Use of network pharmacology and molecular docking to explore the mechanism of action of curcuma in the treatment of osteosarcoma

Curcuma has been used as an adjuvant treatment for osteosarcoma (OS) due to its anticancer compounds. However, the underlying mechanism remains unclear. Therefore, this study aimed to explore the mechanism of action of curcuma in the treatment of OS using network pharmacology and molecular docking. In this study, anticancer compounds were obtained from relevant literature, and curcuma-related targets and OS treatment targets were obtained from public databases. Protein‒protein interaction networks were constructed to screen out the hub genes using the STRING database and Cytoscape software. Cluster analysis of the protein modules was then performed using the Cytoscape MCODE plugin. Furthermore, Gene Ontology enrichment and Kyoto Encyclopedia of Genes and Genomes pathway analyses were performed for common targets among curcuma targets and OS-related targets using the DAVID database. Finally, molecular docking was performed, and the results were verified by Auto dock Tool and PyMOL software. Our research identified 11 potential active compounds, 141 potential therapeutic targets and 14 hub genes for curcuma. AKT1, TNF, STAT3, EGFR, and HSP90AA1 were the key targets closely related to the PI3K/Akt signaling pathways, HIF-1 signaling pathways, ErbB signaling pathways, and FOXO signaling pathways, which are involved in angiogenesis, cancer cell proliferation, metastasis, invasion, and chemotherapy resistance in the microenvironment of OS. Molecular docking suggested that the core compound had a strong affinity for key targets, with a binding energy of less than – 5 kJ/mol. The study showed that curcuma-mediated treatment of OS was a complex process involving multiple compounds, targets, and pathways. This study will enhance the understanding of how curcuma affects the proliferation and invasion of OS cells and reveal the potential molecular mechanism underlying the effect of curcuma on OS lung metastasis and chemotherapy resistance.


Identification of OS-related targets.
We searched for OS-associated targets in three databases, including the GeneCards database (https:// www. genec ards. org/) (accessed on 8th December 2022) 9 , which provides comprehensive, user-friendly information on all annotated and predicted human genes and automatically integrates gene-centric data (e.g. genomic, transcriptomic, proteomic, genetic, clinical, and functional information) from ~ 150 web sources; the OMIM database (https:// omim. org/) (accessed on 8th December 2022) 10 , which is a comprehensive, authoritative compendium of human genes and genetic phenotypes that is freely available and updated daily and contains information on all known Mendelian disorders, over 16 11 . Finally, the targets from each database were merged and labeled as OS-related targets.
Construction and analysis of the protein-protein interaction (PPI) network. STRING (https:// string-db. org/) (accessed on 8th December 2022) 12 is a database that integrates all known and predicted associations between proteins, including both physical and functional interactions. In this study, we used the STRING database to analyze the relationship of common curcuma and OS-related targets, which were considered poten- www.nature.com/scientificreports/ tial therapeutic targets. The PPI network of potential therapeutic targets was constructed using the following conditions: the species was "Homo sapiens", and the reliability was "medium confidence of ≥ 0.4". A confidence score of 0.4 is often used to indicate medium confidence when predicting protein-protein interactions. This particular value was selected due to its ability to balance sensitivity and specificity while also serving as a dependable threshold for identifying potential interactions. Therefore, only proteins with interaction relationship scores greater than or equal to 0.4 are included in the protein network. Subsequently, the PPI network was analyzed using Cytoscape v3. 8.0 software to obtain topology parameters of nodes in the network, including degree value, betweenness centrality (BC) value, and closeness centrality (CC) value. The "degree" indicates the number of nodes in the network that act directly with the node and is commonly used to assess node importance in networks. Analogous to the former, BC and CC are measures of a node's importance in a network based on its ability to interact with other nodes in the network or act as a bridge for information flow between other nodes. Therefore, the hub genes of the PPI network were selected based on the cutoff criteria: first, genes with a degree value (28), BC value (0.00144074408), and CC value (0.485611511) greater than the median value in the entire set were screened out; then, the genes with degree (40), BC (0.016190014), and CC (0.5625) values greater than the median in this gene set were screened out. The MCODE plugin was used for module clustering of the PPI network.
