试验以猪苓为研究对象，综合利用遗传算法-BP神经网络数学模型、色谱和质谱等技术，建立快速筛选猪苓中黄嘌呤氧化酶抑制剂的新方法。首先，利用遗传算法-BP神经网络数学模型优化猪苓中麦角甾醇的提取工艺；其次，以黄嘌呤氧化酶为生物靶分子，应用超滤质谱技术，筛选猪苓中潜在的黄嘌呤氧化酶抑制剂，同时，应用分子对接技术验证超滤实验的准确性；最后，以活性筛选为导向，利用高速逆流色谱技术分离猪苓中黄嘌呤氧化酶抑制剂。结果显示：猪苓中麦角甾醇的最佳提取工艺为乙醇体积分数80%、液料比16 mL/g、提取时间1.4 h、提取次数3次，在此条件下，麦角甾醇得率为2.31%；从猪苓中筛选并分离出4个纯度均在90%以上的黄嘌呤氧化酶抑制剂，其抑制效果强弱顺序为猪苓酮A>猪苓酮B>麦角甾-4,6,8(14),22-四烯-3-酮>麦角甾-7,22-二烯-3-酮。该方法快速、灵敏，为猪苓抗痛风药理作用的进一步研究提供了一定的理论基础。

A new method was developed for the rapid screening of xanthine oxidase inhibitors in Polyporus umbellatus. The novel rapid and sensitive method uses an integrated approach combining a genetic algorithm-back propagation neural network mathematical model, chromatography, and mass spectrometry. Extraction of ergosterol from P. umbellatus was optimized using the genetic algorithm-back propagation neural network mathematical model. Secondly, using xanthine oxidase as the biological target molecule, ultrafiltration mass spectrometry was applied to screen for potential xanthine oxidase inhibitors in P. umbellatus. Molecular docking technology was used to validate the accuracy of the ultrafiltration experiment. Finally, with the guidance of the activity screening, high-speed counter-current chromatography was used to separate the xanthine oxidase inhibitors from P. umbellatus. The optimal extraction conditions for P. umbellatus ergosterol were 80% ethanol by volume, liquid-solid concentration of 16 mL/g, extraction time of 1.4 hours, and three extraction cycles. The optimized yield of ergosterol was 2.31%. Four xanthine oxidase inhibitors with purity >90% were identified and isolated from P. umbellatus. Their inhibitory activity ranked in order of polyporusterone A> polyporusterone B> ergosta-4,6,8(14),22-tetraen-3-one> ergosta-7,22-dien-3-one. The findings providing a theoretical basis for further studies on the anti-gout effects of P. umbellatus.

国家自然科学基金项目（31870336）；吉林省自然科技基金项目（YDZJ202201ZYTS514）