血管性痴呆代谢相关基因及防治中药筛选的生物信息学及实验验证研究  

Bioinformatics and experimental verification research on screening of metabolism-related genes of vascular dementia and traditional Chinese medicines for prevention and treatment

作　　者：张金枝 陈炜[1] 卓桂锋 黄德庆[2] 符钰岚 杨富才 欧阳超 吴林[2] ZHANG Jinzhi;CHEN Wei;ZHUO Guifeng;HUANG Deqing;FU Yulan;YANG Fucai;OUYANG Chao;WU Lin(Zone One of Encephalopathy,the First Affiliated Hospital of Guangxi University of Chinese Medicine,Guangxi Zhuang Autonomous Region,Nanning530023,China;Graduate School,Guangxi University of Chinese Medicine,Guangxi Zhuang Autonomous Region,Nanning530022,China)

机构地区：[1]广西中医药大学第一附属医院脑病一区,广西南宁530023 [2]广西中医药大学研究生院,广西南宁530022

出　　处：《中国医药导报》2024年第29期36-42,共7页China Medical Herald

基　　金：国家自然科学基金资助项目(82160885、82374387);广西中医药大学研究生教育创新计划项目(YCSW2023384、YCBXJ2023009、YCBXJ2023029)。

摘　　要：目的基于代谢差异表达基因(MDEG)探讨血管性痴呆(VD)代谢相关的发病机制及可能的中药防治策略。方法GEO数据集GSE122063的差异表达基因与代谢相关基因取交集后得到MDEGs。随后,基于MDEGs将GEO数据集中VD患者数据进行聚类分型。利用P值排名前10的MDEGs进行机器学习模型构建并筛选排名前5位的MDEGs;根据前5位的MDEGs进行中药预测、四气、五味、归经分析并选取重要中药进行实验验证。抽取18只雄性SD大鼠按照随机数字表法分成分成假手术组、模型组、鹿茸组,每组各6只。假手术组仅分离颈动脉不接扎;模型组与鹿茸组大鼠予双侧颈总动脉结扎;鹿茸组大鼠予中药鹿茸[0.572 g/(kg·d)]灌胃。大鼠造模后进行水迷宫实验、RT-q PCR检测以验证生物信息学结果。结果本研究得到CYP17A1等24个MDEGs。基于MDEGs将VD基因分为2个亚型;极限梯度上升(XGB)模型为最佳学习模型并得到排名前10的MDEGs(CYP17A1、CYP26A1、KCNE4、ADCY7、FMO1、SLC13A5、GBA3、PCK1、ELOVL1、ALOX5AP);通过前5位的MDEGs获得鹿茸等27味具有生物学效应的中药,多归于肝、肾、脾经;五味多属甘、苦味;四气多属温、寒。药物功效多与补虚相关。与假手术组比较,模型组水迷宫实验逃逸潜伏期延长,目标象限停留时间减少,跨越平台次数下降(P<0.01);与模型组比较,鹿茸组逃逸潜伏期缩短,目标象限停留时间延长,跨越平台次数增多(P<0.01)。与假手术组比较,模型组大脑皮质CYP17A1、CYP26A1的m RNA相对表达量降低(P<0.05);与模型组比较,鹿茸组大脑皮质CYP17A1、CYP26A1的m RNA相对表达量上升(P<0.05)。结论CYP17A1等24个MDEGs可能是VD代谢相关发病机制的重要环节。XGB学习模型可以较准确诊断VD。MDEGs可能为揭示VD脾肾亏虚证型的内在机制提供线索。同时,鹿茸等中药有望成为防治VD的潜在药物来源,其中鹿茸可能通过代谢相关基因CYP17A1、CYP26A1发挥抗VD作�Objective To explore the pathogenesis of vascular dementia(VD)related to metabolism based on metabolic differentially expressed gene(MDEG)and the possible prevention and treatment strategies of traditional Chinese medicine.Methods MDEGs were obtained by intersection of differentially expressed genes and metabolism-related genes in GEO data set GSE122063.Then,VD patient data in GEO data set were clustered and classified based on MDEGs.The top ten MDEGs with P-value were used to construct the machine learning model and the top five MDEGs were screened.According to the top five MDEGs,traditional Chinese medicine prediction,four natures,five flavours,channel tropism analysis were carried out,and important traditional Chinese medicine was selected for experimental verification.Eighteen male SD rats were selected and divided into sham operation group,model group,and Cervi Cornu Pantotrichum group according to random number table method,with six rats in each group.In sham operation group,only carotid artery was separated without ligature.Rats in model group and Cervi Cornu Pantotrichum group were treated with bilateral common carotid artery ligation.Cervi Cornu Pantotrichum group rats were given Cervi Cornu Pantotrichum(0.572 g/[kg·d])by intragastric administration.Water maze test and RT-qPCR were performed to verify the bioinformatics results of the rat model.Results In this study,24 MDEGs such as CYP17A1 were obtained.VD gene was divided into two subtypes based on MDEGs.The eXtreme gradient boosting(XGB)model was the best learning model and the top ten MDEGs(CYP17A1,CYP26A1,KCNE4,ADCY7,FMO1,SLC13A5,GBA3,PCK1,ELOVL1,ALOX5AP)were obtained.Through the top five MDEGs,27 kinds of traditional Chinese medicines with biological effects were obtained,which were mostly attributed to liver,kidney,and spleen channels.the five flavours were mostly sweet and bitter.The four natures were mostly warm and cold.The efficacy of the drug was mostly related to tonifying deficiency.Compared with the sham operation group,the escape latency

关 键 词：血管性痴呆 代谢相关基因 中药筛选 证候实质 机器学习 

分 类 号：R255.9[医药卫生—中医内科学]