In terms of CC, the primary associations included the extracellular space, protein-containing complex, and receptor complex. For MF, the main functions involved enzyme binding, identical protein binding, and ubiquitin protein ligase binding.

Furthermore, KEGG pathway enrichment analysis identified 93 pathways, which were ranked by P value. The top 20 pathways were selected for bubble diagram representation (Fig. 3E). The findings revealed that the KEGG enrichment analysis highlighted pathways such as the HIF-1 signalling pathway, the IL-17 signalling pathway, and the PI3K-Akt signalling pathway, among others.

Molecular docking validation

The top selected core proteins, ACTB, AKT1, IL6, TNF, and TP53, were molecularly docked with luteolin, psoralenol, liquiritigenin, and isoliquiritigenin, respectively, from the active ingredients of GBKC. The binding energies were predicted using AutoDock Vina, and thermal energy maps were generated using GraphPad Prism (Fig. 3F). Generally, a binding energy of less than − 5.0 kcal/mol indicates good binding activity, whereas values below − 7.0 kcal/mol signify strong binding activity24. Notably, the lower the binding energy is, the stronger the binding activity, the higher the affinity, and the more stable the conformation. Among them, the binding of ACTB with psoralenol, luteolin, liquiritigenin, and isoliquiritigenin was more stable, with binding energy scores of −10.8 kcal/mol, −10.4 kcal/mol, −9.5 kcal/mol, and − 9.2 kcal/mol, respectively. Additionally, the binding of AKT1 with psoralenol was relatively stable, with a binding energy score of −7.9 kcal/mol. The docking results of the five molecules were visualized using PyMol software (Fig. 4A–E).

Fig. 4
figure 4

Molecular docking results. (A) ACTB-psoralenol (B) ACTB- liquitigenin (C) ACTB-luteolin (D) ACTB- isoliquitigenin (E) AKT1-psoralenol.

Full size image

Molecular dynamics simulations

On the basis of the molecular docking results, we selected the two active components with the strongest binding affinity to ACTB: psoralenol and liquiritigenin. Additionally, we chose to combine psoralenol with AKT1, which has a relatively strong binding affinity. These pairs were then subjected to further molecular dynamics simulations to verify the results. The RMSD serves as a quantitative metric for assessing conformational stability in protein‒ligand complexes and evaluating positional deviations of atomic coordinates from their initial configuration. Lower RMSD values indicate reduced structural fluctuations, reflecting enhanced conformational stability during molecular dynamics simulations25. This parameter is therefore widely employed to monitor the equilibration status of simulated systems. The ACTB–psoralenol, ACTB–liquitigenin and AKT1–psoralenol complex systems all reached equilibrium after 70 ns, and all of them eventually stabilized below 3 Å (Fig. 5A). Thus, these three sets of complexes exhibited high stability. Further analysis revealed that the Rg values and SASA values of the small molecules psoralenol and liquitigenin with the ACTB and AKT1 target proteins fluctuated stably during the simulations, indicating that the target protein‒small molecule complexes remained stable and compact at all times during the simulations (Fig. 5B and C). Hydrogen bonding is suggested to contribute significantly to ligand‒protein binding. In the kinetic simulations, the number of hydrogen bonds formed between the small molecule and the target protein ranged from 0 to 5 (Fig. 5D). Typically, approximately 2 hydrogen bonds were present in the complex, an observation that further suggests good hydrogen bonding ability between the ligand and the target protein. The RMSF characterizes the degree of conformational variability of amino acids in a protein. The ACTB–psoralenol, ACTB–liquitigenin, and AKT1–psoralenol complexes had relatively low RMSF values (mostly below 3 Å) (Fig. 5E). The RMSF plots show that some residues exhibit relatively high fluctuations, with values ranging 6 Å. In ACTB–psoralenol, residues exhibiting high RMSF peaks (Å) include Lys41 and Asp42 within the N-terminal DNase-I binding loop (D-loop) and Glu401 at the C-terminal tail. None of them are involved in the interaction and are not relevant for binding. In AKT1–psoralenol, residues exhibiting high RMSF peaks (4–6 Å) include Val202-Met203-Asp204-Leu205-Ala206 (αD-αE loop) and Thr264-Ile265-Gly266-Ser267 (C-lobe linker). These regions are distal from the binding site. The observed fluctuations are due to intrinsic loop flexibility in nonbinding regulatory zones. None of these high-RMSF regions are directly involved in ligand interactions, nor do they influence binding pocket stability. Thus, they are less flexible and more stable. Therefore, the binding of the ACTB–psoralenol, ACTB–liquitigenin and AKT1–psoralenol complexes are relatively stable due to numerous hydrogen bonds between the ligands and proteins.

