Abstract Chronic low-grade inflammation is a key contributor to the pathogenesis and complications of diabetes, leading to issues such as joint pain, skin disorders, periodontal disease, and neuropathy. Therefore, targeting inflammatory pathways has emerged as a promising therapeutic strategy for both the prevention and management of diabetes and its associated comorbidities. Natural products with dual anti-inflammatory and antidiabetic properties have gained significant interest, with Scoparia dulcis showing notable therapeutic potential. This study aimed to evaluate the efficacy of this herbal medicine in alleviating inflammation in diabetic patients using an integrative in silico approach, incorporating network pharmacology, molecular docking, and molecular dynamics simulations. Initial screening of compounds focused on their ability to inhibit key pathological targets implicated in diabetes-related inflammation. Pathway enrichment analysis revealed significant involvement in the AGE-RAGE signaling pathway, lipid metabolism, atherosclerosis pathways, and the hypoxia-inducible factor 1 (HIF-1) pathway. Ten critical molecular targets were identified, with TNF-α being the most prominent. Molecular docking followed by 200 ns molecular dynamics simulations assessed the binding affinity of TNF-α with the top ten selected compounds, revealing strong and stable interactions with essential active site residues. Furthermore, ADMET analysis and density functional theory (DFT) evaluations highlighted the therapeutic potential of these compounds as promising lead candidates for drug development. Existing literature supports the antidiabetic effects of these bioactive compounds, reinforcing the in silico findings. Thus, Scoparia dulcis represents a potential adjunct or alternative therapy for diabetic patients with chronic inflammation, offering a multifaceted approach to disease management. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-025-06862-5. Keywords: Diabetes mellitus, Inflammation, Network Pharmacology, Docking, Molecular dynamics, DFT Subject terms: Computational biology and bioinformatics, Drug discovery Introduction The globally escalating prevalence of diabetes mellitus (DM), a chronic disease linked to substantially elevated all-cause mortality rates and reduced life expectancy, constitutes a major threat to global public health^[38]1. Although DM is traditionally classified into type 1 and type 2 subtypes, the specific contributions of obesity, adipose tissue, gut microbiota, and pancreatic β-cell function remain subjects of ongoing investigation, as evidenced by numerous completed and active clinical trials^[39]2. The increasing recognition of inflammation’s role in the pathogenesis of both Type 1 and Type 2 DM, and their associated metabolic complications, has spurred considerable interest in the development of anti-inflammatory therapeutic strategies aimed at improving disease prevention and management^[40]3. Chronic, low-grade inflammation is increasingly recognized as a unifying pathogenic mechanism underlying obesity, DM, and cardiovascular disease. This is evidenced by the frequent presence of subclinical inflammation in individuals with diabetes, and the strong correlation between circulating inflammatory biomarkers, including those secreted by adipocytes, and both prevalent and incident diabetes, as well as its major complications, particularly cardiovascular disease^[41]4,[42]5. The primary management of T2DM involves pharmacological interventions utilizing a range of drug classes, such as insulin sensitizers (e.g., biguanides, thiazolidinediones), insulin secretagogues (e.g., sulfonylureas, meglitinides), alpha-glucosidase inhibitors, as well as the more recent incretin-based therapies and sodium-glucose co-transporter 2 (SGLT2) inhibitors. Nevertheless, the prolonged use of these medications is associated with numerous adverse effects, underscoring the critical need for exploring natural therapeutic alternatives as a complementary or adjunctive approach^[43]6. Scoparia dulcis (SD) is a traditionally used medicinal herb in southern Nigeria, China, Brazil, and India, possessing diverse pharmacological activities attributed to its complex phytochemical composition, including nitrogen-containing compounds, flavonoids, diterpenoids, triterpenoids, steroids, phenolics, and aliphatics. Traditional applications range from treating respiratory ailments (colds, coughs, sore throats) to addressing gastrointestinal issues, urinary disorders, and various dermatological conditions (eczema, miliaria), reflecting its documented antidiabetic, anti-inflammatory, anti-arthritic, anti-atherosclerotic, anti-hyperlipidemic, hepatoprotective, antioxidant, and anti-urolithiasis properties relevant to metabolic syndrome management^[44]7. Although earlier research has explored the potential antidiabetic and anti-inflammatory properties of both ethanol and aqueous extracts of SD^[45]8,[46]9, its therapeutic potential is primarily attributed to its dual pharmacological actions targeting hyperglycemia and chronic inflammation—two key pathological features of DM and its associated complications. Growing evidence highlights the intricate interplay between metabolic dysregulation and low-grade systemic inflammation in the pathophysiology and clinical manifestation of type 2 diabetes, positioning SD as a promising candidate for integrated disease management. Notably, the combined anti-inflammatory and glucose-lowering properties of SD suggest its potential to minimize the reliance on multiple herbal formulations often prescribed to diabetic patients with significant inflammatory burden. Rather than administering several distinct herbal preparations—each addressing separate aspects of the disease—SD offers a more unified and holistic therapeutic approach capable of simultaneously regulating metabolic imbalance and inflammatory responses. This strategy not only simplifies treatment regimens but may also improve patient compliance and reduce the risk of herb–herb or herb–drug interactions^[47]10–[48]12. Therefore, the present study seeks to elucidate the specific bioactive constituents of SD, along with their molecular targets and signaling pathways, that contribute to its anti-inflammatory efficacy in the context of diabetes management. Materials and methods Software and web-based databases utilized In the course of this study, the following software tools and databases were utilized for analytical purposes: database—GeneCards, DisGeNet, OMIM, PubChem, ShinyGO v0.82, Protein Data Bank, String version 12.0, SwissADME, UniProt, and CHARM-GUI. Software—Cytoscape 3.8.2, Discovery Studio (DS) version 21.1, GROMACS 2024.1, Gaussian 16, and Visual Molecular Dynamics v1.9.3. Target prediction of compounds Bioactive compounds from SD were compiled from a review of relevant literature^[49]7,[50]13. Drug-likeness, assessed using SWISSADME^[51]14, with high gastrointestinal absorption and a minimum bioactivity score of 0.18, was applied as a filtering criterion prior to further analysis. Predicted target proteins for the selected compounds, identified using the SwissTargetPrediction tool ([52]http://swisstargetprediction.ch/), were integrated with additional targets compiled from an encyclopaedia of traditional Chinese medicine^[53]15. Protein targets associated with DM and inflammation Genes associated with inflammation and DM were identified from the DisGeNET^[54]16, OMIM^[55]17, and GeneCards^[56]18 databases, using the keywords “inflammation” and “DM”, restricted to Homo sapiens. GeneCards results were filtered for relevance scores ≥ 5.0, and redundant entries were removed to generate a consolidated dataset. Evaluation of protein-protein network By utilizing a Venn diagram to overlap the candidate targets with both diseases-related targets, common targets were identified. To identify SD components most relevant to the combined pathologies of inflammation and DM, a compound-target-disease network was constructed and analyzed using Cytoscape v3.8.2, integrating selected compounds, shared targets, and disease-associated targets^[57]19. These shared proteins were then input into the STRING database ([58]https://string-db.org/, version 12.0) to construct a protein-protein interaction (PPI) network, restricting the assessment to “Homo sapiens”. The analysis employed a full-string network type, with a high confidence score threshold of 0.700 and a medium false discovery rate (FDR) stringency of 5%. Protein-protein interaction (PPI) data, imported into Cytoscape v3.8.2 from a TSV file, underwent topological analysis using the CytoNCA plugin. Network properties were assessed via node degree, betweenness centrality (BC), and closeness centrality (CC). Nodes with high BC scores were identified as key mediators of network communication, while those with high CC scores indicated central network locations. Hence, 20 compounds and 10 key targets were subsequently predicted based on degree, selecting nodes exceeding median thresholds for BC and CC^[59]20. GO enrichment and KEGG pathway analysis Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed to characterize the biological roles and functional relevance of the top SD targets associated with inflammation and DM. Using ShinyGO v0.82 ([60]https://bioinformatics.sdstate.edu/go/), GO terms (cellular component, biological process, molecular function) and KEGG pathways were identified. The top 20 enriched GO terms and KEGG pathways, ranked by -log[10] (FDR), are presented as dot bubble and bar plots^[61]21. Protein and ligand preparation The crystal structures of the tumor necrosis factor (TNF-α)^[62]22, interleukin-1β (IL1β)^[63]23, AKT Serine/Threonine Kinase 1 (AKT1)^[64]24, D-glyceraldehyde-3-phosphate dehydrogenase (GAPDH)^[65]25, toll-like receptor 4 (TLR4)^[66]26, signal transducer and activator of transcription 3 (STAT3)^[67]27, tumor protein P53 (TP53)^[68]28, epidermal growth factor receptor (EGFR)^[69]29, proto-oncogene tyrosine-protein kinase (SRC)^[70]30, mitogen-activated protein kinase 3 (MAPK3)^[71]31 proteins were gained from the Protein Data Bank (PDB) with the following PDB IDs: 2AZ5, 5R85, 7NH5, 1U8F, 3FXI, 6NJS, 3ZME, 1M17, 2BDF, and 3FHR. These structures were imported and pre-processed using AutoDock Tools 1.5.6 by removing water molecules and adding hydrogen atoms. Proteins were then subjected to ionization and protonation calculations, followed by energy minimization, optimized for molecular docking. Docking grids were constructed for IL1β, encompassing key residues (TYR24, LEU26, VAL41, LYS63, LEU80, LEU82, LEU110, VAL132, PHE133, PHE146, MET148)^[72]23, and for TLR4, targeting critical residues (PHE440, LEU444, PHE463, LYS388, GLN436)^[73]26. For all remaining proteins, docking grids (0.375 Å spacing) were centered on the native ligand coordinates within the respective crystal structures. Utilizing AutoGrid4, precomputed grid maps were created to calculate the binding energies between specific ligand atom types and the receptor target. Additionally, the structures of the active metabolites were subjected to geometry optimization using density functional theory (DFT) under the B3LYP/6-31G(d, p) level in the gas phase. This optimization was performed using Gaussian 16 software (Gaussian Inc.; Wallingford, CT, USA). The energy-minimized metabolite structures were subsequently employed in the molecular docking studies. Molecular docking Molecular docking is a computational method employed to analyze the interactions between a protein and small molecules based on their geometric compatibility and binding scores. In this study, molecular docking was performed using the AutoDock 4.2 software^[74]32 to calculate the binding affinity between the receptor and the ligand. The docking results were visualized using Discovery Studio 2021, offering a detailed representation of the binding interactions, including hydrogen bonding, hydrophobic interactions, and the participation of specific amino acid residues in the binding event. Molecular dynamics simulations The consistency and dynamics of protein-ligand interactions were evaluated using GROMACS 2024.1 software^[75]33 to determine binding modes^[76]34, with simulations based on initial conformations from molecular docking studies. Protein structures were optimized using the CHARMM-GUI platform^[77]35, and ligands were fully hydrogenated in Discovery Studio Visualizer before generating topology files via the SwissParam web server ([78]http://www.swissparam.ch)^[79]36. The CHARMM36 force field was used to generate protein topologies with the GROMACS pdb2gmx module and was applied throughout all subsequent simulation stages^[80]37. Apo and protein-ligand complex systems were solvated in a cubic box with the TIP3P three-site water model^[81]38,[82]39, maintaining a 10 Å buffer from the box edges. Charge neutralization was achieved through the addition of appropriate counterions^[83]40, and a physiological NaCl concentration of 0.15 M^[84]41 was employed to mimic the cellular environment. After solvation and neutralization, the system’s geometry was optimized through energy minimization using the steepest descent algorithm for 50,000 steps to eliminate steric clashes and ensure structural integrity before MD simulation^[85]42. All simulations employed a constant Number of particles, Volume, and Temperature (NVT) ensemble to maintain a reference temperature of 300 K. Subsequent equilibration at a reference pressure of 1 atm was achieved using a constant Number of particles, Pressure, and Temperature (NPT) ensemble; both NVT and NPT equilibration phases lasted for 100 ps^[86]43. The primary simulation was executed for 200 ns, with trajectories recorded at intervals of 0.01 ns. The outcomes of MD were subsequently assessed to calculate the root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), solvent accessible surface area (SASA) as well as to assess connection between the ligands and main residues through the distribution of hydrogen bonds formation, visualized using VMD software^[87]44. Hydrogen bonds were characterized according to the criterion whereby the donor–hydrogen–acceptor (D–H⋯A) angle exceeds 120°, and the donor–acceptor distance remains within a specified cutoff^[88]45,[89]46. MM-PBSA binding energy The binding-free energy was