In this study, UPLC-Q-TOF-MS was first employed to identify the chemical components in the ethanol extract of SR. Then, the extraction process was optimized using star point design-response surface methodology. Fingerprints of different batches and processed products were established, and chemical markers were screened through a combination of various artificial neural network models. Finally, network pharmacology and molecular simulation techniques were utilized for verification to determine the quality markers.
A total of 35 chemical components in SR were identified, and the optimal extraction process was determined as follows: ultrasonic extraction with 80% methanol at a ratio of 120:1 for 70 minutes, with a soaking time of 30 minutes. Through discriminant analysis using various artificial neural network models, the samples of SR could be classified into two categories based on their growth years: Kuqin (dried roots of older plants) and Ziqin (roots of younger plants). Moreover, the samples within each category could be further clustered according to their origins. The four different processed products of SR could also be distinguished separately. Finally, through the integration of network pharmacology and molecular simulation techniques, it was determined that baicalin, baicalein, wogonin, norwogonin, norwogonin-8-O-glucuronide, skullcapflavone II, hispidulin, 8, 8"-bibaicalein, and oroxylin A-7-O-beta-D-glucuronide could serve as quality markers for SR.
The primary factors affecting the quality of SR were its growth years. The geographic origin of SR was identified as a secondary factor affecting its quality. Processing also had a significant impact on its quality. The selected quality markers have laid the foundation for the quality control of SR, and this research strategy also provides a research paradigm for improving the quality of TCM.