Integrated network pharmacology reveal new key curcumin-binding targets in triple-negative breast cancer Authors: Xiaohang Qi, Hua Zhang, Shucong Wang, Lu Qiao, Li Ma, Kanghuai Zhang, Yan Wang Published: 2026 ABSTRACT PurposeTriple-negative breast cancer (TNBC) is an aggressive malignancy lacking targeted therapies, leading to limited treatment options. Curcumin, a natural polyphenol from Curcuma longa, exhibits promising anticancer activity, but its precise molecular mechanisms and key targets in TNBC remain largely unclear. This study aimed to clarify curcumin’s antitumor mechanisms and identify its key targets in TNBC.MethodsTo clarify curcumin’s antitumor mechanisms in TNBC, an integrated approach was used, including network pharmacology, molecular docking, 500 ns molecular dynamics simulations, free energy landscape analysis, and MM/PBSA binding free energy calculations.ResultsNetwork pharmacology analysis identified AURKA, BRAF (V600E), and CHEK1 as key genes overlapping between curcumin-associated targets and TNBC-related targets. According to molecular docking, Curcumin displayed favorable binding affinities toward these targets, comparable to reference inhibitors. Stable protein–curcumin complexes with improved conformational stability were shown by molecular dynamics simulations. Strong binding free energies have been identified by MM/PBSA calculations for curcumin–CHEK1 (–209.37 kJ/mol) and curcumin–BRAF (V600E) (–192.87 kJ/mol), which are similar to inhibitors that have been clinically proven.ConclusionsOur integrated analysis suggests that curcumin exerts multitarget inhibitory effects in TNBC by concurrently modulating mitotic regulation (via AURKA), MAPK signaling (via BRAF (V600E)), and DNA damage checkpoints (via CHEK1). Among these, CHEK1 emerged as the most thermodynamically stable and conformationally favorable binding target for curcumin, indicating its potential as a crucial mediator of curcumin’s antitumor activity.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s12672-026-05413-2. Background TNBC is one of the most aggressive subtypes of breast cancer, accounting for approximately 10–20% of all breast cancer cases [16]. TNBC is characterized by the lack of human epidermal growth factor receptor 2 (HER2), progesterone receptor (PR), and estrogen receptor (ER). As a result, chemotherapy is still the standard treatment for TNBC because hormonal and HER2-targeted treatments don’t work [12]. Some TNBC subtypes have been granted permits, such as immune checkpoint inhibitors (pembrolizumab, atezolizumab) for PD-L1-positive tumors, PARP inhibitors (olaparib, talazoparib) for BRCA-mutated cases, and the antibody–drug conjugate sacituzumab govitecan for advanced disease [26]. Meanwhile, new targets like epidermal growth factor receptor (EGFR), phosphoinositide 3-kinase (PI3K)/ protein kinase B (AKT)/ mechanistic/mammalian target of rapamycin (mTOR), androgen receptor (AR), and Notch are being studied but are not yet included in standard treatment [52]. However, chemotherapy is linked to high toxicity and resistance risk, and the majority of patients still lack access to efficient, clearly defined targeted therapies [31]. This therapeutic gap highlights the pressing need for innovative, multitargeted approaches that can alter important signaling pathways implicated in the development of TNBC. Natural products have historically acted as a significant source of anticancer medications owing to their exceptional structural diversity and capacity to interact with many molecular targets [14]. This kind of combination makes them very useful for dealing with the complexity and variety of cancer [29]. Curcuma longa, a member of the ginger family, contains a naturally occurring polyphenolic compound called curcumin in its rhizomes, and underground stems [20]. In a struggle against TNBC, curcumin has demonstrated great promise. Curcumin demonstrates strong, multi-targeted anticancer activity against TNBC through altering several signaling pathways, such as PI3K/Akt, Hedgehog/Gli1, JAK/STAT, mTOR, EGFR, and Wnt/β-catenin [3, 7, 10, 25, 27]. Curcumin’s therapeutic potential is further enhanced by advanced formulations and when combined with doxorubicin and carboplatin, indicating that it is a promising treatment option for TNBC that awaits clinical validation [4, 8, 47]. However, the clinical application of curcumin is hindered by poor oral bioavailability, attributed to poor solubility, and rapid metabolism and systemic clearance causing very low levels in plasma [17, 38]. Curcumin is thought to be safe at normally encountered doses; however, high doses or long-term usage can lead to various possible safety issues [15]. These drawbacks point to the importance of having more insights into the mechanism and molecular targets to better explore the potential efficacy of curcumin. Although these studies contributed to clarify curcumin’s comprehensive anticancer potential, they fail to capture the complex nature of its multi-target mode of action, especially when considering TNBC. Beyond the conventional “one drug–one target” model, network pharmacology explores compound–target–pathway interactions by combining systems biology and computational pharmacology to find the complex, multiple-target mechanisms of natural compounds [49]. It facilitates the development of multi-targeted, synergistic therapies with increased efficacy and safety, speeds up drug discovery, and clarifies mechanisms of action by utilizing big data and computational tools [30]. However, network-based predictions by themselves frequently are insufficient to verifying the molecular validity of these interactions. Though molecular docking and molecular dynamics (MD) simulations are combined, the predictive effectiveness of docking and MD’s capacity to capture flexibility and dynamic stability provide a potent and complementary framework for precisely characterizing ligand–protein interactions [13]. The depth and dependability of structure-based drug discovery are increased by this cooperative method, which also refines binding poses, identifies true binders, and provides comprehensive mechanistic and energetic insights [45]. In this work, we present an integrated systematic and mechanistic profiling of curcumin targeting clinically relevant cancer-specific target kinases AURKA, and CHEK1 in TNBC [5, 37]. These selected targets have shown top topological importance in the protein-protein interaction network and proved its role in cell cycle regulation, oncogenic signaling, DNA damage