Unexpected partial correction of metabolic and behavioral phenotypes of Alzheimer’s APP/PSEN1 mice by gene targeting of diabetes/Alzheimer’s-related Sorcs1

Elysse M. Knight, Henry H. Ruiz, Soong Ho Kim, Jessica C. Harte, Wilson Hsieh, Charles Glabe, William L. Klein, Alan D. Attie, Christoph Buettner, Michelle E. Ehrlich, Sam Gandy

Insulin resistance and type 2 diabetes mellitus (T2D) are associated with increased risk for cognitive impairment, Alzheimer's disease (AD) and vascular dementia. SORCS1 encodes a protein-sorting molecule genetically linked to both T2D and AD. The association of SORCS1 with both AD and T2D is sexually dimorphic in humans, with both disease associations showing more robust effects in females. Based on published evidence that manipulation of the mouse genome combining multiple genes related to cerebral amyloidosis, to T2D, or both, might provide novel mouse models with exacerbated amyloid and/or diabetes phenotypes, we assessed memory, glucose homeostasis, and brain biochemistry and pathology in male and female wild-type, Sorcs1 -/-, APP/PSEN1, and Sorcs1 -/- X APP/PSEN1 mice. Male mice with either the APP/PSEN1 or Sorcs1 -/- genotype displayed earlier onset and persistent impairment in both learning behavior and glucose homeostasis. Unlike prior examples in the literature, the behavioral and metabolic abnormalities in male mice were not significantly exacerbated when the two disease model mice (Sorcs1 -/- models T2D; APP/PSEN1 models AD) were crossed. However, female Sorcs1 -/- X APP/PSEN1 mice exhibited worse metabolic dysfunction than Sorcs1 -/- knockout mice and worse memory than wild-type mice. The deletion of Sorcs1 from APP/PSEN1 mutant mice led to no obvious changes in brain levels of total or oligomeric amyloid-beta (Aβ) peptide. In general, unexpectedly, there was a trend for gene targeting of Sorcs1-/- to partially mitigate, not exacerbate, the metabolic and amyloid pathologies. These results indicate that crossing AD model mice and T2D model mice may not always cause exacerbation of both the amyloidosis phenotype and the metabolic phenotype and highlight the unexpected pitfalls of creating mixed models of disease.