Enhanced release of amyloid beta-protein from codon 670/671 "Swedish" mutant beta-amyloid precursor protein occurs in both secretory and endocytic pathways.

The mutation at codons 670/671 of β-amyloid precursor protein (βPP) dramatically elevates amyloid β-protein (Aβ) production. Since increased Aβ may be responsible for the disease phenotype identified from a Swedish kindred with familial Alzheimer's disease, evaluation of the cellular mechanism(s) responsible for the enhanced Aβ release may suggest potential therapies for Alzheimer's disease. In this study, we analyzed Chinese hamster ovary cells stably transfected with either wild type βPP (βPP-wt) or “Swedish” mutant βPP (βPP-sw) for potential differences in βPP processing. We confirmed that increased amounts of Aβ and a β-secretase-cleaved COOH-terminally truncated soluble βPP (βPP) were secreted from βPP-sw cells. As shown previously for βPP-wt cells, Aβ was released more slowly than the secretion of βPP from surface-labeled βPP-sw cells, indicating that endocytosis of cell surface βPP is one source of Aβ production. In contrast, by [S]methionine metabolic labeling, the rates of Aβ and βPP release were virtually identical for both cell lines. In addition, the identification of intracellular βPP and Aβ shortly after pulse labeling suggests that Aβ is produced in the secretory pathway. Interestingly, more Aβ was present in medium from βPP-sw cells than βPP-wt cells after either cell surface iodination or [S]methionine labeling, indicating that βPP-sw cells have enhanced Aβ release in both the endocytic and secretory pathways. Furthermore, a variety of drug treatments known to affect protein processing similarly reduced Aβ release from both βPP-wt and βPP-sw cells. Taken together, the data suggest that the processing pathway for βPP is similar for both βPP-wt and βPP-sw cells and that increased Aβ production by βPP-sw cells arises from enhanced cleavage of mutant βPP by β-secretase, the as-yet unidentified enzyme(s) that cleaves at the NH terminus of Aβ.

In Alzheimer's disease a characteristic pathological finding in the brains of affected individuals is the deposition of amyloid ␤-protein (A␤) 1 in senile plaques (1). A␤ is the 39 -43-amino acid proteolytic cleavage product of the type I integral mem-brane protein ␤-amyloid precursor protein (␤PP). The ␤PP gene is encoded on chromosome 21, and alternative exon splicing produces three major isoforms of 695, 751, or 770 amino acids (2). During constitutive secretion some full-length ␤PP molecules are proteolytically cleaved between lysine and leucine residues at positions 16 and 17 of A␤ ( Fig. 1) by an enzyme termed ␣-secretase (3,4). Cleavage of ␤PP at this position creates a soluble ϳ100 -120-kDa NH 2 -terminal fragment (␤PP s ) (5) and a COOH-terminal membrane-retained fragment of ϳ10 kDa (6). Generation of these fragments by ␣-secretase precludes formation of an intact A␤ sequence from full-length ␤PP.
A␤, however, is known to be released during normal cellular metabolism both in vivo (7,8) and in a number of cell culture systems (9,10). Cleavage of ␤PP at the NH 2 terminus of the A␤ sequence by an enzyme designated ␤-secretase creates a shortened form of ␤PP s and the ϳ12-kDa COOH-terminal fragment (11,12). An additional enzymatic cleavage at the COOH terminus of the A␤ sequence by the as yet unidentified enzyme designated ␥-secretase generates the 4-kDa A␤ peptide. The ␥-secretase enzyme is also hypothesized to generate p3, the 3-kDa NH 2 -terminal piece of the membrane-retained ϳ10-kDa COOH-terminal fragment of ␤PP produced by ␣-secretase cleavage (7)(8)(9)13). In addition to the secretory cleavage, ␤PP can also be processed in an endosomal/lysosomal pathway (14 -17). Although A␤-containing COOH-terminal fragments are generated in lysosomes, evidence suggests that these are not an important source of A␤ (18). Recently, it was shown that cell surface ␤PP molecules can be processed in the endocytic pathway and may be the direct precursors of A␤, presumably by recycling internalized molecules from the cell surface (19).
