Artificial intelligence and network pharmacology based investigation of pharmacological mechanism and substance basis of Xiaokewan in treating diabetes

Abstract

Xiaokewan is a typical Traditional Chinese medicine (TCM) for diabetes and contains various natural chemicals, such as lignans, flavonoids, saponins, polysaccharides, and western medicine glibenclamide. In the current study, a highly efficient system for screening hypoglycemic efficacy constituents of Xiaokewan has been developed with the integration of intelligent data acquisition, data mining, network pharmacology, and computer assisted target fishing. With the combination of background exclusion data dependent acquisition (BE-DDA) and non-targeted precise-and-thorough background-subtraction (PATBS) techniques, a novel workflow has been established for the non-targeted recognition and identification of TCM constituents in vivo, and has been applied to the exposure study of Xiaokewan in rat. In this case, an interesting correlation among drug, target, and disease can be established, by combining the screening or characterization results with the strategy of network pharmacology and multiple computer assisted techniques. Consequently, five main constituents (puerarin, daidzein, formononetin, deoxyschizandrin and glibenclamide) exposed in vivo have been selected as effective hypoglycemic components. Meanwhile, the network pharmacology experimental results showed that these five constituents could act on various drug targets, such as PI3K, PTP1B, MAPK, AKT, TNF, and NF-κB. These five constituents might be involved in the regulation of β-cell function or exhibit inflammation inhibition ability to relieve the pathophysiological process of disease from multiple links. Furthermore, the pharmacological effects of these five constituents have been verified by diabetic zebrafish model. The zebrafish model results showed that the TCM monomer mixture without glibenclamide exhibited similar hypoglycemic activity with Xiaokewan. Although the monomer mixture with glibenclamide showed better activity than Xiaokewan only, the deoxyschizandrin (TCM constituent of Xiaokewan) exhibited best hypoglycemic performance. In summary, the above results indicated that the application of both intelligent recognition technology in mass spectrometry dataset and computerized network pharmacology might provide a pioneering approach for investigating the substance basis of TCM and searching lead compounds from natural sources.

Keywords: Background exclusion data dependent acquisition; Chinese traditional medicine; Diabetes; Metabolites; Network pharmacology; Non-targeted data mining.