A Basic Amino Acid in the Cytoplasmic Domain of Alzheimer’s β-Amyloid Precursor Protein (APP) Is Essential for Cleavage of APP at the α-Site*

In Alzheimer’s disease (AD), the β-amyloid peptide (Aβ) is thought to be produced as a result of the aberrant metabolism of β-amyloid precursor protein (APP). We report that the APP cytoplasmic domain contains a novel and important signal for APP metabolism. A single amino acid mutation that changed arginine at amino acid 747 of APP770 (corresponding to position 672 of APP695) to a non-basic amino acid greatly increased the production of intracellular APP carboxyl-terminal fragment(s) cleaved at β-site(s) (CTFβ), but did not result in increased secretion of Aβ40 and Aβ42. This was not due to a simple intracellular accumulation of CTFβ resulting from a lack of γ-secretase. CTFβ derived from this mutant APP was generated and degraded as efficiently as CTFβ derived from wild-type APP. This result indicates that the increase in the quantity of CTFβ does not always give rise to more Aβ production, as was previously suggested by studies of a familial AD mutation of APP. These findings suggest that APP carrying the substitution mutation at this basic amino acid may be metabolized by another protein secretory pathway. Although these results have not completely elucidated why CTFβ derived from the mutant APP escapes from subsequent cleavage by γ-secretase, analysis of the processing pathway of this mutant APP should provide insights into the pathogenesis of the sporadic type of AD.

essing of APP generates two fragments, a large extracellular amino-terminal domain (sAPP␣) and a truncated carboxyl-terminal fragment (CTF␣), both of which are products of cleavage within the A␤ domain. Another form of proteolytic processing occurring at ␤-site(s) gives rise to limited production of carboxyl-terminal fragments (CTF␤) and A␤ in unidentified intracellular compartments of the protein secretory pathway (14). Previous studies suggest that the ␣-secretase responsible for cleavage of APP at the ␣-site is active in the trans-Golgi network or other end-stage compartments of the protein secretory pathway (15,16). Studies utilizing the APP of familial Alzheimer's disease (FAD), such as APP with the "Swedish" double mutation (K595N/M596L, numbering for the APP695 isoform), suggest that APP cleavage at the ␤-site by ␤-secretase occurs in the medial-Golgi network and in unidentified compartments proximal to the plasma membrane (17,18), although a recent report suggests that the endoplasmic reticulum is the site of generation of A␤42 (but not of A␤40) in neurons (19). Furthermore, it is generally held that the increased production of CTF␤ always gives rise to more A␤, although the molecular mechanisms of CTF␤ production by ␤-secretase from wild-type APP and of subsequent A␤ production following the cleavage of CTF␤ by ␥-secretase are still unknown in sporadic AD patients, who constitute the majority of AD cases.
The APP cytoplasmic domain is thought to be responsible for the intracellular transport of APP. Two internalization signals, NPTY (APP770 amino acids 759 -762 and APP695 amino acids 684 -687) and YTSI (APP770 amino acids 728 -731 and APP695 amino acids 653-656), have been identified (20,21). The NPXY motif in the cytoplasmic domain of membrane proteins is believed to mediate interactions between internalized proteins and the clathrin cage of the clathrin-coated vesicle (22). The YXXI motif conforms to a potential 4-residue tyrosinebased internalization signal consensus sequence (21). Furthermore, the sorting signal(s) in the cytoplasmic domain of APP are thought to play an important role in the distribution of APP within the protein secretory pathway (15), although the sorting signal has not been well analyzed. Therefore, using site-directed mutagenesis, we performed further analysis of the sorting signal(s) in the cytoplasmic domain of APP, which may be responsible for the regulation of APP metabolism such as the production of CTF␤ and/or A␤. Identification and characterization of the signal(s) in the APP cytoplasmic domain are important for our understanding of the molecular mechanism that directs APP into a protein secretory pathway where CTF␤ and/or subsequent A␤ production occurs without an FAD mutation.
In this study, we introduced an alanine-scanning mutation into the cytoplasmic domain of APP and established many cell lines that expressed stably transfected APP carrying a single amino acid mutation in its cytoplasmic domain. Among the cell * This work was supported by the Yamanouchi Foundation for Research on Metabolic Disorders and by a grant from the Program for Promotion of Basic Research Activities for Innovative Biosciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Recipient of a Japan Society for the Promotion of Science Research Fellowships for Young Scientists.
