1. Introduction
Lung adenocarcinoma (LUAD) is the most prevalent histological subtype of lung cancer worldwide, [ 1 ] accounting for approximately 40% of all cases. [ 2 ] The complex pathogenesis of LUAD involves dysregulation of multiple signaling pathways, including wingless-related integration site/beta-catenin, mitogen-activated protein kinase, focal adhesion kinase, protein kinase B (AKT), and transforming growth factor-beta cascades. Overexpression of the PRC1 protein is closely associated with the activation of the Wnt signaling pathway, which enhances the proliferative and invasive capabilities of LUAD cells. [ 3 ] The transcription factor NFE2L2 may contribute to increased cell growth, division, and gene transcription by modulating key genes within the Wnt and mitogen-activated protein kinase signaling pathways. [ 4 ] ITGB4 is highly expressed in LUAD and promotes cancer cell migration through focal adhesion kinase pathway activation. [ 5 ] Additionally, the C1orf131 gene is significantly upregulated in LUAD, and its knockout leads to dysregulation of the AKT signaling pathway, suppressing cell proliferation while inducing apoptosis. [ 6 ] At present, ferroptosis, as a new iron-dependent cell programmed death mode, plays an important role in the occurrence and development of lung cancer. COPZ1 can regulate ferroptosis in LUAD through NCOA4 mediated ferritin phagocytosis. [ 7 ] Notably, pulmonary fibrosis and LUAD share common genetic alterations, and aberrant transforming growth factor-beta signaling may drive the initiation and progression of both diseases. [ 8 , 9 ] Although significant progress has been made in molecular targeted therapy for LUAD in recent years, patients still face major challenges, including limited efficacy, adverse reactions, drug resistance, and the risk of recurrence .[ 10 – 13 ] To overcome these limitations, precision medicine has emerged as a pivotal strategy in LUAD management.
Precision medicine emphasizes tailoring treatment regimens based on individual genomic profiles, thereby enhancing therapeutic efficacy while minimizing unnecessary toxicity. [ 14 ] In LUAD, this approach involves screening for actionable genetic alterations (e.g., epidermal growth factor receptor and anaplastic lymphoma kinase mutations) to guide the selection of targeted agents, [ 15 ] enabling personalized therapeutic interventions that improve treatment success rates. [ 16 ] With advances in molecular profiling technologies, the integration of druggable targets and gene expression signatures into precision medicine holds great promise for further optimizing therapeutic outcomes and prognosis in LUAD patients.
The classical Chinese herbal formula Shengmai San (SMS), composed of ginseng (monarch herb), Ophiopogon japonicus (minister herb), and Schisandra chinensis (assistant herb), [ 17 , 18 ] has been traditionally prescribed for Qi replenishment, Yin nourishment, and pulse stabilization according to traditional Chinese medicine (TCM) theory. [ 19 ] Contemporary pharmacological studies reveal that its bioactive components (e.g., ginsenosides, ophiopogon polysaccharides, and schisandrins) exhibit antitumor effects through modulating cancer cell proliferation/migration [ 20 , 21 ] and remodeling the immune microenvironment. [ 22 , 23 ] The SMS-derived Shengmai Injection, routinely administered as an adjuvant to chemotherapy in non-small cell lung cancer (NSCLC) patients, demonstrates chemosensitizing properties [ 24 ] by enhancing cisplatin cytotoxicity against resistant cells while ameliorating chemotherapy-induced adverse effects .[ 25 ] Nevertheless, the precise mechanistic basis of SMS in LUAD intervention remains elusive.
In recent years, the emergence of technologies such as network pharmacology, bioinformatics, molecular docking, and molecular dynamics simulation has provided powerful tools for systematically elucidating the mechanisms of action of TCM compounds. A multidimensional interaction network of “herb-compound-target-disease” (H-C-T-D) can be constructed by network pharmacology, [ 26 ] through which the core targets and synergistic regulatory mechanisms of TCM compounds in the treatment of LUAD are revealed. Bioinformatics is employed to integrate databases such as The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), enabling the screening of differentially expressed genes (DEGs) in LUAD and conducting clinical prognostic analyses of related genes, thereby validating the potential therapeutic value of relevant targets. Molecular docking is utilized to simulate the binding patterns between active ingredients of TCM and tumor-associated proteins, predicting their interaction sites and binding affinity. Molecular dynamics simulation is further applied to evaluate the dynamic stability of binding complexes, elucidating the molecular mechanisms of drug–target interactions and providing a theoretical basis for the optimization of lead compounds.
This study aims to comprehensively investigate the mechanism by which SMS regulates LUAD at the molecular level through bioinformatics, network pharmacology, molecular docking, and molecular dynamics simulation. The research seeks to uncover potential therapeutic targets and candidate drug compounds, providing a theoretical foundation for precision medicine in LUAD. Figure 1 illustrates the overall workflow, while Figure 2 presents the screening flowchart of core regulatory targets.
2. Methods
2.1. Collection of public data
GEO ( http://www.ncbi.nlm.nih.gov/geo ) database serves as a public repository for high-throughput genomic data. [ 27 , 28 ] Within GEO, dataset GSE19188 [ 29 ] comprises 19 large-cell carcinoma tissue samples, 45 LUAD samples, 27 lung squamous cell carcinoma samples, and 65 adjacent nontumor lung tissue samples. For this study, samples from large-cell carcinoma and lung squamous cell carcinoma were excluded, resulting in a final cohort of 110 tissue samples (45 LUAD and 65 controls). All samples were profiled using the Affymetrix Human Genome U133 Plus 2.0 Array platform ( GPL570 , Affymetrix, Inc.).
2.2. Detection of DEGs
For the analysis of continuous and approximately normally distributed GEO microarray data, the “limma” package (version 3.60.4) in R (version 4.4.1) was employed to identify DEGs in the GEO dataset. The screening criteria were set as an adjusted P value <.05 and |log FC | > 1. [ 30 ] It should be noted that this analysis was exploratory in nature and based on GSE19188 as a single discovery dataset.
2.3. Acquisition of active compounds
SMS, a traditional Chinese herbal formula composed of ginseng, O japonicus , and S chinensis , [ 31 ] was investigated in this study. The compounds of SMS were retrieved from the encyclopaedia of traditional Chinese medicine [ 32 ] database ( http://www.tcmip.cn/ETCM/index.php/Home/Index/ ). Compounds lacking PubChem Compound IDs (CIDs) were excluded, and the remaining CIDs were submitted to the PubChem [ 33 ] database ( https://pubchem.ncbi.nlm.nih.gov/ ) to obtain their physicochemical properties. Subsequently, bioactive compounds satisfying Lipinski rule of 5 [ 34 ] were selected for further analysis.
2.4. Drug targets and target screening
The simplified molecular-input line-entry system (SMILES) information of each bioactive compound was retrieved from the PubChem database. The molecular structures in SMILES format were subsequently submitted to the SwissTargetPrediction [ 35 ] platform ( http://www.swisstargetprediction.ch/ ) to predict potential targets (only targets with probability > 0 were retained [ 36 ] ). As a complementary approach, the Similarity Ensemble Approach [ 37 ] database ( https://sea.bkslab.org/ , filtered for Homo sapiens) was employed for additional target prediction. All predicted targets were standardized using the UniProt [ 38 ] database ( https://www.uniprot.org/ ). Compounds without identifiable targets in either database were excluded from further analysis. Following this screening process, we obtained 54 bioactive compounds from ginseng, 30 from O japonicus , and 50 from S chinensis . The predicted targets of SMS bioactive compounds were then compared with DEGs in LUAD, resulting in 94 intersecting targets.
