Systemic lupus erythematosus (SLE) is a chronic autoimmune disease, which prevalence ranges from 20 to 150 cases per 100,000 population and appears to be increasing as the disease is recognized more readily and survival increases [
Workflow of the systematic strategies to elucidate the mechanisms of curcumin in the treatment of SLE-ONFH. ① Acquisition of drugs and disease targets. ② Collection of curcumin targets for the treatment of SLE-ONFH. ③ Analysis of therapeutic target proteins. ④ Topology of analysis of therapeutic targets. ⑤ The construction of curcumin target network, SLE-ONFH target network, and curcumin-SLE-ONFH target protein-protein interaction (PPI) network. ⑥ GO and KEGG enrichment analysis. ⑦ Gene-pathway network analysis. ⑧ Molecular docking.
The 2D chemical structure of curcumin (Figure
Curcumin chemical structure, CAS number: 458-37-7.
The Swiss target prediction database (
The SLE and ONFH-related targets were identified in the GeneCards database (
The curcumin target network and SLE-ONFH network were constructed and visualized using Cytoscape 3.8.0 software [
Shared targets of curcumin, SLE, and ONFH were obtained by running the Venn script as described above, which was installed in the R 4.0.2 software. They were uploaded to Cytoscape software. The PPI data were obtained from the Database of Interacting Proteins (DIP™), Biological General Repository for Interaction Datasets (Bio GRID), Human Protein Reference Database (HPRD), IntAct Molecular Interaction Database (IntAct), Molecular INTeraction database (MINT), and biomolecular interaction network database (BIND) using the plugin BisoGenet of Cytoscape software [
After uploading the common gene above, the nodes with topological importance in the interaction network were screened by calculating Degree Centrality (DC), betweenness centrality (BC), closeness centrality (CC), Eigenvector Centrality (EC), local average connectivity-based method (LAC), and Network Centrality (NC) with the Cytoscape plugin CytoNCA. At the heart of Cytoscape is a network. A simple network diagram consists of nodes and edges. Each node represents a gene. The node-node connection represents the interaction between these nodes. DC believes that the greater the number of neighbors of a node, the greater its impact. CC was used to calculate the importance of nodes. BC is a kind of centrality measure. EC is an indicator in ranking the importance of nodes in the network. The closer the EC of a node is to the network radius, the closer the node is to the network center. LAC is often used to determine the importance of proteins by assessing the relationship between proteins and their neighbors. NC is a method based on the edge clustering coefficient, which considers not only the centrality of nodes but also the relationship between nodes and adjacent nodes [
GO analysis with the biological process, cellular component, and molecular function was carried out by R software. The result is obtained by running the script installed on R software including “colorspace,” “stringi,” “ggplot2,” “BiocManager,” “clusterProfiler,” and “enrichplot.” The top 20 items were selected and visualized; a Bonferroni-corrected
The gene-pathway network was constructed based on the significantly enriched pathways with genes that regulated these pathways. The topological analysis of 20 pathways and 285 genes was carried out with BC. The squares represented target genes, and the V-shapes represented pathways in the network.
Molecular docking simulations were used to explore the potential interaction between the top 3 genes of the network above. The SDF format of curcumin was downloaded from the PubChem database; Chem 3D software was used to convert the SDF format into mol2 format file; then, the RESB database was used to get the PDB format structure of the top 3 target proteins. After that, solvent molecules and ligands were removed by Pymol software and saved as pdbqt format. Docking simulations were performed in Autodock1.1.2 software, and Discovery Studio 2020 was used to visually analyze the docking conformation at last [
100, 152, and 11 target genes of curcumin were identified by searching the Swiss, HERB, and STITCH databases, respectively, and 240 genes were selected as curcumin target candidate genes through running R software to remove duplicate genes. In order to ensure credibility, the 240 target genes were inputted into the STRING database; by setting the highest confidence (0.9000) and selecting the condition “Homo sapiens,” 201 targets were finally identified after hiding disconnected nodes in the network. The PPI network of them was saved as “tsv” file and visualized by Cytoscape software according to the degree (Figure
Curcumin target network. Note: PPI network of targets regulated by curcumin.
