INTRODUCTION

Amyloid -peptide is now widely believed to play a major role in AD¹ (1-3). A is a heterogeneous peptide derived from proteolytic processing of the -amyloid precursor protein (APP) by -secretase (cleaving at the N terminus) and -secretase (cleaving at the C terminus) (3). Heterogeneity of A occurs at the N terminus (4-11) as well as the C terminus (3, 12). Heterogeneity at both ends of the peptides is known to affect the toxic properties of A peptides. N-terminal-deleted peptides aggregate more rapidly and exhibit enhanced neurotoxicity (6). Moreover, cells expressing the FAD-linked APP A692G mutation close to the -secretase site produce elevated levels of A alternatively cleaved at the N terminus (9). At the C terminus A peptides terminate predominantly at amino acid 40. However, approximately 10% of the peptides are elongated by two amino acids (12, 13). Analysis of the three FAD-associated APP mutations (V717I/G/F; Ref. 3) located close to the -secretase cleavage sites showed that these mutations result in an enhanced production of elongated A (12). Moreover, mutations in the PS genes, which are responsible for approximately 30-40% of all FAD cases (14, 15), also cause elevated levels of the long form of A (16). Furthermore, peptides ending at amino acid 42 are preferentially deposited within senile plaques of patients with sporadic AD, Down's syndrome, as well as FAD (3, 17, 18). Therefore, A peptides terminating at amino acid 42 appear to be very important factors playing a pivotal role in FAD and AD (13). In contrast to the generation of A terminating at amino acid 40, which predominantly occurs at or close to the cell surface (3, 8, 19), very little is known about the subcellular localization of the -secretase cleavage resulting in the generation of A peptides ending at amino acid 42. Since this proteolytic cleavage is very important for initiating the pathological events finally resulting in AD as well as FAD detailed knowledge about its cellular mechanism is required for potential therapeutic strategies involving inhibition of A42 production.

EXPERIMENTAL PROCEDURES

Kidney 293 cells stably transfected with wt APP or the Val/Gly mutation were generated and grown as described (7, 8). Cytoplasmic deletions of APP are described by Haass et al. (8).

Fresh Dulbecco's modified Eagle's medium (high glucose) medium (Life Technologies, Inc.) containing 10% fetal bovine serum (Life Technologies, Inc.), 1% glutamine, and 1% penicillin/streptomycin was added to a confluent monolayer of cells in a 15-cm culture dish. Conditioned media were collected after 18 h. Cell lysates were prepared in 1 ml of lysis buffer as described (20), insoluble material was pelleted by centrifugation at 100,000 × g for 1 h at 4 °C. Supernatants were analyzed by immunoprecipitation or ELISA. In addition the insoluble pellet was re-extracted with formic acid and analyzed by ELISA as well. Extraction of the insoluble pellet revealed additional A. The ratio of A40 to A42 in the formic acid extracted material was very similar to Nonidet P-40-extracted A species. Treatment with brefeldin A (BFA) (4 h), monensin (18 h), and NH₄Cl (18 h) was performed as described (8, 21). Control cultures were incubated under identical conditions.

The highly specific antibodies as well as the ELISA assay used for detection of A1-40, Ax-40, A1-42, and Ax-42 are described by Shinkai et al. (22), Yamatsuji et al. (23), and Suzuki et al. (12). The monoclonal antibodies BNT77 (generated against synthetic A11-28; the epitope is located in A11-16) or BAN50 (generated against synthetic A1-16 (22)) were used as a capture antibody. The specificity of the detecting antibodies BA27 and BC05 was described previously (12).

A postnuclear supernatant was prepared and separated by density centrifugation as described by Leimer et al. (24).

Cells were metabolically labeled as described (7). Cell lysates were immunoprecipitated as described (7) with antibody 3926 (raised to A1-42), antibody BC05, or antibody C7 (raised to the C terminus of APP; Ref. 7).

Quantitation of APP Expression

Cell lysates from metabolically labeled kidney 293 cells stably expressing wt APP or APP with the Val/Gly mutation were immunoprecipitated with antibody C7 (8) raised to the last 20 C-terminal amino acids of APP. Immunoprecipitates were separated on 10% SDS-polyacrylamid gels. Immunoprecipitated full-length APP (N- and N/O-glycosylated APP) was quantitated by phosphorimaging.

RESULTS

To identify intracellular A a postnuclear supernatant from metabolically labeled kidney 293 cells transfected with wt APP was separated by density gradient centrifugation. We then immunoprecipitated potential intracellular A with the polyclonal antibody 3926, which recognizes all species of A, including p3. Interestingly, we observed intracellular A but no p3 in fractions 9-14 (Fig. 1). The lack of p3 is consistent with the knowledge that the -secretase cleavage generating this peptide occurs at or close to the cell surface, which results in the immediate secretion of p3 (8). The presence of intracellular A suggests its production early within the secretory pathway. However, this experiment does not discriminate A terminating at amino acid 40 (A40) or amino acid 42 (A42).