§ To whom correspondence should be addressed. Tel./Fax: 81-3-3814-6937; E-mail: tsuzuki@mayqueen.f.u-tokyo.ac.jp. 1 The abbreviations used are: AD, Alzheimer's disease; A␤, ␤-amyloid peptide; APP, ␤-amyloid precursor protein; sAPP, large extracellular amino-terminal domain truncated at the ␣-site (sAPP␣) and/or the ␤-site (sAPP␤); CTF␣, carboxyl-terminal fragment of APP cleaved at the ␣-site; CTF␤, carboxyl-terminal fragment of APP cleaved at the ␤-site; FAD, familial Alzheimer's disease; PCR, polymerase chain reaction; wt, wild-type; PAGE, polyacrylamide gel electrophoresis; ELISA, enzyme-linked immunosorbent assay. lines, cells that expressed APP carrying a mutation at amino acid 747 of APP770 (position 672 of APP695) generated a greater amount of CTF␤, but this did not result in an increase of A␤ secretion, indicating that the intracellular increase in CTF␤ is not necessarily related to an increase in A␤ production. This effect was not due to a simple intracellular accumulation of CTF␤ resulting from a lack of ␥-secretase. These results suggest that the basic amino acid at position 747 of APP770 (position 672 of APP695) plays an important role in the direction of APP into the normal secretory pathway, in which APP is cleaved preferentially at the ␣-site. Thus, APP carrying this mutation may be metabolized in a different protein secretory pathway distinct from the normal one. Our findings provide a new insight into what regulates the processing of CTF␤ by ␥-secretase and the nature of the compartment in which ␥-secretase cleavage occurs in the case of sporadic AD.

EXPERIMENTAL PROCEDURES
Introduction of Substitution Mutation and Plasmid Construction-cDNA encoding human APP770 (10) was cloned from a ZAP HeLa cDNA library by immunoscreening with anti-APP antibody, G-369 (23). A PCR of a human brain cDNA library (CLONTECH) was performed with primers APP695 cDNA-(681-700) (forward, 5Ј-GTTCCGAGGGG-TAGAGTTTG-3Ј) and APP695 cDNA-(1922-1940) (reverse, 5Ј-GCATC-CATCTTCACTTCAG-3Ј). A 1.2-kilobase PCR product was digested with XcmI/BglII, and the resulting fragment (APP695 cDNA-(990 -1916)) was recombined into the larger fragment from XcmI/BglII digestion of APP770 to delete Kunitz-type protease inhibitor (exon 7) and APP770specific (OX-2, exon 8) domains of APP770 and to construct a complete APP695 cDNA (see Fig. 1a) (5). The cDNAs were subcloned into pcDNA3 (Invitrogen) at HindIII/XbaI sites. A sequence of the APP770 extracellular domain between amino acids 379 and 666 was deleted by exclusion of the XhoI/BglII fragment (p⌬APP770wt) (see Fig. 1a) (24). Site-directed mutations were introduced using PCR (25) with the following primers: primer 1, 5Ј-GCCGCGGTCACCCCAGAGGAGGCCCA-CACCTGTCC-3Ј (the underlined nucleotides were changed to produce an EcoO65I site (T to G) and the Arg-to-Ala mutation (CG to GC)) and primer 2, 5Ј-ATTTAGGTGACACTATAGAATAG-3Ј (SP6 promoter primer), in the presence of plasmid p⌬APP770wt and Pwo DNA polymerase (Boehringer Mannheim). Primer 3, 5Ј-TCTGGGGTGACCGCG-GCGTCAACCTCCACC-3Ј (the underlined nucleotide was changed to produce an EcoO65I site (A to C)), and primer 4, 5Ј-TAATACGACT-CACTATAGGG-3Ј (T7 promoter primer), were used in PCR in the presence of plasmid p⌬APP770wt and Pwo DNA polymerase. Both PCR products were digested with EcoO65I, ligated, and then inserted into pcDNA3 at the HindIII/XbaI sites. The resulting plasmid, p⌬APP770R747A, encodes an amino acid sequence identical to that of ⌬APP770wt except for the substitution of Ala for Arg 747 . The introduction of the EcoO65I site did not change the amino acid sequence of the APP protein. A similar procedure was used to introduce other single amino acid substitutions into the APP cytoplasmic domain. The sites of restriction enzymes and the positions of primers used in this study are shown in Fig. 1a.
