血管性痴呆铜死亡关键差异基因的生物信息学分析及防治中药筛选  

Bioinformatics analysis of cuproptosis in vascular dementia and screening of traditional Chinese medicine for prevention and treatment

作　　者：卓桂锋 陈炜 朱健敏[1] 匡龙娇 廖乃彬 张金枝 苏明阳 吴林[1,3] ZHUO Gui-feng;CHEN Wei;ZHU Jian-min;KUANG Long-jiao;LIAO Nai-bin;ZHANG Jin-zhi;SU Ming-yang;WU Lin(Science Experiment Center,Guangxi University of Traditional Chinese Medicine,Nanning 530022,China;The First Clinical Faculty of Guangxi University of Chinese Medicine,Nanning 530022,China;Guangxi Key Laboratory of Basic Research of Traditional Chinese Medicine,Guangxi University of Traditional Chinese)

机构地区：[1]广西中医药大学科学实验中心,广西南宁530022 [2]广西中医药大学第一临床医学院,广西南宁530022 [3]广西中医药大学广西中医基础研究重点实验室,广西南宁530022

出　　处：《中草药》2023年第21期7120-7129,共10页Chinese Traditional and Herbal Drugs

基　　金：国家自然科学基金项目(82160885);广西中医药大学研究生教育创新计划项目(YCSW2023384);广西中医药大学2015年广西中医基础研究重点实验室项目(KJT15007);广西中医脑病临床研究中心项目(桂科AD20238028);广西高等学校高水平创新团队及卓越学者计划(桂教人才(2020)6号);广西中医药大学第一附属医院学术团队建设项目(院字[2018]146);广西中医药重点学科建设项目(GZXK-Z-20-13)。

摘　　要：目的通过机器学习等生物信息学方法筛选参与血管性痴呆(vascular dementia,Va D)发病机制的铜死亡关键差异表达基因(differentially expressed genes,DEGs),并预测和分析具有防治作用的中药。方法基于GSE33000数据集筛选铜死亡DEGs并分析其相关性;对数据集样本进行聚类分型,应用基因集变异分析(gene set variation analysis,GSVA)分型后通路富集情况;应用加权基因共表达网络分析与Va D关系密切的基因并取交集获得重要基因;构建风险预测列线图模型筛选重要基因的风险因子;基于风险因子构建多种机器学习方法的预测模型并将其与铜死亡DEGs进行相关性分析得到关键基因,并进行防治中药的筛选。结果共获得铜转运ATP酶β(ATPase copper transporting beta,ATP7B)、硫辛酸合成酶(lipoic acid synthetase,LIAS)等9个铜死亡DEGs,其相互之间表现出较强的协同或拮抗效应。根据铜死亡DEGs可将Va D患者分为2种亚型且DEGs在亚型间表达有所差异。分型后GSVA通路富集结果涉及刺猬信号通路等;绿松石模块(37个差异基因)与Va D分型高度相关,其与数据集DEGs交集得到5个重要基因,其中XLOC_005471可能是Va D的风险因子。广义线性模型(generalized linear models,GLM)机器学习模型的预测性能最高。脂酰基转移酶2(lipoyltransferase 2,LIPT2)、二氢脂酰胺S-乙酰基转移酶(dihydrolipoamide S-acetyltransferase,DLAT)、二氢硫辛酰胺脱氢酶(dihydrolipoamide dehydrogenase,DLD)、丙酮酸脱氢酶E1亚基β(pyruvate dehydrogenase E1 subunit beta,PDHB)与金属调节转录因子1(metal regulatory transcription factor 1,MTF1)与风险因子相关性较强,可作为铜死亡关键DEGs,其中MTF1、PDHB的表达与Va D患者年龄负相关(P<0.01)。根据铜死亡DEGs筛选出海蛤壳、鱼鳔胶等29味中药,其四气五味多属寒、温、平,苦、甘,归胃、肾、心、肝经,多为清热补虚药。结论ATP7B、LIAS等9个铜死亡DEGs相互调控作用及其相�Objective Key cuproptosis differentially expressed genes(DEGs)involved in the pathogenesis of vascular dementia(VaD)were selected by machine learning and otherbioinformatics methods.Differentially expressed genes(DEGs)were analyzed.Methods Cuproptosis DEGs were screened based on GSE33000 data set and their correlation was analyzed.Pathway enrichment was performed by cluster typing and gene set variation analysis(GSVA).The weighted gene co-expression network was used to analyze the genes closely related to VaD and the intersection was used to obtain the important genes.The risk prediction nomogram model was constructed to screen the risk factors of important genes.Based on the risk factors,a variety of predictive models of machine learning methods were constructed,and the correlation analysis between them and copper death DEGs was carried out to obtain key genes,and clinical correlation analysis and screening of prevention and treatment of traditional Chinese medicine were carried out.Results A total of nine cuproptosis DEGs including ATPase copper transporting beta(ATP7B)and lipoic acid synthetase(LIAS)were obtained.DEGs showed strong synergistic or antagonistic effect between them.According to cuproptosis DEGs,VaD patients could be divided into two subtypes and the expression of DEGs was different among the subtypes.The enrichment results of GSVA pathway after typing involved hedgehog signaling pathway,etc.The turquoise module(37 differential genes)was highly correlated with VaD typing,and its intersection with the dataset differential genes resulted in five important genes,among which XLOC_005471 may be a risk factor for VaD.The prediction performance of generalized linear models(GLM)machine learning model is the highest.Lipoyltransferase 2(LIPT2),dihydrolipoamide S-acetyltransferase(DLAT),dihydrolipoamide dehydrogenase(DLD),pyruvate dehydrogenase E1 subunit beta(PDHB)and metal regulatory transcription factor 1(MTF1)were strongly correlated with risk factors,and could be used as key DEGs for cuproptosis.The exp

关 键 词：血管性痴呆 铜死亡 清热补虚药 证候实质 机器学习 广义线性模型 海蛤壳 鱼鳔胶 

分 类 号：R285[医药卫生—中药学] Q811.4[医药卫生—中医学]