Evidence that A␤ and ␤PP contribute to the pathogenesis of Alzheimer's disease comes from the findings of missense mutations within and adjacent to the A␤ region of the ␤PP gene in families with autosomal dominant forms of Alzheimer's disease (20). The concurrence of the mutations with the disease phenotype suggests that altered ␤PP function or processing may be pathogenic. A double mutation at amino acids 670 and 671 (␤PP 770 numbering) changing Lys 670 to Asn 670 and Met 671 to Leu 671 (K670N/M671L) was identified in a Swedish pedigree with familial Alzheimer's disease (21). In vitro analyses of transfected cells expressing the Swedish form of ␤PP (12,22) and primary cell cultures of fibroblasts obtained from affected individuals (23) reveal a dramatic increase in A␤ production. However, the mechanism by which A␤ generation is increased has not been elucidated. Furthermore, a detailed analysis of cellular processing of ␤PP with this mutation has not been reported. Because recent studies have implicated the endocytic pathway in A␤ production (19), we speculated that A␤ production may be similarly enhanced in this pathway in cells expressing the K670N/M671L ␤PP 751 mutation.
In this report, biosynthetic analyses confirmed the increase in A␤ production and the abundant secretion of a shorter ␤PP s species by Chinese hamster ovary (CHO) cells stably transfected with the ␤PP 751 K670N/M671L mutation. Furthermore, A␤ generation was increased in both the secretory and endocytic pathways. We postulate that this increase in A␤ production is the result of enhanced proteolytic cleavage of the mutant ␤PP by the ␤-secretase enzyme.
Metabolic Labeling-Confluent cultures of ␤PP-transfected CHO cells were incubated in methionine-free DMEM for 15 min followed by incubation with methionine-free DMEM supplemented with 200 Ci/ml [ 35 S]methionine for 10 min (pulse labeling) or for 2-4 h with 50 -100 Ci/ml [ 35 S]methionine (long labeling). Cells were lysed immediately or incubated with 2-fold unlabeled methionine (chase) in DMEM from 10 min to 4 h. For some experiments, single dishes of confluent cells were pulse labeled and chased at multiple time points with repeated collection of medium to evaluate the incremental secretion of ␤PP products at each time point of the chase period. ␤PPs were immunoprecipitated using ␤PP-specific antibodies and separated by SDS-polyacrylamide gel electrophoresis (using 6 -10% Tris-glycine gels for high molecular weight proteins and 16.5% Tris-Tricine gels for low molecular weight proteins). Gels were either fluorographically enhanced and exposed to x-ray film or dried and exposed on a Phosphor screen (Molecular Dynamics). All experiments reported herein were performed two to six times, and a representative example of each is shown. Where applicable, average values Ϯ S.E. are given.
Assessment of Total ␤PP and ␤PP s -␤PP expression and ␤PP s quantity were determined using parallel triplicate cultures of stably transfected CHO cells expressing wild type ␤PP or "Swedish" ␤PP. One set of cultures was lysed immediately after a 10-min pulse labeling and immunoprecipitated with C7 to determine total ␤PP. The other set was chased for 4 h, and media were collected and immunoprecipitated with MMAb to determine ␤PP s secretion. Samples were separated by SDSpolyacrylamide gel electrophoresis and analyzed by Phosphorimaging. Comparison of ␤PP s levels was made after normalization for total ␤PP expression.
Intracellular ␤PP s -Separate dishes of stably transfected CHO cells, one dish for each chase time, were pulse labeled for 10 min and chased for either 10, 20, 30, or 60 min. Chase media were collected, and cells were washed with ice-cold DPBS and rapidly chilled to 4°C. After washing, DPBS was replaced with DPBS containing 0.1% saponin plus protease inhibitors (leupeptin and Pefabloc, Boehringer Mannheim). Cells were treated with saponin buffer for 40 min at 4°C as described previously (29), to allow the release of intracellular ␤PP s . Saponin buffer was then collected, and both the chase media and saponin buffers were immunoprecipitated using antibody B5 for total ␤PP s and antibody C7 for full-length ␤PP.
Cell Surface Iodination-Derivatized Bolton-Hunter reagent, sulfosuccinmidyl-3-(4-hydroxyphenyl)propionate (sulfo-SHPP, Pierce) was labeled with Na 125 I in the presence of IODO-GEN (Pierce) at room temperature essentially as described (30). After 5 min, the iodination reaction was quenched with p-hydroxyphenylacetic acid (Sigma), diluted with DPBS, and immediately added to chilled washed cells for 40 min at 4°C. After iodination, the cells were extensively washed in DPBS containing 1 mg/ml lysine followed by incubation in prewarmed CHO medium at 37°C. Three independent experiments were performed.