checkpoint regulation in TNBC respectively [9, 19]. Moreover, a mutation specific analysis was carried out on clinically relevant BRAF (V600E) mutant, which is enriched in TNBC [34], to get the mechanistic view of curcumin targeting a therapeutically relevant TNBC specific oncogenic mutant. Several kinase targets have been associated in cancer but the mechanism of action of curcumin in terms of targeting the kinase is yet to be sufficiently explored from structural and dynamic perspective. Thus, in the present study, we have explored multi-target activity of curcumin with the aid of network pharmacology, long time scale (500 ns) MD simulations, free energy landscape (FEL) calculations and MM/PBSA binding energy calculations. Methods Identification of curcumin and TNBC’s potential targets The SwissTargetPrediction databases were used to predict curcumin’s potential molecular targets (https://www.swisstargetprediction.ch/result.php?job=216152288&organism=Homo_sapiens) [6]. Meanwhile, the keyword “triple-negative breast cancer” was used to get TNBC-associated genes from the GeneCards databases (https://www.genecards.org/Search/Keyword?queryString=triple%20negative%20breast%20cancer&geneCategories=ProteinCoding) [40]. Related genes were screened using GeneCards’ protein coding categories and remove genes do not have uniport ID. We assessed on these datasets on 10 October lastly. The VennDiagram package in R was used to find genes that overlapped between curcumin-related targets and TNBC-associated genes. Pathway enrichment analysis and network construction To visualize the relationship between curcumin, its predicted targets, and TNBC-related genes, a compound–target–disease (C–T–D) network was built using Cytoscape (v3.10.3) [35]. The STRING database (v12.0) was used to construct the Protein–Protein Interaction (PPI) network of overlapping targets with an interaction confidence score > 0.4 [42]. To find hub genes based on topological parameters like degree and betweenness centrality, the generated network was examined using Cytoscape. To identify enriched biological processes and signaling pathways, functional enrichment analyses were carried out using Metascape (https://metascape.org), including Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses [50]. The enrichment parameters were set as Minimum overlap = 3, P-value cutoff = 0.01, and minimum enrichment = 1.5. P-values were adjusted using the Benjamini–Hochberg (BH) method, and terms with adjusted p < 0.05 were considered significant. Selecting reference inhibitors and crucial targets AURKA, BRAF (V600E), and CHEK1 were chosen as key targets among the identified hub targets due to their high degree centrality within the PPI network and biological significance to TNBC. Using data from AURKA (PDB ID: 5L8L), the BRAF V600E mutant (PDB ID: 8C7Y), and CHEK1 (PDB ID: 2YEX), protein crystal structures were acquired from the RCSB Protein Data Bank (PDB) [18]. The BRAF (V600E) mutant was specifically selected to perform mutation-specific docking and MD analyses to investigate curcumin interactions with this oncogenic variant. Positive controls included vemurafenib (BRAF (V600E), alisertib (AURKA), and AZD7762 (CHEK1), which are corresponding reference inhibitors. Curcumin’s 3D structure was obtained from PubChem (CID: 969516), and Chem3D was used to minimize its energy. Docking of molecular structure AutoDock Vina (v1.2.3) with default parameters was used for all molecular docking investigations [11]. Hydrogen atoms were added, water molecules and unnecessary ligands were eliminated, and Gasteiger charges were assigned using AutoDockTools to create protein structures. For each protein, the docking grid box was centered at the following coordinates and dimensions: CHEK1: center (13.8546, 34.7526, 10.2607); box size (25.0 × 25.0 × 25.0 Å). CCNA2: center (19.361, 0.9451, 5.1374); box size (25.02 × 24.71 × 40.68 Å). AURKA: center (8.0567, − 14.938, − 19.4916); box size (47.96 × 59.29 × 50.43 Å). BRAF (V600E): center (− 6.6903, 115.258, 13.0859); box size (53.67 × 48.23 × 49.51 Å). The energy range was set to 4 kcal/mol, exhaustiveness to 10, and num-modes to 10 for all docking runs. Docking poses were ranked according to binding affinity, and the top-ranked conformation was selected for subsequent molecular dynamics simulations. Though no specific re-docking validation was performed using the co-crystallized ligand, the protocol of docking was followed according to the commonly reported AutoDock Vina methodology. The grid box was large enough to contain the whole active site and a flexible exploration of the ligand. The value 10 for the exhaustiveness parameter was a compromise between the cost in computing time and the density in the exploration. The relevance of the chosen pose is illustrated by the subsequent molecular dynamic simulations with long timescale. PyMOL (v2.5) and Discovery Studio Visualizer 4.5 were used to visualize binding conformations and binding affinities (kcal/mol) [23]. Simulations of molecular dynamics (MD) GROMACS (v2020.1) was used to perform 500 ns all-atom MD simulations in order to assess the dynamic stability and binding behavior of the curcumin–protein complexes (GROMACS version 2020.1, https://doi.org/10.5281/zenodo.3685919). Proteins were subjected to the AMBERff19SB force field, while Ligand parameters were generated using the AmberTools23 antechamber module with the AMBER Force Field2 (GAFF2) force field and AM1-BCC atomic charges, water model generated using the TIP3P and using ACPYPE to convert [39, 44]. Each system, inhibitors, curcumin complex, and apo protein, was neutralized with counterions after being solvated in a TIP3P water box with 8 Å buffering and neutralized with counterions (Na⁺/Cl⁻). Energy minimization was used to minimize energy until the maximum force was less than 1000 kJ/ mol·nm (1000steps). Subsequently, each system performed 100 ps NVT and NPT equilibrations at 310.15 K and 1 bar. By using the V-rescale thermostat and Parrinello–Rahman barostat with coupling constants of 0.1 ps and 2.0 ps, respectively. Production simulations were run for 500 ns with a 2 fs time step, and trajectories were saved every 10 ps for analysis. All bonds involving hydrogen atoms were constrained using LINCS. Using common GROMACS tools (including gmx rms, gmx rmsf, gmx gyrate) trajectory analyses were performed to determine the Radius of Gyration (Rg), Root Mean Square Deviation (RMSD), and Root Mean Square Fluctuation (RMSF). Principal component analysis (PCA) was utilized to