estimated through the gmx_MMPBSA package, utilizing a single trajectory generated by GROMACS under the CHARMM36 force field framework^[90]47. Free energy calculations are enabled using the MM-PBSA over the last 50 ns, with each nanosecond corresponding to a distinct frame, as described by the following equation^[91]48: graphic file with name d33e550.gif 1 Although MM-GBSA offers greater computational efficiency, MM-PBSA, with its more rigorous theoretical foundation, was deemed more suitable for determining the precise binding free energies. Consequently, MM-PBSA calculations were performed using MD trajectories for each protein-ligand complex^[92]48,[93]49. Pharmacokinetic properties, ADMET characteristics, and toxicity risk assessment The pharmacokinetic characteristics, including absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiles of the natural products, were evaluated using SwissADME^[94]14 and ProTox 3.0^[95]50. ADME or toxicity serves as a critical predictive framework for assessing drug-likeness, chemical properties, lead likeness, and the toxicity potential of novel drugs and phytochemicals, and a comprehensive understanding of these attributes is essential for guiding the selection and optimization of compounds during the early stages of drug development^[96]51,[97]52. By incorporating in silico assessments, it becomes possible to identify promising lead candidates while mitigating the risks associated with toxicity and adverse effects. This methodology not only facilitates a more efficient drug discovery process but also improves the safety profiles of compounds prior to clinical trials. The utilization of computational tools in this context underscores the growing importance of bioinformatics in advancing pharmaceutical sciences and ensuring that only the most viable candidates progress through development pipelines. Conceptual DFT analysis DFT is an essential tool for elucidating molecular and atomic structures through the analysis of molecular orbital energies, providing important insights into the structure-activity relationships of molecules^[98]53,[99]54. Geometry optimizations and frequency calculations for the molecular structures of the compounds were performed using DFT at the B3LYP/6-31G(d, p) level in the gas phase, employing the Gaussian 16 software suite. The absence of imaginary frequencies confirmed that all analyzed compounds corresponded to stable points on the potential energy surface. Additionally, the energy levels of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO), along with reactivity parameters derived from quantum chemistry, were calculated utilizing the following five discrete equations^[100]55: graphic file with name d33e603.gif 2 graphic file with name d33e609.gif 3 graphic file with name d33e615.gif 4 graphic file with name d33e621.gif 5 graphic file with name d33e628.gif 6 Results and discussion PPI network analysis A comprehensive in silico analysis was undertaken to investigate the therapeutic potential of SD compounds in the context of DM and inflammation. Pharmacokinetic profiling of 123 SD compounds, using SwissADME (Table [101]S1), predicted favorable oral bioavailability (≥ 0.55) for 70 compounds, indicating significant intestinal absorption. Subsequent target prediction, integrating data from SwissTargetPrediction and a comprehensive traditional Chinese medicine database, identified a total of 770 unique molecular targets. Comparative analysis, using a Venn diagram (Fig. [102]1A), revealed 122 targets shared between this dataset and consolidated gene sets for DM (2709 genes) and inflammation (895 genes), obtained from GeneCards, OMIM, and DisGeNET databases. Network analysis of these shared targets, employing degree, BC and CC metrics, prioritized 20 key SD components as potential bioactive constituents (Table [103]1; Fig. [104]1B). Further analysis constructed a PPI network encompassing these 122 targets, revealing a complex network architecture comprising 121 nodes (representing target proteins) and 2283 edges (representing protein-protein interactions), with an average node degree of 37.74, as illustrated in Fig. [105]1C. Analysis of this network identified ten key hub proteins—TNF-α, IL1B, AKT1, GAPDH, TLR4, STAT3, TP53, EGFR, SRC, and MAPK3—as central nodes exhibiting the highest degree values (Table [106]2; Fig. [107]1D), highlighting their potential importance in mediating the anti-inflammatory and antidiabetic effects of SD. Fig. 6. [108]Fig. 6 [109]Open in a new tab Backbone atom RMSD of TNF-α in apoprotein form (T0) and in complexes with 10 ligands. (T1). Kaempferol. (T2). Luteolin. (T3). Quercetin B. (T4). Scutellarein. (T5). Cirsimaritin. (T6). Hispidulin. (T7). Acerosin. (T8). Apigenin. (T9). Acacetin. (T10). Morin. (T11). Control ligand. Table 1. Topological analysis of the core active compounds in the compounds-targets network. NO. Degree Betweenness centrality Closeness centrality Name 56 152 0.025 0.389 Kaempferol 40 152 0.025 0.389 Luteolin 55 151 0.025 0.389 Quercetin 41 151 0.024 0.389 Scutellarein 46 145 0.026 0.387 Cirsimaritin 42 145 0.026 0.387 Hispidulin 25 139 0.021 0.384 Acerosin 31 133 0.019 0.382 Apigenin 30 127 0.018 0.380 Acacetin 50 116 0.014 0.377 Morin 65 111 0.048 0.371 Dulcidiol 66 110 0.032 0.370 Dulcinodal 74 109 0.065 0.370 Scopadulcic acid A 61 109 0.046 0.369 Dulcinodal-13-one 59 109 0.051 0.370 Scopadulcic acid B 75 108 0.047 0.374 4-epi-7α-hydroxydulcinodal-13-one 73 108 0.037 0.370 Neo-dulcinol 62 108 0.038 0.369 Scopadulciol 70 107 0.061 0.370 7α-hydroxyscopadiol 16 105 0.101 0.371 Procaine [110]Open in a new tab Table 2. Key characteristics of the top ten proteins within the PPI-network. Name Degree Betweenness centrality Closeness centrality TNF-α 103 0.073 0.876 IL1B 95 0.048 0.828 AKT1 89 0.027 0.795 GAPDH 87 0.025 0.779 TLR4 85 0.029 0.774 STAT3 83 0.019 0.755 TP53 83 0.032 0.755 EGFR 83 0.027 0.764 SRC 82 0.037 0.759 MAPK3 79 0.022 0.745 [111]Open in a new tab Fig. 1. [112]Fig. 1 [113]Open in a new tab Comprehensive network pharmacology investigation of diabetes and inflammation-related targets linked to SD. (A) Venn diagram illustrating overlapping target genes among SD, diabetes, and inflammation. (B) Compound–target interaction network showing the relationships between active compounds (blue diamonds) and intersecting target genes (pink rectangles). (C) PPI network of shared targets constructed using the STRING database. (D) Top 10 core target genes identified through topological analysis. The circular nodes indicate targets that are associated with compounds, with darker colors reflecting a stronger correlation. GO enrichment and KEGG pathway analysis GO enrichment analysis, employing ShinyGO v0.82, was performed on the top 50 SD-associated genes implicated in diabetes and inflammation. The top 20 enriched terms were analyzed across three GO domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF), as depicted in Fig. [114]2. Significant BP enrichment indicated activation of inflammatory and immune response pathways (e.g., responses to chemical, biotic, and endogenous stimuli), consistent with the known functions of TNF-α, IL1β, and TLR4; regulation