Surface Antibody Binding-To determine the amount of cell surface and total ␤PP in stably transfected CHO cells, 5A3 monoclonal antibody Fab fragments were radioiodinated with IODO-GEN to 2-4 Ci/g (19). Confluent cell cultures were chilled and washed. One set of cultures was treated with 0.1% saponin in DPBS for 30 min at 4°C to permeabilize cells gently and permit the labeling of both cell surface and intracellular ␤PP. Parallel cultures for each cell line were treated for 30 min at 4°C with DPBS for evaluation of cell surface ␤PP. Both sets were incubated with radioiodinated antibody at 10 nM in binding medium (RPMI 1640 medium supplemented with 0.2% bovine serum albumin) at 4°C for 1 h, followed by two washes with binding medium and two washes with DPBS. The cells were then lysed with 0.2 M NaOH, and radioactivity was determined by ␥-counting. To calculate specific binding, background levels of radioactivity were determined from parallel cultures of untransfected CHO cells to subtract from the counts obtained from ␤PP-transfected CHO cells. Four separate experiments were performed using triplicate cultures.
Drug Studies-Confluent CHO cells were metabolically labeled with 50 -100 Ci/ml terminal black box represents the signal sequence. ␣, ␤, and ␥ mark the sites of the enzymatic cleavages by ␣-, ␤-, and ␥-secretases, respectively. Also indicated are the ϳ10-kDa fragment (including the p3 region, transmembrane region, and COOH terminus) and the ϳ12-kDa fragment (including the A␤ region, transmembrane region, and COOH terminus). -NPTY-indicates the putative clathrin internalization signal. Horizontal black bars indicate the approximate epitopes of antibodies B5, C7, 6E10, MMAb, R1280 and R1282, and R1736. ent with earlier reports (12,22,(31)(32)(33) all of the ␤PP-sw cell lines released severalfold more of the 4-kDa A␤ peptide than wild type cells, which had comparable levels of ␤PP expression (not shown). Pulse-chase experiments also showed that the timing of appearance and disappearance of labeled ␤PP was essentially identical in ␤PP-wt and ␤PP-sw cells ( Fig. 2A). Both ␤PP-wt and ␤PP-sw cells produced abundant N-glycosylated ␤PP during the 10-min labeling reaction (time 0, Fig. 2A). At 1 h of the chase, higher molecular weight NϩO-glycosylated ␤PP molecules were seen in addition to the N-glycosylated forms. Full-length ␤PP decreased by 2 h; and by 4 h, little full-length ␤PP remained ( Fig. 2A).
Secretion of a shortened ␤PP s species has been reported from a ␤PP chimeric molecule expressing the "Swedish" mutation (31). To confirm this finding with authentic ␤PP molecules, ␤PP s was immunoprecipitated from conditioned media using B5 antibody, which recognizes both ␣and ␤-secretase species of ␤PP s , and two antibodies that recognize only ␣-secretasecleaved ␤PP s (R1736 and 6E10). As observed for untransfected CHO cells (not shown), ␤PP s from transfected CHO cells migrates as a doublet of bands on low percentage polyacrylamide gels. As a result, the higher molecular weight ␤PP s cleaved by ␣-secretase and the slightly lower molecular weight ␤PP s cut by ␤-secretase can best be compared by observing the lower of the two bands of each doublet (Fig. 2B). The ␤PP s from ␤PP-sw cells migrated at an M r consistently lower than that of ␤PP-wt cells, indicating the secretion of a shorter ␤PP s species. Al-though both cell lines secreted comparable levels of total ␤PP s by B5 antibody immunoprecipitation (Fig. 2B), ␤PP-sw cells had dramatically reduced levels of ␣-secretase-cleaved ␤PP s (6 Ϯ 1.3-fold less) than ␤PP-wt cells using antibodies R1736 and 6E10 (Fig. 2B). Consistent with this finding, and as reported by others (12,(31)(32)(33), ␤PP-sw cells also had correspondingly higher levels of ϳ12-kDa COOH-terminal ␤PP fragments (see below).