2.5. Construction of “herbs-compounds-targets-disease” network
The Cytoscape [ 39 ] software (version 3.9.1; managed by the Cytoscape Consortium) was employed to construct a H-C-T-D network, with their interconnections visualized graphically.
2.6. Enrichment analysis and screening of hub targets
The 94 intersecting targets were subjected to Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and visualization using R packages including “clusterProfiler” (version 4.12.2), “enrichplot” (version 1.24.2), “org.Hs.e.g..db” (version 3.19.1), “ggplot2” (version 3.5.2), “dplyr” (version 2.5.0), and “reshape2” (version 1.4.4).
Simultaneously, protein–protein interaction (PPI) network analysis of the 94 intersecting targets was performed on Search Tool for the Retrieval of Interacting Genes/Proteins (version 12.0; maintained by the STRING Consortium, https://string-db.org/ with the confidence threshold set to the default value of 0.4. The obtained PPI network was imported into Cytoscape 3.9.1, and hub targets were screened using the MCODE (version 2.0.3) and cytoHubba (version 0.1) plugins (managed by the Cytoscape Consortium). [ 40 ] For MCODE analysis, default parameters were applied to the entire network: Network scoring with a degree cutoff set to 2, cluster finding using the haircut method, a node score cutoff of 0.2, K-core of 2, and max. depth of 100, identifying key nodes in the 1st cluster module. For cytoHubba analysis, the top 10 key nodes in the network were identified using the maximal clique centrality algorithm. Finally, the intersection of nodes identified by both MCODE and cytoHubba was obtained, yielding the hub targets for SMS treatment of LUAD.
2.7. Validation in the TCGA-LUAD cohort
The transcriptomic data and clinical information of LUAD patient samples (TCGA-LUAD) were retrieved from the TCGA [ 41 ] database ( https://portal.gdc.cancer.gov ). The LUAD dataset in TCGA comprises 541 tumor RNA-seq samples, 59 adjacent nontumor RNA-seq samples, and 522 clinical data samples. For duplicated genes in the TCGA-LUAD transcriptomic data, the maximum expression value of each gene across all samples was retained. Additionally, only protein-coding genes were included in the analysis.
Differential expression analysis of the identified hub targets in TCGA-LUAD transcriptomic data was performed using R packages including DESeq2 (version 1.44.0), tidyverse (version 2.0.0), and ggpubr (version 0.6.0), with a significance threshold of P < .05. It should be noted that the most authentic data in TCGA are discrete RNA-seq count data following a negative binomial distribution. Therefore, this study employed DESeq2, which is specifically designed for analyzing discrete count data that adhere to a negative binomial distribution, to perform differential expression analysis. For survival analysis, clinical samples with a survival time ≥30 days were selected from TCGA-LUAD. Kaplan–Meier survival analysis was conducted using limma (version 3.60.4), survival (version 3.7-0), survminer (version 0.4.9), and ggplot2 (version 3.5.1) to visualize overall survival differences between high-risk and low-risk groups. Genes showing significant survival differences were further validated for clinical relevance.
For clinical correlation analysis, key variables – including age, gender, stage, T, M, and N status – were extracted from TCGA-LUAD clinical data, and samples with missing values were excluded. The analysis was performed using limma (version 3.60.4), ggpubr (version 0.6.0), and ComplexHeatmap (version 2.20.0), with results visualized via heatmaps to identify high-prognostic relevance targets.
2.8. Molecular docking
Compounds linked to high-prognostic relevance targets were identified through the “H-C-T-D” network. The SMILES strings of these compounds were submitted to SwissADME [ 42 ] ( http://www.swissadme.ch/ ) to obtain their absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties. The ADMET data were then filtered according to specific criteria (Table 1 ). The core goal of ADMET screening is to quickly screen out molecules that may fail due to poor pharmacokinetic properties or high toxicity, so as to focus resources on the most promising drug candidates. Topological polarity surface area is a key parameter of compound transmembrane permeability. A lower topological polarity surface area (<140 Å 2 ) usually means that the molecule is more lipophilic and easier to cross the bilayer lipid structure of the cell membrane. [ 43 , 44 ] Log S is the logarithmic representation of water solubility, [ 44 ] where a drug must dissolve in gastrointestinal fluids to be absorbed by the body. A Log S value >−4 is considered the threshold for ensuring basic acceptable solubility. [ 44 ] Estimated solubility and Ali are 2 distinct computational models used to predict a compound’s water solubility. [ 44 ] Using these 2 different models (estimated solubility and Ali) enhances the reliability of predictions while avoiding biases from relying on a single model. Gastrointestinal absorption [ 45 ] is an indicator used to evaluate the degree of gastrointestinal absorption of a compound. High gastrointestinal absorption indicates that the compound has higher potential for oral bioavailability. [ 45 ] P-glycoprotein is a transport protein found in cell membranes, such as those in the gastrointestinal tract and the blood-brain barrier. [ 44 , 46 ] It actively expels foreign substances, including certain drugs, from cells. Because a compound that is not a P-glycoprotein substrate avoids this efflux mechanism, its absorption in the human body is enhanced, leading to improved bioavailability and therapeutic efficacy. [ 44 , 46 ] Cytochrome P450 (CYP) enzymes, including the major isoforms CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4, are a group of key enzymes in the liver responsible for metabolizing most therapeutics. [ 43 , 44 ] A drug candidate that potently inhibits multiple CYPs can cause the accumulation of co-administered drugs, potentially leading to serious adverse effects. [ 44 ] Therefore, to identify “cleaner” molecules with a lower risk of drug-drug interactions, our screening strategy requires that a candidate should not be predicted as an inhibitor of at least 3 out of these 5 critical CYP enzymes.
Table 1
Screening criteria for compound ADMET properties based on the SwissADME platform.
ADMET = absorption, distribution, metabolism, excretion, and toxicity, ESOL = estimated solubility, GI = gastrointestinal, Pgp = P-glycoprotein, TPSA = topological polarity surface area.
| Screening metrics | Required criteria |
| TPSA | <140 Å 2 |
| ESOL log S | >−4 |
| Ali log S | >−4 |
| GI absorption | High |
| Pgp substrate | No |
| CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4 inhibitor | At least 3 non-inhibitory types |
In this study, the 2-dimensional structural files of target ligands were retrieved from the PubChem database. Subsequently, the ligand structures were converted into 3-dimensional models using Chem3D (version 20.0) software and subjected to energy minimization optimization with the MM2 force field to obtain stable conformations. [ 47 ] Finally, the optimized structures were exported in PDB format for subsequent molecular docking analysis.