of apoptotic processes, suggesting a role for TP53; and metabolic processes (responses to organic substances, lipids, and nitrogen compounds), implicating AKT1. CC analysis revealed enrichment in plasma membrane, cytoplasmic, and nuclear components, consistent with the known subcellular localization of TNF-α, IL1β, TLR4, EGFR, AKT1, GAPDH, SRC, MAPK3, STAT3, and TP53. Finally, MF analysis showed significant enrichment of protein kinase activities (serine/threonine and tyrosine kinases), DNA-binding transcription factor activity, and enzyme binding, further supporting the involvement of the aforementioned proteins in key regulatory pathways. Fig. 2. [115]Fig. 2 [116]Open in a new tab Gene Ontology (GO) enrichment analysis of target genes is summarized across three domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). Bubble color intensity, ranging from red (most significant) to blue (least significant), represents -log₁₀(FDR) values. Bubble size reflects the number of genes associated with each pathway. The top 20 pathways identified through KEGG signal pathway enrichment analysis are illustrated in Fig. [117]3^[118]56–[119]59, with the AGE-RAGE signaling pathway displaying significant fold enrichment and a noteworthy false discovery rate, underscoring its critical significance. This pathway is intricately linked to chronic inflammation and diabetes, as elevated levels of advanced glycation end-products can initiate inflammatory responses, contributing to insulin resistance and metabolic dysfunction characteristic of type 2 diabetes. Furthermore, cytokine and inflammatory pathways associated with lipid metabolism and atherosclerosis highlight the impact of inflammatory mediators on these processes, where chronic inflammation, prevalent in obesity and diabetes, disrupts lipid metabolism and exacerbates cardiovascular risks. The hypoxia-inducible factor 1 (HIF-1) signaling pathway also underscores the interplay between hypoxia and inflammation within metabolic pathways, demonstrating that hypoxic conditions can intensify adipose tissue inflammation, impairing insulin signaling and elevating blood glucose levels. Additionally, various pathways, especially those related to pancreatic cancer, elucidate the connection between cancer biology and metabolic diseases, with chronic inflammation serving as a common factor that promotes cancer progression and insulin resistance. Lastly, inferred pathways related to neuroinflammatory mechanisms may influence both metabolic regulation and cancer pathways, suggesting a comprehensive relationship between neurological inflammation and diabetes, thereby necessitating a nuanced understanding of these interrelated health challenges. Fig. 8. [120]Fig. 8 [121]Open in a new tab 3D FEL plots of TNF-α in apoprotein form (T0) and in complexes with 10 ligands. (T1). Kaempferol. (T2). Luteolin. (T3). Quercetin B. (T4). Scutellarein. (T5). Cirsimaritin. (T6). Hispidulin. (T7). Acerosin. (T8). Apigenin. (T9). Acacetin. (T10). Morin. (T11). Control ligand. Fig. 3. [122]Fig. 3 [123]Open in a new tab Dot plot of KEGG enrichment analysis of target genes. The bubble and bar color range from red to blue, representing the decreasing − log10(FDR) value of the pathway, and the size of the circle and the length of the bar indicate the number of target genes associated with each pathway. Docking protocol validation The generated grid was validated through a redocking procedure, wherein the ligand originally bound in the crystal structure was repositioned employing the same grid parameters. Following this, the resulting poses were superimposed, and RMSD was calculated for comparison. The redocking process was successfully performed, shown in Fig. [124]S1. Specifically, redocking of the co-crystallized ligand from TNF-α, AKT1, GAPDH, STAT3, TP53, EGFR, SRC, and MAPK3, yielded a docking score of – 6.056, – 15.609, – 10.552, – 7.26, – 7.362, – 6.947, – 8.108, – 11.503 kcal/mol, in which the superimposition of the docked and co-crystallized ligands resulted in an RMSD of 1.87, 1.45, 1.19, 1.10, 0.72, 0.72, 1.81 and 0.47 Å, which falls within the acceptable threshold of < 2.0 Å. This confirms that the grid generation effectively targeted the inhibitor binding pocket and validates the accuracy of the docking approach used. Molecular docking AutoDock Vina was employed to perform molecular docking simulations aimed at investigating the binding interactions between the top 20 high-degree compounds identified in Scoparia dulcis and the 10 primary target proteins. All core targets exhibited binding affinities below – 5 kcal/mol, indicative of strong ligand–protein interactions, with the exception of procaine (Fig. [125]4). Considering the central role of TNF-α in inflammation and diabetes-related signaling pathways among the identified targets, subsequent analyses prioritized the binding interactions between TNF-α and the top 10 selected compounds. Representative docking poses showing significant TNF-α–ligand interactions were visualized using Discovery Studio, with both three-dimensional and two-dimensional interaction profiles presented in Fig. [126]5 and Fig. S2, respectively. For further statistical analysis of the docking scores and error assessment in molecular docking, the exhaustiveness parameter was varied while maintaining the control ligand constant. The docking experiments, performed in duplicate (n = 2), yielded results with the standard error of the mean (SEM), as detailed in Table S2^[127]53,[128]60. All compounds and control ligand formed pi–pi stacking interactions with the TYR59 residue of TNF-α, while several also engaged in hydrogen bonding with residues such as SER60 and LEU120 or hydrophobic interactions with LEU57, TYR151, and TYR119, as summarized in Table [129]3. These residues are located within TNF-α’s binding hotspot^[130]22, highlighting their functional importance. To gain deeper insights into the binding stability and affinity of these complexes, molecular dynamics simulations were subsequently performed. Fig. 9. [131]Fig. 9 [132]Open in a new tab Electron density maps of HOMO and LUMO of ten selected compounds and control ligand. Fig. 5. [133]Fig. 5 [134]Open in a new tab 3D interactions between top 10 selected compounds and protein TNF-α. (T1). Kaempferol. (T2). Luteolin. (T3). Quercetin B. (T4). Scutellarein. (T5). Cirsimaritin. (T6). Hispidulin. (T7). Acerosin. (T8). Apigenin. (T9). Acacetin. (T10). Morin. (T11). Control ligand. Table 3. Docking results of the ligands with their interactions with TNF-α. Compound Interaction residue Hydrogen bond Other bonds Kaempferol LEU120 TYR59, TYR151 Luteolin LEU120 TYR59 Quercetin SER60, LEU120 TYR59, TYR151, HSD15 Scutellarein TYR59, LEU120 Cirsimaritin LEU57, TYR59, LEU120, GLY121 Hispidulin SER60, LEU120 HSD15, TYR59, TYR151 Acerosin TYR119, TYR151 TYR59, LEU120 Apigenin TYR151 TYR59, LEU120 Acacetin HSD15, TYR59 Morin SER60 HSD15, TYR59, TYR151 307* LEU57, TYR119, LEU120 [135]Open in a new tab *Control ligand. Fig. 4. [136]Fig. 4 [137]Open in a new tab Computational screening via molecular docking of the core targets and bioactive compounds associated with SD.. Molecular dynamics