Separation of a postnuclear supernatant obtained from metabolically labeled kidney 293 cells stably transfected with wt APP on a linear 15-60% sucrose gradient.

To discriminate A40 from A42 species we monitored intracellular as well as secreted A terminating at amino acid 42 or amino acid 40 by using a previously established ELISA assay (12, 22, 23). To detect all species of A generated, we specifically detected A peptides beginning with Asp¹ (A1-40, A1-42) but also A peptides beginning at alternative cleavage sites (Ax-40, Ax-42) which are abundant in kidney 293 cells and might play an important role during the disease (4-11). We first wanted to determine if the ELISA can be used to detect A1-42/Ax-42 or A1-40/Ax-40. As shown in Fig. 2, A peptides terminating at amino acid 42 are indeed detected within lysates of cells stably transfected with wt APP. To prove if we indeed detect intracellular A42, we analyzed cell lysates and conditioned media from cells stably transfected with the V717G mutation, which is known to cause an enhanced production of A 42 (3, 12). Indeed, much higher amounts of A peptides terminating at amino acid 42 were detected in cell lysates from cells stably expressing the APP Val/Gly mutation (Fig. 2). This is consistent with data published previously showing that mutations at position 717 of APP increase the levels of A42 secreted into culture media (12). These results therefore support the specificity of the ELISA for measuring intracellular A42. A terminating at amino acid 40 appears to be under the detection limit of the ELISA (Fig. 2), which might be due to the fact that it is secreted immediately after -secretase cleavage on the cell surface (8). Alternatively, A40 could be present in a biochemically modified form or bound to another protein, which inhibits its detection using our ELISA or extraction protocol. In conditioned media all species of A peptides were detected (Fig. 2). Again, cells transfected with the APP Val/Gly mutation secreted more A peptides (A1-42 as well as Ax-42) terminating at amino acid 42 as compared with wt transfected cells (Fig. 2), which is consistent with data published by Suzuki et al. (12). It is interesting to note that the increased secretion of A species terminating at amino acid 42 by cells transfected with the APP Val/Gly mutation appears to be compensated partially by reduced levels of A40 species (Fig. 2). In conditioned media higher levels of A peptides terminating at amino acid 40 as of those terminating at amino acid 42 were detected ruling out that the ELISA is particularly sensitive for the detection of A42. Therefore, although higher levels of total A terminating at amino acid 40 are detected within conditioned media only Ax-42/A1-42 could be demonstrated within cell lysates indicating that the proteolytic cleavages after amino acid 40 and amino acid 42 occur by different cellular mechanisms. Since the cell lines expressing the APP Val/Gly mutation produced significantly elevated levels of A terminating at amino acid 42, these cells were used to determine the subcellular site of the -secretase cleavage at amino acid 42. 

Intracellular detection of A1-42/x-42.

The detection of intracellular A (Figs. 1 and 2) indicates that at least a subpopulation of A peptides are generated early during protein trafficking. We therefore accumulated APP within the ER by treating cells with BFA and searched for intracellular as well as secreted A species. Consistent with previous results (8, 25), secretion of all types of A was strongly reduced (Fig. 3A). In cell lysates BFA did not inhibit intracellular production of A1-42/x-42 but even caused an increase of these peptides (Fig. 3A). In contrast, BFA treatment did not cause a detectable accumulation of intracellular A1-40 or Ax-40 (Fig. 3A). To prove intracellular accumulation of A42 peptides by an independent method, kidney 293 cells stably expressing APP Val/Gly were metabolically labeled in the presence and absence of BFA, and cell lysates were immunoprecipitated with antibody BC05. As shown in Fig. 3B, A terminating at amino acid 42 accumulated after BFA treatment within cell lysates. Moreover, consistent with the data in Fig. 1 we did not detect p3, although large amounts of p3 terminating at amino acid 42 are known to be produced in kidney 293 cells (26). This is again consistent with the finding that the -secretase cleavage generating p3 occurs predominantly at or close to the cell surface (8). Moreover, the immunoprecipitation independently proves the accumulation of intracellular A peptides terminating after amino acid 42.

Accumulation of APP within the ER results in an increased generation of intracellular A terminating at amino acid 42.