build the free energy landscape (FEL) from the MD trajectories in order to investigate conformational stability and energy minima. Reaction coordinates were derived from the first two principal components (PC1 and PC2). Binding free energy calculation (MM-PBSA) By using the g_mmpbsa package, the binding free energies of curcumin with targets were estimated using the Molecular Mechanics Poisson–Boltzmann Surface Area (MM-PBSA) method [22]. The total binding energy was expressed as:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ {\mathbf{\Delta\mathrm{G}}}_{{{\mathbf{bind}}}} \, = \,{\mathbf{\Delta\mathrm{E}}}_{{{\mathbf{vdW}}}} \, + \,{\mathbf{\Delta\mathrm{E}}}_{{{\mathbf{ele}}}} \, + \,{\mathbf{\Delta\mathrm{G}}}_{{{\mathbf{polar}}}} \, + \,{\mathbf{\Delta\mathrm{G}}}_{{{\mathbf{nonpolar}}}} $$\end{document} ΔEvdW and ΔEele are the van der Waals and electrostatic interaction energies, and ΔGpolar and ΔGnonpolar correspond to the polar and nonpolar solvation energies. Snapshots were extracted from the equilibrated portion of the trajectories, corresponding to the lowest-energy conformational basin identified by FEL analysis. A total of 100 evenly spaced frames were selected to ensure representative sampling of stable conformations while maintaining computational efficiency, following commonly used MM/PBSA practices. The sum of the solvation free energy components and the molecular mechanics energy (electrostatic + van der Waals) is represented by each term in the calculation of the total binding free energy (ΔGbind). To confirm binding stability and thermodynamic favorability, the outcomes were contrasted with those of known inhibitors. Thermal shift assay Recombinant GST-tagged human CHEK1 protein was diluted in phosphate-buffered saline (PBS). Curcumin was dissolved in DMSO and incubated with the purified GST–CHEK1 protein or vehicle control at 37℃ for 30 min. The protein samples were then aliquoted and subjected to a temperature gradient of 37, 39, 41, 43, 45, 47, 49, 51, 53, and 55 °C for 5 min at each temperature point. After heat treatment, samples were immediately cooled on ice and centrifuged to remove aggregated proteins. The soluble fractions were collected and analyzed by SDS–PAGE followed by Western blotting using an anti-GST antibody. Results Identification of potential targets of curcumin in TNBC The SwissTargetPrediction database provided a total of 65 potential curcumin targets. In the meantime, 6282 genes linked to TNBC were gathered from GeneCards. 58 overlapping targets were identified by the intersection of genes related to curcumin and TNBC; these targets were thought to be possible therapeutic targets of curcumin against TNBC (Fig. 1A). A compound–target–disease (C–T–D) network was built with Cytoscape in order to visualize the interactions between curcumin, its putative targets, and TNBC (Fig. 1B). Curcumin was shown as a blue node in this network, linked to 58 target proteins, which were then connected to the orange node, which stood for TNBC. The resulting network demonstrates the pharmacological activity of curcumin’s multitarget and multi-pathway properties. Fig. 1Network pharmacology analysis of curcumin against TNBC. A Venn diagram illustrating the overlap between curcumin-predicted targets (n = 65) and TNBC-associated genes (n = 6282), resulting in 58 common targets. B Compound–target–disease (C–T–D) network showing the interactions among curcumin, TNBC, and the identified targets. The blue node represents curcumin, the orange node represents TNBC, and gray nodes represent the 58 overlapping target genes. Edges indicate the interactions between the compound, targets, and disease Network pharmacology analysis of curcumin against TNBC. A Venn diagram illustrating the overlap between curcumin-predicted targets (n = 65) and TNBC-associated genes (n = 6282), resulting in 58 common targets. B Compound–target–disease (C–T–D) network showing the interactions among curcumin, TNBC, and the identified targets. The blue node represents curcumin, the orange node represents TNBC, and gray nodes represent the 58 overlapping target genes. Edges indicate the interactions between the compound, targets, and disease Functional enrichment analysis Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to better clarify the biological significance of the overlapping targets between curcumin and TNBC. To determine the main biological processes, cellular components, molecular functions, and signaling pathways involved in curcumin’s pharmacological effects, 58 overlapping genes were examined (Fig. 2). The targets were primarily enriched in processes associated with cellular response to nitrogen compound, cell cycle phase transition, inflammatory response, phosphorylation, and regulation of cellular response to stress in the biological process (BP) category (Fig. 2A). These results indicate that curcumin may have anti-tumor effects through regulating inflammatory signaling, cell proliferation, and oxidative stress responses—biological processes that are often dysregulated in TNBC. Significant enrichment was found for the cellular component (CC) category (Fig. 2B) in terms of focal adhesion, mitochondrial membrane, vesicle lumen, membrane raft, transcription regulator complex, and cyclin A2-CDK2 complex. According to these localizations, curcumin-targeted proteins primarily function in cytoplasmic and nuclear signal transduction, mitochondrial regulation, and cell adhesion—cellular processes essential for the growth and metastasis of tumors. Phosphotransferase activity (alcohol group as acceptor), kinase binding, protein heterodimerization activity, MAP kinase activity, and cytokine binding were the primary enriched terms, according to the molecular function (MF) analysis (Fig. 2C). These findings demonstrate the critical role that protein–protein binding and kinase-related interactions play in curcumin’s possible modulation of TNBC signaling networks. Various cancer-associated pathways, such as the cell cycle (hsa04110), p53 signaling pathway (hsa04115), EGFR tyrosine kinase inhibitor resistance (hsa01521), HIF-1 signaling pathway (hsa04066), Apelin signaling pathway (hsa04371), and AGE-RAGE signaling pathway in diabetic complications (hsa04933), were significantly enriched in the KEGG pathway analysis (Fig. 2D). Furthermore, pathways like hormone signaling (hsa04081) and glucagon signaling (hsa04922) were also enriched, indicating that curcumin’s action in TNBC may be influenced by metabolic and endocrine regulatory mechanisms. When taken