simulations MD simulations were conducted over a timescale of 200 ns on ten ligand-TNF-α complexes, in addition to the apoprotein and control ligand, to assess conformational changes and binding stability. RMSD analysis revealed variable degrees of structural stability across the complexes, as illustrated in Fig. [138]6. The apoprotein exhibited an average RMSD of 3.46 Å (Table S5), characterized by two distinct stable phases observed between 10 and 125 ns and 160–200 ns. In comparison, TNF-α in complexes T1, T8, T9, T10, and T11 achieved equilibrium early in the simulation and maintained stable average RMSD values of 2.76 Å, 2.97 Å, 2.60 Å, 2.87 Å, and 2.87 Å, respectively, throughout the simulation period. Complexes T4 and T7 demonstrated transient RMSD peaks of approximately 4 Å, occurring at 50 ns for the scutellarein-bound complex and 30 ns for the acerosin-bound complex, respectively, followed by stabilization. The T5 complex showed an initial spike in RMSD to 4 Å, which rapidly decreased to 2 Å by 40 ns and remained stable thereafter. In the T2 complex, the protein RMSD fluctuated within a 2 Å amplitude during the first 50 ns, then stabilized across three distinct phases: 2.5 Å from 50 to 125 ns, 3.5 Å from 125 to 175 ns, and 2.5 Å from 175 to 200 ns. Similarly, the T3 complex exhibited RMSD fluctuations within a 2 Å range, with two clear stability phases spanning 25–140 ns and 150–200 ns. Notably, the T6 complex displayed the highest degree of fluctuation, with RMSD reaching a maximum of 6 Å. Three short-lived stable phases were observed: 3.5 Å from 0 to 40 ns, 4 Å from 110 to 140 ns, and 6 Å during the final 25 ns. These phases were interspersed with high-amplitude fluctuations (~ 2 Å), indicating significant structural variability and lower overall stability in this complex. With respect to the RMSD of ligands in all 11 protein-ligand complexes, the majority demonstrated average values below 1 Å, with the exception of quercetin and the control ligand, which exhibited RMSD values of 1.26 Å and 4.43 Å, respectively. Interestingly, several ligands displayed a higher amplitude of RMSD, consistently approaching approximately 1 Å for the majority of the simulations, apart from the control ligand, which exhibited the highest amplitude of RMSD at around 3.5 Å. Furthermore, the RMSD profiles for kaempferol and quercetin in their respective complexes with the protein distinctly illustrated various stages, wherein certain stages exhibited fluctuations with very low amplitudes, remaining below 0.3 Å. Additionally, RMSF plot of the TNF-α backbone atom and its complexes, which displayed similar fluctuation patterns, was utilized to characterize the spatial variations of the ligand molecules within the protein structure. As shown in Fig. [139]7A, key residues at the active site of TNF-α, including TYR59, TYR119, GLY121, and TYR151, maintained consistent flexibility, as indicated by RMSF values remaining below 2 Å throughout the entire simulation. The ten docked complexes, along with the apoprotein, as shown in Fig. [140]7B, exhibited Rg values ranging from 16.25 to 17.47 Å with minimal fluctuations, suggesting structural compactness and the absence of major conformational changes throughout the simulation. In contrast, the TNF-α –hispidulin complex demonstrated the most notable variation, with Rg values fluctuating more frequently between 15.9 Å and 16.7 Å during the 50 to 200 ns interval. Across all 11 protein–ligand complexes and the apoprotein, SASA profiles demonstrated consistent behavior during the course of the simulation, fluctuating around an average of 92 nm^2 and ranging from 80.13 to 104.29 nm^2, as shown in Fig. [141]7C. The consistently low and stable SASA values indicate effective encapsulation of the ligands within the protein binding pocket, resulting in reduced solvent exposure. This encapsulation is indicative of strong and stable binding interactions, contributing to enhanced complex stability and binding affinity. Fig. 7. [142]Fig. 7 [143]Open in a new tab RMSF (A), Rg (B), SASA (C) values for the complexes involving the TNF-α protein with ligands. Hydrogen bond occupancy analysis, as presented in Table S6, revealed notable differences in binding affinity among the ligands. All compounds demonstrated strong interactions with TNF-α, characterized by hydrogen bond occupancy values exceeding 70% with key residues^[144]45. All ligands consistently engaged with the TYR59 residue, functioning as both hydrogen bond donors and acceptors—an interaction not captured in the molecular docking results, where TYR59 appeared solely as a hydrophobic contact. This discrepancy highlights a key limitation of molecular docking, which fails to fully account for the dynamic nature of ligand–receptor interactions, more comprehensively characterized through molecular dynamics simulations^[145]61. Quercetin exhibited the highest hydrogen bond occupancy, reaching 259.72%, and also formed unique interactions with TYR151 (94.36%) and GLY121 (87.46%). Kaempferol, cirsimaritin, and acerosin engaged with three key residues—TYR59, TYR119, and TYR151—with each interaction showing occupancy above 97%. In contrast, luteolin formed hydrogen bonds with TYR59 (140.06%) and TYR119 (76.88%), whereas scutellarein interacted with TYR59 (227.22%) and TYR151 (93.77%). Similarly to luteolin, the control ligand exhibited hydrogen bonds with TYR59 (174.23%) and TYR119 (154.89%), representing the highest percentages in comparison to the other compounds. Morin demonstrated the most extensive binding profile, forming hydrogen bonds with four residues: LEU57 (74.43%), TYR59 (258.58%), TYR119 (140.57%), and TYR151 (386.58%), the highest occupancy observed for TYR151 among all ligands. Meanwhile, apigenin, acacetin, and hispidulin interacted exclusively with TYR59, with occupancies ranging from 107.94 to 192.70%. This study employed PCA^[146]62, a well-established dimensionality reduction technique, to investigate the functional dynamics of the TNF-α -ligand complexes, focusing on dominant conformational changes and their relationship to binding stability. PCA of complexes revealed distinct patterns in the distribution of principal components (PC1 and PC2). Complexes T1, T2, T3, T4, T6, T8, T10, and T11 displayed significantly tighter clustering of conformations in the PC1-PC2 space compared to the apoprotein (T0), strongly suggesting the presence of distinct and stable conformers in these complexes, as shown in Fig. S3. In contrast, complexes T7 and T8 exhibited less tightly clustered conformations, while complexes T5 and T9 demonstrated a more dispersed and less dense distribution of conformations, indicating that these complexes underwent greater conformational changes throughout the simulation and may therefore exhibit lower stability compared to the others. Further analysis of the conformational dynamics was conducted using FEL, generated from the first two principal components (PC1 and PC2). These FELs, depicted as three-dimensional surfaces where color intensity corresponds to Gibbs free energy (dark blue-purple: 0–2 kJ/mol; dark red: up to 22 kJ/mol), provided a comprehensive visualization of