Abstract
Given the unclear pathogenesis and insidious progression of Alzheimer's disease (AD), the aim of the present study was to identify reliable diagnostic markers for AD detection using a combination of bioinformatics analysis, animal experiments and clinical patient validation. Gene expression profiles were retrieved from the GSE95587 dataset. Weighted gene co‑expression network analysis combined with four machine learning algorithms identified two signature genes: Serine/Arginine Rich Splicing Factor 1 (SRSF1) and NADH: Ubiquinone oxidoreductase subunit B5 (NDUFB5), and a diagnostic model with moderate efficiency in differentiating AD was established. The AD diagnostic signature genes (SRSF1 and NDUFB5) were associated with specific immune cell infiltration. SRSF1 was significantly enriched in the p38MAPK and AKT1/mTOR signalling pathways. Notably, in an Aβ(1‑42)‑induced mouse model, SRSF1 expression was upregulated in the hippocampus and cerebral cortex. Moreover, in patients with AD, SRSF1 mRNA levels in peripheral blood mononuclear cells showed a strong negative correlation with mini‑mental state examination and Montreal cognitive assessment scores and a positive correlation with clinical dementia rating scores, indicating a notable association between elevated SRSF1 expression and cognitive decline. Furthermore, SRSF1 levels were positively associated with plasma levels of p‑tau217, p‑tau181 and glial fibrillary acidic protein. These findings underscore the strong association between SRSF1 and AD pathology. The newly identified genes, particularly SRSF1, show potential as candidate biomarkers of AD progression and may provide insights into AD pathogenesis, but require further validation in a larger prospective cohort.