3.1. ADME-related characteristic of Tanshinone IIA.
From the TCMSP results, the OB value of Tanshinone IIA is 49.89% and Tanshinone IIA showed a superior DL value 0.40, indicating that Tanshinone IIA can be absorbed effectively and has good druggability. The other properties of Tanshinone IIA, such as MW, AlogP, FASA, TPSA, and RBN were shown (Table 1). Therefore, it is reasonable to use Tanshinone IIA as a drug in HCC treatment.
Table 1
ADME-related properties of Tanshinone IIA
MW
|
AlogP
|
OB (%)
|
Caco-2
|
BBB
|
DL
|
FASA
|
TPSA
|
HL
|
294.3
|
4.66
|
49.89
|
1.05
|
0.70
|
0.4
|
0.31
|
47.28
|
23.56
|
3.2. Overlapping genes of Tanshinone IIA against HCC
We predicted the targets of Tanshinone IIA by TCMSP, SymMap, BATMAN-TCM, STITCH, SWISS, SEA, Pharmmapper servers. The first three databases were searched used the name” Tanshinone IIA”, while the last four databases were obtained according to the structure of Tanshinone IIA. The numbers were 38, 25, 7, 9, 100, 20, and 256, respectively. Furthermore, set union and removed the duplicated genes of all the predicted targets, among which we obtained 384 targets for the next analysis. The mRNA related genes were screened out from the downloaded HCC expressed profile. Further, we obtained a total number of 3391 DEGs between the HCC tissue samples and paracancerous tissue using the edgR method. That is to say, the predicted targets of Tanshinone IIA were 384, and there are 3391 DEGs. Then, Venn diagram showed the intersection genes. Finally, 105 overlapping genes were predicted (Fig. 2).
3.3 WGCNA screened important module and key genes
The expression matrix of 105 DEGs overlapping genes and the related clinical data were prepared for WGCNA analysis. With the module selection criteria (cut height: 0.1, minimum module size: 5), we obtained four modules. The correlation between the blue module and tumor grade was the most important (correlation coefficient = 0.37, P = 2e − 13) (Fig. 3). There are 23 genes in the blue module, in which four key genes (PLK1, KIF11, AURKB, EZH2) were contained in the blue module with a strict criterion of MM > 0.9 and GS > 0.3.
3.4 GO and KEGG analysis of blue module
There are 46 terms involved in BPs, 15 terms in CCs and 10 terms in MFs. In details, the BPs were enriched in mitotic nuclear division, G2/M transition of mitotic cell cycle, cell division, cell proliferation, regulation of cell cycle, spindle organization, response to drug, etc. The CCs were enriched in cytosol, spindle microtubule, nucleoplasm, spindle midzone, nucleus, and chromosome passenger complex,etc. The MFs were enriched in protein kinase binding, protein kinase activity, kinase activity, ATP binding, phosphoprotein phosphatase activity, drug binding, protein serine/threonine kinase activity, etc. From which, we showed the top 10 of BPs, CCs and MFs, respectively (Fig. 4 and Table 2). KEGG pathway analysis showed seven enriched pathways (Fig. 5, Table 3), involving in the cell cycle, progesterone-mediated oocyte maturation, MicroRNAs in cancer, pyrimidine metabolism, viral carcinogenesis, etc. (Table 3). Then, we analyzed the drug-target-pathway interactions (Fig. 6A), and showed the pathway of the cell cycle (Fig. 6B).
Table 2
GO enrichment analysis (p < 0.05)
Category
|
Term
|
Count
|
PValue
|
GOTERM_BP_DIRECT
|
GO:0007067 ~ mitotic nuclear division
|
8
|
1.99E-08
|
GOTERM_BP_DIRECT
|
GO:0000086 ~ G2/M transition of mitotic cell cycle
|
6
|
7.91E-07
|
GOTERM_BP_DIRECT
|
GO:0051301 ~ cell division
|
7
|
4.42E-06
|
GOTERM_BP_DIRECT
|
GO:0008283 ~ cell proliferation
|
7
|
5.71E-06
|
GOTERM_BP_DIRECT
|
GO:0051726 ~ regulation of cell cycle
|
5
|
1.87E-05
|
GOTERM_BP_DIRECT
|
GO:0007051 ~ spindle organization
|
3
|
1.94E-04
|
GOTERM_BP_DIRECT
|
GO:0042493 ~ response to drug
|
5
|
5.95E-04
|
GOTERM_BP_DIRECT
|
GO:0097421 ~ liver regeneration
|
3
|
6.51E-04
|
GOTERM_BP_DIRECT
|
GO:0000079 ~ regulation of cyclin-dependent protein serine/threonine kinase activity
|
3
|
0.001179039
|
GOTERM_BP_DIRECT
|
GO:0046777 ~ protein autophosphorylation
|
4
|
0.001409272
|
GOTERM_CC_DIRECT
|
GO:0005829 ~ cytosol
|
16
|
3.58E-07
|
GOTERM_CC_DIRECT
|
GO:0005876 ~ spindle microtubule
|
4
|
1.96E-05
|
GOTERM_CC_DIRECT
|
GO:0005654 ~ nucleoplasm
|
13
|
2.27E-05
|
GOTERM_CC_DIRECT
|