Timing of Secretion of ␤PP s , A␤, and p3 Is Identical in ␤PP-wt and ␤PP-sw Cells-Since the turnover rate of fulllength ␤PP was essentially the same in the ␤PP-wt and ␤PP-sw cells, we next examined the biosynthetic rate for the generation of ␤PP-secreted products (␤PP s , A␤, and p3). Following a 10min pulse labeling, media were collected from a single dish each from ␤PP-wt and ␤PP-sw cells and reapplied at 10-min intervals to define the incremental release of ␤PP secretion products during the 1st h of the chase period.
The onset of secretion of total ␤PP s was first detectable at 10 min as determined with B5 immunoprecipitation (Fig. 3B). However, at this first time interval only minute amounts of ␤PP s were secreted from both ␤PP-wt and ␤PP-sw cells because the signal could be seen in the 10-min lane only after prolonged autoradiographic exposures (Fig. 3B). ␤PP s became pronounced at 20 min for both cell lines with peak secretion at approximately 30 min (Figs. 3A and 4). The profile of ␤PP s secretion as a function of time was essentially identical for the two cell lines (Fig. 4). In the experiment shown, although ␤PP s secretion by ␤PP-sw cells was lower because of diminished expression of full-length ␤PP, the profile of secretion is essentially identical to that of ␤PP-wt cells. This profile of ␤PP s secretion did not depend on the level of ␤PP expression because other wild type and Swedish cell lines exhibited the same patterns of release (not shown). Furthermore, comparison of ␤PP-wt and ␤PP-sw cells that expressed equivalent levels of ␤PP confirmed that ␤PP s secretion by both cell lines was es- Regarding A␤ release, the timing of A␤ secretion during the 1st h from ␤PP-wt and ␤PP-sw cells was also identical (Figs. 3C and 4). The A␤ signal was first apparent at the 20-min collection time by autoradiography (Fig. 3C) and reached a peak at 30 -40 min. Although no discernible A␤ signal was ever seen on either autoradiograms or Phosphorimages at the 10-min chase time, after long exposures a few Phosphorimage counts higher than background were detected in the 10-min lane (Fig. 4). At each chase time, ␤PP-sw cells consistently released more A␤ than ␤PP-wt cells. The timing of p3 secretion mirrored that of A␤ in both ␤PP-wt and ␤PP-sw cells throughout the chase period (Fig. 3C), although ␤PP-wt cells consistently released more p3 relative to A␤ than did ␤PP-sw cells. Authentication of A␤ (beginning at Asp 1 ) and p3 (beginning at Lys 17 ) was obtained by radiosequencing (not shown), as reported previously (15,19). Thus, a difference in the ratios of ␣-secretaseand ␤-secretase-generated molecules was also reflected by the levels of p3 and A␤ released by these cell lines. Finally, the formation of the ␤-secretase-generated ϳ12-kDa ␤PP COOHterminal fragments preceded the release of A␤ from pulselabeled ␤PP-sw cells (Fig. 2C). After a 10-min labeling with [ 35 S]methionine, the ϳ12-kDa fragment was apparent by the 10-min chase time in ␤PP-sw cells and increased at 20 min (Fig. 2C). Consistent with the above results, A␤ was not apparent in the corresponding media until 20 min of the chase period (Fig. 2C). This earlier generation of the ϳ12-kDa fragment prior to A␤ release, consistently seen in three experiments, indicates a precursor-product relationship between the two molecules.
␤PP s and A␤ Appear to Be Present Intracellularly in ␤PP-wt and ␤PP-sw Cells-The release of ␤PP s at very early chase times suggested that ␤PP may be cleaved by ␤-secretase in the secretory pathway. To determine if soluble ␤PP s was present intracellularly, metabolically labeled cells were treated with 0.1% saponin in buffer. Saponin is a mild detergent that permeabilizes cells but does not solubilize the lipid bilayer (29) and therefore allowed the intracellular ␤PP s to diffuse into the buffer. Essentially no full-length ␤PP was detected in the saponin buffer of treated cells. However, soluble intracellular ␤PP s was recovered from the saponin buffer from both ␤PP-wt and ␤PP-sw cells (Fig. 5, A and B). Intracellular ␤PP s species from ␤PP-wt cells (a finding previously reported by others; see Refs. 34 -36) migrated with an M r consistent with ␣-secretasecleaved molecules (Fig. 5A). A shorter ␤PP s species with an M r identical to secreted ␤PP s and consistent with ␤-secretasecleaved molecules was observed from ␤PP-sw cells ( Fig. 5A; Ref. 32). Furthermore, in pulse-chase experiments a precursorproduct relationship could be demonstrated between intracellular ␤PP s from the saponin-treated cells and ␤PP s secreted into the medium (Fig. 5B).