The 3-dimensional protein structures of high-prognosis-relevant targets were downloaded from the UniProt database (species: Homo sapiens , with X-ray crystallography as the preferred determination method, followed by AlphaFold predictions). For X-ray-determined protein structures, priority was given to selecting those containing catalytic active centers, with higher resolution, and in complex with bound ligands to ensure structural integrity and applicability. For the X-ray-determined protein structures, structural refinement was performed using Modeller (version 10.4; maintained by the Sali Lab at the University of California) based on their amino acid sequences to repair missing side-chain residues. Water molecules and originally bound small-molecule ligands were removed from the crystal structures to obtain a more accurate receptor model.
The small-molecule ligands and target regulatory proteins were imported into AutoDock Tools (version 1.5.7; The Scripps Research Institute) for hydrogen addition, charge calculation, and conversion into the pdbqt format. The processed small-molecule ligands and target receptors were then subjected to molecular docking using AutoDock Vina.
2.9. Molecular dynamics simulation
To validate the stability of molecular docking results, molecular dynamics simulations were performed on strongly bound complexes (binding energy <−7.0 kcal/mol) obtained from the docking studies. The simulations were conducted using GROMACS [ 48 ] (version 2020.6) software for a duration of 100 ns to further assess the rationality and reliability of the docking results between the target protein receptor and the small-molecule drug ligand. Periodic boundary conditions were applied, and the dimensions of the simulation box were optimized by positioning the protein at the center of a cubic box with a minimum distance of 1.0 nm to the box edges, followed by solvation with water molecules. To maintain electrical neutrality in the system, a portion of the solvent water molecules was replaced with Na + and Cl − ions at a concentration of 0.15 mol/L. The CHARMM36m force field was employed to generate the parameters and topology files. Energy minimization of the entire system was carried out using the steepest descent method to eliminate unfavorable atomic contacts or overlaps, ensuring a geometrically reasonable initial structure with optimal solvent molecule orientation. Prior to production molecular dynamics simulations, the system was equilibrated by relaxing the solvent and ions around the protein to achieve convergence under the target conditions. The leap-frog algorithm was utilized for integrating the equations of motion during equilibration. All molecular dynamics simulations were performed under isothermal-isobaric conditions at 303.15 K and 1 bar pressure for a total simulation time of 100 ns. Following the molecular dynamics simulations, trajectory corrections were applied by centering the protein, and subsequent analyses included root mean square deviation (RMSD) and hydrogen bond (Hbond) evaluations.
3. Results
3.1. Determination of DEGs
The “limma” package in R was utilized to analyze 110 LUAD and adjacent normal tissue samples from the GSE19188 dataset. A total of 974 DEGs were identified based on an adjusted P value <.05 and |log FC | > 1. Among these DEGs, 356 genes were found to be up-regulated, while 618 were down-regulated in LUAD (Fig. 3 ).
3.2. Determination of SMS active compounds and related targets
Compounds lacking PubChem CIDs or violating Lipinski rule of 5 were excluded, followed by the removal of those without predicted targets in both SwissTargetPrediction and Similarity Ensemble Approach platforms. Ultimately, 54 active compounds from Panax ginseng, 30 from O japonicus , 50 from S chinensis , and 895 SMS-related targets were obtained ( Table S1 , Supplemental Digital Content 1). By comparing DEGs in LUAD microarray data from GEO, 94 intersecting targets were identified as potential therapeutic targets for SMS in LUAD treatment (Fig. 4 and Table S2 , Supplemental Digital Content 2).
3.3. Construction of the “H-C-T-D” network
The 94 overlapping targets and their associated compounds were used to construct an “H-C-T-D” network (Fig. 5 ). This network comprised 32 compounds from ginseng, 27 from O japonicus , 37 from S chinensis ( Table S3 , Supplemental Digital Content 3), and 3 shared compounds present in multiple herbs (Table 2 ).
Table 2
Bioactive compounds shared among multiple herbal medicines.
ID refers to a custom-defined code. CID = compound ID.
| PubChem compound name | PubChem CID | Drug 1 | Drug 2 | ID |
| Adenosine | 60961 | Ginseng | Ophiopogon japonicus | A |
| [1 S -(1α,4α,7α)]-1,2,3,4,5,6,7,8-octahydro-1,4,9,9-tetramethyl-4,7-methanoazulene | 12313993 | Ginseng | O japonicus | B |
| Besigomsin | 3001662 | Ginseng | Schisandra chinensis | C |
3.4. Enrichment analysis and screening of hub targets
Gene Ontology and KEGG pathway enrichment analyses were conducted on the 94 intersecting targets using R software, with subsequent visualization of both enrichment results and KEGG pathway gene expression heatmaps (Fig. 6 ). The analysis revealed that SMS treatment of LUAD was significantly associated with several biological processes, including regulation of inflammatory response and response to toxic substance. Key cellular components identified were membrane raft and membrane microdomain, while prominent molecular functions were characterized as endopeptidase activity and serine-type endopeptidase activity. Furthermore, the KEGG pathway analysis identified several core pathways, including the tumor necrosis factor signaling pathway, interleukin-17 (IL-17) signaling pathway, cell cycle, and lipid and atherosclerosis pathways. According to the KEGG heatmap, the genes involved in the tumor necrosis factor signaling pathway include FOS, ICAM1, IL-6, MAP3K8, MMP3, MMP9, PTGS2, and SELE. The IL-17 signaling pathway encompasses genes such as FOS, IL-6, MMP1, MMP13, MMP3, MMP9, and PTGS2. The cell cycle pathway involves genes including CCNA2, CCNB1, CCNE2, CDK1, CHEK1, PLK1, and TTK. Meanwhile, the lipid and atherosclerosis pathway is associated with genes such as FOS, ICAM1, IL-6, MMP1, MMP3, MMP9, PPARG, SELE, and SELP.
A PPI network was constructed by inputting 94 overlapping targets into the Search Tool for the Retrieval of Interacting Genes/Proteins database (Fig. 7 A). The PPI network was subsequently analyzed for hub target identification using both MCODE and cytoHubba packages. A total of 26 key nodes were identified by the MCODE package as the 1st cluster module (Fig. 7 B), while the top 10 key nodes in the network were determined by the cytoHubba package using the maximal clique centrality algorithm (Fig. 7 C). Finally, an intersection analysis was performed between the nodes identified by MCODE and cytoHubba, resulting in the identification of 10 hub targets for SMS treatment of LUAD (Table 3 ).
Table 3
Ten hub targets for SMS treatment of LUAD.
LUAD = lung adenocarcinoma, SMS = Shengmai San.
| Gene symbol | UniProt ID | Protein name |
| AURKA | O14965 | Aurora kinase A |
| CCNA2 | P20248 | Cyclin-A2 |
| CCNB1 | P14635 | G2/mitotic-specific cyclin-B1 |
| CDK1 | P06493 | Cyclin-dependent kinase 1 |
| CHEK1 | O14757 | Serine/threonine-protein kinase Chk1 |
| KIF11 | P52732 | Kinesin-like protein KIF11 |
| NEK2 | P51955 | Serine/threonine-protein kinase Nek2 |
| PLK1 | P53350 | Serine/threonine-protein kinase PLK1 |
| TTK | P33981 | Dual specificity protein kinase TTK |
| TYMS | P04818 | Thymidylate synthase |
3.5. Validation of the TCGA-LUAD cohort
A significant differential expression pattern was 1st observed for all 10 hub targets (AURKA, CCNA2, CCNB1, CDK1, CHEK1, KIF11, NEK2, PLK1, TTK, and TYMS) between LUAD tumor samples and non-LUAD tissues (Fig. 8 A–J). This expression pattern was subsequently validated in the GSE19188 dataset, where distinct expression profiles were demonstrated between LUAD specimens and their adjacent normal tissues.