together, these enrichment results suggest that curcumin may modulate multiple pathways and targets to affect drug resistance, oxidative stress response, cell proliferation, and apoptosis in TNBC. This suggests a mechanistic basis for curcumin’s possible therapeutic effects. Fig. 2Functional enrichment analysis of overlapping targets between curcumin and TNBC. A Top 20 significantly enriched Gene Ontology (GO) biological process (BP) terms; B Top 13 cellular component (CC) terms; C Top 13 molecular function (MF) terms; D Top 13 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The x-axis represents the number of enriched genes, while the y-axis lists the corresponding functional terms or pathways. The color scale indicates the level of statistical significance (− log10 p-value), with darker colors representing higher significance Functional enrichment analysis of overlapping targets between curcumin and TNBC. A Top 20 significantly enriched Gene Ontology (GO) biological process (BP) terms; B Top 13 cellular component (CC) terms; C Top 13 molecular function (MF) terms; D Top 13 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The x-axis represents the number of enriched genes, while the y-axis lists the corresponding functional terms or pathways. The color scale indicates the level of statistical significance (− log10 p-value), with darker colors representing higher significance Protein–protein interaction (PPI) network construction A PPI network was created in STRING using the 58 intersecting targets, and Cytoscape was used to visualize it (Fig. 3A). Curcumin-associated TNBC proteins were highly interconnected, as evidenced by the network’s 58 nodes and 338 edges. Based on degree centrality, topological analysis identified a number of hub proteins, such as EGFR, AKT1, BCL2, and STAT3, suggesting that they play important roles in mediating the pharmacological activity of curcumin. Additionally, each node’s degree value was calculated and displayed as a bar chart to further illustrate the relative significance of each target in the network (Fig. 3B). The results demonstrated the crucial functions that AKT1, EGFR, STAT3, BCL2, AURKA, EP300, GSK3B, CDK2, CCNA2, CCNA1, CHEK1, BRAF (V600E), RPS6KB1 play in the curcumin–TNBC interaction network by displaying the lager than or equal to degree 10. It’s important to note that AURKA, CCNA2, CCNA1, CHEK1, BRAF (V600E), and RPS6KB1, have not been thoroughly examined in relation to TNBC or curcumin interaction. These results imply that curcumin might have unknown regulatory effects on TNBC cell cycle and signaling regulation via these targets, which calls for additional verification using molecular docking and molecular dynamics simulation. We examined the expression profiles of AURKA, CCNA2, CCNA1, CHEK1, BRAF (V600E), and RPS6KB1 in breast invasive carcinoma (BRCA) using the TCGA database in order to confirm the significance of these hub targets in breast cancer. Comparing BRCA tissues to normal controls, the results showed that CCNA1, BRAF (V600E), and RPS6KB1 were downregulated, while AURKA, CCNA2, and CHEK1 were significantly upregulated (Figure S1). These results offer more proof that these genes may be involved in the curcumin–TNBC network. Fig. 3Identification of hub genes and protein–protein interaction (PPI) network. A PPI network of the 58 overlapping targets between curcumin and TNBC, constructed using the STRING database and visualized in Cytoscape. Nodes represent proteins, and edges represent protein–protein interactions. Node size and color intensity are proportional to the degree value, indicating the relative importance of each protein within the network. B Bar plot showing the degree centrality of nodes in the PPI network. Hub genes are identified as those with the highest degree values Identification of hub genes and protein–protein interaction (PPI) network. A PPI network of the 58 overlapping targets between curcumin and TNBC, constructed using the STRING database and visualized in Cytoscape. Nodes represent proteins, and edges represent protein–protein interactions. Node size and color intensity are proportional to the degree value, indicating the relative importance of each protein within the network. B Bar plot showing the degree centrality of nodes in the PPI network. Hub genes are identified as those with the highest degree values Molecular docking analysis AURKA, CCNA2, CHEK1, and BRAF (V600E) were chosen for molecular docking with curcumin based on the results of network analysis and differential expression in TNBC. Despite appearing as major nodes in the network as well, CCNA1 and RPS6KB1 were not included in the analysis as, according to a thorough literature review, there is currently no concrete evidence connecting them to TNBC. AURKA, CCNA2, CHEK1, and BRAF (V600E), on the other hand, have been shown to be biologically relevant targets for additional research due to their reported critical roles in tumor cell proliferation, DNA damage response, and oncogenic signaling. To evaluate the binding affinity between curcumin and these four proteins, molecular docking was performed using AutoDock Vina. According to the docking results, curcumin exhibited stable binding to all of the targets. The binding energies were − 7.1 kcal/mol for CCNA2, − 6.8 kcal/mol for CHEK1, − 6.7 kcal/mol for AURKA, and − 6.5 kcal/mol for BRAF (V600E) (Fig. 4A). The one with the lowest binding energy, CCNA2, indicated a stronger propensity to interact with curcumin. Curcumin’s special binding interactions with each protein were further demonstrated by visualizing the docking conformations. Curcumin was found in a distinct hydrophobic pocket in CHEK1 (Fig. 4B), suggesting a common pocket-binding mode. Instead of being inside a distinct pocket, curcumin was found on the protein surface of CCNA2 (Fig. 4C), indicating a weaker or more surface-level interaction. Curcumin was evidently embedded in the binding pocket of AURKA, where it was held in place by numerous hydrogen bonds and hydrophobic contacts (Fig. 4D). Curcumin also filled a surface cavity in BRAF (V600E), which is in line with its capacity to alter kinase activity (Fig. 4E). Fig. 4Molecular docking of curcumin with core targets. A Binding affinities (kcal/mol) of curcumin toward AURKA, CCNA2, CHEK1, and BRAF(V600E) predicted by AutoDock Vina. B Binding mode of curcumin within the ATP-binding pocket of CHEK1, showing key interactions and pocket occupancy. C Docking conformation of curcumin with CCNA2, indicating a surface-level interaction rather than a