We previously showed that beta-amyloid precursor protein (APP) is cleaved not only in the middle of the membrane (gamma-cleavage) but also at novel cleavage sites close to the membrane/cytoplasmic boundary (epsilon-cleavage), releasing APP intracellular domains (AICDs) 49-99 and 50-99. To learn more about the relationship between gamma- and epsilon-cleavage, C-terminally truncated carboxyl-terminal fragments (CTFs) of APP, especially CTFs1-48 and 1-49 (the postulated products that are generated by epsilon-cleavage), were transiently expressed in CHO cells. Most importantly, the cells expressing CTF1-49 secreted predominantly amyloid beta-protein (Abeta) 40, while those expressing CTF1-48 secreted preferentially Abeta42. This supports our assumption that epsilon-cleavage precedes Alphabeta production and that preceding epsilon-cleavage determines the preference for the final Abeta species. The gamma-secretase inhibitors, L-685,458 and DAPT, suppressed Abeta production from CTF1-49. Regarding Abeta production from CTF1-48, L-685,458 suppressed it, but DAPT failed to do so. A dominant negative mutant of presenilin 1 suppressed the production of Abeta40 and 42 from both CTFs1-48 and 1-49. These data should shed significant light into the mechanism of Abeta production.