Detection of Intracellular Carboxyl-terminal Fragments of APP-293 cells were transfected with plasmid DNA, and several independent clones stably expressing APP were isolated for each construct. Identical results for APP metabolism were obtained from several independent clones, although the level of APP expression varied among the clones. The cells were lysed, and the CTFs of APP were immunoprecipitated (26) using a polyclonal anti-APP cytoplasmic domain antibody, UT-421, or a monoclonal anti-A␤ antibody, 2D1 (24). The immunoprecipitates were subjected to SDS-PAGE (15% (w/v) polyacrylamide) using a modified version (27) of the procedure of Laemmli (28). Samples were transferred to a nitrocellulose membrane (BA83, Schleicher & Schull), and the membrane was probed with antibody UT-421 and 125 I-protein A (IM144, Amersham Pharmacia Biotech), and then analyzed by autoradiography using a Fuji BAS 2000 imaging analyzer or by ECL (Amersham Pharmacia Biotech).
Pulse-Chase Study-Pulse-chase labeling of cells was carried out with [ 35 S]methionine (1 mCi/ml; NEG-072, NEN Life Science Products). 293 cells expressing APP695wt and APP695 R672A were labeled metabolically for 30 min, followed by a chase period as indicated. The chase was initiated by replacing the labeling medium with medium containing excess unlabeled methionine. CTF and APP were immunoprecipitated from the cell lysate using antibody UT-421 and separated by SDS-PAGE (15% (w/v) polyacrylamide) (27). sAPP was isolated from the medium by immunoprecipitation with monoclonal anti-APP extracellular domain antibody 22C11 (Boehringer Mannheim) and separated by SDS-PAGE (7.5% (w/v) polyacrylamide) (26). Radioactivity in CTF␣, CTFs␤, sAPP, and APP was analyzed using the Fuji BAS 2000 imaging analyzer.
ELISA Analysis-The conditioned medium from cells (2 ϫ 10 6 cells) was collected 18 -20 h after a medium change. A␤40 and A␤42 were quantified using sandwich ELISA with three types of A␤-specific monoclonal antibodies as described previously (24). Briefly, wells were coated with the A␤40 (4D1) or A␤42 (4D8) end-specific monoclonal antibodies (0.3 g of IgG in phosphate-buffered saline (140 mM NaCl and 10 mM sodium phosphate, pH 7.2)), washed with phosphate-buffered saline containing 0.05% (v/v) Tween 20 (washing buffer), blocked with bovine serum albumin (3% (w/v) in phosphate-buffered saline), and washed with washing buffer, and then a sample (100 l) diluted suitably with washing buffer containing 1% (w/v) bovine serum albumin (dilution buffer) was incubated together with a standard amount of synthetic A␤-(1-40) and A␤-(1-42) peptides (synthesized at the W. M. Keck Foundation Biotechnology Resource Laboratory, Yale University). After washing, wells were treated with biotinized 2D1 (12. 5 ng in dilution buffer), washed, and incubated with 100 l of a streptavidin-horseradish peroxidase complex (1:2000 dilution; RPN1051, Amersham Pharmacia Biotech). The plates (96 wells) were further washed, and 100 l of ABTS solution (KPL 5062-01, Kirkegaard & Perry Laboratories Inc., Gaithersburg, MD) was added to the wells and incubated at room temperature. Reactions were stopped by adding 100 l of 1% (w/v) SDS, and the absorbance at 405 nm was measured.