Subsequently, the correlation between the 10 hub targets and survival prognosis was examined (Fig. 9 A–J). The cohort was evenly distributed, with 246 and 247 patients assigned to the high- and low-expression groups, respectively, for each gene. Significantly reduced survival probabilities ( P < .001) with lower hazard ratios (HR range: 0.54–0.59) were observed in the high-expression groups of CCNA2, CCNB1, CDK1, KIF11, NEK2, PLK1, TTK, and TYMS, indicating their elevated expression was associated with poorer prognosis. Although comparatively weaker statistical significance was demonstrated for AURKA ( P = .045, HR = 0.74) and CHEK1 ( P = .003, HR = 0.64), their high expression levels were still correlated with reduced survival rates. Furthermore, a progressive decline in surviving patient numbers was consistently noted in high-expression groups over time, providing additional evidence for the negative prognostic impact of elevated expression of these genes.
Furthermore, the clinical correlations of the 10 hub targets were analyzed using the TCGA database (Fig. 10 and Table 4 ). Highly significant associations ( P < .001) were observed for KIF11 and PLK1 with gender, stage, and T-stage. CDK1 and CHEK1 were found to be most strongly associated with tumor stage and T-stage ( P < .001). TTK demonstrated highly significant correlations with gender and T-stage ( P < .001), while AURKA showed a highly significant association with gender ( P < .001). CCNA2 and CCNB1 exhibited highly significant differences across tumor stages ( P < .001). Moderate associations were identified for CDK1 with gender and N-stage, and for CHEK1 with age. KIF11 was shown to have a moderate correlation with N-stage, and PLK1 displayed moderate associations with M-stage and N-stage. Although NEK2 and TYMS were also found to have some correlations with gender and tumor stage, these associations were relatively weaker compared to other genes. Since CDK1, CHEK1, KIF11, PLK1, and TTK were all demonstrated to have high correlations with 2 or more clinical features, these targets could be considered as highly relevant prognostic markers for LUAD.
Table 4
Clinical correlation statistics for the 10 hub targets.
* P < .05. ** P < .01. *** P < .001.
| Gene | Age | Gender | Stage | T | M | N |
| AURKA | * | *** | ** | - | ** | - |
| CCNA2 | - | ** | *** | ** | ** | * |
| CCNB1 | ** | ** | *** | ** | * | ** |
| CDK1 | - | ** | *** | *** | * | ** |
| CHEK1 | ** | - | *** | *** | - | * |
| KIF11 | - | *** | *** | *** | - | ** |
| NEK2 | * | ** | ** | ** | * | - |
| PLK1 | - | *** | *** | *** | ** | ** |
| TTK | - | *** | ** | *** | * | * |
| TYMS | - | * | ** | - | * | - |
3.6. Molecular docking
Forty-four compounds associated with the 5 high-prognostic relevance targets (CDK1, CHEK1, KIF11, PLK1, and TTK) were identified through the “H-C-T-D” network. After ADMET screening, 7 bioactive compounds were obtained, as shown in Table 5 (ADMET data provided in Table S4 , Supplemental Digital Content 4).
Table 5
Seven compounds screened by ADMET.
ID refers to a custom-defined code. RS12 is derived from ginseng. MD6, MD10, MD12, MD14 and MD25 are derived from Ophiopogon japonicus . WWZ16 is derived from Schisandra chinensis . ADMET = absorption, distribution, metabolism, excretion, and toxicity, CID = compound ID.
| Compound name | PubChem CID | ID |
| 7-Hydroxy-2,5-dimethyl-4 H -1-benzopyran-4-one | 5316891 | RS12 |
| N -trans-feruloyltyramine | 5280537 | MD6 |
| Paprazine | 5372945 | MD10 |
| ( E )- N -[(2 S )-2-hydroxy-2-(4-hydroxyphenyl)ethyl]-3-(4-hydroxyphenyl)prop-2-enamide | 11185637 | MD12 |
| ( Z , S )-jasmololone | 12304693 | MD14 |
| Jasmolone | 5374699 | MD25 |
| (2 S )-2-[(2 R ,5 S )-5-ethenyl-5-methyloxolan-2-yl]-6-methylhept-5-en-3-one | 101796751 | WWZ16 |
Prior to docking, the protein structures of PLK1, CDK1, CHEK1 and TTK were derived from X-ray crystallographic determinations, with the docking box center being defined by the centroid of their native small-molecule ligand binding sites. For the KIF11 protein, whose experimentally determined X-ray structure was unavailable, a predicted structural model generated by AlphaFold was utilized, and the docking box center was established based on the amino acid residue cluster constituting the ATP-binding site. The target names, sources of 3-dimensional protein structures, and corresponding docking box center parameters are summarized in Table 6 .
Table 6
Information on high-prognostic relevance targets, their corresponding protein structures, and docking box center parameters.
| Gene | UniProt ID | PDBID | AlphaFold ID | X | Y | Z |
| CDK1 | P06493 | 6GU2 | N/A | 328.83 | 213.27 | 192.53 |
| CHEK1 | O14757 | 2YEX | N/A | 1.97 | 37.18 | 10.63 |
| PLK1 | P53350 | 2RKU | N/A | −47 | 12.2 | 16.67 |
| TTK | P33981 | 4JS8 | N/A | 0.11 | 17.69 | 9.9 |
| KIF11 | Q6P9P6 | N/A | Q6P9P6 | 2.81 | 10.36 | 15.89 |
Molecular docking was performed between the bioactive compounds and their corresponding high-prognostic relevance targets (Table 7 ), and the protein-compound docking complexes were visualized (Fig. 11 ). Higher absolute values of binding energy indicated greater complex stability, where binding energies below −5.0 and −7.0 kcal/mol conventionally represent moderate and strong binding affinities, respectively. [ 49 ]
Table 7
Molecular docking results of high-prognostic relevance targets with corresponding compounds.
ID refers to a custom-defined code. RS12 is derived from ginseng. MD6, MD10, MD12, MD14, MD25 are derived from Ophiopogon japonicus , WWZ16 is derived from Schisandra chinensis . Hbond = hydrogen bond.
| ID | Target | Binding energy (kcal/mol) | Hbond residues (length) | π–-π interaction residues (length) |
| RS12 | CDK1 | −7.5 | LYS-33 (2.7), GLU-81 (2.5), ASP-146 (2.7) | N/A |
| RS12 | PLK1 | −7.3 | CYS-67 (3.8), CYS-133 (1.9) | PHE-183 (4.1) |
| MD6 | CDK1 | −8.8 | GLU-51 (2.1), LEU-83 (2.5), PHE-147 (2.7) | PHE-80 (4.0) |
| MD6 | CHEK1 | −8.3 | GLU-17 (2.4), GLU-85 (2.6), CYS-87 (2.4) | N/A |
| MD10 | CDK1 | −9.1 | GLU-51 (3.0), ASP-146 (3.8), PHE-147 (2.4) | TYR-15 (5.1), PHE-80 (4.0) |
| MD10 | CHEK1 | −8.3 | GLU-17 (3.6), GLU-85 (2.7), CYS-87 (3.1) | N/A |
| MD12 | CHEK1 | −8.3 | GLU-17 (2.5), CYS-87 (2.3), ASP-148 (2.8) | N/A |
| MD14 | CDK1 | −6.6 | LEU-83 (2.3) | N/A |
| MD14 | CHEK1 | −6.3 | CYS-87 (3.5) | N/A |
| MD14 | TTK | −5.8 | GLN-670 (2.5) | N/A |
| MD25 | CDK1 | −6.5 | LEU-83 (2.2) | N/A |
| WWZ16 | KIF11 | −5.8 | ARG-23 (2.4) | N/A |
All 7 bioactive compounds demonstrated binding energies below −5 kcal/mol with their respective targets (Table 7 ). PyMOL (version 3.1.3; Schrödinger, Inc.) was utilized for interaction visualization and 3D interaction diagram generation. Notably, Paprazine exhibited optimal binding affinity toward CDK1 among all evaluated compounds.
