Candidate genes associated with AD
A total of 1932 significant genes were collect from the TTD and Drugbank databases after deleting repetitions (Supplementary file, Table S1).
Putative major chemical ingredients in the 4 herbs of QYG
28 ingredients with OB≥30%, DL≥0.18 and BBB≥-0.3 from TCMSP, BATMAN-TCM and TCMID were selected. Besides, some compounds out of the condition but with pharmacological activities and high contents, including catalpol, rehmaglutin A, rehmaglutin B, rehmaglutin D, adenosine, acteoside, daucosterol, echinacoside, martynoside, rehmaionoside B and rehmaionoside C from Rehmanniae Radix (RR), polyporenic acid C, pachymic acid, poricoic acid A, poricoic acid B, poricoic acid C, poricoic acid D, poricoic acid DM, tumulosic acid and ergosterol from Poria Cocos (PC), ginsenoside Rh2, panaxynol, 20(S)-protopanaxadiol, ginsenoside La, ginsenoside Ro, ginsenoside Rc, ginsenoside Re, ginsenoside rf, gypenoside LXIX, sanchinoside C1, Panaxytriol, 20(R)-ginsenoside Rg2, 20-(S)-Ginsenoside Rg3, ginsenoside Rb1, and ginsenoside Rg1 from Ginseng Radix et Rhizoma of Panax Ginseng C. A. Mey (PG), and fructose, glucose, maltose, pyridoxine, acetylcholine and sucrose from MEL of Apis Cerana (AC), were also regarded as putative active ingredients. The final candidate compounds were showed in Supplementary file, Table S2.
Network construction of the anti-AD targets of QYG
According to pharmacophore matching, some statistical factors and similarity measures, 1270 targets of QYG for Homo. sapiens were obtained. In which, RR contained 333, PC contained 184, PG contained 594, and AC contained 159 (Supplementary file, Table S3). And compound-target-disease network for the 4 herbs were constructed, respectively (RR-Targets, Figure 2A; PC-Targets, Figure 2B; PG-Targets, Figure 2C and AC-Targets, Figure 2D). The results revealed that 14 major ingredients in RR, 15 in PC, 34 in PG and 7 in AC may play important roles in anti-AD.
As shown in Figure 2A, 123 AD-related targets were associated with 14 ingredients in RR and the analysis of network revealed that β-sitosterol (41 targets) was predicted as the major ingredient in RR. FGF1 and FGF2 were forecasted as the major targets of RR for the treatment of AD. As shown in Figure 2B, the PC network included 93 targets. Among the ingredients of PC, polyporenic acid C (48 targets) was regarded as a major ingredient. PTPN1, HSD11B1, NR3C1, POLB and NR1H3 were predicted as major targets of PC for the treatment of AD. The PG network was showed in Figure 2C, containing 240 targets. Gomisin A (61 targets) was predicted as the major ingredient of PG for the treatment of AD, and FGF1, FGF2, VEGFA and STAT3 were predicted as the major targets. For AC network, 7 bioactive ingredients in AC were validated to bind with 55 AD-related targets, in which sucrose (23 targets) was forecasted as the major ingredient and CA2, FGF1, CHRM2, FGF2, CHRM1 and VEGFA were the major target (Figure 2D).
As shown in Figure 3, the compound-target-disease interaction network was constructed. The ingredients of QYG were described by diamond nodes, and the direct targets of QYG for AD treatment were encoded by circular nodes. Only the targets with higher values of “degree” (above two-fold of the median value) “betweenness centrality” and “closeness centrality” (above the median value), and “average shortest path length” (below the median value) were identified as the candidate targets of QYG for AD. Ultimately, 17 direct targets were screened (Table 1).
The characteristics of QYG in the treatment of AD
As shown in Figure 4, the ingredients directly to the targets were obtained via the compound-target-disease interaction network. Multi-ingredients from RR, PC, PG and AC regulated almost all direct targets. The results indicated that the anti-AD effect of QYG was played through regulating multi-targets by multi-ingredients.
GO enrichment analysis for targets
To investigate the biological functions of the direct targets of QYG for AD, the gene GO biological process (BP) was performed by the Cytoscape plugin ClueGO (Figure 5A). As shown in Figure 5B, 9 most significantly enriched GOBP terms (p≤0.05) were listed. Among the 9 GOBP terms, adenylate cyclase-inhibiting G-protein coupled acetylcholine receptor signaling pathway was the most prominent.
Target-pathway network and analysis
The functional annotation tool of DAVID was used for KEGG pathway enrichment analysis. Ultimately, three signaling pathways were obtained, including PI3K-Akt signaling pathway, regulation of actin cytoskeleton pathway and insulin resistance pathway (Table 2).