ABSTRACT: Recent microarray studies in the hippocampus of rodents or Alzheimerâs disease (AD) subjects have identified a substantial number of cellular pathways/processes correlated with aging and cognitive decline. However, the temporal relationships among these expression changes or with cognitive impairment have not been studied in depth. Here, using Affymetrix microarrays, immunohistochemistry and Morris water maze cognitive testing across 5 age groups of male F344 rats (n=9-15/group, one microarray per animal), we systematically analyzed the temporal sequence and cellular localization of aging changes in expression. These were correlated with performance scores on the hippocampus-dependent Morris Water Maze task. Significant microarray results were sorted in to Early, Intermediate, Midlife, and Late patterns of expression, and functionally categorized (Early- downregulated neural development, lipid synthesis and energy-utilization; upregulated ribosomal synthesis, growth, stress/inflammatory, lysosome and protein/lipid degradation. Intermediate- increased defense/inflammatory activation and decreased transporter activity; Midlife- downregulated energy-dependent signaling and neurite growth, upregulated astroglial activation, Ca2+-binding, cholesterol/lipid trafficking, myelinogenic processes and additional lysosome/inflammation; Late- further recruitment of genes in already-altered pathways). Immunohistochemistry revealed a primarily astrocytic localization of the processes upregulated in midlife, as well as increased density of myelin proteins. Evidence of cognitive impairment first appeared in the 12-month-old group (midlife) and was increased further in the 23-month-old group, exhibiting the highest correlations with some upregulated genes related to cholesterol transport (e.g., Apoe, Abca2), protein management and ion binding. Some upregulated genes for inflammation (Il6st) and myelinogenesis (Pmp22) also correlated with impairment. Together, the data are consistent with a novel sequential cascade model of brain aging in which metabolic alterations early in maturity are followed by inflammation and midlife activation of an astrocyte-centered cholesterol trafficking pathway that stimulates oligodendrocyte remyelination programs. Importantly, this cholesterol trafficking pathway also may compete for astroglial bioenergetic support of neurons, in turn, leading to downregulation of energy-dependent pathways needed to sustain cognitive functions. Experiment Overall Design: Fisher 344 rats aged 3 (n = 9), 6 (n = 9), 9 (n = 9), 12 (n = 9), or 25 (n = 15; but see below) months were behaviorally characterized on the Morris water maze. Briefly, animals were placed in a circular water bath chamber with a partially submerged platform. Three times per day for three days, animals were placed at randomized starting points with the goal of locating the hidden platform. Generally, animals became better at locating the platform, and by the third day of training, were reliably swimming directly to the platform despite randomized starting loci. On the fourth day, the platform was removed and the degree to which the animals focused their searches on the area that previously contained the platform was measured. On the last day of behavior, animals were given a cue trial in which the platform was raised above the water level so that the rats could see it. Measures taken from the behavior study included path length (how long did the animal swim prior to locating the platform/ platform area), latency to platform in seconds, and velocity (centimeters/ second; swim speed). Two aged animals (E2 and E12) were removed from the study for poor health, reducing the 23 month cohort (n = 13). Twelve to twenty four hours after the last behavioral session, animals were killed and their hippocampi removed. The right hippocampus went to immunohistochemistry for quantification and localization of selected protein products. The left hippocampus was dissected, the CA1 region removed and frozen in dry ice prior to microarray processing. mRNA was extracted from each animal's tissue by standard Affymetrix protocols (see Blalock et al., 2003) and hybridized to individual microarrays (RAE 230A; N = 49 arrays in total). Results were processed with the MAS5 algorithm and subjected to âall probe setâ scaling with a target intensity of 500. Probe sets were filtered for presence (> = 5 presence calls study wide) and gene symbol level annotation prior to statistical analysis. 1-ANOVA across age was used to identify aging-related probe sets/genes (p <= 0.05). Aging-related genes were subjected to post hoc template correlation to determine the pattern of age-related change, and further, within the 12 and 23 month old groups, Pearson's correlation with behavioral scores was used to identify genes whose expression levels correlated significantly (p <= 0.05) with behavior. Twelve and 23 month old groups were chosen for this correlation analysis as they showed the greatest variability among similarly aged animals, suggesting that these ages show important cognitive transition phases during aging-related cognitive decline. Functional grouping analysis was performed using the DAVID suite of on-line tools.