3.7. Molecular dynamics simulation
According to the molecular docking results, a total of 7 docking complexes with binding energies <−7 kcal/mol were identified, indicating strong binding activity. These 7 docking complexes were subjected to molecular dynamics simulation using GROMACS software to analyze the dynamic changes in RMSD and Hbonds. RMSD was employed to evaluate the degree of structural deviation of the protein from its initial conformation during the simulation. The number and stability of protein-ligand Hbonds were found to directly influence binding affinity.
The molecular dynamics simulation results (Figs. 12 – 18 ) demonstrated that the RMSD values of all 7 compound-target protein complexes exhibited overall stable trends, with relatively stable Hbond counts. Based on these findings, the primary active components of SMS in the treatment of LUAD were determined to be 7-hydroxy-2,5-dimethyl-4 H -1-benzopyran-4-one, N -trans-feruloyltyramine, paprazine, and ( E )- N -[(2 S )-2-hydroxy-2-(4-hydroxyphenyl)ethyl]-3-(4-hydroxyphenyl)prop-2-enamide. The core regulatory targets were identified as CDK1, CHEK1, and PLK1.
![Figure 18. Molecular dynamics simulation results of the MD12 and CHEK1 complex (MD12: ( E )- N -[(2 S )-2-hydroxy-2-(4-hydroxyphenyl)ethyl]-3-(4-hydroxyphenyl)prop-2-enamide). (A)…](/library/25226/assets/ae8b38b50594f000ccd1f55a1ca20abdad90591f80f23ee12b3adb3d04ac5dba.webp/)
4. Discussion
LUAD, a common subtype of NSCLC, is characterized by complex and diverse pathological features, which are primarily reflected in molecular subtypes, histological subtypes, tumor microenvironment (TME), and metabolic abnormalities. LUAD exhibits significant molecular heterogeneity and can be classified into multiple molecular subtypes based on driver gene mutations, [ 50 ] and dysregulated signaling pathways. [ 51 ] These subtypes are closely associated with patient prognosis. [ 52 ] Histologically, LUAD can be categorized into lepidic (indolent), micropapillary, and solid (aggressive) subtypes, among others. Each subtype demonstrates distinct clinical manifestations and prognostic outcomes. LUAD with micropapillary components is recognized as a more aggressive subtype and is associated with poorer prognosis. [ 53 , 54 ] The TME plays a critical role in the initiation and progression of LUAD. Within the TME of LUAD, various immune and stromal cells are present, and their proportions and functions exhibit significant spatial heterogeneity across different tumor regions. [ 55 ] Metabolic reprogramming is widely observed in LUAD, with dysregulation of glycolysis and lipid metabolism pathways being particularly implicated in tumorigenesis and progression. These metabolic abnormalities not only influence tumor cell proliferation and survival but are also closely linked to the immune microenvironment of the tumor. [ 56 , 57 ]
The complex and heterogeneous pathological features of LUAD are closely associated with prognosis; therefore, treatment strategies based on pathological characteristics are expected to enhance therapeutic precision and efficacy. TCM is characterized by its multi-component, multi-target integrated regulatory effects in LUAD treatment, demonstrating unique advantages in modulating immune function, improving patients’ quality of life, and alleviating the side effects of chemotherapy and radiotherapy. In this study, an integrated approach combining network pharmacology, bioinformatics, molecular docking, and molecular dynamics simulation was employed to identify 4 key bioactive compounds in SMS and 3 critical core targets for LUAD intervention, providing valuable insights for potential candidate drugs and precision-targeted therapeutic strategies. Specifically, within the bioinformatics analysis, the “H-C-T-D” network and TCGA cohort were utilized to achieve the identification of core regulatory targets. Based on the findings, CDK1, CHEK1, and PLK1 were identified as the core regulatory targets of SMS in LUAD treatment and were significantly associated with poorer survival rates in LUAD patients. CDK1 (cyclin-dependent kinase 1) has been recognized as a crucial therapeutic target in various cancers. [ 58 , 59 ] It was demonstrated that elevated CDK1 expression in LUAD correlates with altered relative abundance of tumor-infiltrating immune cells, including B cells, CD4+ T cells, and macrophages, which may further influence tumor progression and patient prognosis. [ 60 ] CHEK1 (serine/threonine protein kinase Chk1) plays a crucial role in various cancers. It was found that certain genetic variations of CHEK1 are significantly associated with overall survival in NSCLC patients. [ 61 ] Additionally, CHEK1 plays a significant role in multiple myeloma, where its high expression is linked to cell proliferation and drug resistance and may promote bone lesion formation by activating chromosomal instability. [ 62 ] PLK1 (serine/threonine-protein kinase PLK1) was shown to promote LUAD pathogenesis and progression through multiple mechanisms, including the formation of an immunosuppressive TME. [ 63 ] Additionally, PLK1 may activate the DIO2-TH signaling pathway via phosphorylation of the aryl hydrocarbon receptor, thereby facilitating epithelial-mesenchymal transition (EMT) and metastatic events in LUAD. [ 64 ]
The critical compound 7-hydroxy-2,5-dimethyl-4 H -1-benzopyran-4-one (RS12), a benzopyranone derivative, was found to primarily target PLK1 and CDK1, thereby influencing immune cell abundance and regulating EMT and metastatic events. Benzopyranone derivatives have been reported to exhibit potent antiproliferative activity in human breast cancer cell lines [ 65 ] ; however, their immunomodulatory effects in cancer require further investigation. Two other key compounds, N -trans-feruloyltyramine (MD6) and paprazine (MD10), were shown to form highly stable complexes with CDK1 and CHEK1, suggesting their roles in modulating immune cell abundance and affecting tumor cell proliferation. MD6, a key member of the ferulic acid (FLA) family, shares a core FLA derivative structure. It has been demonstrated that FLA suppresses the proliferation and metastatic potential of breast cancer cells through modulation of the EMT process. [ 66 ] FLA also blocks the phosphoinositide 3-kinase/Akt pathway, induces G0/G1 arrest, and promotes apoptosis in osteosarcoma cells, [ 67 ] which is consistent with the findings of the present study. Previous studies have demonstrated that MD10, when combined with indomethacin or diclofenac, enhances apoptosis in breast cancer cells through an endoplasmic reticulum stress-dependent mechanism. [ 68 ] ( E )- N -[(2 S )-2-hydroxy-2-(4-hydroxyphenyl)ethyl]-3-(4-hydroxyphenyl)prop-2-enamide (MD12), an enamide derivative, primarily acts on CHEK1 to regulate tumor cell proliferation and apoptosis. Studies have identified that acrylamide-containing olmutinib derivatives exhibit promising bioactivity against epidermal growth factor receptor kinase and phosphoinositide 3-kinase kinase, demonstrating anticancer potential [ 69 ] and warranting further investigation.