These results suggested that A␤ can be formed within the secretory pathway. Indeed, intracellular A␤ appeared to be present in both ␤PP-wt and ␤PP-sw cell lysates labeled for 4 h (Fig. 5C). Preabsorption of R1280 antibody with the A␤ 1-40 peptide totally eliminated immunoprecipitation of A␤ from the cell lysates by R1280 antibody (Fig. 5C), and no 4-kDa band was observed from the same lysate using antibody C7. Treatment with trypsin prior to immunoprecipitation did not diminish the A␤ signal (not shown), indicating that A␤ was present inside the cells. In addition, cells pulse labeled with [ 35 S]methionine followed by a 20-min or 30-min chase had both A␤ and p3 isolated from cell lysates (not shown). Thus, the immunoprecipitated A␤ had not been derived from secreted molecules present on the extracellular plasma membrane at the time of cell lysis. Furthermore, the appearance of these intracellular A␤ and p3 molecules after short pulse-chase intervals provides indirect evidence of their production in the secretory and not the endosomal/lysosomal pathway. Nevertheless, A␤ and p3 bands were visualized only after 8 -10 weeks of autoradiographic exposure, suggesting that only very low levels of A␤ were ever present intracellularly. The minute amounts of intracellular A␤ precluded definitive identification by amino acid radiosequencing.
␤PP from the Cell Surface Contributes to A␤ Production in Both ␤PP-wt and ␤PP-sw Cells-To determine if the endocytic pathway contributed to A␤ production in ␤PP-sw cells, release of A␤ was analyzed after selective cell surface radioiodination. Consistent with an earlier report (19), little radiolabeled A␤ was secreted within the first 10 min by either cell line, but considerable A␤ was released by 2 h from both ␤PP-wt and ␤PP-sw cells (Fig. 6A), with ␤PP-sw cells releasing greater than 2-fold more A␤ than ␤PP-wt cells at the 2-h collection time (2.4 Ϯ 0.4). The timing of A␤ release following labeling of cell surface ␤PP was essentially the same in the two cell lines (Fig.  6A). However, in sharp contrast to the timing of A␤ secretion, the majority of the iodinated ␤PP s was released within the first 5 min of incubation at 37°C from both ␤PP-wt and ␤PP-sw cells (Fig. 6B). In addition, these profiles of both A␤ and ␤PP s release are distinctly different from the timing observed for A␤ and ␤PP s release observed using [ 35 S]methionine labeling (Figs. 3 and 4).
Two additional observations are noteworthy from these experiments. First, ␤PP s derived from cell surface ␤PP by ␤PP-sw cells had an M r compatible with ␣-secretase-cleaved ␤PP s (Fig.  6B). A lower M r ␤-secretase-cleaved ␤PP s species was not readily apparent after surface labeling. However, resolution of the labeled bands is significantly less distinct from an iodine signal because of radiographic intensification, and minor differences may be undetectable. Second, we consistently observed more full-length ␤PP on the surface of ␤PP-wt cells than ␤PP-sw cells (Fig. 6C) expressing the same amount of ␤PP. To confirm and quantitate this difference, the levels of cell surface and total ␤PP were measured by an antibody binding assay using radioiodinated antibody 5A3 Fab fragments, which bind to an extracellular ␤PP epitope (19). Treatment with 0.1% saponin permitted labeling of both cell surface and intracellular ␤PP. Multiple repetitions of this experiment showed that ␤PP-sw cells had approximately 50% less cell surface ␤PP than ␤PP-wt cells (49.8% Ϯ 0.7, p Ͻ 0.0001). Interestingly, ␤PP-sw cells showed more of the COOH-terminal ϳ12-kDa fragment and less of the ϳ10-kDa fragment than ␤PP-wt cells (Fig. 6C) present on the cell surface.
