3.1 Candidate compound screening for ginseng
Retrieved from TCMSP, there were 190 components related to ginseng in total. Under the screening thresholds of OB ≥30% and DL ≥0.18, 22 active ingredients were selected for further analysis (Table 1). Among them, there are five active ingredients with higher OB: Celabenzine (101.88%), Aposiopolamine (66.65%), Frutinone A (65.9%), Inermin (65.83%) and Girinimbin (61.22%). These components play important roles in the pharmacological activities of ginseng.
3.2 Potential target prediction for ginseng
Among the 22 active components, 5 of them did not find corresponding targets, while the remaining 17 components obtained 109 potential targets after removing duplicates. All active compounds, their targets and the interactions between them are presented in the compound-compound target network (Fig. 2), which is composed of 126 nodes (17 components and 109 targets) and 246 edges. The pink nodes represent drug targets while the blue nodes represent compounds, and the edges represent the interactions between them. Degree, a topological parameter describing the importance of a node, stands for the number of edges connecting to the node. We used it to further determine the importance of each active component. Through analysis, we found that RS5 exhibited the largest number of potential targets connections (Kaempferol, degree = 59), followed by RS3 (Beta-sitosterol, degree = 36), RS2 (Stigmasterol, degree = 31), RS17 (Fumarine, degree = 24) and RS4 (Inermin, degree = 16). The same active ingredient can act on different targets, and different active ingredients can also act on the same target, which fully reflects the multi-component, multi-target action characteristics of ginseng.
3.3 Collection of therapeutic targets of ginseng for depression
A total of 11478 disease targets were obtained from GeneCards with “depression” as the keyword. Among them, 975 depression-related genes meet the requirement of relevance score > 7.32 and ginseng have 47 potential targets for depression (Fig. 3a). Compound-target-disease network (Fig. 3b) with 63 nodes and 102 edges linked 16 compounds and 47 target genes related to depression. Among the 16 candidate compounds, RS5 exhibited the largest number of potential anti-depression targets connections (Kaempferol, degree = 24), followed by RS3 (Beta-sitosterol, degree = 18), RS2 (Stigmasterol, degree = 14), RS17 (Fumarine, degree = 9) and RS10 (Frutinone A, degree = 8). For the 47 potential anti-depression targets, the network showed PTGS2 had the largest number of compound-target interactions, followed by SCN5A, GABRA1, CHRNA7 and SLC6A4, while the remaining 42 targets showed interactions with up to three compounds.
3.4 Protein-protein interaction analysis
After removing a free target, the PPI network (Fig. 4) contains 46 nodes and 259 edges, with average node degree of 11 and average local clustering coefficient of 0.533. Target size and color are used to reflect the degree, while edge thickness and color are used to reflect the combine score. The important targets were painted red and located centrally in the network. AKT1 (degree = 25), CASP3 (degree = 20), NOS3 (degree = 19), TNF (degree = 19), PPARG (degree = 18), SLC6A4 (degree = 18), ACHE (degree = 18), IL1B (degree = 17), PTGS2 (degree = 17) and MAOA (degree = 16) were the top ten genes regarding their degree. AKT1 also has the highest closeness centrality (0.67), indicating the faster the signal is transferred to other nodes. Due to the highest betweenness centrality (0.14), AKT1 is considered a bottleneck node in monopolistic position between modules in the network and more suitable as a therapeutic target.
3.5 GO biological process and KEGG pathway enrichment analysis
Through DAVID database, we obtained 305 GO terms (226 BP terms, 31 CC terms and 48 MF terms) and 68 pathways in total. With FDR < 0. 05 as the screening condition, 32 GO terms (22 BP terms, 6 CC terms and 4 MF terms) were selected (Fig. 5). GO analysis revealed that target genes were majorly associated with the biological process of response to drug (GO:0042493), positive regulation of cell proliferation (GO:0008284), positive regulation of nitric oxide biosynthetic process (GO:0045429), response to hypoxia (GO:0001666), positive regulation of ERK1 and ERK2 cascade (GO:0070374), cellular response to organic cyclic compound (GO:0071407), response to nicotine (GO:0035094), regulation of insulin secretion (GO:0050796), cellular response to hypoxia (GO:0071456), response to ethanol (GO:0045471) and etc. The 6 cell component terms were involved in plasma membrane (GO:0005886), integral component of plasma membrane (GO:0005887), membrane raft (GO:0045121), neuron projection (GO:0043005), postsynaptic membrane (GO:0045211) and Caveola (GO:0005901). Among them, plasma membrane and integral component of plasma membrane accounted for the largest proportion, with 29 and 15 targets respectively. Depending on the outcomes of GO enrichment, the enriched molecular function ontologies were dominated by protein homodimerization activity (GO:0042803), enzyme binding (GO:0019899), heme binding (GO:0020037), extracellular ligand-gated ion channel activity (GO:0005230). Under the conditions of P < 0.01 and FDR < 0.05 (Table 2), we found that pathways in neuroactive ligand-receptor interaction (hsa04080), African trypanosomiasis (hsa05143), HIF-1 signaling pathway (hsa04066), Leishmaniasis (hsa05140), Toxoplasmosis (hsa05145), Serotonergic synapse (hsa04726), Tuberculosis (hsa05152) and Malaria (hsa05144) are the interaction pathways that ginseng exerts combined antidepressant effects (Fig. 6).