Several limitations of this study warrant consideration. First, although a 100 ns [ 70 , 71 ] molecular dynamics simulation is common practice and sufficient for achieving local equilibrium, this timescale may be inadequate to capture rare conformational transitions or complete binding/dissociation kinetics, which are essential for a comprehensive understanding of complex stability and binding mechanisms. Beyond this computational constraint, our DEG screening relied on a single transcriptomic profile, and the lack of independent external validation beyond TCGA limits the generalizability of our findings. Moreover, the network pharmacology and docking approaches employed herein are inherently subject to database completeness, scoring function approximations, and the attendant risk of false-positive identifications. Importantly, no wet-lab experiments have been performed to corroborate any of the in silico predictions. Consequently, our results should be interpreted as exploratory and hypothesis-generating, serving to prioritize candidates for future experimental testing rather than to provide definitive therapeutic evidence. All prioritized targets and compounds explicitly await independent experimental confirmation.
5. Conclusion
The 4 compounds contained in SMS, including paprazine, were demonstrated to possess favorable drug-like properties (complying with Lipinski rule of 5 and ADMET characteristics), suggesting their potential as candidate drugs for LUAD treatment. Further in vitro and in vivo pharmacodynamic validation and safety evaluation of these components should be conducted, which may provide a scientific basis for either single-component targeted drug development or active fraction extraction from TCM formulations.
The core regulatory targets, CDK1, CHEK1, and PLK1 identified in this study were demonstrated to be key cell cycle regulatory proteins, whose overexpression was closely associated with abnormal tumor cell proliferation. The expression levels of CDK1, CHEK1, and PLK1 were significantly correlated with the prognosis of LUAD patients, indicating their potential application as prognostic biomarkers. Based on computational predictions derived from network pharmacology analyses, a combination regimen incorporating active components of SMS with standard chemotherapy or targeted therapeutics [ 13 , 72 , 73 ] was proposed for LUAD patients exhibiting high expression of CDK1/CHEK1/PLK1 and developing resistance to existing treatments. It is emphasized that this combination proposal is hypothesis‑generating and prediction‑derived, rather than a direct finding of the present study. Additionally, because the active components of SMS were also found to ameliorate the TME through modulation of cell cycle pathways, further research on immunotherapy combining SMS with PD‑1/PD‑L1 inhibitors was recommended to evaluate its potential immunomodulatory effects. These predictive findings may provide potential drug candidates and combination strategies for precision therapy in LUAD. It should be noted that these findings are computational predictions that await experimental validation and may inform future investigations.
Author contributions
Conceptualization: Xiaoling Zhou.
Data curation: Xiaohui Zhang, Xinyou Zhang.
Formal analysis: Xiaoling Zhou, Diyao Wu.
Methodology: Xiaoling Zhou, Diyao Wu, Xiaohui Zhang.
Software: Xiaoling Zhou, Diyao Wu, Xinyou Zhang.
Visualization: Xiaoling Zhou.
Writing – original draft: Xiaoling Zhou, Diyao Wu, Xiaohui Zhang.
Writing – review & editing: Diyao Wu, Xinyou Zhang.
Abbreviations: ADMET absorption, distribution, metabolism, excretion, and toxicity AKT protein kinase B, CIDs compound IDs DEGs differentially expressed genes EMT epithelial-mesenchymal transition FLA ferulic acid GEO Gene Expression Omnibus Hbond hydrogen bond H-C-T-D herbs-compounds-targets-disease HR hazard ratios IL interleukin KEGG Kyoto Encyclopedia of Genes and Genomes LUAD lung adenocarcinoma NSCLC non-small cell lung cancer PPI protein–protein interaction RMSD root mean square deviation SMILES simplified molecular-input line-entry system, SMS Shengmai San TCGA The Cancer Genome Atlas TCM traditional Chinese medicine TME tumor microenvironment.
This study was supported by The National Natural Science Foundation of China (No. 82360992), Natural Science Foundation of Jiangxi Province (No. 20224BAB206115), Key Research Laboratory of Cancer Formula and Syndrome Information Data Mining of Jiangxi Provincial Administration of Traditional Chinese Medicine (No. ZDYJS20202202), Key Project of Science and Technology Plan of Jiangxi Provincial Administration of Traditional Chinese Medicine (No. 2022Z007), Development Plan of Science and Technology Innovation Team at Jiangxi University of Chinese Medicine (No. CXTD22015).
All data employed in this analysis were sourced from publicly accessible online repositories. For this reason, ethical review was not applicable.
The authors have no conflicts of interest to declare.
All data generated or analyzed during this study are included in this published article (and its supplementary information files).
Supplemental Digital Content is available in the online version of this article ( http://dx.doi.org/10.1097/MD.0000000000050440 ).
How to cite this article: Zhou X, Wu D, Zhang X, Zhang X. Exploring the mechanism of Shengmai San in treating lung adenocarcinoma based on bioinformatics and molecular dynamics simulation. Medicine 2026;105:35(e50440).
Contributor Information
Xiaoling Zhou, Email: 20194001@jxutcm.edu.cn.
Diyao Wu, Email: 20201011@jxutcm.edu.cn.
Xiaohui Zhang, Email: xinyouzhang@163.com.
References
[1] Zhang L, Xiong Y, Zhang J, Feng Y, Xu A. Systematic proteome-wide Mendelian randomization using the human plasma proteome to identify therapeutic targets for lung adenocarcinoma. J Transl Med. 2024;22:330.
[2] Wei X, Li X, Hu S, Cheng J, Cai R. Regulation of ferroptosis in lung adenocarcinoma. Int J Mol Sci. 2023;24:14614.
[3] Zhan P, Zhang B, Xi G-M, et al. PRC1 contributes to tumorigenesis of lung adenocarcinoma in association with the Wnt/β-catenin signaling pathway. Mol Cancer. 2017;16:108.
[4] Liu S, Wang X, Qin W, Genchev GZ, Lu H. Transcription factors contribute to differential expression in cellular pathways in lung adenocarcinoma and lung squamous cell carcinoma. Interdiscip Sci. 2018;10:836–47.
[5] Zhang S, Liu C, Liu D, et al. Integrin β4 regulates cell migration of lung adenocarcinoma through FAK signaling. Mol Biotechnol. 2025;67:496–509.
[6] Wei Z, Zhao Y, Cai J, Xie Y. The nucleolar protein C1orf131 is a novel gene involved in the progression of lung adenocarcinoma cells through the AKT signalling pathway. Int J Mol Sci. 2024;25:6381.
[7] Wu A, Yang H, Xiao T, Gu W, Li H, Chen P. COPZ1 regulates ferroptosis through NCOA4-mediated ferritinophagy in lung adenocarcinoma. Biochim Biophys Acta Gen Subj. 2024;1868:130706.
