The causes of cerebral accumulation of amyloid beta-protein (Abeta) in most cases of Alzheimer's disease (AD) remain unknown. We recently found that homozygous deletion of the insulin-degrading enzyme (IDE) gene in mice results in an early and marked elevation of cerebral Abeta. Both genetic linkage and allelic association in the IDE region of chromosome 10 have been reported in families with late-onset AD. For IDE to remain a valid candidate gene for late-onset AD on functional grounds, it must be shown that partial loss of function of IDE can still alter Abeta degradation, but without causing early, severe elevation of brain Abeta. Here, we show that naturally occurring IDE missense mutations in a well-characterized rat model of type 2 diabetes mellitus (DM2) result in decreased catalytic efficiency and a significant approximately 15 to 30% deficit in the degradation of both insulin and Abeta. Endogenously secreted Abeta(40) and Abeta(42) are significantly elevated in primary neuronal cultures from animals with the IDE mutations, but there is no increase in steady-state levels of rodent Abeta in the brain up to age 14 months. We conclude that naturally occurring, partial loss-of-function mutations in IDE sufficient to cause DM2 also impair neuronal regulation of Abeta levels, but the brain can apparently compensate for the partial deficit during the life span of the rat. Our findings have relevance for the emerging genetic evidence suggesting that IDE may be a late-onset AD-risk gene, and for the epidemiological relationships among hyperinsulinemia, DM2, and AD.