Genetic Substrates of Brain Vulnerability and Resilience in APOE2 Mice Transitioning from Midlife to Old Age

The interplay between genotype, age, and sex may play an important role in determining successful versus pathological aging. Predicting brain networks vulnerable to age associated decline, or conversely identifying brain networks and associated molecular pathways that promote resilience will inform on treatment and prevention strategies. APOE allelic variation modulate brain vulnerability and cognitive resilience during aging and Alzheimer′s disease (AD). The APOE4 allele confers the most risk and has been intensely studied with respect to the control APOE3 allele. The APOE2 allele has been less studied, and the mechanisms by which it confers cognitive resilience and neuroprotection remain largely unknown. Mouse models with targeted replacement of the murine APOE gene with the human major APOE2 alleles provide useful tools to address such questions. Our goals were to identify changes during a critical period of middle to old age transition, and to examine whether female sex confers specific vulnerability. Our results support that age, but not female sex was important in modulating learning and memory estimates based on Morris water maze metrics, including the total distance swam to a target, the distance in the target quadrant, and the absolute winding number characterizing the shape of the swim trajectory. Diffusion weighted MRI provided estimates for atrophy, microstructural changes through fractional anisotropy, and cellularity changes through the return to origin probability from ensemble average diffusion propagators. We detected a small but significant 3% global brain atrophy due to aging, reflected by regional atrophy in the hippocampal commissure (15%), fornix, and cingulate cortex 24 (9%). Females had larger regional volumes relative to males for the bed nucleus of stria terminalis, subbrachial nucleus, postsubiculum (≈10%), and claustrum (>5%), while males had larger volumes for the brachium of the superior colliculus, orbitofrontal cortex, frontal association cortex, and the longitudinal fasciculus of pons (>9%). Age promoted atrophy in both white matter (anterior commissure, corpus callosum, etc.), and gray matter, in particular in olfactory cortex, frontal association area 3, thalamus, hippocampus and cerebellum. A negative age by sex interaction was noted for the olfactory areas, piriform cortex, amygdala, ventral hippocampus, entorhinal cortex, and cerebellum, suggesting faster decline in females. We used vertex screening to find associations between connectome and traits, and sparse multiple canonical correlation analysis to integrate our analyses over connectomes, traits, and RNA-seq. Brain subgraphs favored in males included the secondary motor cortex and superior cerebellar peduncle, while those for females included hippocampus and primary somatosensory cortex. Connectivity loss during transition from middle to old age affected the hippocampus and primary somatosensory cortex. We validated these subgraphs using neural networks showing increased accuracy for sex prediction from 81.9% when using the whole connectome, to 94.28% when using the subgraph estimated through vertex screening. We hypothesized that blood gene expression may help track brain and behavior changes with age and sex. The largest fold change for age related genes was for Cpt1c (log2FC = 7.1), involved in oxidation, transport of long-chain fatty acids into mitochondria, lipid metabolic processes. Arg1, a critical regulator of innate and adaptive immune responses (log2FC = 4.9) also showed age specific changes. Amongst the sex related genes, the largest FC were observed for Maoa (log2FC = 4.9) involved in the degradation of the neurotransmitters serotonin, epinephrine, norepinephrine, and dopamine, and implicated in response to stress. Four genes were common for age and sex related vulnerability: Myo1e (log2FC = −1.5), Creld2 (log2FC = 1.4), Ptprt (log2FC = 2.9), and Pex1 (log2FC = 3.6). Genes with the highest weight after connectome filtering included Ankzfp1, Pex1, Cep250, Nat14, Arg1, and Rangrf. Connectome filtered genes pointed to pathways relate to stress response, transport, and metabolic processes. We propose modeling approaches using sparse canonical correlation analysis to relate quantitative traits to vulnerable brain networks, and blood markers for biological processes. Our study investigated the APOE2 impact on neurocognition, brain networks, and biological pathways during aging. Identifying such brain and gene networks may help reveal targets for therapies that support successful aging. ### Competing Interest Statement The authors have declared no competing interest.