Fig. 5
figure 5

Molecular dynamics simulation results of ACTB-Psoralenol and ACTB-liquitigenin and AKT1-Psoralenol complexes. (A) RMSD value (B) Rg value (C) SASA value (D) Number of hydrogen bonds between small molecules and target proteins in kinetic processes. (E) RMSF value.

Full size image

GMFA analysis

The five putative core targets identified by network pharmacology, AKT1, TNF, IL6, ACTB, and TP53, were expanded by the GMFA method to obtain GMFA-ED, which contains 105 genes in this dataset (Fig. 6A). The target genes in this dataset were subjected to PPI network analysis using the STRING database (Fig. 6B). A total of 140 pathways were obtained from the KEGG pathway enrichment analysis of GMFA-ED, which were sorted according to P value, and the top 30 pathways were selected to draw a bar graph (Fig. 6C). The results revealed that the HIF-1 signalling pathway, the IL-17 signalling pathway, and the PI3K‒Akt signalling pathway were significantly enriched before and after expansion, and after expansion, five new signalling pathways were identified, including the TNF signalling pathway, the NF‒kB signalling pathway, the p53 signalling pathway, the adipocytokine signalling pathway, and the NOD‒like receptor signalling pathway.

Fig. 6
figure 6

GMFA analysis results. (A) GeneMANIA Functional Association (GMFA) network analysis illustrated functionally relevant genes for the 5 core targets and created a database of extended potential targets (GMFA-ED). (B) PPI network analysis of target genes in the GMFA-ED dataset. (C) Bar graph of KEGG pathway enrichment analysis of GMFA-ED showing the top 30 pathways.The KEGG pathway annotations in this figure were retrieved from the KEGG database (www.kegg.jp/kegg/kegg1.html).

Full size image

Discussion

In TCM, COPD is typically categorized under the terms “cough,” “wheezing,” and “lung distension.” It is believed that this deficiency continues to exist while the disease is stabilizing. The onset of external pathogens, disruption of qi flow, accumulation of phlegm and blood stasis manifest as excess symptoms during acute exacerbations. In the early stages of the disease, the primary impact is on the lungs, spleen, and kidneys. However, in advanced stages or during acute exacerbations, symptoms may extend to the heart. Therefore, treatment strategies should adhere to the principle of addressing both the root and branch of the disease, employing methods of “tonifying deficiency, resolving phlegm, promoting circulation, and regulating the lungs”26. In TCM, “qi” conceptualizes the dynamic functional activities that sustain life processes. Qi deficiency is clinically correlated with reduced physiological resilience, which presents as fatigue, metabolic dysregulation, and immune imbalance. The pathological mechanisms of COPD primarily involve qi deficiency in the lungs, spleen, and kidneys, along with the interplay of phlegm dampness and blood stasis. At any stage of COPD, deficiency of the lungs, spleen, and kidneys constitutes a fundamental pathological basis, whereas internal obstruction of phlegm dampness and blood stasis represent significant pathological changes. Therefore, the treatment of COPD should integrate multiple systems of regulation: lung dysfunction (reflecting impaired immune defence and mucociliary clearance), spleen deficiency (manifesting as metabolic‒nutritional imbalance), and kidney instability (indicating HPA axis dysregulation). When pulmonary inflammation and oxidative stress are addressed, concurrent modulation of gastrointestinal barrier function and nutrient absorption is critical; when respiratory‒digestive synergy is optimized, adrenal‒glucocorticoid homeostasis must be coordinated to mitigate systemic inflammation27. GBKC comprises Dangshen, Baizhu, Wuweizi, Zhi Gancao, Fuling, Maidong, and Buguzhi. This formula is designed to tonify qi (enhance physiological resilience), nourish the lungs (support respiratory epithelial function), strengthen the spleen (improve metabolic homeostasis), and warm the kidneys (regulate renal‒adrenal axis activity) to facilitate qi intake, thereby addressing the integrated functions of the lungs, spleen, and kidneys28.

