Integrating peripheral blood and brain transcriptomics to identify immunological features associated with Alzheimer’s disease in mild cognitive impairment patients

BACKGROUND: Immune system dysfunction has been proven to be an important pathological event in Alzheimer’s disease (AD). Mild cognitive impairment (MCI), as a transitional stage between normal cognitive function and AD, was an important research object for the screening of early diagnostic markers and therapeutic targets for AD. However, systematic assessment of peripheral immune system changes in MCI patients and consistent analysis with that in the CNS were still lacking. METHODS: Peripheral blood transcriptome data from the AddNeuroMed Cohort (n = 711) was used as a training dataset to assess the abundance of 24 immune cells through ImmuCellAI and to identify MCI-related immune signaling pathways and hub genes. The expression level of the immune hub gene was validated in peripheral blood (n = 587) and brain tissue (78 entorhinal cortex, 140 hippocampi, 91 temporal cortex, and 232 frontal cortex) validation datasets. Finally, reliable immune hub genes were applied for Gene Set Enrichment Analysis and correlation analysis of AD pathological characteristics. RESULTS: MCI patients have early changes in the abundance of various types of immune cells in peripheral blood, accompanied by significant changes in NF-kB, TNF, JAK-STAT, and MAPK signaling pathways. Five hub immune-related differentially expressed genes (NFKBIA, CD4, RELA, CASP3, and HSP90AA1) were screened by the cytoHubba plugin in Cytoscape and the least absolute shrinkage and selection operator (LASSO) regression. Their expression levels were significantly correlated with infiltration score and the abundance of monocytes, natural killer cells, Th2 T cells, T follicular helper cells, and cytotoxic T cells. After validation with independent datasets derived from peripheral blood and brain, RELA and HSP90AA1 were identified as two reliable immune hub genes in MCI patients and had consistent changes in AD. The Gene Set Enrichment Analysis (GSEA) showed that their expression levels were closely associated with Alzheimer’s disease, JAK-STAT, calcium signaling pathway, etc. In addition, the expression level of RELA was positively correlated with β- and γ-secretase activity and Braak stage. The expression level of HSP90AA1 was negatively correlated with α- and β-secretase activity. CONCLUSION: Immune system dysfunction was an early event in AD. It provides a new target for the early diagnosis and treatment of AD.