Integrative Bioinformatics and Machine Learning Identify Iron Metabolism Genes MAP4, GPT, and HIRIP3 as Diagnostic Biomarkers and Therapeutic Targets in Alzheimer's disease

Xiaoqiong An¹Xiangguang Zeng²Zhenzhen Yi³Manni Cao⁴Yijia Wang²Wenfeng Yu^2,5*Zhenkui Ren^1*

• ¹Department of Laboratory Medicine, The Second People's Hospital of Guizhou Province, Guiyang, Guizhou Province, China
• ²Key Laboratory of Molecular Biology, Guizhou Medical University, Guiyang, 550001, Guizhou, P.R. China, Guiyang, Guizhou Province, China
• ³Department of Laboratory Medicine ,Qianxinan People's Hospital, Xingyi, Guizhou, China
• ⁴Center for Tissue Engineering and Stem Cell Research, Guizhou Medical University, Guiyang, Guizhou Province, China
• ⁵Key Laboratory of Human Brain bank for Functions and Diseases of Department of Education of Guizhou Province, Guizhou Medical University, Guiyang 550025, Guizhou, P.R. China, Guiyang, Guizhou Province, China

Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline, memory impairment, and the accumulation of pathological markers such as amyloid-beta plaques and neurofibrillary tangles. Recent evidence suggests a role for dysregulated iron metabolism in the pathogenesis of AD, although the precise molecular mechanisms remain largely undefined. To address this, we utilized an integrative bioinformatics approach that combines weighted gene co-expression network analysis (WGCNA) with machine learning techniques, including LASSO regression and Generalized Linear Models (GLM), to identify hub genes associated with AD from transcriptomic data derived from postmortem prefrontal cortex samples (GSE132903, comprising 97 AD cases and 98 controls). To assess changes in the immune microenvironment, single-sample gene set enrichment analysis (ssGSEA) was employed. Furthermore, pathway enrichment analysis and gene set variation analysis (GSVA) were performed to uncover the underlying biological mechanisms driving these alterations. Protein validation was carried out in APP/PS1 transgenic mice through Western blotting. Three genes related to iron metabolism-MAP4, GPT, and HIRIP3-are identified as strong biomarkers. The GLM classifier showed high diagnostic accuracy (AUC=0.879). AD samples had increased immune activity, with more M1 macrophages and neutrophils, indicating neuroinflammation. MAP4 and GPT were linked to Notch signaling and metabolic issues. In APP/PS1 mice, MAP4 decreased, while GPT and HIRIP3 increased. This analysis highlights these genes as diagnostic biomarkers and therapeutic targets, connecting iron balance, neuroinflammation, and metabolic problems in AD. The immune profile suggests potential for immunomodulatory treatments, enhancing understanding of AD and aiding precision diagnostics and therapies.