Incorporating genetic diversity to improve alignment of mouse models to human Alzheimer’s disease

Abstract

Previous work by our labs, and others, reveal the benefits of incorporating diverse mouse strains to better model human Alzheimer's disease (AD). For instance, mouse models of AD on genetically distinct wild-derived mouse strains demonstrated robust differences in immune response to amyloid and neurodegeneration, better representing the phenotypic spectrum of AD than previous models. To further investigate the importance of genetic diversity, as part of MODEL-AD, we generated AD-relevant mouse models on Collaborative Cross (CC) lines background, a recombinant inbred mouse panel created from eight highly diverse founder strains.C57BL/6J (B6J) mice homozygous for humanized APOE4 allele and carrying mutant APP and PS1 alleles (APPswe , PS1de9 ) were crossed to five CC strains that were selected based on variation in AD-relevant genes. Cohorts of male and female B6JCCnF1.APOE4.APP/PS1 mice and controls were aged to 8 months (a mid-point for amyloid deposition in B6.APP/PS1 mice) and brain hemispheres processed for RNA-Seq and neuropathological assessment. Expectation Maximization for Allele Specific Expression (EMASE) tool was used to quantify transcriptome-wide expression at allelic level. Differential expression and KEGG pathway enrichment analysis were performed to determine altered molecular processes in CC strains. Strain specific transcriptional signatures were mapped to co-expression gene modules of human Late-Onset AD (LOAD) from the AMP-AD consortium. Neuropathological assessment included characterization of plaque deposition, Tau pathology, glial activation, neurodegeneration and cerebrovascular health.RNA-Seq analysis revealed up-regulation of neuroinflammation related pathways (lysosome, complement and coagulation cascades) in all CC strains as an effect of mutant APP and PS1 transgenes in presence of APOE4 allele. On the other hand, we observed significant CC strain and sex specific variation in immune (Fc gamma R-mediated phagocytosis) and metabolic (glycosphingolipid biosynthesis) response to amyloid pathogenesis. Alignment with human LOAD signatures revealed high strain and sex specific variation in correlation with synaptic signaling, cell cycle and DNA repair related gene modules. Overall, novel AD mouse models that incorporated CC lines demonstrated increased alignment to human AD-like transcriptomic signatures compared to B6J.This work builds on previous work highlighting the strengths of including multiple diverse mouse strains to better model complex neurodegenerative disorders such as AD and related dementias.