Alternative splicing of human insulin-degrading enzyme yields a novel isoform with a decreased ability to degrade insulin and amyloid beta-protein.

Deletion of insulin-degrading enzyme (IDE) in mice causes accumulation of cerebral amyloid beta-protein (Abeta), hyperinsulinemia, and glucose intolerance. Together with genetic linkage and allelic association of IDE to Alzheimer's disease (AD) and type 2 diabetes mellitus (DM2), these findings suggest that IDE hypofunction could mediate human disease. To date, no coding mutations have been found in the canonical isoform of IDE, suggesting that pathological mutations could exist in undiscovered exons or regulatory regions, including untranslated regions (UTRs). However, neither isoforms arising from alternative splicing nor the UTRs have been described. Here, we systematically characterize human IDE mRNAs, identify a novel splice form, and compare its subcellular distribution, kinetic properties, and ability to degrade Abeta to the known isoform. Six distinct human IDE transcripts were identified, with most of the variance attributable to alternative polyadenylation sites. In the novel spliceoform, an exon we designate "15b" replaces the canonical exon "15a", and the resultant variant is widely expressed. Subcellular fractionation, immunofluorescent confocal microscopy, and immunogold-electron microscopy reveal that the 15b-IDE protein occurs in both cytosol and mitochondria. Organelle targeting of both isoforms is determined by which of two translation start sites is used, and only those isoforms utilizing the second site regulate levels of secreted Abeta. 15b-IDE can exist as a heterodimer with the 15a isoform or as a homodimer. The apparent K(m) values of recombinant 15b-IDE for both insulin and Abeta are significantly higher and the k(cat) and catalytic efficiency markedly lower than those of 15a-IDE. In accord, cells coexpressing beta-amyloid precursor protein (APP) and 15b-IDE accumulated significantly more Abeta in their media than those expressing APP and 15a-IDE. Our results identify a novel, catalytically inefficient form of IDE expressed in brain and non-neural tissues and recommend novel regions of the IDE gene in which to search for mutations predisposing patients to AD and DM2.