We also treated cells with monensin, which is known to inhibit APP trafficking late within the secretory pathway (21). As BFA, monensin blocks secretion of all types of A (Fig. 4), which is consistent with previous findings (8, 21). However, monensin causes a significant intracellular accumulation of Ax-42 as well as Ax-40 (Fig. 4). Due to the prolonged treatment with monensin (18 h versus 4 h with BFA), the total amount of both peptides is significantly higher as compared with BFA treatment (Fig. 3A). Since Ax-40 is not increased upon BFA treatment this indicates that the -secretase cleavage after amino acid 40 can occur within the Golgi but not within the ER. The latter observation is consistent with the data presented by Koo and Squazzo (19) and Gabuzda et al. (21) showing that small amounts of A peptides are generated prior to reinternalization. Based on our data, these peptides appear to represent alternatively cleaved A peptides. The data in Figs. 3 and 4 suggest that the proteolytic cleavage at amino acid 42 can occur at least to some extend within the ER and the Golgi network.

Inhibition of APP transport through the Golgi results in a preferential accumulation of intracellular Ax-40 and Ax-42.

To further prove that endosomal/lysosomal processing of APP is not a prerequisite for A42 formation we treated cells transfected with the APP Val/Gly mutation with increasing concentrations of NH₄Cl to inhibit vesicular acidification. As shown in Fig. 5A NH₄Cl treatment did not inhibit intracellular A1-42 generation but even caused a dose-dependent increase in the generation of Ax-42. This indicates not only that endosmal lysomal processing is not a prerequisite for A42 generation but might also suggest that A42 could be degraded within endosomes and lysosomes. Moreover, inhibition of reinternalization by expressing the Y687A APP mutation (19) or by deleting the entire cytoplasmic tail (Ref. 8; data not shown) also did not inhibit intracellular generation of A42 peptides (Fig. 5B). These results therefore indicate that endosomal/lysosomal targeting of APP is not a prerequisite for A42 generation.

Inhibition of intracellular acidification and reinternalization does not block intracellular A1-42 and Ax-42 generation.

DISCUSSION

Our results suggest that A peptides can be cleaved at their C termini by distinct proteolytic mechanisms located within different subcellular compartments of kidney 293 cells. The identification of a novel cell biological mechanism for A42 generation in peripheral cells might also be relevant for neuronal cells. This is supported by the observations that all APP and PS mutations elevate A42 production in peripheral cells, primary neurons, as well as in human brain and brains from transgenic mice overexpressing PS or APP mutations (14, 15). However, cellular toxicity appears to be restricted to neuronal cells as observed in the AD brain (3). The -secretase cleaving at amino acid 40 predominantly occurs close to or at the cell surface (8, 19), whereas the protease cleaving at position 42 can act within the ER and Golgi. This is consistent with the recent finding that the -secretase cleavage at position 40 and 42 can be inhibited differentially with the protease inhibitor MDL 28170 (26, 27). Moreover, as discussed by Tischer and Cordell (28) the cleavage at position 42 might be facilitated within the ER and early Golgi, because the lower membrane thickness may allow proteases located on the cytoplasmic site of the ER to cleave at position 42, a hypothesis supported by our data.

Our results also indicate that alternative N-terminal cleavages might occur preferentially within the secretory pathway. Moreover, -secretase appears to be active to some extend within the ER as well. Most of the alternative N-terminal cleavages (Ax-40/42) are detected with antibody BNT77, which identifies an epitope between amino acids 11-16. This might suggest that most of the Ax-40/42 species detected in the ELISA represent alternatively cleaved A peptides starting at Glu¹¹. Indeed, relatively high levels of this truncated A peptide were previously detected in media of kidney 293 cells expressing APP (7, 8). This peptide might also play an important role in the pathogenesis of AD as well (4).

The cleavage at position 42 within the ER and Golgi is of particular interest, since mutant PS proteins are known to cause enhanced production of the long form of A (14, 15, 16) and are located within the ER and Golgi as well (29-32). One should therefore assume that the pathological action of the mutant PS proteins occurs to some extend within the ER and Golgi, which is supported by our findings that the proteolytic cleavage of A at amino acid 42 occurs within the same compartments where the PS proteins were located previously. Moreover, the detection and accumulation of intracellular A terminating at amino acid 42 leads to the possibility that the toxic effects of this highly amyloidogenic peptide are exerted prior to its secretion. This might lead to neuronal cell death as well as the generation of a nidus for further A aggregation, which is then followed by amyloid plaque formation. Interestingly, intracellular amyloid fibers were recently detected in transgenic mice developing the typical pathological lesions found in human AD brains (33). Such a putative mechanism might also solve the apparent discrepancy that A42 levels determined in biological fluids do not reach the critical concentration required for aggregation, since intracellular accumulation of A42 could very well lead to high local concentrations of A42 sufficient for its precipitation.