糖尿病肾病中铜死亡相关基因的生物信息学分析及靶向中药预测

Bioinformatics analysis and targeted Traditional Chinese Medicines prediction of cuproptosis⁃associated genes in diabetic nephropathy

目的通过生物信息学方法分析与糖尿病肾病(DN)有关的铜死亡相关基因,并预测靶向作用于这些基因的中药。方法从GEO数据库中下载与DN相关的基因表达数据集(GSE96804和GSE30529),并从相关文献中获取铜死亡相关基因。基于GSE96804数据集,筛选DN中差异表达铜死亡相关基因(DECAGs),并分析DECAGs之间表达量的相关性;根据DECAGs的表达量对DN样本进行聚类分型分析和基因集变异分析(GSVA)。基于聚类分型结果进行加权基因共表达网络分析,获取疾病关键基因模块和核心基因。利用“limma”程序包获取GSE96804数据集中的差异表达基因,然后与核心基因取交集以获得交集基因。基于GSE96804数据集及交集基因,通过3种机器学习模型筛选疾病特征基因。通过受试者工作特征曲线,基于GSE96804数据集评估特征基因对DN的诊断价值,并基于GSE30529数据集进行验证。通过分析疾病特征基因与DECAGs之间的相关性获得关键DECAGs。在CoreMine^(TM)MEDICAL数据库,针对DECAGs进行中药预测。结果共获得7个DECAGs,其中DBT、PDHA1、FDX1表达量与GCSH的表达量呈正相关,DLAT、PDHA1表达量与FDX1的表达量呈正相关,GLS表达量与DLAT的表达量呈负相关。基于DECAGs可将DN样本可分为两簇(C1簇和C2簇)。GSVA结果显示,与C1簇相比,NOD样受体(NLR)、Notch和Toll样受体(TLR)等信号通路在C2簇中富集。棕色模块(含369个核心基因)与基于DECAGs的聚类分型高度相关。共获得114个交集基因,通过3种机器学习模型最终筛选出疾病特征基因G6PC。在GSE96804、GSE30529数据集中,G6PC表达量诊断DN的曲线下面积分别为0.987、0.900。PDHA1、FDX1、DBT、GCSH与G6PC具相关性,故作为关键DECAGs。根据DECAGs筛选出168味中药,中药四气以寒性、温性和平性为主,五味以苦味、甘味为主,常见归经为肝经、脾经和肾经,功效以补虚类、清热和活血化瘀最常见。其中,丹参、青风藤、紫草、茯神、白术、苦豆草、蛇床子、艾叶均至少靶向2个DECAGs。结论7个DECAGs及其相关的NLR、Notch、TLR等信号通路可能是DN铜死亡相关发病机制的重要环节,其中,PDHA1、FDX1、DBT和GCSH是关键DECAGs,G6PC是DN的疾病特征基因。丹参、青风藤、紫草等中药可能通过调控铜死亡治疗DN。

Objective To analyze the cuproptosis-associated genes related to diabetic nephropathy(DN)by bioinformatics method,and to predict targeted Traditional Chinese Medicines on these genes.Methods Data sets(GSE96804 and GSE30529)of genes expressions related to DN were downloaded from the database of GEO,and cuproptosis-associated genes were obtained from relevant literature.Based on GSE96804 data set,differentially expressed cuproptosis-associated genes(DECAGs)of DN were screened,and the correlation of expressions between DECAGs was analyzed.The cluster typing analysis and gene set variation analysis(GSVA)were performed on DN samples according to DECAGs expressions.Weighted gene co-expression network analysis was performed based on the results of cluster typing for acquiring disease-critical gene module and core genes.Differentially expressed genes of GSE96804 data set were obtained by employing the limma program package,and then the intersection genes were obtained by intersecting with core genes.Disease-defining genes were screened through 3 categories of machine learning models based on GSE96804 data set and intersection genes.Diagnostic value of defining genes on DN was evaluated based on GSE96804 data set through the receiver operating characteristic curve,and the validation was performed based on GSE30529 data set.Key DECAGs were obtained through analyzing the correlation of disease-defining genes with DECAGs.In database of CoreMine^(TM) MEDICAL,Traditional Chinese Medicine prediction for DECAGs was conducted.Results A total of 7 DECAGs were obtained,therein DBT,PDHA1,and FDX1 expressions positively correlated with GCSH expression,DLAT and PDHA1 expressions positively correlated with FDX1 expression,and GLS expression negatively correlated with DLAT expression.DN samples could be assigned to two clusters(C1 cluster and C2 cluster)based on DECAGs.GSVA results revealed that compared with C1 cluster,NOD-like receptor(NLR),Notch,Toll-like receptor(TLR),and other signaling pathways were enriched in C2 cluster.Brown module(containing 369 core genes)was highly correlated with cluster typing based on DECAGs.A total of 114 intersection genes were obtained,and disease-defining gene G6PC was finally screened out through 3 categories of machine learning models.In the data sets of GSE96804 and GSE30529,areas under the curve of G6PC expression for diagnosing DN were 0.987 and 0.900,respectively.PDHA1,FDX1,DBT,and GCSH were correlated with G6PC,which therefore could be regarded as key DECAGs.According to DECAGs,a total of 168 flavors of Traditional Chinese Medicines were screened out,the four properties of Traditional Chinese Medicine were mainly cold,warm and moderate,the five flavors were mainly bitter and sweet,meridians mainly belonged to liver meridian,spleen meridian,and kidney meridian,and the effects were mainly categories of restoring qi,clearing heat and promoting blood circulation to remove stagnation,therein Radix salviae,Caulis sinomenii,Lithospermum erythrorhizon,Fu Shen,Atractylodes macrocephala,Sophora alopecuroides,Cnidii fructus,and Folium artemisiae argyi all targeted at least 2 DECAGs.Conclusion Seven DECAGs and their related NLR,Notch,TLR and other signaling pathways may be important links to cuproptosis-associated pathogenesis of DN,among which PDHA1,FDX1,DBT and GCSH are key DECAGs,and G6PC is the disease-defining gene of DN.Radix salviae,Caulis sinomenii,Lithospermum erythrorhizon,and other Traditional Chinese Medicines can treat DN through regulating cuproptosis.