Functional enrichment and pathway analysis. All potential therapeutic targets were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis via the DAVID database (https:// david. ncifc rf. gov/) (accessed on 10th December 2022) 13 to identify the related pathways and related GO terms, including those in the biological process (BP), molecular function (MF), and cellular component (CC) categories. The pathways and GO terms with applicable thresholds of p < 0.05 were considered significant and were retained. Additionally, Bioinformatics (http:// www. bioin forma tics. com. cn/) (accessed on 13th December 2022) was used to visualize the GO and KEGG enrichment analysis results in a bar graph of signaling pathways and a bubble plot of GO categories. The network of active compounds-potential targets-signaling pathways was visualized using Cytoscape v3.8 software.
Verification with molecular docking. To analyze the relationship between the key targets of curcuma and the core compound, a molecular docking approach was employed. The 3D structure of the core compound was obtained from the PubChem database and imported into Chem3D software for energy minimization, resulting in the pdb file of small molecules. The 3D structures of the core target were downloaded from the PDB database (http:// www. rcsb. org/) (accessed on 11th December 2022) 14 , and we selected the protein 3D structure with high quality based on the following criteria: Source organism is "Homo sapiens", Refinement resolution within 2.5A, and complete protein structure sequence with information of small molecule ligands. Subsequently, water molecules and the original ligand of the proteins obtained were removed using PyMOL software. AutoDock Tool 1.5.7 software was used to hydrogenate, calculate charge and combine nonpolar hydrogen, and the ligands and receptors were stored as PDBQT files. All the flexible bonds of the small molecule ligand were set to be rotatable. Consequently, the Lamarckian genetic algorithm was used with a maximum eval number as the medium, and the Ga Crossover mode "twopt" was used to perform rigid docking and identify the binding activity of ligand and receptors. Finally, the reliability of the docking results was assessed by calculating the root mean square deviation (RMSD) value using PyMOL software.

Results
Compounds and curcuma-related targets. Curcuma compounds with anticancer activity were obtained by searching the relevant literature 5,14 . Using the SwissADME tool, we screened eleven active curcuma compounds with high gastrointestinal (GI) absorption and drug-like properties (Table 1). These compounds included curcumin, curcumol, Ar-turmerone, alpha-turmerone, beta-turmerone, curdione, cyclocurcumin, germacrone, furanodiene, calebin A, and furanodienone, which may be the material basis for curcumin's antiosteosarcoma (OS) effect. β-Sesquiphellandrene, δ-elemene, and δ-elemene were excluded due to their low GI absorption. In addition, the potential targets of each candidate component were predicted using the SwissTarget Prediction database (Supplementary Table S1). We then removed any duplicate targets and merged the remaining targets to obtain a total of 270 active compound-related targets.
Potential therapeutic targets of curcuma used in the treatment of OS. Data on "osteosarcoma, " which was used as the keyword, were retrieved from various databases. A total of 3688 targets were searched, including 3655 from GeneCards, 11 from OMIM, and 97 from PharmGKB (Supplementary Table S2). Of the curcuma-and OS-related targets, 141 common targets were identified and considered potential therapeutic targets of curcuma used in the treatment of OS, as shown in the Venn diagram (Fig. 2).
PPI network visualization and analysis. The STRING database was used to analyze the relationship between 141 potential therapeutic targets (Supplementary Table S3). Cytoscape v3.8.0 software was used for the visualization of the PPI network. According to the visualization, the PPI network included 141 nodes and 1279 edges, and the average node degree was 18.1 (Fig. 3). The nodes represent the individual proteins, while the edges represent the connections between the proteins. The greater the degree value, the more significant the role of the protein in the network. After filtering using the above criteria ( Fig. 4) www.nature.com/scientificreports/ ulation processes, such as human signal transcription and protein phosphorylation. AKT1, TNF, STAT3, EGFR, and HSP90AA were the top five nodal targets; they showed a strong association with other potential therapeutic targets and play an important role in the treatment of OS. The MCODE plugin was used for the cluster analysis of protein groups, and six cluster modules (Fig. 5) were obtained. Twelve important protein targets, including AKT1, TNF, STAT3, and EGFR, were clustered in the top two functional modules. AKT1 and TNF were identified as the core nodes in each of the two functional modules, suggesting that these two key targets significantly influence the effects of curcuma against OS.