In the present study, the UPLC-MS/MS method was utilized to identify the four active ingredients of GBKC, namely, luteolin, psoralenol, liquiritigenin, and isoliquiritigenin. Notably, in the current sample, the contents of the four active components exhibited a wide range of concentrations. Importantly, high-potency components, even at low concentrations, may exert profound biological effects. Additionally, it is plausible that low-abundance components with similar pharmacological activities could collectively exert effects through “concentration summation” in vivo, thereby contributing to the overall efficacy of TCM29. More importantly, the efficacy of TCM is essentially a result of “multicomponent–multitarget” synergism. Although the in vitro activity of a single low-abundance component may be underestimated, when multiple components target the same signalling pathway, their superimposed effects may produce a “threshold-breaking” phenomenon. However, further pharmacokinetic studies are needed to verify these hypotheses. Furthermore, a limitation of the present UPLC–MS/MS method is its restricted chromatographic resolution, which may result in partial co-elution or overlapping peaks with structurally related analytes or matrix-derived components30. Such phenomena can compromise analytical specificity and introduce quantitative bias through ion suppression or enhancement31,32. Although the use of multiple reaction monitoring (MRM) transitions enhances analyte identification, it cannot fully eliminate the risk of false positives or inaccurate quantification due to unresolved co-elution. In this study, these issues were mitigated by optimizing the gradient profile, monitoring multiple transitions where feasible, and incorporating quality control samples; however, the possibility of residual overlap cannot be excluded.

Studies have shown that luteolin is a natural tetrahydroxyflavonoid. CHEN et al.33 reported that luteolin was able to reduce the DNA-binding activity of nuclear transcription factor-κB (NF-κB) and activator protein-1 (AP-1), inhibit the expression of TNF-α, IL-6, inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2), and eliminate oxygen free radicals (ROS). The overloading of ROS and excessive release of inflammatory factors constitute the main pathogenesis of COPD. In addition, other studies have demonstrated that luteolin can alleviate inflammation, oxidative stress, and tissue damage in COPD models induced by cigarette smoke. This is achieved by inhibiting the NOX4-mediated NF-κB signalling pathway, modulating the TRPV1 and CYP2A13/NRF2 signalling pathways, downregulating the expression of related harmful proteins, and upregulating the expression of protective proteins34,35. Therefore, luteolin may have therapeutic effects on COPD. Isoliquiritigenin, an isoflavonoid extracted from the traditional Chinese medicine Glycyrrhiza glabra, reduces the number of macrophages and neutrophils as well as the levels of TNF-α and IL-1β in the BALF and reduces the activity of myeloperoxidase (MPO) and MDA levels in lung tissues by regulating the expression of the Nrf2 and NF-κB pathway proteins, thus alleviating oxidative stress and excessive inflammation in the treatment of COPD36. Therefore, it is hypothesized that the above key active compounds can act as anti-inflammatory and antioxidant agents, improve airway remodelling, improve the health of the lung tissue, and mitigate damage in the treatment of COPD.

The PPI network identified five putative core targets, AKT1, TNF, IL6, ACTB, and TP53, which are suggested as candidate targets for the treatment of COPD. In the lungs, ACTB (β-actin) is expressed in airway epithelial cells, where it maintains the structural integrity of the airways. Additionally, this protein is present in alveolar cells, contributing to their normal role in gas exchange. Mutations in ACTB affect the structure and function of actin, leading to abnormal respiratory cell motility and airway inflammation. AKT1 is one of the 3 serine/threonine protein kinases involved in the PI3K–Akt-mediated signalling pathway and is closely related to the pathogenesis of COPD37. In addition, its activation inhibits the transcriptional activity of FOXO3a and induces the p53/p21 pathway to play an important role in cigarette-induced bronchial epithelial cellular senescence38, which is closely associated with cigarette smoking in patients with COPD; cigarette smoke exposure is a major aetiological factor in COPD. IL-6 regulates multiple aspects of the immune response mediated by CD4 T cells, such as cytokine secretion, soluble IL-6 receptor (sIL-6R) production, and inhibitory activity. These factors further participate in the exacerbation of lung injury by affecting mucus secretion, matrix accumulation and the release of granulocyte proteases39. Studies have demonstrated that various inflammatory cells, such as macrophages, play important roles in the pathophysiology of COPD40,41. Macrophages are a major source of cytokines, such as TNF-α and IL-6, which activate macrophages and induce the secretion of adipokines, thereby exerting a proinflammatory effect42. In previous studies, GBKC effectively alleviated asthma and chronic bronchitis-mediated airway inflammation and remodelling by reducing inflammatory cell infiltration and suppressing the expression of proinflammatory cytokines (TNF-α and IL-6)15,16.