Production of Intracellular Carboxyl-terminal Fragments (CTF␣ and CTF␤) in Cells That Express APP Carrying a Single
Amino Acid Mutation in Its Cytoplasmic Domain-It is well known that APP is cleaved preferentially at the ␣-site to produce CTF␣ (Fig. 1a) (29). Intracellular products of APP cleaved at the ␤-site are extremely rare in healthy cells if APP does not carry an FAD mutation such as the Swedish double amino acid substitution (30). However, the production of intracellular CTF␤ (Fig. 1a) and the secretion of A␤ increase in the sporadic type of AD even if the resident APP gene carries an entire sequence (31,32). In the case of sporadic AD, one of the most reasonable explanations for the increased production of CTF␤ and the secretion of A␤ is abnormality in the protein secretory pathway. To explore the signals that direct APP into the protein secretory pathway, we introduced a series of single amino acid substitutions into the 735-755-amino acid region of the cytoplasmic domain of ⌬APP770, which lacks the 287-amino acid region between amino acids 379 and 666 of APP770, but undergoes identical intracellular metabolism as endogenous APP in transfected cells (24). The mutants were then examined for their level of production of APP carboxyl-terminal fragments, CTF␣ and CTF␤ (Fig. 1). The two known functional internalization signal sequences, NPTY and YTSI (underlined in Fig. 1b), were excluded from the analysis because they have already been well characterized (21). 293 cells (human transfected primary embryonal kidney) were transfected with wild-type (p⌬APP770wt) and mutant (p⌬APP770mt) plasmids (see "Experimental Procedures" for plasmid construction), and then several independent clones of cells expressing stably transfected ⌬APP770 were isolated. CTF was recovered from 293 cells expressing wild-type (⌬APP770wt) or mutant (⌬APP770mt) APP by immunoprecipitation with polyclonal anti-APP cytoplasmic domain antibody UT-421 (epitope(s) exists within APP770-(759 -770)) and was detected by immunoblot analysis using the same antibody (Fig.  2). CTF␣, which is a product of cleavage by ␣-secretase, was detected in 293 cells expressing ⌬APP770wt, (Fig. 1a (part ii) and arrows in Fig. 2 (a and b)). The CTF␣ from endogenous APP was below the detectable levels (data not shown). CTF␤ with a higher molecular mass than expected was detected ( Fig.  1a (part iii) and arrowheads in Fig. 2 (a and b)) in 293 cells expressing the ⌬APP770 R747A (substitution of Ala for Arg 747 ) mutant protein. A weak CTF␤ band (20 -30% of the amount of CTF␣ in terms of radioactivity) was also detectable in cells expressing ⌬APP770wt after longer exposure of the autoradiogram (Fig. 2, b and c) or after metabolic labeling of protein with [ 35 S]methionine (see Fig. 5a). A moderate amount of CTF␤ (ϳ50% of the amount of CTF␣ in terms of radioactivity) was also detectable in cells expressing ⌬APP770 L749A (substitution of Ala for Leu 749 ) and ⌬APP770 K751A (substitution of Ala for Lys 751 ) mutants (arrowheads in Fig. 2 (b and c)). Although the differences in the production level of CTF␣ among the substitution mutants reflected the expression level of ⌬APP770mt in the cell, none of the alanine-scanning mutants, except for ⌬APP770 R747A, containing a substitution within the 735-755-amino acid region of APP770, led to greater CTF␤ production (ϳ130% of the amount of CTF␣) when compared with ⌬APP770wt (see the ratio CTF␤/CTF␣ in Fig. 2c and data not shown for the site between Val 735 and Thr 743 ). This result suggests that the argininyl residue at amino acid 747 of APP770, or possibly the domain consisting of 5 amino acids from Arg 747 to Lys 751 of APP770, plays an important role in APP cleavage at the ␣-site.