well-defined binding pocket. D Binding mode of curcumin within the catalytic pocket of AURKA. E Binding mode of curcumin within the ATP-binding pocket of BRAF(V600E). Surface representations are shown in the right panels for each complex to illustrate ligand positioning within or outside the binding pocket Molecular docking of curcumin with core targets. A Binding affinities (kcal/mol) of curcumin toward AURKA, CCNA2, CHEK1, and BRAF(V600E) predicted by AutoDock Vina. B Binding mode of curcumin within the ATP-binding pocket of CHEK1, showing key interactions and pocket occupancy. C Docking conformation of curcumin with CCNA2, indicating a surface-level interaction rather than a well-defined binding pocket. D Binding mode of curcumin within the catalytic pocket of AURKA. E Binding mode of curcumin within the ATP-binding pocket of BRAF(V600E). Surface representations are shown in the right panels for each complex to illustrate ligand positioning within or outside the binding pocket Molecular dynamics simulations of curcumin and target complexes RMSD, RMSF, Rg analysis CCNA2 was not included in MD simulations since curcumin was discovered to bind to the surface of CCNA2 instead of its active pocket. To investigate the dynamic stability and conformational changes of the curcumin–protein complexes, 500 ns MD simulations were performed on the remaining three targets: AURKA, BRAF (V600E), and CHEK1. Nine systems were produced by simulating the apo forms of each protein and the corresponding positive inhibitors (Alisertib for AURKA, Vemurafenib for BRAF (V600E), and AZD7762 for CHEK1) for comparison [2, 19, 33]. All simulated complexes achieved equilibrium in roughly 100–150 ns, as shown by the RMSD, RMSF, and Rg trajectories (Figure S2), and they all retained overall structural stability for the duration of the 500 ns simulation. All three trajectories (apo, curcumin-bound, and alisertib-bound) for the AURKA systems (Figure S2A) stabilized during the simulation and reached equilibrium in about 100 ns. The curcumin–AURKA complex showed RMSD values of 0.25–0.35 nm, which were marginally lower than those of the apo form. This suggests that curcumin binding reduced large-scale conformational fluctuations and improved structural stability. Curcumin binding might stabilize AURKA similarly to its known inhibitor, according to the RMSD of the Alisertib complex, which showed a similar trend. After 150 ns, the RMSD values for the curcumin and Vemurafenib complexes in the BRAF (V600E) systems (Figure S2B) stabilized at 0.3–0.4 nm, whereas the apo form showed somewhat greater fluctuations, peaking at 0.45 nm. In line with a clearly defined binding mode in the kinase pocket, this suggests that ligand binding, whether curcumin or vemurafenib, decreased backbone mobility and increased conformational rigidity. The RMSD trajectories of the curcumin- and AZD7762-bound systems for CHEK1 (Figure S2C) quickly converged (< 0.3 nm) and stayed steady for the full 500 ns. A slightly larger deviation (~ 0.35 nm) was seen in the apo system, indicating that curcumin binding encouraged a more stable and rigid protein conformation. Out of all the complexes, CHEK1–curcumin showed the smallest RMSD variations overall, indicating a particularly stable and tight binding. A summary plot of the mean RMSD, RMSF, and Rg values was created in order to compare the overall stability of all systems in more detail (Figure S2D). The results demonstrated that, while retaining comparable compactness (Rg), curcumin-bound systems showed consistently lower RMSD and RMSF values than apo and inhibitor-bound complexes. Its superior dynamic stability was confirmed by the fact that CHEK1-curcumin showed the lowest RMSD of all. According to RMSF values, in all systems, residues in the enzyme and binding domains stayed mostly rigid, while the largest fluctuations took place in the terminal or loop regions. Curcumin binding may constrain these motifs similarly to Alisertib, as evidenced by the slight flexibility differences between the curcumin- and inhibitor-bound systems of AURKA’s αC-helix and activation loop, which are important regulatory regions in kinase activity. In BRAF (V600E), loop segments distant from the active site were represented by modest peaks in RMSF between residues 550 and 620. In contrast to the apo form, curcumin binding reduced these motions, which further suggests stabilizing effects on the protein core. All systems showed almost constant Rg values over time, with AURKA (~ 1.9 nm), BRAF (V600E) (~ 1.95 nm), and CHEK1 (~ 1.9–2.0 nm). The Rg is a measure of the global compactness of the protein structure. Interestingly, Rg was generally slightly lower in ligand-bound complexes than in apo forms, suggesting a more compact structure upon ligand association. The fact that Rg varied only slightly (< 0.05 nm) over all trajectories indicates that there was no substantial unfolding or structural expansion, indicating that all systems were well-equilibrated and conformationally stable throughout the simulation. While taken together, RMSD, RMSF, and Rg analyses show that curcumin formed compact and stable complexes with AURKA, BRAF (V600E), and CHEK1 that were on level with or even marginally more stable than those of their respective positive inhibitors. These results suggest that curcumin not only binds to these kinases’ functional pockets but also provides conformational stability, which may be the basis for its ability to inhibit signaling pathways linked to TNBC. FEL analysis The first two principal components (PC1 and PC2) obtained from PCA were used to build the FEL in order to further investigate the conformational stability and energy states of each system (Figure S3). In the absence of a ligand, the Apo system for AURKA (Figure S3A) displayed a comparatively wide and distributed energy basin, suggesting high conformational flexibility. The energy surfaces showed a single dominant deep basin upon binding with either Curcumin (Figure S3B) or the positive control Alisertib (Figure S3C), indicating that ligand binding stabilized the protein structure and constrained the conformational space. Interestingly, the energy depth of the Curcumin–AURKA complex was comparable to that of the Alisertib–AURKA complex, suggesting similar conformational stability. The Apo form (Figure S3D) in the BRAF (V600E) systems showed several shallow minima, which were indicative of structural fluctuations throughout the simulation. One noticeable deep minimum was seen in the Curcumin-BRAF (V600E) complex (Figure