There is growing evidence suggesting that cholesterol metabolism is linked to susceptibility to Alzheimer's disease by influencing amyloid -protein (A) metabolism. However, the precise cellular linkage sites between cholesterol and A have not yet been clarified. To address this issue, we investigated Niemann-Pick type C (NPC) model cells and NPC mutant cells, which showed aberrant cholesterol trafficking. We observed a remarkable A accumulation in late endosomes of both NPC model cells and mutant cells where cholesterol accumulates and a significant accumulation in the NPC mouse brain. This A accumulation was independent of its constitutive secretion and production through an endocytic pathway. In addition, it is characterized by a marked predominance of A42 and insolubility in SDS, suggesting the presence of aggregated A in late endosomes. Most importantly, A accumulation is coupled with the cholesterol levels in late endosomes. Thus, late endosomes of NPC cells are a novel pool of aggregated A42 as well as cholesterol, suggesting a direct interaction between aggregated A and cholesterol.Formation of senile plaques composed of amyloid -protein (A) 1 is one of the hallmarks of Alzheimer's disease (AD) (1). A, a small protein of M r ϳ4,000, is cleaved out sequentially with -and ␥-secretases from a large membrane protein called -amyloid precursor protein (APP), and secreted into the extracellular space. There are two major species of A that are defined by their C-terminal lengths; A40 terminates at Val-40, and A42 terminates at Ala-42. The latter has a much stronger tendency to aggregate into fibrils and accounts for less than 10% of secreted A. However, A42 is the most predominant form found in senile plaques and is considered the initial species to be deposited in aged and AD brains (2, 3).Three causative genes, APP, presenilin 1 (PS1), and PS2 have been identified in early onset familial AD (1), and their mutations appear to converge on an increased production of A42. However, the pathological mechanisms of sporadic AD, explaining more than 90% of AD patients, remain unknown. Interestingly, there is growing evidence for a significant linkage between A and cholesterol metabolism. First, cholesterol loading or depletion affects A generation both in vitro (4,5) and in vivo (6). Second, the levels of total cholesterol and LDL-cholesterol, but not those of HDL-cholesterol, in the serum correlate with the amount of A42 in AD brains (7). Third, aggregated A preferentially binds to cholesterol in vitro (8). Fourth, a substantial fraction of intracellular A is localized in the detergent-insoluble membrane domain that is rich in glycosphingolipid and cholesterol (9, 10). Fifth, the presence of an E4 isoform of apolipoprotein E, an essential molecule for cholesterol metabolism, is a strong risk factor for developing AD (11). Sixth, high dietary cholesterol accelerates AD-related pathologies, including A deposition, in a transgenic mouse model (12). However, no detailed mechanisms for clarifying ...