Subcellular compartment and molecular subdomain of beta-amyloid precursor protein relevant to the Abeta 42-promoting effects of Alzheimer mutant presenilin 2.

Increased production of amyloid beta peptides ending at position 42 (Abeta42) is one of the pathogenic phenotypes caused by mutant forms of presenilins (PS) linked to familial Alzheimer's disease. To identify the subcellular compartment(s) in which familial Alzheimer's disease mutant PS2 (mt PS2) affects the gamma-cleavage of betaAPP to increase Abeta42, we co-expressed the C-terminal 99-amino acid fragment of betaAPP (C100) tagged with sorting signals to the endoplasmic reticulum (C100/ER) or to the trans-Golgi network (C100/TGN) together with mt PS2 in N2a cells. C100/TGN co-transfected with mt PS2 increased levels or ratios of intracellular as well as secreted Abeta42 at similar levels to those with C100 without signals (C100/WT), whereas C100/ER yielded a negligible level of Abeta, which was not affected by co-transfection of mt PS2. To identify the molecular subdomain of betaAPP required for the effects of mt PS2, we next co-expressed C100 variously truncated at the C-terminal cytoplasmic domain together with mt PS2. All types of C-terminally truncated C100 variants including that lacking the entire cytoplasmic domain yielded the secreted form of Abeta at levels comparable with those from C100/WT, and co-transfection of mt PS2 increased the secretion of Abeta42. These results suggest that (i) late intracellular compartments including TGN are the major sites in which Abeta42 is produced and up-regulated by mt PS2 and that (ii) the anterior half of C100 lacking the entire cytoplasmic domain is sufficient for the overproduction of Abeta42 caused by mt PS2.

Alzheimer's disease (AD) 1 is a progressive dementing disorder characterized pathologically by a massive loss of cortical neurons and an accumulation of two types of fibrillar lesions: i.e. amyloid deposits composed of amyloid ␤ peptides (A␤) and tau-rich paired helical filaments (1). A␤ is produced from ␤-amyloid precursor proteins (␤APP) through sequential cleavages by proteases originally termed ␤and ␥-secretases (1, 2); ␤-secretase has recently been identified as a novel aspartyl protease, BACE (3)(4)(5). Deposition of ␤-amyloid is considered to be closely related to the pathogenesis of AD because (i) deposition of A␤ is a neuropathological change relatively specific to AD; (ii) the diffuse type of senile plaque composed of highly aggregable A␤42 species (6,7), as opposed to A␤40 that comprises the major portion of the secreted form of A␤ (8,9), is the initial lesion of AD pathology; and (iii) mutations in genes coding for ␤APP (10 -14) or presenilin 1 (PS1) (15) or 2 (PS2) (16) are linked to some pedigrees of autosomal dominantly inherited familial AD (FAD), and these mutations increase the production of A␤42 species (12)(13)(14)(17)(18)(19)(20). Mutations in PS genes that code for multipass integral membrane proteins account for the majority of early onset FAD. Studies in knockout mice or invertebrates demonstrated that PS is involved in ␥-cleavage of ␤APP (2,21,22) as well as in site 3 cleavage of the Notch receptor (2,(23)(24)(25), both of which occur within the membrane or at the junction with cytoplasm, although it has not been clear if PS is a co-factor for ␥-cleavage or if PS is identical to ␥-secretase. However, recent data showing that transition state analogue ␥-secretase inhibitors directly and exclusively bound fragment forms of PS strongly support the hypothesis that PS represents the catalytic subunits of ␥-secretase (26 -28).
