The distinct effects of BACE-2 on Aβ and CTF species starting either at Asp1 or at Glu11 suggest that these BACE-1-mediated cleavages occur sequentially. To address this question, cells were labeled with [35S]methionine in the presence or absence of inhibitors of protein forward transport. We used BFA to inhibit protein transport from the endoplasmic reticulum to Golgi compartments or monensin to inhibit transport from the Golgi to later secretory compartments (Haass et al. 1993; Wild-Bode et al. 1997). Holo-βAPP and its proteolytic processing products were isolated from cell lysates and conditioned media. Immunoprecipitation of intracellular βAPP with antibody 5313 directed against the βAPP ectodomain revealed two major variants of βAPP in untreated cells, representing immature and mature βAPP in both BACE-1 and BACE-2 expressing cells (Fig. 6a, left panels). BFA and monensin treatment led to the detection of βAPP variants migrating between mature and immature βAPP due to aberrant glycosylation (Haass et al. 1993; Wild-Bode et al. 1997). BFA and monensin also blocked secretion of APPs in all cell lines analyzed (data not shown), indicating inhibition of protein forward transport. In BACE-1-transfected cells, BFA treatment induced the accumulation of an additional intracellular βAPP derivative. This species was detected by the neoepitope-specific antibody 192wt that selectively recognizes APPs-β but not holo-βAPP [Seubert et al. 1993; Fig. 6(a), upper right panel]. Moreover, antibody 6E10 directed against amino acids 1–16 of the Aβ domain did not recognize this species (data not shown), indicating that βAPP is selectively cleaved at the β-site within the endoplasmic reticulum. Monensin treatment led to an increased accumulation of intracellular APPs-β recognized by antibody 192wt (Fig. 6a). These results indicate that BACE-1 can cleave βAPP at the β-site in early secretory compartments, namely in the endoplasmic reticulum and Golgi. Intracellular APPs-β was not observed in BACE-2-expressing cells (Fig. 6a, lower panels), consistent with the distinct specificity of BACE-2 in cleavage of βAPP between Phe19 and Phe20. We therefore analyzed the levels of intracellular APPs-α/α′ in BACE-1- or BACE-2-expressing cells. Only low amounts of intracellular APPs-α/α′ were detected in untreated cells expressing BACE-1 (Fig. 6b, upper panel), because of predominant cleavage of βAPP at the β-site and residual α-secretase cleavage at or close to the cell surface followed by rapid secretion of APPs-α/α′ (Sisodia et al. 1990). Significantly higher levels of intracellular APPs-α/α′ were detected in untreated cells expressing BACE-2 (Fig. 6b, lower panel). Cell treatment with BFA strongly inhibited production of intracellular APPs-α/α′, also indicating that cleavage of βAPP by BACE-2 predominantly occurred in post-endoplasmic reticulum compartments. Indeed, the detection of high amounts of APPs-α/α′ in cells treated with monensin demonstrated that BACE-2 can efficiently cleave βAPP in the Golgi compartment. To prove this, we next analyzed the levels of βAPP CTFs. In BACE-1-expressing cells, CTFAsp1 and predominantly CTFGlu11 were detected (Fig. 6c, left panel). BFA treatment strongly increased the ratio of CTFAsp1/CTFGlu11. While the generation of CTFGlu11 was strongly reduced, significant amounts of CTFAsp1 were produced. Again, these data indicate that BACE-1-mediated cleavage of βAPP at Asp1 can occur in the endoplasmic reticulum, while cleavage at Glu11 is very inefficient in this compartment. By using βAPP bearing the K670N/M671L double mutation (‘Swedish mutation’), we also observed cleavage at Asp1 within the endoplasmic reticulum even without overexpression of BACE-1, therefore indicating endogenous β-secretase activity in this compartment (data not shown). Interestingly, treatment of BACE-1-expressing cells with monensin allowed generation of CTFGlu11, indicating that this cleavage can occur in the Golgi (Fig. 6c, left panel). In BACE-2-transfected cells, BFA strongly inhibited the cleavage of βAPP as indicated by the significant reduction of CTF-α/CTFPhe20 (Fig. 6c, right panel). However, treatment with monensin allows generation of βAPP CTF-α/CTFPhe20 to some extent, indicating that BACE-2 can cleave βAPP within the Golgi apparatus. Taken together these results demonstrate that cleavage of βAPP by BACE-2 between Phe19 and Phe 20 and by BACE-1 between Tyr10 and Glu11 can occur within the same subcellular compartments, namely the Golgi and later secretory compartments, while cleavage of the Met–Asp bond at the beginning the Aβ domain mediated by BACE-1 can also occur earlier within the endoplasmic reticulum. Pulse–chase analysis of βAPP containing the K670N/M671L ‘Swedish’ double mutation revealed that CTFAsp1 was generated earlier as CTFGlu11, also indicating sequential cleavages of βAPP by BACE-1 first at Asp1 and then at Glu11 (Fig. 6d).