Drug Treatments Affect ␤PP-wt and ␤PP-sw Cells Similar- Since ␤PP s from CHO cells migrates as a doublet, the higher molecular weight ␤PP s cleaved by ␣-secretase (␣ at arrow) and the slightly lower molecular weight ␤PP s cut by ␤-secretase (␤ at arrow) can best be appreciated by observing the lower of the two bands. The shorter ␤PP s species is observed both intracellularly (intra) and secreted into the medium (sec) of ␤PPsw cells. The faint bands that run below 97 kDa in the saponin lanes are degradation products. Molecular weights determined from prestained standards are indicated on the right. Panel B, antibody B5 immunoprecipitations of intracellular and secreted ␤PP s from ␤PP-wt and ␤PP-sw cells following a 10-min pulse with [ 35 S]methionine and 10 -60-min chase. Intracellular ␤PP s was immunoprecipitated from saponin buffers; secreted ␤PP s was obtained from chase media. Panel C, immunoprecipitation of ␤PP-wt and ␤PP-sw control (cont) cell lysates after 4 h [ 35 S]methionine labeling with antibody R1280 or R1280 that had been preabsorbed (abs) with the A␤ 1-40 peptide. The positions of the ϳ12-kDa COOH-terminal fragments and A␤ are indicated at arrows on the left. wt ϭ ␤PP-wt cells; sw ϭ ␤PP-sw cells.
FIG. 6. Release of A␤ and ␤PP s from cell surface-iodinated ␤PP molecules. Panel A, immunoprecipitation of A␤ with antibody R1280 from chase media of ␤PP-wt and ␤PP-sw cells following iodination of cell surface ␤PP. The timing of release of A␤ was the same from both cell lines. Molecular weights determined from prestained standards are indicated. Panel B, rapid release of ␤PP s was observed from both ␤PPwt and ␤PP-sw cells after surface iodination and immunoprecipitation by antibody B5. Panel C, immunoprecipitation of cell lysates with antibody C7 after iodination revealed more full-length ␤PP on the surface of ␤PP-wt cells than ␤PP-sw cells. ␤PP-sw cells, however, had more iodinated ϳ12-kDa COOH-terminal fragments and fewer ϳ10-kDa fragments than ␤PP-wt cells. wt ϭ ␤PP-wt cells; sw ϭ ␤PP-sw cells.
ly-To assess further whether the processing of ␤PP molecules is similar in ␤PP-wt and ␤PP-sw cells, both cell lines were treated with a variety of compounds known to affect protein processing. Control lanes reveal the higher production of both A␤ and ϳ12-kDa COOH-terminal ␤PP fragments from ␤PP-sw cells (Fig. 7A). Treatment with brefeldin A, a drug that blocks the maturation of proteins by collapsing the Golgi into the endoplasmic reticulum, inhibited the production of both A␤ and the ϳ12-kDa fragment in both ␤PP-wt and ␤PP-sw cells (Fig. 7A). Treatments that alkalinize intracellular vesicles were also used because A␤ generation in cultured cells appears to require an acidic compartment (8,13). To determine if ␤-secretase cleavage of both ␤PP-wt and ␤PP-sw cells occurs in an acidic intracellular compartment, cells were exposed to chloroquine or bafilomycin A1 (37). In response to chloroquine presented during the 2-h chase, both ␤PP-wt and ␤PP-sw cells released 30%-60% less A␤ than untreated controls (Fig. 7A). As anticipated (10,13), chloroquine also dramatically elevated the level of COOH-terminal fragments in both cell lines (Fig. 7A). Exposure of cells to bafilomycin A1, a drug that specifically inhibits vacuolar H ϩ -ATPases and thus prevents vesicular acidification (37), produced a ϳ60% decrease in A␤ release from both ␤PP-wt and ␤PP-sw cells (as measured by Phosphorimaging, Fig. 7A). A corresponding increase in ϳ10-kDa COOHterminal fragments was observed following exposure to bafilomycin A1 (Fig. 7A), suggesting that less ␤PP was cleaved by ␤-secretase in the presence of the drug. Consistent with this interpretation is the finding of a dramatic decrease in the amount of ϳ12-kDa fragments in ␤PP-sw cells when bafilomycin was presented using a short pulse-chase paradigm (Fig.  7B). This reduction of the ϳ12-kDa fragment (Fig. 7B) was seen at a time when abundant ϳ12-kDa COOH-terminal fragments were normally observed from ␤PP-sw cell lysates following a 10-min pulse-labeling (Figs. 2C and 7B). In sum, the processing of ␤PP and release of A␤ by CHO cells expressing either wild type or mutant ␤PP were similarly affected by these drug treatments. DISCUSSION A double mutation in the ␤PP gene from a Swedish kindred with familial Alzheimer's disease is invariably linked with Alzheimer's disease (21). All cells reported to date which express the ␤PP mutation produce dramatically more A␤ peptide than do cells expressing wild type ␤PP (12,22,23,(31)(32)(33). Since excess A␤ production may be causally related to the Alzheimer's phenotype in individuals affected with the "Swedish" mutation (21), it is important to evaluate the mechanism by which A␤ is produced from ␤PP with this alteration. In this study we performed a detailed analysis of the biosynthetic processing of ␤PP in ␤PP-wt and ␤PP-sw CHO cells.
