Excessive cerebral accumulation of the 42-residue amyloid ␤-protein (A␤) is an early and invariant step in the pathogenesis of Alzheimer's disease. Many studies have examined the cellular production of A␤ from its membrane-bound precursor, including the role of the presenilin proteins therein, but almost nothing is known about how A␤ is degraded and cleared following its secretion. We previously screened neuronal and nonneuronal cell lines for the production of proteases capable of degrading naturally secreted A␤ under biologically relevant conditions and concentrations. The major such protease identified was a metalloprotease released particularly by a microglial cell line, BV-2. We have now purified and characterized the protease and find that it is indistinguishable from insulin-degrading enzyme (IDE), a thiol metalloendopeptidase that degrades small peptides such as insulin, glucagon, and atrial natriuretic peptide. Degradation of both endogenous and synthetic A␤ at picomolar to nanomolar concentrations was completely inhibited by the competitive IDE substrate, insulin, and by two other IDE inhibitors. Immunodepletion of conditioned medium with an IDE antibody removed its A␤-degrading activity. IDE was present in BV-2 cytosol, as expected, but was also released into the medium by intact, healthy cells. To confirm the extracellular occurrence of IDE in vivo, we identified intact IDE in human cerebrospinal fluid of both normal and Alzheimer subjects. In addition to its ability to degrade A␤, IDE activity was unexpectedly found be associated with a time-dependent oligomerization of synthetic A␤ at physiological levels in the conditioned media of cultured cells; this process, which may be initiated by IDE-generated proteolytic fragments of A␤, was prevented by three different IDE inhibitors. We conclude that a principal protease capable of down-regulating the levels of secreted A␤ extracellularly is IDE.Converging lines of evidence support the hypothesis that progressive cerebral accumulation of the 40 -42-residue amyloid ␤-proteins (A␤s) 1 is an early, invariant, and necessary step in the pathogenesis of Alzheimer's disease (AD). As a result, there is growing interest in decreasing cerebral A␤ levels as a therapeutic and preventative approach to the disease. A␤ is generated by endoproteolysis of the ␤-amyloid precursor protein (APP) and secreted constitutively by most mammalian cells throughout life. Whereas many studies have examined the proteolytic processing of APP and the mechanisms of A␤ production, almost nothing is known about how A␤ peptides are normally degraded and cleared following their secretion. We recently screened the conditioned media of several different cell lines for A␤-degrading activity and found that the principal such activity was conferred by a nonmatrix metalloprotease that was released by microglial cells and other cells and efficiently degraded both endogenous and synthetic A␤ (1). The release of the protease from microglial cells was augmented by activating the cells with lipopolysa...

Filamentous aggregates of the 40-42-residue amyloid beta-protein (A beta) accumulate progressively in the limbic and cerebral cortex in Alzheimer's disease, where they are intimately associated with neuronal and glial cytopathology. Attempts to model this cytotoxicity in vitro using synthetic peptides have shown that monomeric A beta is relatively inert, whereas aggregated A beta reproducibly exerts a variety of neurotoxic effects. The processes that mediate the conversion of monomeric A beta into a toxic aggregated state are thus of great interest. Previous studies of this conversion have employed high concentrations (10(-5)-10(-3) M) of synthetic A beta peptides under nonbiological conditions. We report here the detection of small amounts (< 10(-9) M) of SDS-stable A beta oligomers in the culture media of Chinese hamster ovary cells expressing endogenous or transfected amyloid beta-protein precursor genes. The identity of these oligomers (primarily dimers and trimers) was established by immunoprecipitation with a panel of A beta antibodies, by electrophoretic comigration with synthetic A beta oligomers, and by amino acid sequencing. The oligomeric A beta species comprised approximately 10-20% of the total immunoprecipitable A beta in these cultures. A truncated A beta species beginning at Arg 5 was enriched in the oligomers, suggesting that amino-terminal heterogeneity can influence A beta oligomerization in this system. Addition of Congo red (10 microM) during metabolic labeling of the cells led to increased monomeric and decreased oligomeric A beta. The ability to detect and quantitate oligomers of secreted A beta peptides in cell culture should facilitate dynamic studies of the critical process of initial A beta aggregation under physiological conditions.

The progressive aggregation and deposition of amyloid beta-protein (Abeta) in brain regions subserving memory and cognition is an early and invariant feature of Alzheimer's disease, the most common cause of cognitive failure in aged humans. Inhibiting Abeta aggregation is therapeutically attractive because this process is believed to be an exclusively pathological event. Whereas many studies have examined the aggregation of synthetic Abeta peptides under nonphysiological conditions and concentrations, we have detected and characterized the oligomerization of naturally secreted Abeta at nanomolar levels in cultures of APP-expressing CHO cells [Podlisny, M. B., Ostaszewski, B. L., Squazzo, S. L., Koo, E. H., Rydell, R. E., Teplow, D. B., and Selkoe, D. J. (1995) J. Biol. Chem. 270, 9564-9570 (1); Podlisny, M. B., Walsh, D. M., Amarante, P., Ostaszewski, B. L., Stimson, E. R., Maggio, J. E., Teplow, D. B., and Selkoe, D. J. (1998) Biochemistry 37, 3602-3611 (2)]. To determine whether similar species occur in vivo, we probed samples of human cerebrospinal fluid (CSF) and detected SDS-stable dimers of Abeta in some subjects. Incubation of CSF or of CHO conditioned medium at 37 degrees C did not lead to new oligomer formation. This inability to induce oligomers extracellularly as well as the detection of oligomers in cell medium very early during the course of pulse-chase experiments suggested that natural Abeta oligomers might first form intracellularly. We therefore searched for and detected intracellular Abeta oligomers, principally dimers, in primary human neurons and in neuronal and nonneural cell lines. These dimers arose intracellularly rather than being derived from the medium by reuptake. The dimers were particularly detectable in neural cells: the ratio of intracellular to extracellular oligomers was much higher in brain-derived than nonbrain cells. We conclude that the pathogenically critical process of Abeta oligomerization begins intraneuronally.