The GO enrichment results included broad biological terms. Negative regulation of the apoptotic process is pathologically critical in COPD, where cigarette smoke-induced alveolar epithelial apoptosis directly contributes to emphysematous lung destruction and progressive respiratory failure43. The enrichment of this term suggests that GBKC may counteract apoptotic cascades. The extracellular space is pathologically central to COPD, wherein unchecked protease activity (e.g., neutrophil elastase, MMPs) degrades structural proteins, leading to emphysema and airflow obstruction44. The enrichment of this term implies that GBKC may regulate extracellular proteolytic homeostasis, which is consistent with its predicted interactions with inflammation-related targets (e.g., IL6 and TNF) in our network. According to the results of the KEGG pathway enrichment analysis, GBKC mainly regulates the HIF-1 signalling pathway, the PI3K–Akt signalling pathway and the IL-17 signalling pathway in the treatment of COPD. IL-17 affects inflammation, airway remodelling, and mucus hypersecretion during COPD development. Clinical studies in animal models of COPD have confirmed the efficacy of anti-IL-17 antibodies in reducing airway inflammation and remodelling45. Thus, targeting IL-17 has great potential in the treatment of COPD. The HIF-1 signalling pathway is abnormally activated in COPD patients, and the overexpression of HIF-1α, VEGF, and VEGFR2-related proteins is significantly associated with decreased pulmonary function, diminished quality of life, and the progression of COPD46. The HIF-1 signalling pathway may be a potential target for the treatment of COPD. The PI3K‒Akt signalling pathway, an important driver of COPD progression, is involved in the regulation of the airway inflammatory response in COPD by affecting inflammatory cells, inflammatory factors, and downstream targets (including NF-κB, mTOR, and ROS)47. Therefore, the PI3K‒Akt signalling pathway may play a key role in attenuating inflammatory responses, ameliorating oxidative stress, modulating glucocorticoid resistance and enhancing lung function48,49. The mechanism of COPD treatment by GBKC potentially involves multiple signalling pathways, and improvements in lung signs may be achieved through the modulation of multiple pathways.

In molecular docking analysis, binding energies below − 7.0 kcal/mol reflect robust binding affinity, and the predicted interactions between the identified active compounds and core COPD targets met this threshold. Specifically, ACTB demonstrated particularly stable interactions with psoralenol, luteolin, liquiritigenin, and isoliquiritigenin, with binding energy scores of −10.8, −10.4, −9.5, and − 9.2 kcal/mol, respectively. Furthermore, AKT1 exhibited favourable binding with psoralenol, characterized by a binding energy of −7.9 kcal/mol, suggesting potential ligand‒receptor interactions. Notably, ACTB consistently has the strongest binding energy, reflecting its structurally permissive hydrophobic pocket that accommodates diverse ligand chemotypes (e.g., the furan ring of psoralenol and the polyphenolic structure of luteolin). AKT1 and TP53 showed moderate yet stable binding, likely driven by their flexible active-site architectures—AKT1’s adaptable kinase domain and TP53’s DNA-binding motif-derived pocket. In contrast, TNF and IL6 displayed weaker and narrower affinity ranges, which is attributable to their rigid cytokine–receptor interfaces that restrict ligand diversity. This pattern underscores how target structural flexibility (ACTB > AKT1/TP53 > TNF/IL6) and ligand–target property matching (lipophilicity, hydrogen bonding potential) shape binding promiscuity, providing a structural basis for prioritizing multitarget interventions in COPD.