Our results showed that, as anticipated, ␤PP-sw cells released substantially more A␤ than ␤PP-wt cells. Interestingly, the timing of onset and the duration of A␤ secretion during the 1st h following a short pulse labeling were coincident with p3 release for both cell lines. Only the amounts of A␤ and p3 varied between ␤PP-wt and ␤PP-sw cells. Furthermore, treatments known to decrease A␤ in ␤PP-wt cells (8,13,38) also affected ␤PP-sw cells. We interpret our data to suggest that the pathway of A␤ production is similar for ␤PP-wt and ␤PP-sw cells. In contrast, however, the timing of A␤ release differed substantially depending on whether cells were [ 35 S]methionine-labeled or surface-iodinated. In both cell lines A␤ was released with a shorter time course from [ 35 S]methionine-labeled cells than from cells that were surface-iodinated. This difference in the timing of A␤ secretion leads us to propose that A␤ is generated in both the secretory and endocytic pathways from both ␤PP-wt and ␤PP-sw cells.
A number of observations suggest that A␤ is generated in the secretory pathway (10,39). First, the timing of secretion of ␤PP s , A␤, and p3 was essentially identical at early chase times in both cell lines. Specifically, within the first 30 min in a short pulse-chase experiment, the profiles of ␤PP s , A␤, and p3 secretion were remarkably similar. This chase paradigm was chosen specifically to reveal the incremental release of these early secretory products. Second, permeabilization of [ 35 S]methionine pulse-labeled cells followed by immunoprecipitation with a ␤PP midregion antibody (B5) showed that intracellular soluble ␤PP s was present in both ␤PP-wt (34 -36) and ␤PP-sw cells as reported previously (32). The major intracellular species of soluble ␤PP s from ␤PP-sw cells had a lower M r than ␤PP s from ␤PP-wt cells, consistent with production by ␤-secretase cleavage. Significantly, intracellular ␤PP s was present before abundant ␤PP s was secreted into the culture medium, thus demonstrating a precursor-product relationship. Third, the ϳ12-kDa COOH-terminal fragment of ␤PP and A␤ showed a precursorproduct relationship, with the ϳ12-kDa molecules apparent 10 min prior to the appearance of A␤. Moreover, consistent with a recent report (12), this COOH-terminal ϳ12-kDa fragment was specifically increased in ␤PP-sw cells compared with ␤PP-wt cells. Fourth, our data suggest that intracellular A␤ is present in both ␤PP-wt and ␤PP-sw cells. Based on results from trypsin digestion using a short pulse-chase paradigm, A␤ in cell lysates did not appear to represent extracellular A␤ attached to the cell surface or to be derived from the lysosomal pathway. However, the exceedingly small amount of intracellular A␤ suggests that A␤ turnover and secretion are rapid. This is consistent with the earlier postulation that A␤ is released from cells soon after it is formed and suggests that ␥-secretase cleavage occurs at or near the cell surface (19). Previously, intracellular A␤ has only been detected in neurons (40). Thus our preliminary findings suggest that the pathways of A␤ production in neurons and non- neuronal cells may be more similar than was previously thought.
