Shared genetic components between brain regions identify potential therapeutic targets for Alzheimer’s disease by integrative proteomic analysis

Background Genome‐wide association studies (GWAS) have identified >30 genetic loci associated with Alzheimer’s disease (AD). However, how those variants impact protein expression in different brain regions remain elusive. Large‐scale quantitative proteomic datasets of AD postmortem brain tissues have become available with the breakthrough of high‐throughput mass spectrometry‐based technologies. Thus, we utilized these datasets to understand brain region‐specific molecular pathways underlying AD pathogenesis and their potential drug targets. Method We extended our multi‐omics network‐based tool, Edge‐Weighted Dense Module Search of GWAS (EW_dmGWAS), to integrate genetic variants with brain region‐specific proteomic profiles. We performed this analysis by leveraging publicly available AD GWAS statistics of 472,868 individuals (Schwartzentruber et al., 2021) and protein expression profiles covering 190 parahippocampal gyrus (PHG) samples from the Mount Sinai Brain Bank (MSBB) and 192 dorsolateral prefrontal cortex (DLPFC) samples from the Religious Orders Study and Memory and Aging Project (ROSMAP). The resulting modules were evaluated using a scale‐free network index. We highlighted key overlapped genes between region‐specific top module genes (TMGs) and further examined region‐specific differentially co‐expressed genes to identify investigational AD drug targets. Gene set enrichment analyses were performed to assess the functions and cell‐type‐specificity of region‐specific TMGs. Result The EW_dmGWAS analyses prioritized 52 TMGs from PHG and 58 TMGs from DLPFC. The TMGs in both regions were found to be enriched in excitatory neurons through cell‐type‐specific enrichment analysis (combined p‐value = 0.03 in PHG; combined p‐value = 0.09 in DLPFC). We found four key overlapped genes with high genetic risk (gene‐level FDR < 0.05), CLU, PICALM, PRRC2A, and NDUFS3. In addition, the differentially co‐expressed drug target analysis pinpointed three potentially drug‐targetable genes, APP, SNCA, and VCAM1, in PHG but none in DLPFC. Additional enrichment analyses of TMGs revealed region‐specific biological processes and drug signatures, including disulfiram in PHG (‐log2FDR = 9.5). Conclusion Our network‐based analyses of PHG and DLPFC proteomic profiles prioritized potential key shared genetic factors, which were further proved to play possible roles in the excitatory neurons in AD pathology. Our drug signature analyses validated three known investigational drugs tar