To further analyze the function of the amino acid at position 747 of APP770, we substituted another basic amino acid, lysine (⌬APP770 R747K), and an acidic amino acid, glutamic acid (⌬APP770 R747E) for Arg 747 (Fig. 4a). Identical substitution mutations were also introduced into the entire length of APP695, a neuron-specific isoform (Fig. 4b), as described under "Experimental Procedures." The 293 cells stably expressing these APP proteins with a mutation at amino acid 747 of ⌬APP770 (position 672 of APP695) were also isolated and analyzed for intracellular CTF␤ production. The production of the heterogeneous CTF␤ species was also clearly observed when Arg 747 of ⌬APP770 (Arg 672 of APP695) was replaced by Glu (⌬APP770 R747E and APP695 R672E), but not when replaced by Lys (⌬APP770 R747K and APP695 R672K) (arrowheads in Fig. 4). Because only small quantities of CTF␤ 3 and CTF␤ 4 were usually detected (Figs. 2 and 3), all CTF␤ species cannot be observed in Fig. 4. The substitution of the acidic amino acid at position 747 of ⌬APP770 (position 672 of APP695) with a basic amino acid seemed to have more drastic effects on the production of the CTF␤ species compared with that seen after substitution with alanine. The slight upward mobility shift of the substitution mutants of CTF␣ (arrows) and CTF␤ (arrowheads) may be due an alteration of the charge of CTF (Figs. 2 and 4) because this shift was not observed in ⌬APP770 R747K or APP695 R672K, where the basic amino acid was replaced by other basic amino acids (Fig. 4). There was no difference between the ⌬APP770 and APP695 isoforms with respect to these mutations. This result indicates that the basic amino acid at position 747 of ⌬APP770 (position 672 of APP695) is essential for the cleavage of APP at the ␣-site and that the substitution of a non-basic amino acid with a basic amino acid alters APP metabolism in the protein secretory pathway.
Intracellular Generation and Degradation of CTFs-To examine for eventual differences between the intracellular generation and degradation of CTF␣ and CTF␤ deriving from APP695 R672A and those derived from APP695wt, a pulsechase study was performed (Fig. 5). Cells expressing APP695 R672A and APP695wt were labeled with [ 35 S]methionine and chased for the indicated times (Fig. 5a). The radioactivities of CTF␣ and CTF␤ 1-4 were quantified, and the relative ratios with respect to the maximum level were calculated (Fig. 5b for CTF␣ and Fig. 5c for CTF␤ [1][2][3][4] ). CTF␣ (squares) and CTF␤ (circles) derived from APP695 R672A (closed symbols) and APP695wt (open symbols) were produced and degraded at almost the same rate (Fig. 5, b and c). These results clearly indicate that both CTF␣ and CTF␤ derived from APP695 R672A are generated and degraded in the same way as those derived from APP695wt, except for APP695 R672A, which was more efficiently cleaved by ␤-secretase. Thus, the increase in the level of CTF␤ resulting from the APP695 R672A mutation, which was detected by immunoblot analysis (Figs. 2-4), is not due to a simple intracellular accumulation of CTF␤. The CTFs from APP695 R672A, as well as those from APP695wt, are subject to further degradation.
Secretion of sAPP from Cells Expressing APP695 R672A-Because the large extracellular amino-terminal domain truncated at ␣or ␤-sites (sAPP␣/␤) is believed to be always secreted into the medium, we examined the secretion of sAPP␣/␤ derived from APP695 R672A and compared it with that derived from APP695wt. The culture medium of the pulse-chase studies (Fig. 5) was collected; sAPP␣/␤ was recovered by immunoprecipitation with anti-APP extracellular domain antibody 22C11; and the immunoprecipitates were subjected to SDS-PAGE (7.5% (w/v) polyacrylamide), followed by autoradiography (Fig. 6a). It was not possible to distinguish sAPP␤ from sAPP␣ on SDS-PAGE as described previously (33). The arrow in Fig. 6a indicates a band containing both sAPP␣ and sAPP␤. The sAPP␣/␤ from APP695 R672A (closed circles) appeared slightly earlier than that from APP695wt (open circles) in the medium (Fig. 6b). The secretion of sAPP␣/␤ reached the maximum level after 3 h (APP695 R672A) or after 5 h (APP695wt). The slight difference may be due to the fact that sAPP from APP695 R672A probably contains more sAPP␤ than sAPP from APP695wt. This result suggests the possibility that sAPP␤ may be secreted earlier than sAPP␣, although we did not detect any remarkable differences between the metabolic rates of intracellular APP695 (Fig. 6c) and CTFs (Fig. 5). The secretion level of sAPP␣/␤ from APP695 R672A seemed to be low when compared with the level of APP expression (Fig. 6a). Therefore, the relative ratio of the level of sAPP␣/␤ to that of intracellular APP was determined at different time points during the pulsechase experiment (Fig. 6d). The secretion level of sAPP␣/␤ from APP695 R672A was lower (ϳ50% at 3 h and ϳ30% at 5 h of APP695wt) compared with that from APP695wt when the values were standardized with the level of APP expression (Fig.  6d). This may be a reason why the secretion of sAPP from APP695 R672A reached a plateau at 3 h. This result suggests that sAPP␣/␤ from APP695 R672A is not always secreted into the medium and that a certain amount of sAPP␣/␤ from  [1][2][3][4] from APP695wt (left) and APP695 R672A (right). b and c, the incorporation of radioactivity into CTF was determined as described under "Experimental Procedures," and the relative ratios of the levels of CTF␣ (b) and CTF␤1-4 (c) to maximum levels, which were assigned a reference value of 1.0, were calculated. Results are the average of duplicate assays. Error bars are indicated. APP695 R672A is subject to intracellular degradation prior to secretion.