Further molecular dynamics simulations revealed that the binding of the ACTB–psoralenol, ACTB–liquitigenin and AKT1–psoralenol complexes was stable due to numerous hydrogen bonds between the ligands and proteins. Notably, the MD simulation results revealed distinct ligand-specific binding dynamics that underpin the biological relevance of these interactions. The ACTB–psoralenol/liquiritigenin and AKT1–psoralenol complexes exhibited remarkable stability (RMSD < 3 Å after 70 ns), a feature attributed to their robust hydrogen-bond networks (2–5 bonds) and compact conformational states (stable Rg and SASA profiles). This stability aligns with ACTB’s role as a cytoskeletal scaffold: the rigid binding of psoralenol/liquiritigenin to ACTB’s hydrophobic pocket (RMSF < 3 Å) may modulate actin filament dynamics, potentially influencing cell motility and airway remodelling in COPD50. Since β-actin (ACTB) is a major component of the cytoskeleton, the regulation of AKT1 can indirectly affect ACTB expression or its dynamic aggregation status51. In our MD study, the stable binding of ligands to AKT1 and ACTB potentially modulated their functions. Future research could consider several enhancements to provide a more comprehensive analysis of the protein‒ligand interactions. These include reporting the buried solvent-accessible surface area (BSA) between the protein and ligand, including apo (unbound) protein simulations to provide a clearer baseline for interpreting ligand-induced flexibility, and incorporating representative structural snapshots or 2D interaction maps to highlight key ligand‒protein contacts and hydrogen bond occupancies.

The GMFA was extended beyond the five putative core targets identified by network pharmacology, AKT1, TNF, IL6, ACTB, and TP53, to obtain the GMFA-ED dataset, and the results of the KEGG pathway enrichment analysis of GMFA-ED revealed that three signalling pathways, namely, HIF-1, IL-17, and PI3K–Akt, were significantly enriched before and after extension. In addition, the extension also identified five additional significantly enriched signalling pathways, the TNF, NF-κB, p53, adipocytokin, and NOD-like receptor signalling pathways, which enriched the discovery of NP. The p53 signalling pathway contributes to COPD pathogenesis by mediating cigarette smoke-induced cellular damage responses; it promotes DNA damage-triggered senescence in lung epithelial cells and impairs antioxidant defences, accelerating emphysema progression52. Concurrently, NOD-like receptor signalling activates the NLRP3 inflammasome through recognition of damage-associated molecular patterns, including mitochondrial ROS and extracellular ATP; this triggers caspase-1-dependent IL-1β maturation and neutrophil recruitment, which drive persistent airway inflammation during exacerbations53. Notably, these pathways form a self-sustaining cycle in which p53-mediated cellular damage releases molecular patterns perpetuating NOD-like receptor activation, thus fuelling progressive tissue destruction and chronic inflammation in COPD. The novel identification of these signalling pathways as possible targets of GBKC in COPD not only supports the reliability of the core targets predicted by NP but also further reveals the potential molecular mechanisms of drug action, providing important clues for an in-depth understanding of the synergistic therapeutic effects of GBKC’s multiple targets and pathways.

This study is the first to explore the potential mechanism of action of GBKC in the treatment of COPD. Nevertheless, this study has several limitations that should be acknowledged. Owing to the limitations of network pharmacology, research on the mechanism of GBKC in treating COPD has only been conducted through data mining, with a lack of COPD animal models and in vivo analyses. Thus, the predicted involvement of multiple pathways (e.g., the PI3K–AKT pathway) requires validation via in vivo and in vitro experiments to confirm their functional relevance. However, independent evaluation of bioactive compounds in isolation neglects potential synergism or antagonism among GBKC components. Future studies should integrate multicompound pharmacokinetic models to assess holistic effects. Current clinical evidence derives from regionally restricted trials in Chinese populations with TCM syndrome-based diagnoses. Therefore, multicentre trials across diverse ethnicities are needed to fully determine the clinical value of GBKC in treating COPD.

Conclusion

In summary, the results of the present study suggest that GBKC may act on targets AKT1, TNF, IL6, ACTB and TP53 through active ingredients such as liquiritigenin, isoliquiritigenin, luteolin, and psoralenol, thus regulating various pathways, such as the HIF-1 signalling pathway, the PI3K–Akt signalling pathway, and the IL-17 signalling pathway, to exert its effects in treating COPD.