Screening active compounds and potential targets
The compounds were collected in the TCMSP and screened by OB ≥30% and DL≥ 0.18. A total of 115 compounds from YZG were screened, and among which, 72 compounds that had targets were retained. Of these compounds, 23 were from Huangqin, 12 from Qiancao, 11 from Zhizi, 9 from Mohanlian, 7 from Baishao, 7 from Cebaiye, 6 from Diyu, 6 from Mudanpi, 6 from Puhuang, 6 from Daji, 5 from Huaihua, 5 from Quanshen, 4 from Xianhecao, 2 from Dihuang, and 2 from Danggui. Compounds such as (+)catechins, sitosterol, kaempferol and quercetin exist in a variety of TCM in YZG, suggesting that these compounds may be necessary for therapeutic effects.
Compound-putative target network construction
We used Cytoscape to construct a compound–compound target network that comprised 285 nodes (72 compound nodes and 213 target nodes) and 654 edges. The blue nodes represent the compound molecules, and the orange nodes represent the compound targets. Each edge represents the interaction between the compound and its target (Fig. 2). In the network, the degree of a node indicates the number of routes it takes to connect to other nodes. According to the topological properties of the network, nodes with a larger degree were screened for further analysis. These nodes, together with more connected compounds or targets, play a pivotal role in the network and may be key compounds or targets. In this network, there is a phenomenon in which a compound corresponds to multiple targets, and several compounds share a common target simultaneously. This reflects the YZG mechanism that comprises multiple components and targets, which is in accordance with the characteristics of many TCM prescriptions and drugs that have a curative effect on the treatment. The number of targets of 47.2% compounds was 10 or more, and the number of targets for seven compounds was 20 or higher. Because AA can act on a variety of targets, it may be the crucial compound of YZG that exerts a hemostatic effect. Regarding the targets, the top three were PTGS2, PTGS1 and HSP90AB1, which can interact with 46, 35, and 27 compounds, respectively. These three targets may be essential for the actions of the compounds. According to the degree of the compound, we finally chose the 13 more important compounds, and the details were shown in Table 1.
Construction of the PPI network for the corresponding disease
A total of 100 targets of ODUB-related diseases were retrieved from the TTD and PharmGKB databases, and the genes were entered into the STRING database for obtaining a protein interaction, and finally, 126 disease-related targets were obtained from STRING database (see Fig. 3).Three topological features of each node in the network were calculated to find the major nodes. We selected the targets with the top five grades of degree, betweenness and closeness as the key disease targets, which may be studied in further research. The 5 nodes with degree≥18 were TNF, IL10, BDKRB2, F3, and PTGS2; the 5 nodes with betweenness≥0.06 were TNF, PTGS2, IL10, SYK, and CXCR; and the 5 nodes with closeness=1 were GABRA1, GABRB2, GABRA5, GABRA2, and GABRA3, and these were considered genes essential to the development of ODUB.
PPI network of targets for YZG against ODUB
By merging compound-target and PPI networks for ODUB, we obtained 29 overlapping protein targets, which can be considered the potential therapeutic targets. Using these overlapping targets, we searched the STRING database and found, and a total of 17 interacting secondary proteins that were associated with potential therapeutic targets. As a result, the network was composed of 46 nodes and 155 edges (Fig. 4).
We applied the bottleneck algorithm of the plug-in CytoHubba in Cytoscape software and found a total of 15 hub–bottleneck genes. There were four genes, PTGS2, CASP3, TNF, and PPARG, with a score of ≥5 and accompanied by red dots. The score of 11 genes, namely, ALOX5, CASP7, CYP3A4, FAS, F3, SYK, RIPK1, TRADD, NCOA2, AR, and PPARGC1A, was ≥1, and these were represented by orange dots (Fig. 5). These may be key genes that play a vital role in the pathway, implicating that we should focus on their respective pathways and conduct a detailed analysis.