Secretion of A␤ from Cells Expressing APP695mt-If CTF␤ is always cleaved by ␥-secretase, cells expressing APP695 R672A should generate a greater amount of A␤, as the pulse-chase study of Fig. 5 indicates that CTF␤ from APP695 R672A did not accumulate intracellularly, instead but was subject to further metabolism. To examine this possibility, the amount of A␤ in the medium was analyzed with sandwich ELISA using three types of monoclonal anti-A␤ antibodies: 2D1, which recognizes the human-specific amino acid sequence FRH (amino acids 600 -602) of APP695 in the A␤ sequence; 4D1 (epitope is the terminal of A␤40), which recognizes APP derivatives truncated at A␤40; and 4D8 (epitope is the terminal of A␤42), which recognizes APP derivatives truncated at A␤42 (24). Combinations of antibodies 2D1 and 4D1 or 4D8 in sandwich ELISA allows the specific measurement of the amount of A␤40 or A␤42 secreted into the medium. We established additional 293 cell lines that stably expressed APP695 carrying an FAD mutation and APP695 carrying both the FAD mutation and an R672A substitution. We did this because it is well known that APP carrying an FAD mutation is subject to active ␤-site cleavage and subsequent secretion of A␤ species following the cleavage of CTF␤ by ␥-secretase (30,34,35). When 293 cells (ϳ2 ϫ 10 6 cells) expressing APP695wt were cultured in 3 ml of medium for 24 h and 100 l of the medium was quantified for A␤, usually 40 -50 fmol of A␤40 and 20 -30 fmol of A␤42 were detected. Because the amount of A␤ secretion is affected by the level of APP expression in the established cell lines, we have indicated the secretion level of A␤ as a ratio of the level of APP expression that was determined from the intracellular APP content (Fig. 7). The APP695 R672A mutation suppressed the production of A␤40 more efficiently than that of A␤42. The data suggest that the APP695 R672A mutation does not contribute to the generation the A␤40 and A␤42, at least in 293 cells, despite the fact that a greater amount of CTF␤ was generated intracellularly. This result also indicates that the increase in CTF␤ does not always result in increased secretion of A␤ and that CTF␤ derived from the APP695 R672A mutation is subject to degradation prior to cleavage by ␥-secretase.
The Swedish mutation (APP695sw), which is thought to enhance the production of A␤ by increasing its sensitivity to ␤-secretase (30,34,36,37), caused ϳa 23-fold increase in A␤40 and a 6-fold increase in A␤42 production. Another FAD mutation, the "London (Hardy)" mutation (APP695 V642F) (38,39), also caused a remarkable increase in A␤40 and A␤42 secretion. The results of this study are consistent with these previous reports describing the effect of the FAD mutation on A␤ production (35). Interestingly, the level of A␤40 secretion from cells carrying a double mutation of R672A and FAD, Swedish R672A/sw and London R672A/V642F, was less than that from cells carrying the original FAD mutation alone (Fig. 7a). This result indicates that the R672A mutation induces a decrease in the A␤40 secretion from cells carrying the FAD mutation, as in the case for wild-type APP. An identical double mutation of R672A and FAD caused no remarkable decrease in A␤42 secretion (Fig. 7b), which suggests that the secretion of A␤42 may FIG. 6. Secretion rate of sAPP, intracellular metabolic rate of APP, and secretion level of sAPP. sAPP␣/␤ was recovered from the medium, and APP was recovered from the cell lysate of the pulse-chase experiment (see Fig. 5) by immunoprecipitation. a, shown are autoradiograms of sAPP␣/␤ from APP695wt (left) and APP695 R672A (right). b and c, the incorporation of radioactivity into sAPP␣/␤ and APP was determined as described under "Experimental Procedures," and the relative ratios of the levels of sAPP␣/␤ (b) and APP (c) are indicated relative to the maximum levels, which were assigned a reference value of 1.0. d, the level of sAPP␣/␤ secretion was normalized to the level of APP expression and assigned an sAPP/APP ratio. Results are the average of duplicate assays, and error bars are indicated.