the range of conformational freedom and stability across all 11 complexes (Fig. [147]8). The apoprotein (T0) displayed a deep and expansive energy minimum, indicative of limited conformational flexibility and substantial structural stability throughout the simulation. Complexes T1, T4, T7, and T10 displayed relatively smaller low-energy regions compared to T0, indicating that while these systems achieved stable equilibrium states, their conformational stability was slightly lower than that of T0. In contrast, complexes T5, T8, and T9 showed the most restricted energy basins, implying reduced conformational stability during the simulation. Meanwhile, complexes T2, T3, T6, and T11 exhibited multiple distinct minima separated by high-energy barriers, indicating the presence of several stable conformational states; however, their overall equilibrium stability appeared to be lower than that of T0. MM-PBSA analysis for binding free energy calculation MM-PBSA binding free energy calculations were utilized to quantitatively evaluate the interaction strength between the target protein TNF-α and the top ten selected ligands, as well as the control ligand, as detailed in Table [148]4. These calculations were based on 200 ns molecular dynamics simulations. Among the ligands, the control ligand demonstrated the most favorable binding energy of – 18.98 kcal/mol, followed by acerosin with – 16.78 kcal/mol, luteolin with – 14.95 kcal/mol, and hispidulin with – 14.31 kcal/mol. Apigenin showed the least favorable binding energy (– 7.46 kcal/mol), while the remaining ligands displayed binding free energies ranging from – 12.77 to – 9.11 kcal/mol. The ΔG[bind] values provide a reliable metric for identifying the strongest receptor antagonists. The primary objective of this study was to evaluate the interactions between selected ligands and TNF-α. An integrated analysis encompassing complex stability, hydrogen bond occupancy with key residues, PCA, FEL and MM-PBSA binding free energy confirmed that all ligands possess strong inhibitory potential against TNF-α. Table 4. The calculation of binding free energy results of 10 top selected ligands with TNF-α. Complex with ligand Kcal/mol ΔE[vdW] ΔE[elec] ΔG[PB] ΔG[SA] ΔG[gas] ΔG[sol] ΔG[bind] Kaempferol – 16.30 ± 3.48 – 5.06 ± 4.50 14.24 ± 6.11 – 1.99 ± 0.30 – 21.36 ± 6.50 12.25 ± 5.89 – 9.11 ± 3.11 Luteolin – 15.06 ± 4.65 – 23.52 ± 16.99 25.64 ± 10.50 – 2.01 ± 0.26 – 38.58 ± 15.62 23.63 ± 10.47 – 14.95 ± 5.99 Quercetin – 19.75 ± 1.91 – 4.71 ± 4.14 14.34 ± 4.77 – 2.16 ± 0.10 – 24.46 ± 4.90 12.18 ± 4.72 – 12.28 ± 2.34 Scutellarein – 17.73 ± 3.84 – 3.33 ± 4.10 11.47 ± 3.33 – 1.99 ± 0.29 – 21.07 ± 5.84 9.48 ± 3.14 – 11.59 ± 3.64 Cirsimaritin – 18.46 ± 3.33 – 4.60 ± 5.11 12.50 ± 4.31 – 2.21 ± 0.30 – 23.06 ± 5.73 10.29 ± 4.14 – 12.77 ± 2.97 Hispidulin – 22.29 ± 2.60 – 2.26 ± 3.54 12.67 ± 3.21 – 2.43 ± 0.13 – 24.55 ± 3.49 10.24 ± 3.20 – 14.31 ± 2.32 Acerosin – 28.47 ± 3.17 – 2.97 ± 3.72 17.59 ± 3.95 – 2.93 ± 0.15 – 31.44 ± 5.21 14.66 ± 3.88 – 16.78 ± 3.38 Apigenin – 12.77 ± 6.55 – 7.03 ± 7.86 13.96 ± 8.44 – 1.62 ± 0.72 – 19.79 ± 12.37 12.34 ± 7.89 – 7.46 ± 5.63 Acacetin – 17.10 ± 8.35 – 10.74 ± 10.10 19.23 ± 10.82 – 2.01 ± 0.85 – 27.84 ± 14.76 17.22 ± 10.20 – 10.62 ± 5.47 Morin – 19.09 ± 1.82 – 10.73 ± 4.43 20.17 ± 3.21 – 2.31 ± 0.11 – 29.83 ± 4.09 17.86 ± 3.17 – 11.97 ± 2.61 Control ligand – 27.80 ± 4.79 – 3.44 ± 3.15 15.41 ± 3.72 – 3.14 ± 0.43 – 31.25 ± 6.65 12.26 ± 3.46 – 18.98 ± 4.47 [149]Open in a new tab Kaempferol has been shown to exert inhibitory effects on key pro-inflammatory mediators, including TNF-α, IL6, IL1β, and nuclear factor-kappa B (NF-κB), thereby highlighting the therapeutic potential of anti-inflammatory strategies in the management of metabolic disorders and diabetes^[150]63. Similarly, luteolin significantly mitigates TNF-α-induced inflammatory responses in endothelial cells by modulating the Akt/MAPK/NF-κB signaling pathway^[151]64. Quercetin downregulates TNF-α expression, suggesting its clinical relevance in enhancing host immune defense against various infections. Notably, its regulatory effects are primarily observed in CD4^+ T lymphocytes and CD14^+ monocytes^[152]65. In a murine model of lipopolysaccharide (LPS)-induced acute lung injury (ALI), scutellarein significantly decreased IL-6, and TNF-α levels in bronchoalveolar lavage fluid, while alleviating pulmonary damage and inhibiting neutrophil infiltration^[153]66. Cirsimaritin was found to suppress the production of IL6, TNF-α, and nitric oxide (NO) in LPS-stimulated RAW264.7 macrophages in a concentration-dependent manner^[154]67. Hispidulin demonstrated dose-dependent anti-neuroinflammatory effects in microglial cells by reducing the levels of NO, reactive oxygen species (ROS), inducible nitric oxide synthase (iNOS), TNF-α, IL1β, IL6, and prostaglandin E2 (PGE2)^[155]68. Apigenin significantly modulated TNF-α-induced MUC5AC mucin production and gene expression in airway epithelial cells^[156]69. Additionally, acacetin inhibited microglial activation in the cerebral cortex following subarachnoid hemorrhage and reduced the expression of TNF-α and IL6^[157]70. Morin was also shown to inhibit the gene expression of TNF-α, IL6, and IL1β in a dose-dependent manner in LPS-stimulated bovine mammary epithelial cells^[158]71. Collectively, these studies demonstrate the anti-inflammatory properties of the key compounds identified in SD, with the exception of acerosin, for which no related literature was found concerning its inhibition of TNF-α. However, MD simulations indicate that acerosin may possess significant potential to inhibit TNF-α. Elevated concentrations of pro-inflammatory cytokines (e.g., TNF-α and IL-6) and adipokines (e.g., leptin and resistin) play a critical role in aggravating insulin resistance, thereby reinforcing a pathological cycle of metabolic and immune dysregulation. This dynamic interaction contributes to the chronic low-grade inflammatory state that underlies the pathogenesis of type 2 diabetes mellitus (T2DM), ultimately disrupting insulin signaling pathways and impairing glucose homeostasis^[159]72. The TNF-α/TNF receptor axis, in particular, has emerged as a promising therapeutic target beyond traditional glycemic and blood pressure control, with potential implications for mitigating the progression of diabetic nephropathy to end-stage renal disease^[160]73. Mechanistically, TNF-α disrupts insulin receptor substrate (IRS) signaling, thereby diminishing cellular glucose uptake, while elevated free fatty acids associated with obesity further potentiate inflammation and reduce insulin sensitivity^[161]74. Moreover, sustained elevation of inflammatory cytokines such as TNF-α and IL-6 in T2DM contributes to the downregulation of insulin receptor signaling, promoting a chronic inflammatory milieu and intensifying insulin resistance^[162]75. TNF-α negatively influences insulin signaling and is linked to insulin resistance in type 2 diabetes; thus, downregulating TNF-α may increase phosphorylation of insulin receptor substrate 1 and improve insulin responsiveness in hepatic cells^[163]76. Collectively, the present findings emphasize the central role