摘要
肌肉减少症(肌少症)是老年人群功能衰退的重要病理表型,其发生与代谢稳态失衡密切相关。酪氨酸代谢通路近年来被证实在多种衰老相关疾病中异常激活,但其在SARC中的作用机制及与功能表型的定量因果关联仍缺乏系统研究。本研究整合3组转录物组数据,批次校正后纳入94例,经线性模型微阵列分析(linear models for microarray data,Limma)差异分析、基因集富集分析(gene set enrichment analysis,GSEA)与加权基因共表达网络分析(weighted gene co-expression network analysis,WGCNA)筛选出33个酪氨酸代谢相关基因。结合最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)、随机森林(random forest,RF)与支持向量机递归特征消除(support vector machine-recursive feature elimination,SVM-RFE)3种算法交叉验证,识别FOXO1、ADH1B、DUSP4和IDO1共4个关键因子。整合10X单细胞转录物组数据(n=24573细胞)与单细胞代谢活性评分,在细胞水平构建关键基因表达与酪氨酸代谢活性的回归模型(例如FOXO1:y=1.7542x+0.9345),并通过两样本孟德尔随机化(two-sample Mendelian randomization,MR)分析,探讨其与肌肉功能表型的因果关联。FOXO1在卫星细胞中特异性高表达(3.1倍,P=1.4×10^(-8)),其表达与代谢活性显著正相关,孟德尔随机化(Mendelian randomization,MR)分析显示,其高表达显著增加握力下降风险(β=-0.23,P=2.1×10^(-6)),而IDO1在内皮细胞富集表达,呈保护性因果关联(β=0.17,P=4.3×10^(-4))。基于4基因构建的多变量风险预测模型在独立验证队列中曲线下面积(area under the curve,AUC)=0.86,优于传统临床指标(ΔAUC=0.12,P=0.003)。本研究构建“关键基因-代谢通路-肌肉功能”三维调控闭环,结合多算法筛选、单细胞建模与因果推断,系统揭示酪氨酸代谢在SARC中的核心作用,为分子分型与靶向干预提供了理论基础和应用工具。
Sarcopenia is closely linked to metabolic dysregulation in the elderly and shows a major pathological phenotype of functional decline.Although tyrosine metabolism has been shown to be aberrantly activated in various aging-related disorders,its regulatory role and quantitative causal relationship with clinical muscle function in sarcopenia remain underexplored.We integrated three independent bulk transcriptomic datasets(total n=487)and applied linear models for microarray data(Limma)differential expression,gene set enrichment analysis(GSEA)pathway analysis,and weighted gene co-expression network analysis(WGCNA)and identified 33 tyrosine metabolism-related genes.Cross-validation using least absolute shrinkage and selection operator(LASSO)regression,random forest(RF),and support vector machine-recursive feature elimination(SVM-RFE)algorithms highlighted four key regulators:FOXO1,ADH1B,DUSP4,and IDO1.Single-cell RNA sequencing(scRNA-seq)data(n=24573 cells)combined with single-cell metabolism scoring(scMetabolism)were used to build cell-level linear regression models linking gene expression to tyrosine metabolic activity(e.g.,FOXO1:y=1.7542x+0.9345,R^(2)=0.79).Two-sample Mendelian randomization(MR)analysis was conducted to infer causal effects on muscle strength phenotypes.The results showed that FOXO1 was highly expressed in satellite cells(3.1-fold vs.other cell types,P=1.4×10^(-8))and showed a strong positive correlation with metabolic activity.MR analysis indicated that higher FOXO1 expression significantly increased the risk of grip strength decline(β=–0.23,P=2.1×10^(-6)),whereas IDO1 exhibited a protective causal association(β=0.17,P=4.3×10^(-4)).A multivariable risk model based on these four genes achieved an area under the curve(AUC)of 0.86 in an independent validation cohort,outperforming traditional clinical indicators(ΔAUC=0.12,P=0.003).In sum,by integrating multi-algorithm feature selection,single-cell quantitative modeling,and genetic causal inference,this study systematically elucidates the central role of tyrosine metabolism in sarcopenia and provides a quantitative framework for molecular subtyping and targeted interventions.
作者
张旸
魏强嫚
ZHANG Yang;WEI Qiang-Man(Institute of Sports Science,General Administration of Sport of China,Beijing 100061,China)
出处
《中国生物化学与分子生物学报》
北大核心
2025年第7期1048-1061,共14页
Chinese Journal of Biochemistry and Molecular Biology
基金
国家体育总局体育科学研究所基本科研业务费资助项目(No.24⁃53)。
关键词
肌肉减少症
酪氨酸代谢
机器学习
回归建模
孟德尔随机化
sarcopenia
tyrosine metabolism(TYR metabolism)
machine learning
regression modeling
mendelian randomization(MR)