[8] Li H, Wang W, Huang Z, et al. Exploration of the shared genes and signaling pathways between lung adenocarcinoma and idiopathic pulmonary fibrosis. J Thorac Dis. 2023;15:3054–68.
[9] Saito A, Horie M, Micke P, Nagase T. The role of TGF-β signaling in lung cancer associated with idiopathic pulmonary fibrosis. Int J Mol Sci . 2018;19:3611.
[10] Xu J, Tian L, Qi W, Lv Q, Wang T. Advancements in NSCLC: from pathophysiological insights to targeted treatments. Am J Clin Oncol. 2024;47:291–303.
[11] Ren Q, Zhang P, Lin H, et al. A novel signature predicts prognosis and immunotherapy in lung adenocarcinoma based on cancer-associated fibroblasts. Front Immunol. 2023;14:1201573.
[12] Yu X-Y, Wang M, Qian J-J. CMaf-inducing protein promotes LUAD proliferation and metastasis by activating the MAPK/ERK pathway. Evidence-based complementary and alternative medicine. Evid Based Complement Alternat Med. 2022;2022:1–11.
[13] Wang N, Zhao Q, Huang Y, et al. Lnc-TMEM132D-AS1 as a potential therapeutic target for acquired resistance to osimertinib in non-small-cell lung cancer. Mol Omics. 2023;19:238–51.
[14] Chae YK, Pan AP, Davis AA, et al. Path toward precision oncology: review of targeted therapy studies and tools to aid in defining “actionability” of a molecular lesion and patient management support. Mol Cancer Ther. 2017;16:2645–55.
[15] Girard N. Place adjuvant targeted therapies during localized NSCLC. Bull Cancer. 2025;112:3S46–51.
[16] Matsuoka T, Yashiro M. Molecular-targeted therapy toward precision medicine for gastrointestinal caner: current progress and challenges. World J Gastrointest Oncol. 2021;13:366–90.
[17] Zhang B-R, Liu X-H, Ling Y-T, et al. Effectiveness and safety of shengmai san for viral myocarditis: a systematic review and meta-analysis of randomized controlled trials. Cardiovasc Ther. 2024;2024:2127018.
[18] Ouyang Y, Tang L, Hu S, et al. Shengmai san-derived compound prescriptions: a review on chemical constituents, pharmacokinetic studies, quality control, and pharmacological properties. Phytomedicine. 2022;107:154433.
[19] Ma J, Yu Z, Chen T, et al. The effect of Shengmai injection in patients with coronary heart disease in real world and its personalized medicine research using machine learning techniques. Front Pharmacol. 2023;14:1208621.
[20] Kim J-W, Jung S-Y, Kwon Y-H, et al. Ginsenoside Rg3 attenuates tumor angiogenesis via inhibiting bioactivities of endothelial progenitor cells. Cancer Biol Ther. 2012;13:504–15.
[21] Nasser MI, Zhu S, Chen C, Zhao M, Huang H, Zhu P. A comprehensive review on schisandrin B and its biological properties. Oxid Med Cell Longev. 2020;2020:2172740.
[22] Dong F, Qu L, Duan Z, He Y, Ma X, Fan D. Ginsenoside Rh4 inhibits breast cancer growth through targeting histone deacetylase 2 to regulate immune microenvironment and apoptosis. Bioorg Chem. 2023;135:106537.
[23] Xiong H, Han X, Cai L, Zheng H. Natural polysaccharides exert anti-tumor effects as dendritic cell immune enhancers. Front Oncol. 2023;13:1274048.
[24] Duan B, Xie J, Rui Q, Zhang W, Xi Z. Effects of Shengmai injection add-on therapy to chemotherapy in patients with non-small cell lung cancer: a meta-analysis. Support Care Cancer. 2018;26:2103–11.
[25] Sun Y, Chen Y, Xu M, Liu C, Shang H, Wang C. Shenmai injection supresses glycolysis and enhances cisplatin cytotoxicity in cisplatin-resistant A549/DDP Cells via the AKT-mTOR-c-Myc signaling pathway. Biomed Res Int. 2020;2020:9243681.
[26] Bing P, Zhou W, Tan S. Study on the mechanism of astragalus polysaccharide in treating pulmonary fibrosis based on “drug-target-pathway” network. Front Pharmacol. 2022;13:865065.
[27] Li Y, Gu J, Xu F, Zhu Q, Ge D, Lu C. Transcriptomic and functional network features of lung squamous cell carcinoma through integrative analysis of GEO and TCGA data. Sci Rep. 2018;8:15834.
[28] Barrett T, Wilhite SE, Ledoux P, et al. NCBI GEO: archive for functional genomics data sets – update. Nucleic Acids Res. 2013;41:D991–995.
[29] Hou J, Aerts J, Hamer B D, et al. Gene expression-based classification of non-small cell lung carcinomas and survival prediction. PLoS One. 2010;5:e10312.
[30] Liu T, Wu H, Wei J. The construction and validation of a novel ferroptosis-related gene signature in Parkinson’s disease. IJMS. 2023;24:17203.
[31] Shang H, Wang T, Pang D, et al. The Shengmai San exerts therapeutic effects on type 2 diabetes by modulating the PI3K/AKT/GSK3B signaling pathway. Diabetes Metab Syndr Obes. 2025;18:985–1004.
[32] Xu H-Y, Zhang Y-Q, Liu Z-M, et al. ETCM: an encyclopaedia of traditional Chinese medicine. Nucleic Acids Res. 2019;47:D976–82.
[33] Kim S, Chen J, Cheng T, et al. PubChem 2025 update. Nucleic Acids Res. 2025;53:D1516–25.
[34] Azme E, Hasan MM, Ali ML, et al. Computational identification of potential natural terpenoid inhibitors of MDM2 for breast cancer therapy: molecular docking, molecular dynamics simulation, and ADMET analysis. Front Chem. 2025;13:1527008.
[35] Daina A, Michielin O, Zoete V. SwissTargetPrediction: updated data and new features for efficient prediction of protein targets of small molecules. Nucleic Acids Res. 2019;47:W357–64.
[36] Khanal P, Zargari F, Far BF, et al. Integration of system biology tools to investigate huperzine a as an anti-Alzheimer agent. Front Pharmacol. 2021;12:785964.
[37] Keiser MJ, Roth BL, Armbruster BN, Ernsberger P, Irwin JJ, Shoichet BK. Relating protein pharmacology by ligand chemistry. Nat Biotechnol. 2007;25:197–206.
[38] UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025. Nucleic Acids Res. 2025;53:D609–17.
[39] Shannon P, Markiel A, Ozier O, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498–504.
[40] Lin Y, Liao J, Chong Y. Identification of the shared gene signatures and biological mechanism in type 2 diabetes mellitus and hepatocellular carcinoma. Clin Exp Hepatol. 2025;11:34–44.
[41] Wan Y, Jiang H, Liu Z, et al. Exploring the molecular mechanisms of huaier on modulating metabolic reprogramming of hepatocellular carcinoma: a study based on network pharmacology, molecular docking and bioinformatics. Curr Pharm Des. 2024;30:1894–911.
[42] Daina A, Michielin O, Zoete V. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci Rep. 2017;7:42717.