of inflammatory cytokines, particularly TNF-α, as promising molecular targets for therapeutic intervention in diabetes-associated inflammatory sequelae. The convergence of existing experimental data with the in silico analyses conducted herein provides suggestive evidence that SD may elicit beneficial effects in attenuating inflammation within diabetic populations. Nevertheless, it is pertinent to acknowledge inherent limitations in the current in silico investigation, despite the utilization of molecular docking and dynamics simulations to examine SD compound interactions with key proteins. Although TNF-α studies used a co-crystalized ligand showing TNF-α inhibition, docking and dynamics studies are inherently limited in their ability to definitively determine the functional outcome of these interactions. Specifically, while these computational methodologies effectively assess binding affinity and structural stability, they do not conclusively discern whether interactions with proteins beyond TNF-α—such as AKT1, GAPDH, TLR4, STAT3, TP53, EGFR, and SRC—result in agonistic or antagonistic modulation. Due to an incomplete representation of intricate downstream signaling cascades and comprehensive cellular responses within simulation parameters, further experimental validation in vitro or in vivo remains crucial. Such experimental work is essential to comprehensively confirm the functional consequences of these interactions and fully elucidate their potential role in modulating inflammatory and diabetic pathophysiological mechanisms^[164]77,[165]78. Pharmacokinetic properties, ADME characteristics, and toxicity risk assessment Prior to consideration for in vivo studies, potential drug candidates can undergo virtual screening for assessment of their physicochemical, pharmacokinetic, drug-likeness, and medicinal chemistry properties based on established guidelines. An in silico ADMET analysis of ten selected SD compounds revealed generally favorable predicted pharmacokinetic profiles, with all compounds satisfying Lipinski’s Rule of 5 and exhibiting predicted non-substrate behavior for P-glycoprotein, suggesting a high probability of acceptable oral bioavailability, as detailed in Table [166]5. Conversely, the control ligand also satisfied Lipinski’s criteria but was predicted to be a substrate for P-glycoprotein. Moreover, the in silico analysis also predicts that all compounds will interact with CYP1A2, CYP2D6, and CYP3A4 enzymes, suggesting the likelihood of significant metabolism via these isoforms. Given the crucial role that cytochrome P450 (CYP450) superfamily members play in xenobiotic metabolism and drug clearance, such interactions may lead to clinically relevant drug-drug interactions. Specifically, the inhibition or induction of these enzymes by SD compounds could alter the metabolism and elimination of co-administered drugs, potentially resulting in altered drug efficacy or increased risk of adverse effects due to toxicant accumulation^[167]79. Furthermore, these compounds are not predicted to readily permeate the blood-brain barrier, which may limit their utility in treating central nervous system disorders. Table 5. ADME characteristics of selected SD constituents and control ligand. Compound Hydorgen bond acceptors Hydorgen bond donors TPSA BBB permeant P-glycoprotein substrate Lipinski Ghose Veber Egan Muegge CYP1A2 CYP2C19 CYP2C9 CYP2D6 CYP3A4 Kaempferol 6 4 111.13 No No 0 0 0 0 0 Yes No No Yes Yes Luteolin 6 4 111.13 No No 0 0 0 0 0 Yes No No Yes Yes Quercetin 7 5 131.36 No No 0 0 0 0 0 Yes No No Yes Yes Scutellarein 6 4 111.13 No No 0 0 0 0 0 Yes No No Yes Yes Cirsimaritin 6 2 89.13 No No 0 0 0 0 0 Yes No Yes Yes Yes Hispidulin 6 3 100.13 No No 0 0 0 0 0 Yes No No Yes Yes Acerosin 8 3 118.59 No No 0 0 0 0 0 Yes No Yes Yes Yes Apigenin 5 3 90.9 No No 0 0 0 0 0 Yes No No Yes Yes Acacetin 5 2 79.9 No No 0 0 0 0 0 Yes No Yes Yes Yes Morin 7 5 131.36 No No 0 0 0 0 0 Yes No No Yes Yes Control ligand 7 0 41.62 No Yes 1 4 0 1 1 No No No Yes Yes [168]Open in a new tab To model the safety and potential toxicological risks of drugs, each natural compound underwent in silico toxicity evaluations. These evaluations, performed on ten selected constituents of SD, demonstrated a predominantly favorable profile. The majority of the compounds were categorized as toxicity class V, signifying a low level of predicted general toxicity, as shown in Table [169]6. Quercetin and control ligand stands out as belonging to toxicity class III, potentially indicating a slightly higher general toxicity compared to the other compounds. Notably, all compounds are predicted to be inactive concerning hepatotoxicity and neurotoxicity, with the exception of the control ligand, which is predicted to have an effect on neurotoxicity. However, the in silico analyses also predict activity related to nephrotoxicity and respiratory toxicity for all compounds, which raises the need for further scrutiny through targeted studies. Finally, the activity of all compounds has been shown with various activities relating carcinogenicity, immunotoxicity, mutagenicity, and cytotoxicity. In particular, it is important to conduct targeted in vitro and in vivo studies to assess the safety of these compounds, especially given the limited scope of endpoints considered in this analysis and to validate the present computationally predicted toxicity profiles. Table 6. Toxicity profiles of selected SD constituents and control ligand. Compound Toxicity class Hepatotoxicity Neurotoxicity Nephrotoxicity Respiratory- toxicity Cardiotoxicity Carcinogenicity Immunotoxicity Mutagenicity Cytotoxicity Kaempferol V Inactive Inactive Active Active Inactive Inactive Inactive Inactive Inactive Luteolin V Inactive Inactive Active Active Inactive Active Inactive Active Inactive Quercetin III Inactive Inactive Active Active Inactive Active Inactive Active Inactive Scutellarein V Inactive Inactive Active Active Inactive Active Inactive Active Inactive Cirsimaritin V Inactive Inactive Active Active Inactive Inactive Active Inactive Inactive Hispidulin V Inactive Inactive Active Active Active Inactive Inactive Inactive Inactive Acerosin V Inactive Inactive Active Active Active Inactive Active Inactive Inactive Apigenin V Inactive Inactive Active Active Inactive Inactive Inactive Inactive Inactive Acacetin V Inactive Inactive Active Active Active Inactive Inactive Active Inactive Morin V Inactive Inactive Active Active Inactive Inactive Active Inactive Inactive Control ligand III Inactive Active Inactive Active Inactive Inactive Active Inactive Inactive [170]Open in a new tab Conceptual DFT analysis The energy optimization of all ten selected compounds from SD was performed using DFT, and their optimized coordinates, which reflect the geometries of the lowest energy states, are outlined in Table S3. Additionally, the optimization of the selected compounds met the convergence criteria of the Gaussian 16 software package (see Table S4), confirming that these structures are minima on the potential energy surface. The minimum electronic energy value observed for the control ligand was – 50407.48 eV. In comparison, acerosin and cirsimaritin