GO:0051233 ~ spindle midzone
|
3
|
2.35E-04
|
GOTERM_CC_DIRECT
|
GO:0005819 ~ spindle
|
4
|
4.01E-04
|
GOTERM_CC_DIRECT
|
GO:0030496 ~ midbody
|
4
|
4.84E-04
|
GOTERM_CC_DIRECT
|
GO:0005634 ~ nucleus
|
15
|
0.001007924
|
GOTERM_CC_DIRECT
|
GO:0032133 ~ chromosome passenger complex
|
2
|
0.0060221
|
GOTERM_CC_DIRECT
|
GO:0000922 ~ spindle pole
|
3
|
0.007573146
|
GOTERM_CC_DIRECT
|
GO:0045120 ~ pronucleus
|
2
|
0.008421234
|
GOTERM_MF_DIRECT
|
GO:0019901 ~ protein kinase binding
|
8
|
3.29E-07
|
GOTERM_MF_DIRECT
|
GO:0004672 ~ protein kinase activity
|
6
|
8.27E-05
|
GOTERM_MF_DIRECT
|
GO:0016301 ~ kinase activity
|
5
|
2.42E-04
|
GOTERM_MF_DIRECT
|
GO:0005524 ~ ATP binding
|
9
|
3.83E-04
|
GOTERM_MF_DIRECT
|
GO:0004721 ~ phosphoprotein phosphatase activity
|
3
|
0.001551551
|
GOTERM_MF_DIRECT
|
GO:0008144 ~ drug binding
|
3
|
0.004358836
|
GOTERM_MF_DIRECT
|
GO:0035174 ~ histone serine kinase activity
|
2
|
0.006500008
|
GOTERM_MF_DIRECT
|
GO:0004725 ~ protein tyrosine phosphatase activity
|
3
|
0.007429052
|
GOTERM_MF_DIRECT
|
GO:0004674 ~ protein serine/threonine kinase activity
|
4
|
0.012329852
|
GOTERM_MF_DIRECT
|
GO:0004712 ~ protein serine/threonine/tyrosine kinase activity
|
2
|
0.037142678
|
Table 3
Category
|
Term
|
Count
|
PValue
|
KEGG_PATHWAY
|
hsa04110:Cell cycle
|
7
|
4.73E-07
|
KEGG_PATHWAY
|
hsa04914:Progesterone-mediated oocyte maturation
|
5
|
6.38E-05
|
KEGG_PATHWAY
|
hsa05206:MicroRNAs in cancer
|
6
|
6.57E-04
|
KEGG_PATHWAY
|
hsa00240:Pyrimidine metabolism
|
4
|
0.002136
|
KEGG_PATHWAY
|
hsa04114:Oocyte meiosis
|
4
|
0.002797
|
KEGG_PATHWAY
|
hsa05161:Hepatitis B
|
4
|
0.005934
|
KEGG_PATHWAY
|
hsa05203:Viral carcinogenesis
|
4
|
0.015293
|
3.5 PPI network of the blue module
The network of 23 genes from STRING was saved and visualized in Cytoscape 3.6.1. From the network analysis, we chose the parameter of the degree as a screen criterion. Genes with a degree above 10 were regarded as high connectivity genes in the network. As a result, 10 genes met the requirement. The PPI network was shown in Fig. 7.
3.6 Identify the hub genes
We obtained the final hub genes from the intersection between the four key genes from the blue module and the ten genes with a degree above 10 from PPI network analysis, namely PLK1, KIF11, and AURKB.
3.7 Molecular docking
The computational docking simulation is based on the structure to explore the interplay between ligand and proteins [22]. In this study, there were 20 modes for the interactions between the Tanshinone IIA and a target (AURKB, KIF11 or PLK1) using AutoDock Vina. The first mode with the lowest affinity, which indicated a strong binding capacity between Tanshinone IIA and we chose the candidate targets for further visualization. The affinity is -10.3 kcal/mol between Tanshinone IIA and AURKB. The affinity is -8.8 kcal/mol between Tanshinone IIA and KIF11. The affinity is -10.9 kcal/mol between Tanshinone IIA and PLK1. Then we watched the docking results in PyMOL then (Fig. 8).
3.8 Validation of hub genes
To further validate the importance of AURKB, KIF11, and PLK1, we explored the mRNA expression, protein levels, survival curve, and ROC curve. Compared with the normal tissues, the mRNA expression of AURKB, KIF11 and PLK1 was increased in the HCC as shown by UALCAN (Fig. 9A). The DNA methylation levels of AURKB, KIF11, and PLK1 were decreased in cancer (Fig. 9B). By comparing the survival curves, the three hub genes showed a significant prognostic effect for HCC (Fig. 9C). The expression levels increased in the advanced tumor stages and pathological grades (Fig. 10). Time-dependent ROC curves of AURKB (1 year AUC = 0.674, 3 years AUC = 0.616, 5 years AUC = 0.579) KIF11 (1 year AUC = 0.711, 3 years AUC = 0.645, 5 years AUC = 0.586) and PLK1 (1 year AUC = 0.735, 3 years AUC = 0.667, 5 years AUC = 0.611) were drawn from the TCGA database (Fig. 11). ROC curve indicated that AURKB, KIF11, and PLK1 demonstrated good diagnostic biomarkers. The gene expression levels were also validated using Oncomine4.5 platform. As a result, the expression was increased significantly in the five cancer livers than the normal (Fig. 12). Besides, we validated the protein levels using the HPA database, and the cancer tissues showed stronger immunohistochemical staining (Fig. 13).