Topological analysis of curcumin target network.
Name | ASPL | BC | CC | Clustering coefficient | Degree |
---|---|---|---|---|---|
TP53 | 2.015625 | 0.11128227 | 0.49612403 | 0.20244898 | 50 |
SRC | 2.05729167 | 0.1213118 | 0.48607595 | 0.18454106 | 46 |
STAT3 | 2 | 0.08600855 | 0.5 | 0.23809524 | 43 |
AKT1 | 2.078125 | 0.08456637 | 0.48120301 | 0.20557491 | 42 |
MAPK3 | 2.06770833 | 0.0426935 | 0.4836272 | 0.26666667 | 40 |
MAPK1 | 2.05208333 | 0.07932813 | 0.48730964 | 0.25 | 40 |
EP300 | 2.125 | 0.09219908 | 0.47058824 | 0.20769231 | 40 |
JUN | 2.0625 | 0.05135675 | 0.48484848 | 0.28609987 | 39 |
RELA | 2.16666667 | 0.03133128 | 0.46153846 | 0.3125 | 32 |
VEGFA | 2.11979167 | 0.04222026 | 0.47174447 | 0.27513228 | 28 |
TNF | 2.25 | 0.03547444 | 0.44444444 | 0.29365079 | 28 |
FOS | 2.1875 | 0.02564237 | 0.45714286 | 0.33903134 | 27 |
HSP90AA1 | 2.13541667 | 0.04893392 | 0.46829268 | 0.2991453 | 27 |
MAPK14 | 2.18229167 | 0.02577391 | 0.45823389 | 0.31054131 | 27 |
MYC | 2.27604167 | 0.01608063 | 0.43935927 | 0.38666667 | 25 |
RB1 | 2.29166667 | 0.02207007 | 0.43636364 | 0.4 | 25 |
CCND1 | 2.27604167 | 0.01473079 | 0.43935927 | 0.40333333 | 25 |
EGFR | 2.22916667 | 0.03631959 | 0.44859813 | 0.30434783 | 24 |
CDKN1A | 2.34375 | 0.01530251 | 0.42666667 | 0.46014493 | 24 |
CDK1 | 2.53645833 | 0.01897448 | 0.39425051 | 0.43478261 | 23 |
ESR1 | 2.19270833 | 0.05987742 | 0.45605701 | 0.37944664 | 23 |
JAK2 | 2.40625 | 0.01035399 | 0.41558442 | 0.29437229 | 22 |
IL6 | 2.40625 | 0.0188487 | 0.41558442 | 0.37662338 | 22 |
CASP3 | 2.44791667 | 0.03709723 | 0.40851064 | 0.1991342 | 22 |
CXCL8 | 2.36458333 | 0.03987748 | 0.42290749 | 0.33766234 | 22 |
CASP8 | 2.40625 | 0.02257322 | 0.41558442 | 0.28095238 | 21 |
MMP9 | 2.48958333 | 0.0374213 | 0.40167364 | 0.29473684 | 20 |
CCNA2 | 2.77083333 | 0.00477339 | 0.36090226 | 0.47894737 | 20 |
CDK2 | 2.57291667 | 0.00859466 | 0.38866397 | 0.52105263 | 20 |
PTK2 | 2.328125 | 0.0222 | 0.4295302 | 0.26315789 | 19 |
After removing the duplication, we obtained 4338 SLE-related targets and 300 ONFH-related targets from the GeneCards and OMIM databases, respectively. In order to improve the comprehensiveness and credibility of the data, these targets were all included in the current study. Finally, 170 intersection genes related to SLE-ONFH were obtained by running the Venn map script installed in R software, and 36 genes related to SLE, ONFH, and curcumin were also received, as shown in Figure
Venn diagram of curcumin, SLE, and ONFH intersection targets.
PPI network of SLE-ONFH; the closer to the center of the circle, the darker the color, and the more likely it is to be a potential target.