GO enrichment analyses.
To determine the biological mechanisms of curcuma against OS, we performed GO enrichment analysis on 141 potential therapeutic targets of curcuma used in the treatment of OS using the DAVID database. The analysis was divided into three categories: BP, MF, and CC. We obtained 295 BP, 80 MF, and 42 CC terms, and the top 10 GO terms were visualized in the bubble plot (Fig. 6). The bubble size and color indicate the enrichment gene and p value, respectively. A darker color indicates a smaller p value, a larger bubble, and a higher number of enriched therapeutic genes within that GO item, indicating that the identified GO term is more strongly associated with the treatment of OS than other GO terms. Our analysis revealed that the main enriched BP categories were the response to drugs, positive regulation of transcription from RNA polymerase II promoter, negative regulation of apoptotic process, protein phosphorylation, cell proliferation, and inflammatory response. The main enriched MFs were protein binding, ATP binding, and protein kinase activity. The core targets, such as protein kinases, transcriptases, signaling molecules, and cytokines, were involved in biological metabolism, as supported by the BP and MF results. CC analysis revealed that the nucleus, plasma membrane, KEGG pathway enrichment analyses. The metabolic pathways of potential therapeutic targets for the treatment of OS were identified through KEGG pathway enrichment analyses. Using the DAVID database, we obtained 87 signaling pathways. We plotted the top 30 signaling pathways in a bar graph (Fig. 7), sorted by their P values from smallest to largest (Supplementary Table S4). The results showed that the key targets were enriched in the HIF-1 signaling pathways, PI3K-Akt signaling pathways, and chemokine signaling pathways. To further analyze the relationship between active compounds, potential therapeutic targets, and critical pathways, we constructed a visual network using Cytoscape v3.8.2 (Fig. 8). The visual network comprised 182 nodes and 707 edges. Curcumin, which was closely connected with 48 potential therapeutic targets and had the highest degree value (degree = 48) among candidate compounds, showed a strong association with several core OS targets (AKT1, EGFR, EP300, and STAT3), and was considered the most important compound. Therefore, we selected it as the key compound for molecular docking with key targets, including AKT1, TNF, STAT3, and EGFR, via Auto Dock Tool4.2 software.   Table 3. Additionally, curcumin interacted with specific residues by forming hydrogen bonds between residues including GLU278 and ALA230 of AKT1; GLY24, GLN25, ALA22, and LYS65 of TNF; GLN361 and LYS363 of STAT3; and GLU762, LYS745, ASP855, and MET793 (Fig. 9).

Discussion
OS is a primary malignancy associated with malignant mesenchymal cells with a high propensity for local invasion and early metastasis. Currently, the standard treatment for OS is surgery combined with chemotherapy. In recent years, the treatment of OS has progressed rapidly, and many treatment modalities, such as neoadjuvant chemotherapy combined with limb-sparing surgery, radiation, gene therapy, immunotherapy, and TCM, have emerged and have greatly improved the survival rate of patients with OS within 5 years 15  www.nature.com/scientificreports/ is also a commonly applied TCM with anti-inflammatory, anticoagulant, analgesic, antitumor, hypolipidemic, and other pharmacological effects. Clinically, the antitumor effect of curcuma is particularly prominent. However, the precise mechanism underlying its anticancer activity against OS remains unclear.
In the present study, we utilized network pharmacology to predict the mechanism underlying the therapeutic effects of curcuma for treating OS. We identified and screened 11 active antitumor compounds, including curcumin, curcumol, and calebin A, which serve as the material basis of anti-OS activities. In clinical practice, these compounds have been associated with apoptosis and drug resistance in tumor cells. Specifically, curcumin, curcumol, and calebin A exert anti-OS effects through multiple mechanisms. For example, curcumin is the main active compound of curcuma and can exert its antitumor effect by inhibiting tumor cell proliferation, regulating the energy metabolism of tumor cells, inhibiting tumor invasion and local angiogenesis, inducing tumor cell apoptosis and autophagy, inhibiting tumor immune escape, and reversing tumor cell resistance 18 . In a randomized controlled study of MG-63 cisplatin-resistant OS cell lines, Chen et al. found that curcumin inhibits autophagy activation in MG-63 cisplatin-resistant OS cells and promotes apoptosis in cisplatin-induced OS-resistant cells 19 . In addition, curcuminol, a noncurcumin compound of curcuma, inhibits the proliferation of OS cells by inducing autophagy 20 . Calebin A, an anticancer-active compound isolated from curcuma, inhibits osteoclasts induced by tumor cells by inhibiting the nuclear factor kappa B (NF-κB) pathway 21 .