exhibited higher electronic energies, measured at – 35294.91 eV and – 30132.59 eV, respectively, as outlined in Table [171]7. The remaining compounds exhibited energy values ranging from – 30041.62 to − 25948.51 eV, indicating that more negative energy values are associated with a reduced requirement for external energy in ligand–receptor reactivity. Furthermore, these compounds demonstrated dipole moment values ranging from 3.57 to 7.43 Debye, indicating significant potential for dipole-dipole interactions within their molecular structures. Table 7. Quantum reactivity parameters of 10 selected potential compounds and control ligand. Compound Gibbs free energy (eV) Molecular dipole moment (Debye) E[HOMO] (eV) E[LUMO] (eV) Energy gap (eV) Absolute hardness (η) Global softness (σ) Electro negativity (χ) Chemical potential (Pi) Global electrophilicity (ω) Kaempferol – 27994.94 4.66 – 6.09 – 2.15 3.94 1.97 0.25 4.12 – 4.12 4.31 Luteolin – 27995.29 4.99 – 6.28 – 2.20 4.08 2.04 0.25 4.24 – 4.24 4.40 Quercetin – 30041.75 5.69 – 6.09 – 2.20 3.90 1.95 0.26 4.14 – 4.14 4.41 Scutellarein – 27995.32 3.57 – 6.17 – 2.23 3.95 1.97 0.25 4.20 – 4.20 4.47 Cirsimaritin – 30132.59 7.43 – 6.10 – 2.16 3.93 1.97 0.25 4.13 – 4.13 4.34 Hispidulin – 29064.02 3.57 – 6.25 – 2.19 4.05 2.03 0.25 4.22 – 4.22 4.39 Acerosin – 35294.91 5.68 – 5.97 – 2.12 3.85 1.93 0.26 4.04 – 4.04 4.24 Apigenin – 25948.51 3.85 – 6.32 – 2.20 4.12 2.06 0.24 4.26 – 4.26 4.41 Acacetin – 27017.35 5.55 – 6.27 – 2.16 4.11 2.06 0.24 4.21 – 4.21 4.31 Morin – 30041.62 7.21 – 6.01 – 1.92 4.08 2.04 0.24 3.97 – 3.97 3.85 Control ligand – 50407.48 4.96 – 5.66 – 1.90 3.76 1.88 0.27 3.78 – 3.78 3.80 [172]Open in a new tab Molecules with a narrow energy gap exhibit enhanced biological activity, increased chemical reactivity, and decreased stability compared to those with a wider energy gap^[173]53. Accordingly, the band energy gap of phytochemicals and quantum chemical reactivity metrics derived from E[HOMO] and E[LUMO] were calculated and are summarized in Fig. [174]9; Table [175]7. Among these compounds, the control ligand demonstrated the lowest ΔE value of 3.76 eV. This was followed closely by acerosin, quercetin, and cirsimaritin, which exhibited ΔE values of 3.85, 3.90, and 3.93 eV, respectively. The remaining compounds displayed similar ΔE values around 4.0 eV, with a range of 3.94 eV to 4.12 eV. Quercetin has been demonstrated to enhance oral glucose tolerance and improve pancreatic β-cell function in insulin secretion while reducing the release of pro-inflammatory markers such as IL1β, IL6, and TNF-α^[176]80. Cirsimaritin has been shown to reduce elevated serum glucose levels in diabetic rats^[177]81, and kaempferol has been found to increase plasma insulin levels and decrease blood glucose levels in rats with streptozotocin-induced diabetes^[178]63. The diabetes induced by high glucose levels and dexamethasone may be ameliorated by apigenin and luteolin through the enhancement of glucose utilization and glycogen synthesis, along with the suppression of ROS and AGEs production^[179]82. Acacetin has been shown to improve insulin resistance in obese mice and counteract lipotoxicity in pancreatic β-cells by mitigating oxidative stress^[180]83. Additionally, hispidulin stimulates the secretion of glucagon-like peptide-1 and mitigates hyperglycemia in mice with streptozotocin-induced diabetes^[181]84. In an animal model of streptozotocin-induced DM, the administration of morin significantly reduced blood glucose levels while increasing insulin levels^[182]85. In contrast, although acerosin and scutellarein have not been studied in relation to diabetes, their lower ΔE indicates potential bioactivity that merits further exploration. Furthermore, the global electrophilicity (ω) values of these compounds, which range from 3.80 to 4.47 eV, suggest their potential to interact with biological macromolecules such as proteins^[183]46. Conclusion This investigation conducted a comprehensive exploration of the bioactive constituents and molecular mechanisms underlying the therapeutic potential of SD in alleviating inflammation associated with DM. By employing an integrative approach that combined network pharmacology, molecular docking, and molecular dynamics simulations, the study aimed to elucidate the pharmacological foundation of SD’s efficacy. The research identified ten key compounds—kaempferol, luteolin, quercetin, scutellarein, cirsimaritin, hispidulin, acerosin, apigenin, acacetin, and morin—as major bioactive constituents. These compounds were predicted to interact with essential protein targets involved in inflammatory and metabolic pathways, including TNF-α, IL1β, AKT1, GAPDH, TLR4, EGFR, STAT3, TP53, SRC, and MAPK3, highlighting their therapeutic significance in inflammatory diabetes. Among these targets, TNF-α was identified as the most central and functionally relevant node within the interaction network, exhibiting strong binding affinity with the aforementioned compounds through MD analysis. Pathway enrichment analysis further indicated that the active compounds modulate several critical signaling pathways, particularly the AGE-RAGE signaling pathway associated with diabetic complications, lipid metabolism, atherosclerosis, and the HIF-1 signaling pathway. Additionally, the primary compounds were subjected to ADME and and toxicity evaluations, which indicated that they exhibited similar properties and could potentially serve as suitable drug candidates. The HOMO-LUMO analysis identified acerosin as the most biologically active and chemically reactive compound due to its lower ΔE value, suggesting its potential to engage in multiple interactions with biological macromolecules. Collectively, these findings provide a scientific foundation for the continued exploration and clinical application of SD as a complementary or alternative strategy for mitigating chronic inflammation in diabetic patients. Future research should focus on validating the effects of SD on this pathology through rigorous in vitro and in vivo studies to confirm its therapeutic potential. Electronic supplementary material Below is the link to the electronic supplementary material. [184]Supplementary Material 1^ (1.5MB, docx) Author contributions CrediT authorship contribution statementN.-T. Pham: conceptualization, investigation, and writing original draft ; H.-G. Le: validation, writing original draft, and visualization; Plan T.-T.T: validation and visualization; P.V. Luu: validation and visualization; B.-R. Peng: validation and visualization; L.-Y. Chen: validation and visualization; Y.-C. Chang: validation and visualization; K.-H. Lai: conceptualization, writing review & editing, visualization, supervision, and funding acquisition. Funding The grants that supported this work were from the Ministry of Education Taiwan (DP2-TMU-114-C-06), and the National Science and Technology Council of Taiwan (NSTC 111-2320-B-038-040-MY3, 113-2321-B-255-001, 113-2628-B-038-009-MY3, and 114-2326-B-038-002-MY3). Data availability Data is provided within the manuscript or supplementary information files. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References