According to the network pharmacology analysis method, 36 overlapping genes for curcumin, SLE, and ONFH were defined as potential genes (Table
The potential targets of curcumin-SLE-ONFH (36 intersection genes).
Gene names | Annotation | Degree |
---|---|---|
TP53 | Tumor protein p53 | 35 |
VEGFA | Vascular endothelial growth factor A | 34 |
IL6 | Interleukin 6 | 33 |
TNF | Tumor necrosis factor | 32 |
EGFR | Epidermal growth factor receptor | 31 |
CASP3 | Caspase 3 | 31 |
ESR1 | Estrogen receptor 1 | 29 |
MMP9 | Matrix metallopeptidase 9 | 29 |
CCND1 | Cyclin D1 | 28 |
TGFB1 | Transforming growth factor beta 1 | 27 |
IL1B | Interleukin 1 beta | 27 |
CXCL8 | C-X-C motif chemokine ligand 8 | 27 |
FGF2 | Fibroblast growth factor 2 | 26 |
SRC | SRC protooncogene, nonreceptor tyrosine kinase | 26 |
MMP2 | Matrix metallopeptidase 2 | 26 |
SERPINE1 | Serpin family E member 1 | 26 |
KDR | Kinase insert domain receptor | 25 |
TLR4 | Toll like receptor 4 | 25 |
PPARG | Peroxisome proliferator-activated receptor gamma | 25 |
PECAM1 | Platelet and endothelial cell adhesion molecule 1 | 25 |
TIMP1 | TIMP metallopeptidase inhibitor 1 | 24 |
MMP3 | Matrix metallopeptidase 3 | 24 |
HIF1A | Hypoxia inducible factor 1 subunit alpha | 24 |
NGF | Histone acetyltransferase GCN5 | 21 |
EP300 | E1A binding protein p300 | 19 |
COL1A1 | Collagen type I alpha 1 chain | 18 |
MPO | Myeloperoxidase | 17 |
SOD1 | Superoxide dismutase 1 | 15 |
NOS2 | Nitric oxide synthase 2 | 15 |
TGFBR1 | Transforming growth factor beta receptor 1 | 15 |
ABCB1 | ATP binding cassette subfamily B member 1 | 14 |
CYP19A1 | Cytochrome P450 family 19 subfamily A member 1 | 11 |
CYP3A4 | Cytochrome P450 family 3 subfamily A member 4 | 8 |
TNFRSF10A | TNF receptor superfamily member 10a | 8 |
BAX | BCL2 associated X, apoptosis regulator | 6 |
PCNA | Proliferating cell nuclear antigen | 6 |
PPI network of curcumin-SLE-ONFH; the top three potential genes sorted by degree were TP53, VEGFA, and IL6.
To reveal the mechanisms underlying curcumin’s effects on SLE-ONFH and obtain candidate genes more comprehensively and accurately, topological analysis of 36 potential genes was performed through the plugin BisoGenet and CytoNCA of Cytoscape; a network consisting of 3653 nodes and 87728 edges is presented in Figure
Topological analysis of 36 potential genes; 285 targets were finally identified.
GO analysis of 285 candidate targets was analyzed based on biological process (BP), cellular component (CC), and molecular function (MF). 2,364 GO terms were finally enriched, 1,973 in biological process, 200 in cellular component, and 191 in molecular function. The top 20 terms are shown in Figure
GO analysis of 285 candidate targets, BP, CC, and MF.
The pathways that are significantly influenced by curcumin in the process of treating SLE-ONFH were identified by KEGG pathway analysis; the analysis method is similar to the GO analysis by running the corresponding script in R software. The top 20 significant pathways (
(a) KEGG analysis of 285 candidate targets; (b) cell cycle signaling pathway. The central gene p53 plays a pivotal role in the pathway.
The enrichment pathways corresponding to genes after topology.
