Prefrontal cortex transcriptomes distinguish Alzheimers disease from asymptomatic amyloid pathology, neuropeptide loss, immune-complement activation and glial remodelling
Alzheimer's disease (AD) is now defined biologically by cortical amyloid-{beta} (A) and tau (T) pathology, yet the transcriptional programmes that separate symptomatic AD (A+T+) from amyloid-positive, tau-negative pathology ageing (A+T-) and from pathology-free controls (A-T-) remain incompletely resolved in the association cortex. We re-analysed bulk RNA-sequencing data from the dorsolateral prefrontal cortex (DLPFC) of 15 donors (BioProject PRJNA1017546; 87 sequencing runs aggregated to donor level): Normal (n = 2), pathology ageing (n = 5) and AD (n = 8). Differential expression was tested with DESeq2 (adjusted P < 0.05, |log2 fold change| > 1), and up- and down-regulated gene sets were analysed separately by Gene Ontology (GO) and KEGG over-representation. APOE and TREM2 p.Arg47His contrasts were treated as exploratory. AD versus Normal yielded 341 differentially expressed genes (DEGs; 223 up, 118 down) and AD versus pathology ageing yielded 496 (358 up, 138 down), whereas pathology ageing versus Normal yielded only six. Genes reduced in AD were dominated by neuropeptide and hormone signalling (VGF, VIP, CRH, GRP, TAC1, PENK; neuroactive ligand-receptor interaction, adjusted P = 8.8 x 10-5; neuropeptide hormone activity, adjusted P = 6.3 x 10-8), most of which also differed from pathology ageing. Genes increased in AD were enriched for interferon-{gamma} response, myeloid activation, adaptive immunity and classical complement (C1QA, C1QB, C1QC). Relative to pathology ageing, AD additionally showed induction of gliogenesis, oligodendrocyte differentiation and myelination programmes (gliogenesis, 21 genes, adjusted P = 2.6 x 10-6; SOX10, MYRF) and of the microglial purinergic receptor gene P2RX7 (log2FC = 1.46, adjusted P = 1.5 x 10-5). All six pathology-ageing DEGs, including the myeloid genes VSIG4 and CD163, were also DEGs in AD, and 50 DEGs shared by the two AD contrasts moved in the same direction. Genotype-stratified analyses were underpowered or confounded with diagnosis. In the DLPFC, AD is therefore characterised by a graded loss of neuropeptidergic signalling, myeloid-complement activation that begins at the pathology-ageing stage, and an AD-specific glial/oligodendrocyte response. These signatures nominate candidate biomarkers and mechanisms for early intervention, but the small cohort demands replication in larger, covariate-annotated datasets.