Regarding the endocytic processing of ␤PP, cells labeled by selective cell surface iodination confirmed that ␤PP-sw cells produced more A␤ from cell surface precursors than did ␤PP-wt cells. However, the timing of A␤ release after surface labeling was essentially identical for both ␤PP-wt and ␤PP-sw cells. As shown previously for cells expressing wild type ␤PP (19), A␤ generated from surface-labeled molecules was released more slowly than ␤PP s by ␤PP-sw cells. These profiles of A␤ and ␤PP s release from surface-labeled molecules are dramatically different from the [ 35 S]methionine labeling experiments in which A␤ and ␤PP s were released simultaneously. Interestingly, ␤PP s and A␤ release from [ 35 S]methionine pulse-chase experiments showed that ␤PP s secretion peaked at ϳ30 min followed by a sharp decrease, whereas A␤ release continued at the same level until later chase times. We interpret the sustained A␤ release into the medium at a time when ␤PP s secretion decreased (40 -50 min) to represent the addition of newly generated A␤, derived from the endocytic pool, after the contribution of the secretory pool of A␤ has peaked. Therefore, our data indicate that A␤ can be derived from both the secretory and endocytic pathways and that more A␤ is formed within each pathway by ␤PP-sw cells.
Our studies have defined a number of similarities in ␤PP processing between ␤PP-wt and ␤PP-sw cells. First, the timing of secretion of ␤PP s , A␤, and p3 is essentially the same for both cell lines within the 1st hour following a 10-min pulse label. Second, various drug treatments decrease A␤ in both ␤PP-wt and ␤PP-sw CHO cells. Third, intracellular ␤PP s species and A␤ appear to be present in both cell lines. Fourth, both secretory and endocytic pathways appear to contribute to A␤ generation and release. Fifth, both ␤PP-wt and ␤PP-sw cells secrete primarily ␣-secretase-cleaved ␤PP s from surface-labeled ␤PP. Thus, within the limits and sensitivity of our experimental system, the timing and the pathway of A␤ secretion appear to be identical in ␤PP-wt and ␤PP-sw cells. Only the amounts of A␤ and ␤-secretase-cleaved precursors differed in ␤PP-wt and ␤PP-sw cells. Our data and interpretation are therefore consistent with the results of previous investigators who have suggested that the "Swedish" mutation at the NH 2 terminus of A␤ enhances ␤-secretase cleavage (12,22,23). This altered ␤-secretase cleavage produces abundant ␤-secretase-cleaved ␤PP s in the secretory pathway in ␤PP-sw cells, leading to excess A␤ production. However, it remains unclear at present which pathway, secretory or endocytic, plays the greater role in A␤ production.
␤PP-sw cells did show some differences in ␤PP processing from ␤PP-wt cells. In addition to the increase in ␤-secretasecleaved products described above, there was a 50% reduction in the amount of cell surface ␤PP in ␤PP-sw cells. Concomitantly, there was an increase in the ϳ12-kDa membrane-retained ␤PP fragments present on the cell surface of ␤PP-sw cells. Whether this increase in ϳ12-kDa fragments is sufficient to account for the decrease in full-length ␤PP molecules at the cell surface of ␤PP-sw cells is unclear. Because secreted ␤PP s levels are similar between ␤PP-wt and ␤PP-sw cells, the reduction in fulllength ␤PP at the surface of ␤PP-sw cells suggests that the amount of ␤PP targeted to the cell surface may represent a minor fraction of the total ␤PP processed in the secretory pathway. Otherwise, one would expect to see a substantial increase in ␤PP s released into the medium from ␤PP-sw cells, which was not detected. Furthermore, this interpretation is also consistent with reports of other cell types that express little or no ␤PP on the cell surface (34 -36).
In summary, our data suggest that there is a similar mech-anism for A␤ generation in both ␤PP-wt and ␤PP-sw cells. The increased A␤ production from ␤PP-sw cells appears to result from enhanced ␤-secretase cleavage of the mutant ␤PP in both the secretory and endocytic pathways. A recent report has demonstrated altered ␤PP processing in mutant ␤PP molecules with natural or designed mutations in codon 692 (41), whereas another report demonstrated an increased percentage of longer A␤ peptides from ␤PP with codon 717 mutations (42). Taken together, it appears that FAD ␤PP mutations lead to pleiotropic effects on ␤PP and A␤ metabolism. The Alzheimer phenotype associated with these dominant mutations may therefore result from different cellular perturbations that specifically modify ␤PP processing.