We predicted anticancer-active compounds and OS-related targets in curcuma from a public database and found that curcuma has 141 interrelated common targets against OS, which induce cell apoptosis and proliferation. By analyzing the interactions between potential therapeutic targets, we found that these targets are clustered in these six protein modules. that have various biological functions. For example, module one is related to the G2/M transition of the mitotic cell cycle, protein phosphorylation, and regulation of signal transduction by p53 class mediator. Moreover, module two is related to positive regulation of transcription from the RNA polymerase II promoter, response to hypoxia, positive regulation of transcription, and DNA templating. Moreover, these 141 potential therapeutic targets are closely related to other targets; therefore, they are considered "hubs" in the therapeutic network. These 12 core proteins play a synergistic role involving the molecular mechanisms of proliferation and apoptosis, angiogenesis, metastasis, invasion, and drug resistance of OS cells. In particular, found that AKT1 gene knockout significantly inhibited the migration and proliferation of OS cells, highlighting its importance as an oncogene in OS cell growth and differentiation 23 . TNF-α is also associated with OS, and several related clinical studies have shown that the serum TNF-α level in patients with OS is significantly higher than that in healthy individuals 24,25 . Additionally, TNF-α is related to the pulmonary metastasis of OS. In vitro and in vivo studies have shown that TNF-α inhibitors reduce the expression of chemokine receptors and effectors to decrease cellular activity and pulmonary metastasis in OS cells, thus preventing pulmonary metastasis of OS 26 . STAT3 is another signaling and transcriptional activator involved in cell differentiation, growth, and survival with various biological functions, including embryogenesis, immunity, hematopoiesis, and cell migration. As a proto-oncogene, STAT3 is commonly overexpressed in OS tissues and cell lines, resulting in the decreased expression of the tumor suppressor gene PARK2 27 . The use of STAT3 inhibitors blocks downstream signaling and protein synthesis in OS cells and plays an active role in the treatment of OS 28 . EGFR is involved in the proliferation, differentiation, and migration of OS cells, and its expression increases the resistance of OS cells to certain anticancer drugs 29 . HSP90AA1 is a molecular chaperone known to function as an essential regulator of autophagy in OS cells and promotes autophagy through the PI3K/Akt/mTOR and JNK/P38 signaling pathways, inhibiting apoptosis and increasing chemotherapeutic tolerance in these cells 30 . This particular response is also thought to be a key mechanism of drug resistance in OS cells. Consequently, curcuma might act on these targets to treat OS. GO analysis provides an in-depth definition and description of the functions of potential therapeutic target proteins and genes, reflecting the potential mechanism of curcuma in combatting OS treatment resistance. Some important BPs, such as the regulation of apoptosis, cell proliferation, and inflammatory response, are associated with these potential therapeutic targets. In the OS microenvironment, bone, stromal, vascular, and immune cells, along with the mineralized extracellular stroma, coexist and coordinate to support various physiological processes, such as bone lysis, vascular reconstruction, and immunomodulation, which are associated with OS progression. In addition, the inflammatory response is closely related to the occurrence and progression of bone cancer, and several of the key targets identified in this study, including cytokines (such as TNF and PTGS2) and inflammatory reaction enzymes, are involved in regulating the inflammatory microenvironment of OS 31,32 . This environment depends on the activity of various protein kinases, steroid hormone receptors, RNA polymerase II transcription factors, kinases, and MFs, such as protein, enzyme, protein kinase, and energy substance binding. Moreover, these functions and BPs are also associated with specific cellular substructures, such as the nucleus, plasma membrane, cytosol, and nucleoplasm.   