[43] El-Fagal SF, El-Helw EAE, El-Bordany EA, Ghareeb EA. Antioxidant activity, molecular docking, and modeling pharmacokinetics study of some benzo[f]quinoline candidates. Sci Rep. 2025;15:16522.
[44] Kibet S, Kimani NM, Mwanza SS, Mudalungu CM, Santos CBR, Tanga CM. Unveiling the potential of ent-kaurane diterpenoids: multifaceted natural products for drug discovery. Pharmaceuticals (Basel). 2024;17:510.
[45] Taher MA, Islam MA, Tasmi SF, et al. Phytochemical profiling and bioactivity validation of Moringa oleifera leaves: antimicrobial, antidiarrheal, analgesic, and in silico insights. PLoS One. 2025;20:e0332048.
[46] Abd-ElSalam R, Khaled NA, Ibrahim MA. Modeling the functionalized genistein-hyoscyamine derivatives. Sci Rep. 2025;15:16662.
[47] Zamani M, Mokarram P, Jamshidi M, Siri M, Ghasemi H. Molecular modelling of resveratrol derivatives with SIRT1 for the stimulation of deacetylase activity. Curr Comput Aided Drug Des. 2024;20:943–54.
[48] Abuagla HA, Adam KM, Elangeeb ME, et al. Impact of nonsynonymous single nucleotide polymorphisms in PROCR gene on protein stability and thrombotic risk: a molecular dynamic approach. Front Genet. 2025;16:1580993.
[49] Liu L, Jiao Y, Yang M, Wu L, Long G, Hu W. Network pharmacology, molecular docking and molecular dynamics to explore the potential immunomodulatory mechanisms of deer antler. Int J Mol Sci . 2023;24:10370.
[50] Ruiz-Cordero R, Ma J, Khanna A, et al. Simplified molecular classification of lung adenocarcinomas based on EGFR, KRAS, and TP53 mutations. BMC Cancer. 2020;20:83.
[51] Daemen A, Cooper JE, Myrta S, et al. Transcriptional subtypes resolve tumor heterogeneity and identify vulnerabilities to MEK inhibition in lung adenocarcinoma. Clin Cancer Res. 2021;27:1162–73.
[52] Jeon Y, Choi K-C, Park YN, Kim Y-J. Identification of molecular subtypes and prognostic traits based on chromosomal instability phenotype-related genes in lung adenocarcinoma. Cancers (Basel). 2024;16:3818.
[53] Zhang S, Xu Y, Zhao P, et al. Integrated analysis of genomic and immunological features in lung adenocarcinoma with micropapillary component. Front Oncol. 2021;11:652193.
[54] Huang J, Li J, Zheng S, et al. Tumor microenvironment characterization identifies two lung adenocarcinoma subtypes with specific immune and metabolic state. Cancer Sci. 2020;111:1876–86.
[55] Wang Y, Li X, Peng S, et al. Single-cell analysis reveals spatial heterogeneity of immune cells in lung adenocarcinoma. Front Cell Dev Biol. 2021;9:638374.
[56] Huo C, Zhang M-Y, Li R, Liu T-T, Li J-P, Qu Y-Q. Glycolysis define two prognostic subgroups of lung adenocarcinoma with different mutation characteristics and immune infiltration signatures. Front Cell Dev Biol. 2021;9:645482.
[57] Zhang X, Li W, Liu T, Guo H, Sun Q, Li B. Heterogeneity of lipid metabolism and its clinical and immune correlation in lung adenocarcinoma. Curr Med Chem. 2024;31:1561–77.
[58] Linton A, Cheng YY, Griggs K, et al. An RNAi-based screen reveals PLK1, CDK1 and NDC80 as potential therapeutic targets in malignant pleural mesothelioma. Br J Cancer. 2018;118:e13.
[59] Lu X, Pang Y, Cao H, et al. Integrated screens identify CDK1 as a therapeutic target in advanced gastrointestinal stromal tumors. Cancer Res. 2021;81:2481–94.
[60] Li S, Li H, Cao Y, et al. Integrated bioinformatics analysis reveals CDK1 and PLK1 as potential therapeutic targets of lung adenocarcinoma. Medicine (Baltim). 2021;100:e26474.
[61] Mu R, Liu H, Luo S, et al. Genetic variants of CHEK1, PRIM2 and CDK6 in the mitotic phase-related pathway are associated with nonsmall cell lung cancer survival. Int J Cancer. 2021;149:1302–12.
[62] Gu C, Wang W, Tang X, et al. CHEK1 and circCHEK1_246aa evoke chromosomal instability and induce bone lesion formation in multiple myeloma. Mol Cancer. 2021;20:84.
[63] Kong Y, Li C, Liu J, et al. Single-cell analysis identifies PLK1 as a driver of immunosuppressive tumor microenvironment in LUAD. PLoS Genet. 2024;20:e1011309.
[64] Li C, Allison DB, He D, et al. Phosphorylation of AHR by PLK1 promotes metastasis of LUAD via DIO2-TH signaling. PLoS Genet. 2023;19:e1011017.
[65] Musa MA, Cooperwood JS, Khan MOF, Rahman T. In-vitro antiproliferative activity of benzopyranone derivatives in comparison with standard chemotherapeutic drugs. Arch Pharm (Weinheim). 2011;344:102–10.
[66] Zhang X, Lin D, Jiang R, Li H, Wan J, Li H. Ferulic acid exerts antitumor activity and inhibits metastasis in breast cancer cells by regulating epithelial to mesenchymal transition. Oncol Rep. 2016;36:271–8.
[67] Wang T, Gong X, Jiang R, Li H, Du W, Kuang G. Ferulic acid inhibits proliferation and promotes apoptosis via blockage of PI3K/Akt pathway in osteosarcoma cell. Am J Transl Res. 2016;8:968–80.
[68] Wongsakul A, Lertnitikul N, Suttisri R, Jianmongkol S. N-trans-p-coumaroyltyramine enhances indomethacin- and diclofenac-induced apoptosis through endoplasmic reticulum stress-dependent mechanism in MCF-7 cells. Anticancer Res. 2022;42:1833–44.
[69] Hu X, Tang S, Yang F, et al. Design, synthesis, and antitumor activity of olmutinib derivatives containing acrylamide moiety. Molecules. 2021;26:3041.
[70] Manzoor H, Khan MU, Shabbir CA, et al. Computational identification and evaluation of novel PD-L1 inhibitors for cancer immunotherapy. Sci Rep. 2025;15:33787.
[71] Aouiffat Y, Bakchiche B, Benaceur F, et al. Seasonal variation in the phenolic compounds of Algerian Cistus creticus leaf extracts: an in silico and in vitro study. Sci Rep. 2025;15:33887.
[72] Wen C, Li Y, Huang Y, et al. CircSETD3 mediates acquired resistance to gefitinib in non-small lung cancer cells by FXR1/ECT2 pathway. Int J Biochem Cell Biol. 2023;154:106344.
[73] Lin X, Liao Y, Chen X, Long D, Yu T, Shen F. Regulation of oncoprotein 18/stathmin signaling by ERK concerns the resistance to taxol in nonsmall cell lung cancer cells. Cancer Biother Radiopharm. 2016;31:37–43.