Predictive Gene Expression Signatures for Alzheimer's Disease Using Post-Mortem Brain Tissue

Ashley H. Duche¹Oliver Tan¹Andrius Baskys²Rachita K Sumbria¹Moom Rahman Roosan^1*

• ¹Pharmacy Practice Department, Chapman University School of Pharmacy, Irvine, CA, United States
• ²College of Health Sciences and Specialty Clinics, Clinical Genomics and Pharmacogenetics Service, Western University of Health Sciences,, Pomona,CA, United States

Background Alzheimer's Disease (AD) is a progressive neurodegenerative disorder characterized by amyloid-beta (Aβ) plaques and tau protein aggregates in the brain. These pathological features manifest in specific regions, but mechanisms rendering some areas more susceptible to early AD-related changes remain poorly understood. To address this, we developed predictive gene expression signatures to explore molecular mechanisms underlying regional vulnerability to AD pathology. Methods Post-mortem brain tissues from participants of the Religious Orders Study and Memory and Aging Project (ROSMAP), Mayo Clinic, and Mount Sinai Brain Bank (MSBB) were used to derive gene expression signatures from six brain regions affected at varying stages of AD progression. Differential gene expression analysis identified genes with altered expression patterns which were used to develop predictive gene signatures using Adaptive Signature Selection and InteGratioN (ASSIGN) to predict pathway activity. Predictions of AD activity were validated against known AD status across clinical markers of AD pathology, including Aβ plaque deposition, tau aggregates, cognitive assessments, and clinical diagnoses. Dysregulation of key biological pathways was then analyzed using g:Profiler and ClueGO. Additionally, genetic and sociodemographic factors impacting AD prediction were assessed, and potential drug repurposing candidates identified using Connectivity Map (CMAP). Results Predictive gene expression signatures from six AD-affected brain regions distinguished AD activity in control and AD post-mortem brain tissue, corresponding to clinical markers of disease severity. The signatures revealed common underlying mechanisms of regional vulnerability, including upregulation of extracellular matrix (ECM)-related processes and downregulation of hormonal signaling pathways. Notably, S100A4 was consistently upregulated across all regions, while CRH expression was downregulated except in the cerebellum. Additionally, findings underscored the influence of APOE genotype (e3/e4) and sex on disease progression. Drug repurposing analysis identified FGFR inhibitors, specifically orantinib and bromodomain inhibitors, as promising therapeutic candidates. Conclusion Molecular signatures underlying regional vulnerability to AD provide a framework for understanding genetic and systemic factors in disease progression. Findings highlight specific molecular pathways, including ECM-related processes and hormonal regulation, as key drivers of susceptibility. Finally, identified drug repurposing candidates present promising therapeutic avenues for further investigation.