www.nature.com/scientificreports/ Our KEGG enrichment analysis identified multiple signaling pathways involved in the anti-OS mechanism of curcuma. The core gene targets were enriched in the PI3K/Akt, HIF-1, and FOXO signaling pathways, indicating that these signaling pathways are extremely important. By intervening in these signaling pathways, the BPs in which the core targets are involved can be mostly mediated. Moreover, these signaling pathways are closely related to the development and metastasis of OS. Therefore, curcuma may play an anti-OS role by interfering with tumor cell proliferation, angiogenesis, metastasis, invasion, and chemotherapy resistance. The PI3K/AKT signaling pathway is an important carcinogenic pathway in human cancers, including OS. It is involved in multiple pathological processes, such as tumorigenesis, proliferation, invasion, cell cycle progression, apoptosis, angiogenesis, metastasis, and chemotherapy resistance in OS. In this pathway, AKT phosphorylation significantly upregulates the expression of the oncogene c-Myc and promotes the proliferation and migration of OS cells. It also inhibits the programmed death of OS cells, which is closely related to OS tissue growth. Additionally, this pathway is associated with lung metastasis in patients with OS 33 . The HIF-1 signaling pathway is closely related to the migration and invasion of OS cells. It is involved in the angiogenesis of OS tissues and promotes the proliferation of OS cells. HIF-1, as a biomarker of OS, activates the AKT/cyclin D1 signaling cascade to promote the growth, migration, and invasion of OS cells in vivo. In a mouse model, overexpression of HIF-1 promoted the growth of tumor blood vessels and the growth and development of LM8 OS cells 34,35 . FOXOs are known tumor suppressors that promote tumor cell apoptosis and inhibit cell proliferation. In the metabolic environment of OS, FOXOs reduce osteoblast generation by inhibiting the classical Wnt/β-catenin signaling pathway 36 . The chemokine signaling pathway plays a key role in tumorigenesis and development. The CXCL12/CXCR4 axis activates multiple pathways involved in angiogenesis, metastasis, and survival in the tumor microenvironment and promotes tumor cell invasion and metastasis 37 . The CCL5/CCR5 axis promotes OS cell migration and supports angiogenesis 38 , while the ErbB signaling pathway is associated with OS pathogenesis. Cruz-Ramos et al. found that the application of EGFR blockers reduced the activity of OS cells, decreasing the probability of metastasis 39 . Moreover, Wang found that EGFR is a therapeutic target for inhibiting the invasion of OS, which is related to the unfavorable prognosis of OS. In the ErbB signaling pathway, EGF and its receptor activate MMP9, a protease involved in the proliferation of tumors, resulting in the invasion of cancer cells 40 . The thyroid hormone signaling pathway is of great significance in osteogenesis. In vitro studies by Liu et al. revealed that overexpression of thyroid hormone receptor-interacting proteins promoted the proliferation and migration of OS cells 41 .
Overall, the results based on network pharmacology and molecular docking presented in this study provide important insights into the potential anticancer mechanisms of curcuma by highlighting its ability to interact with multiple targets involved in cancer development and progression. Despite these findings, there are still some limitations to this study. The development of disease involves a pathological process. The curative effect of the compound differs significantly at each stage, suggesting that OS stage should be considered in treatment selection. The therapeutic effects of different doses of curcuma on the treatment of OS should be considered. Additionally, since the study was based on database analysis and computational approaches, there is a lack of www.nature.com/scientificreports/ experimental findings to validate the results of our research. Therefore, further experimental studies on these pathways and targets are necessary to validate the mechanisms of action.

Conclusion
Our research, based on network pharmacology and molecular docking, revealed that curcuma-mediated treatment of OS was a complex process involving multiple compounds, targets, and pathways. Through the PI3K-Akt, HIF-1, and ErbB signaling pathways, curcuma regulates key targets that are involved in angiogenesis, cancer cell proliferation, metastasis, invasion, and chemotherapy resistance, including AKT1, TNF, STAT3, EGFR and HSP90AA1, in the treatment of OS. These targets and pathways not only provide a pharmacological basis for the clinical treatment of OS but also provide a theoretical basis for the development of new drugs that interfere with lung metastasis and drug resistance in OS.