基于随机森林的紫苏抗氧化活性成分分析

Analysis of Antioxidant Components from Perilla frutescens Based on Random Forest Method

为探究紫苏的抗氧化活性成分,采用不同极性的溶剂,通过索氏提取法制备紫苏叶、紫苏梗、紫苏子的15种提取物,利用高效液相色谱法检测各提取物的组成,建立清除羟基自由基、清除超氧阴离子自由基、清除DPPH自由基3个体外抗氧化活性实验检测各提取物的抗氧化活性,以随机森林算法结合Pearson相关性分析建立各提取物色谱峰面积和活性的数学模型,筛选对抗氧化活性贡献较大的组分并进行质谱分析。结果表明,15种提取物中含有原儿茶醛、迷迭香酸、咖啡酸等化合物且含量不同,在3个活性实验中,不同极性提取物体外抗氧化活性的EC 50值(mg/mL)具有一定差异,紫苏叶的石油醚提取物体现出清除自由基的优势。数学计算及质谱分析结果表明,与EC 50值均呈正相关并且特征重要度较大的组分为木犀草素(LU),揭示紫苏中对体外抗氧化活性贡献较大的成分为LU。

In order to investigate the antioxidants of Perilla frutescens,fifteen extracts of Perilla folium,Perilla caulis and Perilla fructus were prepared by soxhlet extraction with different polar solvents,and their compositions were determined by high performance liquid chromatography(HPLC).The antioxidant activities of extracts were determined by the antioxidant activity models of hydroxyl radical scavenging,superoxide anion radical scavenging and DPPH radical scavenging.A mathematical model of chromatographic peak areas and activities of extracts was established by using random forest algorithm combined with Pearson correlation analysis.The component with higher contribution to antioxidant activity was screened and analyzed by mass spectrometry(MS).The results indicated that the fifteen extracts contained different contents of protocatechuic aldehyde,rosmarinic acid,caffeic acid and other compounds.The EC 50 values(mg/mL)of antioxidant activity in vitro were different under the three active models,and the petroleum ether extracts from Perilla folium displayed an advantage of scavenging free radicals.Mathematical calculation and MS analysis showed that luteolin(LU)was the component exhibiting positive correlation with EC 50 value and high characteristic importance,revealed that LU was the main component of Perilla frutescens.contributing to the antioxidant activity in vitro,which provided a reference for the development of natural antioxidants.