m6A-related ferroptosis gene expression and its association with immune infiltration in Alzheimer’s disease: machine learning and molecular biology validation

Abstract

BACKGROUND: Alzheimer’s disease (AD) is a neurodegenerative disorder. Although β-amyloid and Tau proteins are core biomarkers for AD diagnosis, their heterogeneity and diagnostic limitations necessitate the exploration of novel biomarkers for disease diagnosis and treatment. OBJECTIVE: To analyze the interaction between N6-methyladenosine (m6A) epitranscriptomic modifications and ferroptosis genes in AD using machine learning, bioinformatics analysis, and experimental validation, to identify characteristic genes for AD pathogenesis, and to reveal their association with immune microenvironment regulation, thereby providing novel biomarkers for early diagnosis and precise treatment of AD. METHODS: Genomic data of human hippocampal tissues from GSE5281, GSE48350 (training sets), and GSE33000 (validation set) in the GEO database were integrated. Differentially expressed m6A regulators in AD were screened in the training sets, and the correlation between m6A and ferroptosis genes was assessed to identify ferroptosis-related differentially expressed genes associated with m6A. Support vector machine recursive feature elimination combined with Boruta feature selection was used to determine AD characteristic genes. Gene set enrichment analysis was performed to dissect functional modules of characteristic genes. A logistic regression model combined with receiver operating characteristic curves was constructed to evaluate the diagnostic efficacy of characteristic genes in the validation set. Single-sample gene set enrichment analysis was applied to quantify immune cell infiltration levels and analyze their regulatory association with characteristic genes. Transcription factor/miRNA-mRNA regulatory networks were predicted using ENCORI, miRWalk 3.0, and NetworkAnalyst databases. Potential therapeutic compounds were screened via the CTD database. qRT-PCR and western blotting were used to validate characteristic genes in hippocampal tissues of APP/PS1 double-transgenic mice. RESULTS AND CONCLUSION: (1) Two significantly differentially expressed m6A regulators, Wilms tumor 1 associated protein (WTAP) and methyltransferase-like protein 14 (METTL14), were identified, with 16 ferroptosis-related genes associated with them. (2) Machine learning identified five core characteristic genes: fumarate hydratase (FH), aspartate aminotransferase (GOT1), HRas proto-oncogene (HRAS), metallothionein 3 (MT3), and SET domain containing 1B (SETD1B). (3) Characteristic genes were functionally enriched in oxidative phosphorylation, Huntington disease, Parkinson disease, fatty acid degradation and metabolism, and proteasome signaling pathways. (4) The logistic regression diagnostic model achieved area under the curve values of 0.873 and 0.904 in the training and validation sets, respectively, indicating excellent diagnostic efficacy. (5) Immune microenvironment analysis showed that HRAS was significantly correlated with chemokine receptor family and plasmacytoid dendritic cell infiltration levels. (6) A regulatory network comprising 5 mRNAs, 37 miRNAs, and 142 transcription factors was constructed, and 71 potential therapeutic drugs were predicted. (7) Experimental validation showed that mRNA and protein expression of GOT1, HRAS, and SETD1B in the hippocampus of APP/PS1 mice were significantly different (P < 0.05 or P < 0.01), consistent with bioinformatics analysis. (8) The results reveal that FH, GOT1, HRAS, MT3, and SETD1B can serve as characteristic genes for AD; immune infiltration correlation analysis suggests that HRAS may serve as a potential immunotherapeutic marker for AD, providing a theoretical basis for early diagnosis and targeted therapy.

1. Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and neuronal loss, with pathological hallmarks including extracellular β-amyloid plaques and intracellular neurofibrillary tangles composed of hyperphosphorylated Tau protein. Despite significant advances in understanding AD pathogenesis, the disease remains incurable, and current diagnostic biomarkers, such as cerebrospinal fluid Aβ and Tau, have limitations in sensitivity and specificity, particularly in early stages. Therefore, there is an urgent need to identify novel biomarkers and therapeutic targets to improve early diagnosis and intervention.

Recent studies have highlighted the roles of epigenetic modifications, particularly N6-methyladenosine (m6A), the most abundant internal modification in eukaryotic mRNA, in various neurological disorders. m6A modification is dynamically regulated by writers, erasers, and readers, and influences mRNA stability, splicing, translation, and localization. Ferroptosis, a form of regulated cell death driven by iron-dependent lipid peroxidation, has been implicated in AD pathology. However, the interplay between m6A modification and ferroptosis in AD remains largely unexplored. This study aims to investigate the relationship between m6A-related ferroptosis genes and immune infiltration in AD using bioinformatics and experimental validation, with the goal of identifying novel diagnostic biomarkers and therapeutic targets.