Scholarlink

<p>Autism Spectrum Disorder (ASD) pathogenesis remains unclear, with mitochondrial dysfunction implicated as a key contributor. Reliable mitochondrial-related diagnostic biomarkers are lacking, hindering early detection and mechanistic studies. This study integrated transcriptomic data from postmortem ASD cortical tissues (GSE28521 for training; GSE64018 for validation) with mitochondrial-related genes (MRGs) from MitoCarta3.0. Mitochondrial pathways were investigated using gene set enrichment analysis (GSEA). Candidate ASD-mitochondria (ASD-MIT) genes were identified by combining differential expression analysis, weighted gene co-expression network analysis (WGCNA), and MRGs. Machine learning algorithms (LASSO, Random Forest, and SVM-RFE) were applied to screen hub genes. Diagnostic performance was evaluated using a linear predictive model, an artificial neural network (ANN), and a nomogram. Single-sample GSEA (ssGSEA) was used to assess associations between hub genes and mitochondrial pathway activity. Biological validation included qPCR in BTBR mice and protein localization analysis using the Human Protein Atlas (HPA). GSEA revealed significant downregulation of mitochondrial pathways in ASD. 22 candidate ASD-MIT genes were identified, from which three hub genes—<i>IDH3A</i>, <i>MRPL2</i>, and <i>CHCHD4</i>—were consistently selected by all three machine learning models. The three-gene panel demonstrated strong diagnostic ability (AUC = 0.910), confirmed by the ANN model (AUC = 0.903). The nomogram achieved excellent predictive accuracy (C-index = 0.964). Importantly, ssGSEA analysis showed that these genes were strongly associated with mitochondrial pathway activity, particularly mitochondrial calcium ion transport. qPCR validated significant downregulation of <i>Idh3a</i> and <i>Mrpl2</i> in BTBR mice, and HPA confirmed mitochondrial localization and brain expression. This study identifies a mitochondrial gene signature associated with ASD and highlights <i>IDH3A</i>, <i>MRPL2</i>, and <i>CHCHD4</i> as promising diagnostic biomarkers. These findings advance understanding of mitochondrial dysfunction in ASD pathogenesis and further suggest that disruption of mitochondrial Ca<sup>2</sup>⁺–energy coupling may represent a key mechanistic feature of the disease.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mitochondrial Gene Signature Reveals Novel Diagnostic Biomarkers for Autism Spectrum Disorder

  • 期刊论文

  • 发表日期：

    2026年1月1日

查看全文
  • Jieyu Wang,
  • Zeyu Cheng,
  • Mingyuan Liu,
  • Yuting Zhang,
  • Yi Jiang,
  • Tianyu Liu,
  • Junyu Ren,
  • Lijie Wu,
  • Mingyang Zou,
  • Caihong Sun

摘要

Autism Spectrum Disorder (ASD) pathogenesis remains unclear, with mitochondrial dysfunction implicated as a key contributor. Reliable mitochondrial-related diagnostic biomarkers are lacking, hindering early detection and mechanistic studies. This study integrated transcriptomic data from postmortem ASD cortical tissues (GSE28521 for training; GSE64018 for validation) with mitochondrial-related genes (MRGs) from MitoCarta3.0. Mitochondrial pathways were investigated using gene set enrichment analysis (GSEA). Candidate ASD-mitochondria (ASD-MIT) genes were identified by combining differential expression analysis, weighted gene co-expression network analysis (WGCNA), and MRGs. Machine learning algorithms (LASSO, Random Forest, and SVM-RFE) were applied to screen hub genes. Diagnostic performance was evaluated using a linear predictive model, an artificial neural network (ANN), and a nomogram. Single-sample GSEA (ssGSEA) was used to assess associations between hub genes and mitochondrial pathway activity. Biological validation included qPCR in BTBR mice and protein localization analysis using the Human Protein Atlas (HPA). GSEA revealed significant downregulation of mitochondrial pathways in ASD. 22 candidate ASD-MIT genes were identified, from which three hub genes—IDH3A, MRPL2, and CHCHD4—were consistently selected by all three machine learning models. The three-gene panel demonstrated strong diagnostic ability (AUC = 0.910), confirmed by the ANN model (AUC = 0.903). The nomogram achieved excellent predictive accuracy (C-index = 0.964). Importantly, ssGSEA analysis showed that these genes were strongly associated with mitochondrial pathway activity, particularly mitochondrial calcium ion transport. qPCR validated significant downregulation of Idh3a and Mrpl2 in BTBR mice, and HPA confirmed mitochondrial localization and brain expression. This study identifies a mitochondrial gene signature associated with ASD and highlights IDH3A, MRPL2, and CHCHD4 as promising diagnostic biomarkers. These findings advance understanding of mitochondrial dysfunction in ASD pathogenesis and further suggest that disruption of mitochondrial Ca2⁺–energy coupling may represent a key mechanistic feature of the disease.