GO function and KEGG pathway enrichment analyses
To elucidate the biological functions of these major hubs, we analyzed the candidate targets by performing a GO enrichment analysis, and the top 29 significant GO entries (P-value<0.05) were chosen according to the P-value, as shown in Fig. 6. The results indicated that the major hubs were significantly involved in multiple biological processes, primarily including inflammatory response, blood coagulation, hemostasis, coagulation, regulation of blood coagulation, regulation of hemostasis, fever generation, regulation of heat generation, platelet activation, ovulation, positive regulation of hemostasis, positive regulation of coagulation. The results of the cellular component and molecular function analyses showed that important genes exist in the cell body, GABA-ergic synapses, whose functions at the molecular level mainly included heme binding and estrogen receptor binding. The processes of blood coagulation, hemostasis, positive regulation of hemostasis, ovulation, positive regulation of coagulation, response to gonadotropin, heme binding, and estrogen receptor binding are related to abnormal menstrual bleeding in women. Thus, we speculated that YZG exerted its pharmacological effects on ODUB by simultaneously involving these biological processes and molecular functions. With the help of DAVID bioinformatics resources, 70 KEGG pathways with P-values ≤0.05 were also collected and analyzed. The top 15 crucially significant pathways were selected for further study and are displayed in the advanced bubble diagram (Fig. 7). These pathways were associated with the TNF signaling pathway (hsa04668), IL-17 signaling pathway (hsa04657), hepatitis B (hsa05161), hepatitis C (hsa05160), influenza A (hsa05164), C-type lectin receptor signaling pathway (hsa04625), T-cell receptor signaling pathway (hsa04660), Pertussis (hsa05133), AA metabolism (hsa00590), GABA-ergic synapse (hsa04727), sphingolipid signaling pathway (hsa04071), serotonergic synapse (hsa04726), complement and coagulation cascades (hsa04610), Toll-like receptor signaling pathway (hsa04620), and apoptosis (hsa04210). Detailed information about the top 15 significant pathways was supplemented in Supplementary Fig. 7. It is worthy of note that the hub genes we previously found, was mostly enriched the AA metabolism, serotonergic synapses, complement and coagulation cascades, C-type lectin receptor signaling pathway, apoptosis, TNF signaling pathway, and IL-17 signaling pathway, which suggested that the compounds in YZG may treat ODUB through the above pathways by acting on related targets. Importantly, by combining the obtained information of compound-target and hub genes, we found that PTGS1, PTGS2, ALOX5, CASP3, LTA4H, F7 and F10 were considered key targets because they also exist in important pathways related to bleeding (Fig. 8).
Molecular docking verification
Through molecular docking experiments, the numerical results of the binding energies from Vina docking were collected. The docking details are shown in Table 2, and the binding energy values of the 13 important compounds in YZG with their main targets are all less than -5 kcal/mol (20.9 kJ/mol), which suggests that the binding is significant; the less energy that is required, the more stable the binding. We selected the targets that bind to the most stable compound to display the 3D results, which included PTGS1-β-sitosterol (affinity=-10.6 kcal/mol), PTGS2-Hinokinin (affinity=-9.8 kcal/mol), ALOX5-AA (affinity=-5.1 kcal/mol), CASP3-luteolin (affinity=-7.4 kcal/mol), F7-isorhamnetin (affinity=-8.4 kcal/mol), F10-rivularin (affinity=-8.0 kcal/mol), and LTA4H-xyloidone (affinity=-8.5 kcal/mol). In Fig.9, the red circle represents a small-molecule compound, and each small-molecule compound is bound to a large-molecule protein. The stick graph on the right describes the specific form of this interaction. The yellow dashed line in the figure represents hydrogen bonding; there are fewer hydrophobic residues around the compound, and these residues are mainly bound to the target by electrostatic interactions (hydrogen bonding).
Drug-compound-target-pathway network construction
To holistically explain the interaction mechanisms among the different drugs, compounds and main targets in YZG, a drug-compound-target-pathway network was constructed established based on Cytoscape software. Red indicates the drugs; yellow indicates the compounds; pink indicates the potential targets; and green indicates the pathways. The red lines are connected to the potential therapeutic pathway, which has the highest degree of value, its key targets and the components with the highest docking scores (Fig.10). In YZG, flavonoids, coumarins, and organic acids are the compounds that are most closely related to crucial potential targets. Existing animal experiments have shown that flavonoids are the main active ingredients of Platycladus orientalis（L.）Franco（Cebaiye）, Cirsium japonicum Fisch.ex DC.（Daji）, Cirsium setosum（Willd.）MB.（Xiaoji） in cooling and hemostatic effects. The effective active ingredients of TCM for clearing heat and stopping bleeding also contain organic acids, phenols, and tannins. TCMs, including such as Cebaiye, Daji, Xiaoji, Rubia cordifolia L. (Qiancao), and Sanguisorba officinalis L. (Diyu) can achieve hemostasis by shortening bleeding and clotting times. In contrast, Qiancao are mainly used to increase the number of platelets and enhance their aggregation to ensure hemostasis [37-39]. However, the molecular mechanism of these compounds in treating uterine bleeding and heavy menstrual flow is unclear, which will be the focus of the discussion.