FIG. 7. Secretion of A␤ from cells expressing APP695wt and
APP695mt. A␤40 (a) and A␤42 (b) in the medium were quantified by sandwich ELISA with A␤40 (4D1) and A␤42 (4D8) end-specific antibodies. To estimate the level of APP695 expression, APP from cells stably transfected with plasmids was immunoprecipitated, detected by immunoblotting with antibody UT-421 following SDS-PAGE (7.5% (w/v) polyacrylamide gel), and quantified using a Fuji BAS 2000 imaging analyzer (data not shown). The level of APP695 expression (a relative ratio) was normalized to the amount of APP from a clone expressing APPwt. The amount of A␤ (fmol/100 l of medium) was divided by the relative APP695 ratio and is expressed as the A␤/APP695 ratio. wt, cells expressing APP695wt; R672A, cells expressing the APP695 R672A mutation; V642F, cells expressing APP695 V642F (London mutation); R672A/V642F, cells expressing a double mutation of APP695 R672A and APP695 V642F; sw, cells expressing the Swedish mutation (a double mutation of APP695 K595N and APP695 M596L); R672A/sw, cells expressing a double mutation of APP695 R672A and the Swedish mutation. Results are the average of several independent studies with independent clones (n ϭ 4 -7), and error bars indicate S.D. *, p Ͻ 0.01; ***, p Ͻ 0.001. be different from that of A␤40. Furthermore, these observations support the notion that APP carrying the R672A mutation is cleaved in a secretory pathway that differs from the normal secretory pathway, where APP carrying the FAD mutation is cleaved. DISCUSSION Previous studies in cells expressing APP with a truncated cytoplasmic domain have not been able to clarify the function of this cytoplasmic domain in detail because the cytoplasmic domain is thought to contain multiple signal domains regulating APP metabolism (15,40). Therefore, we tried to explore the intracellular sorting signals in the APP cytoplasmic domain with an alanine-scanning mutation, a single amino acid substitution of alanine for the original amino acid. We focused on the production of intracellular CTFs in cells expressing APP carrying the alanine-scanning mutation because an alteration of the protein secretion system should first of all induce CTF␤ generation and because the APP cytoplasmic domain may function to direct APP into the normal secretory pathway. In this study, we found that the basic amino acid at position 747 of APP770 (position 672 of APP695) in the cytoplasmic domain plays an important role in the cleavage of APP at the ␣-site. This basic amino acid seems to be essential for the normal metabolism of APP in the normal secretory pathway, in which APP is cleaved preferentially by ␣-secretase rather than by ␤-secretase when APP does not carry the FAD mutation (Fig.  8). The increased production of CTF␤ derived from APP695 R672A did not result in increased secretion of A␤ and sAPP. The most likely interpretation of these observations is that CTFs␤ do not migrate to the compartment of the normal secretory pathway and are subject to degradation prior to cleavage by ␥-secretase (Fig. 8). The alternative possibility, that the argininyl residue at amino acid 672 of APP695 is critical for ␥-secretase recognition, does not seem plausible because the production of A␤42, at least, was not affected by the R672A mutation of APP695 and because a double mutation of R672A and FAD increased A␤ secretion above the level seen with the R672A mutation alone. Therefore, we postulate a default secretory pathway in which APP carrying the R672A mutation is cleaved by ␤-secretase as well as by ␣-secretase (Fig. 8).