S3E), suggesting that Curcumin binding maintained a particular conformational state. Although slightly wider than Curcumin’s, the Vemurafenib–BRAF (V600E) complex (Figure S3F) also showed a deep single basin, indicating that Curcumin induces a more compact conformational ensemble than the known inhibitor. Higher flexibility was indicated by the Apo form’s (Figure S3G) distributed and less defined energy surface for CHEK1. On the other hand, the AZD7762–CHEK1 (Figure S3I) and Curcumin–CHEK1 (Figure S3H) complexes, showed a single narrow and deep energy basin, suggesting increased structural rigidity and stability upon ligand binding. Curcumin binding decreases conformational heterogeneity and improves the structural stability of AURKA, BRAF (V600E), and CHEK1, just like their respective positive control inhibitors, according to the FEL results, which were in agreement with RMSD, RMSF, and Rg analyses. Binding free energy analysis 100 snapshots evenly extracted from the 100 ns with lowest FEL of each trajectory were used for MM/PBSA calculations to further measure Curcumin’s binding affinities with the chosen target proteins. Figure 5 displays the determined binding free energies (ΔG_binding) for the Curcumin–target complexes and the positive controls that correspond to them. The thermodynamically favorable binding was confirmed by the negative ΔG_binding values displayed by all complexes. The curcumin–target systems with the best affinities were CHEK1 (–209.37 kJ/mol) and BRAF (V600E) (–192.87 kJ/mol), whereas AURKA (–135.64 kJ/mol) had a moderate binding energy. Curcumin may interact with these kinases in a functionally relevant way, as evidenced by the fact that its ΔG_binding toward BRAF (V600E) and CHEK1 was similar to that of the clinically validated inhibitors Vemurafenib (–247.23 kJ/mol) and AZD7762 (–177.12 kJ/mol), respectively. In accordance with the hydrophobic and hydrogen-bonding properties of the ligand–protein interfaces, the decomposition of energy components revealed that van der Waals (ΔG_vdw) and electrostatic (COU) interactions were the main contributors to binding stability. Together with the MD and FEL analyses, these results provide compelling evidence for AURKA, BRAF (V600E), and CHEK1 as putative binding sites via which curcumin may have anti-TNBC effects. Fig. 5Binding free energy analysis of curcumin and reference inhibitors with target kinases. Binding free energies (ΔG_binding) of curcumin and corresponding reference inhibitors (alisertib for AURKA, vemurafenib for BRAF(V600E), and AZD7762 for CHEK1) were calculated using the MM/PBSA method. Comparative analysis highlights the relative binding affinities and interaction patterns of curcumin with each target in relation to clinically relevant inhibitors Binding free energy analysis of curcumin and reference inhibitors with target kinases. Binding free energies (ΔG_binding) of curcumin and corresponding reference inhibitors (alisertib for AURKA, vemurafenib for BRAF(V600E), and AZD7762 for CHEK1) were calculated using the MM/PBSA method. Comparative analysis highlights the relative binding affinities and interaction patterns of curcumin with each target in relation to clinically relevant inhibitors Binding mode analysis of curcumin and reference inhibitors with CHEK1, BRAF (V600E), and AURKA The representative lowest-energy conformations from the FEL landscapes were chosen out, and interaction analyses were carried out for every protein–ligand complex in order to better clarify the molecular interactions between Curcumin and its main target proteins (Fig. 6). Curcumin formed numerous hydrogen bonds and hydrophobic contacts with residues like Leu15, Val23, Lys38, Leu84, Cys87 and Leu137 in the CHEK1–Curcumin complex (Fig. 6A), indicating a strong binding affinity and possible inhibitory activity. Curcumin was stably located within the catalytic pocket of CHEK1. The docking site’s dependability was confirmed by the reference inhibitor AZD7762 (Fig. 6B), which likewise bound deeply within the same pocket and interacted with Leu15, Val23, Ala36, Leu84, Cys87, Leu137 and Asp148. Curcumin occupied the ATP-binding pocket in the BRAF (V600E)–Curcumin complex (Fig. 6C), through hydrogen bonding and π–π interactions with residues Ile463, Leu505, Leu514, Trp531, Phe583 and Asp594. In line with its established inhibitory mechanism against BRAF (V600E), the FDA-approved inhibitor Vemurafenib (Fig. 6D) bound to a similar region through hydrogen bonding and π–π interactions with residues Phe468, Val471, Ala481, Trp531, Cys532 and Phe595. Similarly, Curcumin formed stable interactions with residues Leu139, Ala160, Ala213, Arg220 and Leu263 after embedding itself in the active pocket of AURKA (Fig. 6E). The dependability of the docking results was confirmed by the control compound Alisertib (Fig. 6F), which displayed similar binding localization and interaction patterns, including hydrogen bonds and hydrophobic interactions with Lys143, Phe144, Phe165, Gln168, Ala172 and Val174. Fig. 6Comparative binding modes of curcumin and reference inhibitors with TNBC-related kinases. A Curcumin–CHEK1 complex showing the overall ligand position (left), surface representation of the pocket (middle), and 2D interaction diagram highlighting key residues (right). B AZD7762–CHEK1 complex for comparison, illustrating similar binding orientation and overlapping key interactions. C Curcumin–BRAF(V600E) complex showing ligand placement in the catalytic pocket, surface view, and 2D interaction map. D Vemurafenib–BRAF(V600E) complex demonstrating comparable binding region and interactions as curcumin. E Curcumin–AURKA complex showing ligand docking pocket and key residue interactions. F Alisertib–AURKA complex presented for comparison, highlighting overlap with curcumin in the binding site Comparative binding modes of curcumin and reference inhibitors with TNBC-related kinases. A Curcumin–CHEK1 complex showing the overall ligand position (left), surface representation of the pocket (middle), and 2D interaction diagram highlighting key residues (right). B AZD7762–CHEK1 complex for comparison, illustrating similar binding orientation and overlapping key interactions. C Curcumin–BRAF(V600E) complex showing ligand placement in the catalytic pocket, surface view, and 2D interaction map. D Vemurafenib–BRAF(V600E) complex demonstrating comparable binding region and interactions as curcumin. E Curcumin–AURKA complex showing ligand docking pocket and key residue interactions. F Alisertib–AURKA complex presented for