Term | Description | |
---|---|---|
hsa04110 | Cell cycle | |
hsa05203 | Viral carcinogenesis | |
hsa05161 | Hepatitis B | |
hsa05160 | Hepatitis C | |
hsa05220 | Chronic myeloid leukemia | |
hsa05171 | Coronavirus disease—COVID-19 | |
hsa05215 | Prostate cancer | |
hsa05166 | Human T-cell leukemia virus 1 infection | |
hsa05167 | Kaposi sarcoma-associated herpesvirus infection | |
hsa05169 | Epstein-Barr virus infection | |
hsa05163 | Human cytomegalovirus infection | |
hsa04120 | Ubiquitin mediated proteolysis | |
hsa05131 | Shigellosis | |
hsa04210 | Apoptosis | |
hsa04218 | Cellular senescence | |
hsa05205 | Proteoglycans in cancer | |
hsa04919 | Thyroid hormone signaling pathway | |
hsa04010 | MAPK signaling pathway | |
hsa05212 | Pancreatic cancer | |
hsa05170 | Human immunodeficiency virus 1 infection |
The gene-pathway network was constructed based on the significantly enriched pathways and genes that regulated these pathways, which are presented in Figure
Gene-pathway network of SLE-ONFH.
Molecular docking was applied to validate the binding action mode of the top three genes (TP53, VEGFA, and IL6), according to the network analysis. The results revealed that they can interact with curcumin; the combination of curcumin with them is shown in Figure
(a) Molecular docking between curcumin and TP53 (Affinity, -6.5). (b) Molecular docking between curcumin and VEGFA (Affinity, -5). (c) Molecular docking between curcumin and IL6 (Affinity, -5.9).
SLE is an autoimmune disease with multiple organ involvement and multiple antibodies. It is an urgent problem in the field of medicine. High disease activity of SLE is closely related to osteonecrosis especially ONFH, which has become the main cause of disability in SLE patients. In the whole process of disease development, immunity, inflammation, apoptosis, and angiogenesis are important factors [
Curcumin has been found to be useful in the treatment of SLE and ONFH, but the central genes and significant pathways of curcumin against SLE-ONFH are not clear. Through the target interaction network, we found 36 potential targets as described in Results; there were 16 targets (TP53, VEGFA, IL6, TNF, EGFR, CASP3, ESR1, MMP9, CCND1, TGFB1, IL1B, CXCL8, FGF2, SRC, MMP2, and SERPINE1) with connectivity greater than the median degree. The TP53 gene is at the core of the whole network, which means that it is a central target. p53 protein is a tumor suppressor that inhibits the growth of aberrant cells; functional p53 is believed to sense DNA damage and, subsequently, to induce DNA repair, growth arrest, or apoptosis of the aberrant cell [
At present, the treatment of SLE depends on the organ involved in the disease. Patients with mild symptoms can use low-dose corticosteroids; however, moderate and severe SLE may require higher doses of corticosteroids or other immunosuppressive agents. The adverse effects and limitations of hormones and immunosuppressive agents make TCM become a candidate therapeutic drug because of their unique advantages. The characteristic of SLE is that the complement system is activated; curcumin could inhibit the complement cascade. Studies have found that curcumin inhibits the increase of matrix protein, glial fibrillary acidic protein, and vimentin in the hippocampus of lupus mice [
The data that support the findings of this study are available from the corresponding authors upon request.
The authors report no conflicts of interest regarding this work.
Pan Kang worked on the investigation, methodology, writing (original draft), resources, and software. Zhiming Wu performed the investigation and methodology. Haibin Wang and Hongyu Tang performed the funding acquisition, project administration, and writing (review and editing). Yue Zhong, Zihao Wang, and Kun Xu worked on picture and software. Chi Zhou, Shaochuan Huo, Hai Guo, Songtao Li, Lingyun Liu, and Shuai Chen worked on data curation and advice. Pan Kang and Zhiming Wu contributed equally to this work.
This study was supported by the National Natural Science Foundation of China (Nos.: 81774339 and 82074462) and project funded by Guangzhou Science and Technology Plan Project (No.: 201707010319). Thanks are due to Mr. Wang Haibin, for the care and help he gave to the first author in the study and work.