The present observation also showed that CTF␤ does not always generate A␤ and that sAPP is not always secreted, although the increased production of CTF␤ is thought to result in increased secretion of A␤ and sAPP in the case of APP carrying the FAD mutation (Fig. 8). The decreased secretion of sAPP derived from APP695 R672A may be due to the intracellular degradation of sAPP in the default secretory pathway. CTFs from the APP695 R672A mutation, as well as CTFs from APP695wt, do not accumulate in the cell and are degraded. This may be the reason why CTF␤ does not generate A␤ in the case of the APP695 R672A mutation. These facts may be relevant to A␤ production in the case of the sporadic type of AD. The molecular mechanism that results in APP695 migration into a default secretory pathway that would be enriched in ␤-secretase in the R672A mutant has not been elucidated. Further analyses of the intracellular metabolism of APP695 R672A will be important for our understanding of what regulates ␥-secretase processing of CTF␤ and determines ␥-secretase activation in the default secretory pathway. APP695 R672A may be directed into a default secretory pathway by a putative cytoplasmic factor that can associate with the APP cytoplasmic domain sequence around Arg 747 of FIG. 8. Putative secretory pathways of intracellular APP metabolism. Cells are postulated to have at least two protein secretory pathways, the normal (upper) and default (lower) secretory pathways of APP secretion. In healthy (normal) cells, APP is generally directed into the normal secretory pathway, in which APP is cleaved preferentially at the ␣-site. An FAD mutation such as the Swedish mutation (**) enhances the amyloidogenic metabolism of APP, the cleavage of APP at the ␤-site, in the normal secretory pathway. Then, CTF␣/␤ is always cleaved by ␥-secretase, and the resulting A␤ and p3 fragment are secreted from the cell. sAPP␣/␤ is also secreted. On the other hand, the APP770 R747A/APP695 R672A mutation is thought to direct APP into the default secretory pathway due to inefficient function of putative cytoplasmic factor(s). APP in the default secretory pathway is cleaved at the ␤-site as well as the ␣-site. The resulting sAPP␣/␤ is not always secreted, and CTF␣/␤ is not always subject to subsequent cleavage by ␥-secretase. Some sAPP␣/␤ and CTF␣/␤ is degraded intracellularly prior to secretion. When the regulatory steps of ␥-secretase in the default secretory pathway become aberrant for some reason such as aging, the relatively large amount of CTF␤ will be further cleaved by ␥-secretase prior to degradation, and an increased amount of A␤ will be generated. This may explain how the pathogenic state of sporadic AD continues to develop.
APP770 and Arg 672 of APP695. The protein may play a role in the direction of APP into the normal secretory pathway (Fig. 8). The effect of the APP770 R747A (APP695 R672A) mutation on CTF␤ production probably results from interference with the binding of this hypothetical factor to the APP cytoplasmic domain. One recent report, which indicates that transgenic mice expressing CTF␤ develop extracellular A␤ with age (41), may be in support of this idea. Aging may induce a loss of function of the cytoplasmic factor and activate ␥-secretase activity in a default secretory pathway. Another recent report, which indicates that knockout mice for presenilin-1 have increased levels of CTF␤, but decreased secretion of A␤ (42), may also be relevant to our present results. The basic amino acid (Arg 672 ) of APP695 may play an important role in the APP secretory pathway related to presenilin-1.
Furthermore, when compared with the CTF␤ production in cells expressing APP695 R672A, secretion of A␤ into the medium was reduced slightly (A␤40) or did not change (A␤42). This result is not necessarily cell-type specific, at least in non-neuronal cells, because identical results, i.e. increased CTF␤ but unchanged secretion of A␤, were obtained when APP695 R672A was stably expressed in Chinese hamster ovary cells (data not shown). Double mutants of APP695 R672A and FAD also suppressed the secretion of A␤40, but not A␤42, when compared with the FAD-only mutant, indicating that the mechanism of A␤42 production is different from that of A␤40 production. It has been shown clearly that the FAD mutation increases the secretion of both A␤40 and A␤42, which are thought to be generated in the normal secretory pathway (30,34,36,37). Therefore, these results suggest that the cleavage of APP695 R672A by ␤-secretase occurs in a compartment different from that used for the cleavage of APPwt and APP carrying the FAD mutation at their ␤-sites (Fig. 8). Further analysis of the metabolism of APP695 R672A in this default secretory pathway should contribute to our understanding of the molecular mechanism of APP processing in the sporadic type of AD pathogenesis.