comparison, highlighting overlap with curcumin in the binding site Curcumin can bind to CHEK1 Purified GST-tagged CHEK1 protein was used in an in vitro cellular thermal shift-like assay to confirm the computational prediction that CHEK1 is a novel molecular target of curcumin. Protein thermal shift was evaluated by immunoblot (IB) analysis after the protein was incubated with or without curcumin (20 µM) and exposed to a temperature gradient between 37 °C and 55 °C (Fig. 7). GST-CHEK1 showed a significant temperature-dependent decrease in soluble protein levels in the absence of curcumin (DMSO control), with a sharp decrease noted at temperatures above 45 °C, indicating progressive thermal denaturation. Curcumin-treated samples, on the other hand, demonstrated significantly improved thermal stability over the same temperature range, retaining a significantly larger percentage of soluble CHEK1 at higher temperatures. Together, these results demonstrate that curcumin directly interacts with CHEK1 and increases its thermal stability in vitro, providing experimental evidence to support CHEK1 as a bona fide target of curcumin. Fig. 7Curcumin can bind to CHEK1. A In vitro thermal shift assay was performed with purified GST-tagged CHEK1 protein in the presence or absence of Curcumin (20 µM), followed by immunoblotting (IB) analysis. B Quantitation and fitting of the data in (A) are shown, along with the ΔTm value Curcumin can bind to CHEK1. A In vitro thermal shift assay was performed with purified GST-tagged CHEK1 protein in the presence or absence of Curcumin (20 µM), followed by immunoblotting (IB) analysis. B Quantitation and fitting of the data in (A) are shown, along with the ΔTm value Discussion Lack of expression of ER, PR, and HER2 defines TNBC as one of the most aggressive and resistant breast cancer subtypes, where the primary therapy is chemotherapy, however accompanied by profound toxicity and poor efficacy. Thus, systematic identification of therapeutically relevant targets and drug candidates remains essential. A range of naturally occurring agents with multiple-target activities and low toxicity, such as curcumin-a polyphenol isolated from Curcuma longa L. have been identified as promising candidates. However, the exact molecular targets and mechanisms of curcumin for TNBC remain not fully understood. To understand how curcumin works in TNBC, a combination network pharmacology and molecular simulation was used to systematically identify and characterize potential targets and mechanisms of curcumin for the development of novel drugs against the clinically and biologically relevant kinases AURKA, BRAF (V600E), and CHEK1 controlling the proliferations, invasion, metastasis, and drug resistance in TNBC ([19]; [24, 46, 48]. With respect to all inhibitors, curcuumin bound AURKA, BRAF (V600E) and CHEK1, forming favorable Docking scores (-6.7 kcal/mol for AURKA, -6.5 kcal/mol for BRAF (V600E) and-6.8 kcal/mol for CHEK1) which are slightly better or very close to the reference inhibitor (Alisertib, Vemurafenib and AZD7762). On analysis of the protein bound structure curcumin docked within binding pocket and formed similar interactions as its respective inhibitor, primarily through hydrogen bond and hydrophobic interactions. These results suggest that curcumin may bind within the active sites of these kinases with moderate affinity, potentially overlapping with inhibitor binding modes. Based on the findings, it can be stated that curcumin may exhibits potential inhibitory activity against the kinase domains of AURKA, BRAF (V600E) and CHEK1. Therefore, it is feasible to propose that curcumin could inhibit the activities associated with these oncogenic kinases in TNBC; namely mitotic progression (AURKA), MAPK mediated cell proliferation (BRAF) and check point regulation (CHEK1). AURKA is overexpressed in TNBC and is an important mediator for the proliferation and development of tumor, development of chemotherapy resistance and tumor’s poor prognosis [19]. AURKA induces proliferation, cancer stemness, EMT and immune evasion of cancer cells [41]. Inhibiting AURKA by allisertib has shown the potential to inhibit tumor growth and stimulate the development of tumor-specific anti-tumor immunity [43]. Combining AURKA inhibition and targeting PD-L1 or CHEK1 leads to additive effect in antitumor treatment [32]. Therefore, AURKA kinase is a clinical target in TNBC, which explains our idea of targeting this protein with curcumin. In addition, curcumin binds within the same ATP pocket that alisertib targets, according to the docking results. This suggests that curcumin may partially mimic aspects of AURKA inhibition and thereby suppress mitotic progression and proliferation in TNBC. BRAF (V600E) is a serine/threonine kinase in the MAPK pathway that stimulates cell growth and survival [36]. The BRAF (V600E) mutation is not a very frequent in TNBC, but it accounts for a subset of these tumors that are sensitive to vemurafenib and a clinical niche that is targeted with curcumin [33]. Also, wild-type BRAF-associated hyperactivity of the MAPK pathway is implicated in the aggressively growing phenotypes of TNBC [21, 33]. Here we utilized BRAF (V600E) mutant for docking and MD simulations in an attempt to characterize the interaction and modulation effect by curcumin. However, a direct comparison with the wild-type BRAF was not performed in this study, which may limit mutation-specific interpretation. Docking studies revealed that curcumin interacts in the catalytic pocket. This is in agreement with the literature which indicated the inhibition of TNBC cells proliferation by curcumin and suggest a mechanism of TNBC-mediated tumor progression. In TNBC models, the CHEK1 inhibitor AZD7762 enhances DNA-damage–induced apoptosis and synergizes with other therapies [28, 51]. Curcumin docked to the same ATP-binding site, suggesting it may similarly disrupt CHEK1-mediated checkpoint signaling and promote DNA damage accumulation in TNBC cells. 