Target collection and construction of active ingredient-target maps
Through the TCMSP database, 20 active compounds and 186 targets for Huangqi, 51 active compounds and 141 targets for Leigongteng, and 9 active ingredients and 26 targets for Yiyiren were screened. In addition, 39 compounds and 96 targets of Wugong that have been forecasted or confirmed were gained through the relevant literature 20,21. The compositional information of herbs is shown in Supplementary Table 1. All targets were summarized, duplicate targets and standardised names were removed and corrected to get a total of 286 target genes (Fig. 3B, Supplementary Table 2).
Proven targets for the therapy of osteoarthritis were obtained from seven databases including Disgenet (1827), DrugBank (150), GeneCards (1835), Malacards (63), OMIM (6), PharmGkb (9), and TTD (32). Disease targets were converted into target-human gene IDs in the UniProt database, and duplicates were removed, yielding 2924 targets (Fig. 3A, Supplementary Table 2).
We generated the XFC-component-target network to better comprehend the intricate interactions between components and their related targets. Furthermore, there were 374 nodes (4 plants, 84 ingredients, 286 target) in the network (1236 edges). Components appeared to regulate numerous targets to generate therapeutic effects (Fig. 3C). Particularly, nine substances, including quercetin, kaempferol, nobiletin, stigmasterol, 7-O-methylisomucronulatol, triptolide, hederagenin, formononetin, and scolopendrine, which acted on 136,54,37,36,36,32,31,31, and 31 targets, specifically, become the key active compounds for the XFC because of their significant roles in this network. Further, we built a network diagram of these 9 key active compounds with their targets (Fig. 3D).
Discovery of the fundamental mechanisms of XFC in the course of OA.
To probe the underlying mechanism of XFC for the treatment of OA, the candidate targets of XFC were intersected with the therapeutic targets of OA by Venn diagrams, and 172 common targets were obtained (Fig. 3B, Supplementary Table 2). For network topology analysis, the 172 common objectives were imported into the STRING 11.0 database (Fig. 4A). The six network topology parameters for the PPIs were calculated using the CytoNCA plugin as a basis for scoring. Key targets were identified as frequent targets with topological parameters above the median. We carried out three screenings. The final screening threshold was then set as follows: degree > 37, eigenvector > 0.208, LAC > 40.807, betweenness > 2.412, closeness > 0.824, and network > 32.121 (Fig. 4B-4D).
Ten core targets were finally obtained, including interleukin 1 beta (IL1B), mitogen-activated protein kinase 14 (MAPK14), interleukin 1 alpha (IL1A), matrix metallopeptidase 13 (MMP13), the tumour necrosis factor (TNF), NFKB inhibitor alpha (NFKBIA), collagen type II alpha 1 chain (COL2A1), interleukin 4 (IL4), TNF receptor associated factor 2 (TRAF2), and vascular endothelial growth factor A (VEGFA), which has topological significance and likely plays a crucial role (Fig. 4D).
The top ranked results were filtered according to the physiological and pharmacological importance of the enrichment analysis. The numbers of GOBP, GOCC and GOMF were 2385, 53 and 171, respectively. The figure shows the enrichment results with the top 10 p-values as indicated by graphical p-value bars (Fig. 5A). GOBP primarily contained the regulation of mononuclear cell proliferation, the response to tumour necrosis factor, and the response to inflammatory response; GOCC included the collagen-containing extracellular matrix, and the RNA polymerase II transcription regulator complex, and so on; and GOMF consisted of metallopeptidase activity, extracellular matrix binding, nuclear receptor activity.
In addition, we identified 179 related signaling pathways (Fig. 5B). The enrichment results demonstrated that XFC for OA mainly involved the following: TNF signaling pathway, osteoclast differentiation, NF-kappa B signaling pathway, cytokine-cytokine receptor cytokine-cytokine receptor interaction. Based on the above outcomes, it is likely that the key mechanisms of XFC treatment of OA involve regulation of the TNF pathway, inflammatory response, and extracellular matrix metabolism. Some targets are closely related to these signaling pathways and regulate the aforementioned signaling pathways separately, as shown in Fig. 5C. Interestingly, the ten core targets appeared more frequently.