500ns MD simulation were performed to characterize the stability of the binding between curcumin and proteins. RMSD, RMSF, Rg showed that the binding of curcumin has stabilized kinase structure and decreased local fluctuations in the catalytic site of all three kinases (especially CHEK1). FEL showed that protein-curcumin complex has binding with favorable conformations, and CHEK1-curcumin complex has most compact and lowest energy basin and has strongest binding. MM/PBSA analysis also confirmed that curcumin has good binding with the three proteins and had high binding energy in the CHEK1-curcumin complex (which was found comparable to that of AZD7762) [1]. While no simulations were run as separate replica simulations, the very long simulation times (500ns) and stable nature across various trajectory examinations suggest that the conclusions drawn from the system are very sound. However, simulations run as separate replica simulations in the future would only lend more weight to the statistical validity of these results. Therefore, the conclusions derived from MD simulations should be interpreted with appropriate caution. According to these molecular docking, MD simulation, FEL and MM/PBSA calculation, we propose that curcumin may exert multi-target effects for anti-TNBC through stabilize the kinases structures in the inactive-like conformation, block ATP binding site and so on. More importantly, CHEK1 can be considered a potentially relevant target identified in this integrative analysis, it blocks the checkpoint signals and enhances the sensitivity of TNBC to DNA damage agents. This finding not only support the experimental data we have, but also further develop the multiple targeting kinase inhibitory activity of curcumin on TNBC, apart from PI3K/Akt, MAPK, JAK/STAT, Wnt/-catenin signaling pathway [3, 7]. This mechanism is consistent with the evidence that TNBC sensitivity to radiation and chemotherapy can be effectively increased by dual inhibition of cell cycle checkpoints and DNA repair. Thermal stability assays are widely used to assess direct ligand–protein interactions based on the principle that ligand binding often stabilizes the native conformation of a protein, thereby increasing its resistance to thermal denaturation. In our assay, curcumin treatment resulted in a clear stabilization of GST-tagged CHEK1, as reflected by the increased soluble protein fraction at elevated temperatures. This thermal protection effect suggests possible direct binding of curcumin to CHEK1 rather than an indirect or nonspecific interaction. Further lead optimization and structural modification will be investigated in considering native curcumin’s drawbacks, such as poor solubility, low systemic stability, and rapid metabolism. Target affinity, bioavailability, and metabolic stability may all be enhanced by the development of curcumin derivatives with optimized substituents. Finding effective curcumin-based kinase inhibitors will be accelerated up even more with computational modeling and experimental validation of synthetic analogs. The thermal shift assay was performed using purified protein in vitro, which does not fully recapitulate the cellular environment. Future studies using cellular CETSA, enzymatic activity assays, or mutational analysis of the predicted binding site will be necessary to further confirm the functional relevance of the curcumin–CHEK1 interaction in cells. Furthermore, the complete experimental validation process will be guided by these computational results. In future, initially, in vitro experiments, such as IC₅₀ determination, cell proliferation and colony formation assays, and DNA damage analysis, will warrant to confirm curcumin’s inhibitory effects. In particular, these investigations will shed light on curcumin’s effects on checkpoint signaling and apoptosis induction, as well as its dose-dependent cytotoxic profile and mechanistic effects on TNBC cell lines. Curcumin’s antitumor efficacy, pharmacokinetics, and biosafety will then be evaluated in vivo xenograft experiments in mice, offering vital preclinical evidence for translational potential. Taken together, this study provides an example of how network pharmacology and dynamic simulations can be used to clarify the multi-target mechanisms of natural products. And these findings clarify the molecular mechanism of curcumin’s multitarget inhibitory action against TNBC, and they also provide a useful theoretical and methodological reference for the compound’s future development as a natural lead. Curcumin and its analogs may eventually progress toward clinically effective treatments for TNBC through the future integration of computational modeling, medicinal chemistry, and biological validation. Conclusions The potential target of curcumin in TNBC include AURKA, BRAF (V600E) and CHEK1. MD simulation and molecular docking revealed the stability of interaction of curcumin with each target, especially the best fit interaction was revealed for CHEK1 and this molecule showed highest binding affinity for curcumin based on calculation. The computational results suggest the effects of curcumin on several signaling pathways like response to DNA damage, MAPK pathway and mitotic pathway. Nevertheless, these results are predictions and these assumptions are required to be validated experimentally. The structure of curcumin needs to be optimized in order to get better effective curcumin derivatives for therapy in TNBC, therefore, curcumin can serve as a lead compound. Limitation Meanwhile, this study also has several limitations. First, the analysis was based on publicly available databases and computer simulations, and the size of the dataset depended on the coverage and quality of the selected databases. Second, although network pharmacology, molecular docking, and molecular dynamics simulations suggested potential curcumin binding targets in triple-negative breast cancer, most findings were derived from computer predictions. Besides the thermal displacement experiment confirming the interaction between curcumin and CHEK1, further experimental validation, including cellular and in vivo functional experiments, is needed. Therefore, this study provides a mechanistic and hypothesis-generating framework worthy of further biological and clinical investigation. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2. Supplementary Material 3. Supplementary Material 3. Supplementary Material 4. Figure S1 AURKA, CCNA2, CCNA1, CHEK1, BRAF (V600E), and RPS6KB1 are expressed at different levels in breast invasive carcinoma (BRCA). Figure S2 Molecular dynamics simulation analyses of curcumin–target complexes. Figure S3 FEL analysis of AURKA, BRAF (V600E), and CHEK1 systems based on principal components PC1 and PC2. Supplementary Material 4. Figure S1 AURKA, CCNA2, CCNA1, CHEK1, BRAF (V600E), and RPS6KB1 are expressed at different levels in breast invasive carcinoma (BRCA). Figure S2 Molecular dynamics simulation analyses of curcumin–target complexes. Figure S3 FEL analysis of AURKA, BRAF (V600E), and CHEK1 systems based on principal components PC1 and PC2.