Although there is ample evidence that mutations in PS genes increase the production of A␤42 (29 -32), the intracellular compartment(s) in which mutant forms of PS interact with ␤APP and promote ␥-cleavage at the A␤42 position has not been clearly identified. Generation of intracellular A␤42 has been shown to occur in endoplasmic reticulum (ER) of cultured neurons (33,34) or in human embryonic kidney 293 cells (35), whereas the trans-Golgi network (TGN) (36) or endocytic pathway (37,38) also are implicated in the generation of secretable A␤42. Although the ER localization of PS dovetails with the former data, others have suggested that Golgi may be related to the abnormal effect of mt PS1 to increase A␤42 (39,40). Furthermore, subdomains in ␤APP proteins that are required for this interaction with mt PS to increase production of A␤42 have not been definitively identified. In this study, we studied the intracellular compartment and intramolecular subdomain of ␤APP that are relevant to the abnormal effects of mutant PS2 to affect ␥-cleavage and increase production of A␤42. For these purposes, we expressed modified forms of a C-terminal fragment of ␤APP tagged with targeting signals to specific com-partments or harboring deletion of defined cytoplasmic subdomains. We show here that TGN and other late intracellular compartments are the major sites where mt PS2 up-regulates A␤42 production and that the cytoplasmic domain of ␤APP is dispensable for the overproduction of A␤42 caused by mt PS2.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmids-A cDNA encoding the Cterminal 99 amino acids of ␤APP fused to a signal peptide for rat preproenkephalin cDNA (C100) was previously described (32,41). C100 peptides tagged with sorting signals to ER or TGN were generated using a C100 cDNA constructed in mammalian vector p91023 as a template. Briefly, oligonucleotides encoding the rat preproenkephalin signal peptide were used as a sense polymerase chain reaction primer: 5Ј-TTTAAGCTTCCACCATGGCGCAGTTCCTG-3Ј. The following oligonucleotides encoding the last four C-terminal amino acid residues of C100 (i.e. QMQN) followed by the signal sequences KKLN (for ER) or SDYQRL (for TGN) (42) were used as antisense polymerase chain reaction primers: 5Ј-CCCGGATCCCTAATTCAGATTATTGTTCTGCA-TCTG-3Ј for C100/ER, 5Ј-CCCGGATCCCTAGAGCCGCTGATAATCG-AAGTTCTGCATCTG-3Ј for C100/TGN, and 5Ј-AAAGGATCCCTAGTT-CTGCATCTG-3Ј for C100 (without signal motif). Amplification of cDNAs was performed using PfuTurbo DNA polymerase (Strategene). Amplified DNA fragments were digested with HindIII and BamHI and ligated into a pcDNA3.1-Hygro vector (Invitrogen). ␤APP695 tagged with KKLN or SDYQRL motifs were constructed by ligating the EcoRI/XbaI fragments of tagged C100 with those of ␤APP695 cDNA in pcDNA3.
Cell Culture, Transfection, and Caspase Inhibitor Treatment-Mouse neuro2a (N2a) neuroblastoma cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and penicillin/streptomycin at 37°C in a 5% CO 2 atmosphere as described (32,43,44). Transient transfection of C100 cDNAs into N2a cells and co-expression of C100 cDNAs in N2a cells stably expressing human PS2 cDNAs (43) were performed using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. Expression of transfected proteins was enhanced by treatment with 10 mM butyric acid for 24 h prior to harvesting cells or culture supernatants. For the inhibition of caspase activities, cells were treated with a 100 M concentration of a pancaspase inhibitor, zVAD-fmk, for 24 h prior to analysis by Western blotting.
Immunoblot Analysis of ␤APP or PS2 Derivatives and Cell-associated A␤-Cells were lysed in 2% SDS sample buffer and briefly sonicated. Samples were separated by SDS-polyacrylamide gel electrophoresis using a Tris-Tricine gel system, transferred to polyvinylidene difluoride membrane (Millipore Corp.), and probed with monoclonal antibodies BAN50 (specific for human A␤1-16) for the detection of C100 derivatives. A rabbit polyclonal antibody anti-G2N4 raised against a recombinant protein corresponding to the N-terminal residues 2-59 of human PS2 was used to probe PS2 and its derivatives. For the detection of A␤ in RIPA-soluble fractions of cell lysates, samples were initially lysed in RIPA (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate), and the supernatants after centrifugation at 15,000 ϫ g for 5 min were immunoprecipitated by BAN50 using protein G-agarose and then analyzed by immunoblotting with BA27, BC05, or BAN50, using previously described procedures (44,45,47). Extraction of cell-associated A␤ by formic acid was performed as described (46). Briefly, cell pellets confluently grown in two 10-cm dishes were solubilized by ultrasonication followed by incubation in 100 l of 70% formic acid at room temperature for 30 min. Supernatants after centrifugation at 100,000 ϫ g for 20 min were desiccated and then solubilized in 100 l of SDS sample buffer. Samples containing A␤ were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and reacted with antibodies after boiling (44,45,47). The immunoblots were developed using an ECL system (Amersham Pharmacia Biotech) or Immunostar (Wako Pure Chemicals).
Immunofluoresence Microscopy-Transiently transfected N2a cells were cultured on glass coverslips for 48 h. Cells were fixed by incubation with phosphate-buffered saline (10 mM phosphate buffer, pH 7.4) containing 4% paraformaldehyde for 30 min at room temperature and then permeabilized and blocked with PBS-TB (phosphate-buffered saline containing 150 mM NaCl, 0.1% Triton X-100 and 3% bovine serum albumin) for 30 min at room temperature. Coverslips were then incubated with primary antibodies (a rabbit polyclonal antibody, C4, against the C terminus of ␤APP (48) and monoclonal antibodies specific for BiP or adaptin-␥) for 2 h followed by an incubation with a mixture of fluorescein isothiocyanate-conjugated anti-rabbit IgG and Texas Redconjugated anti-mouse IgG antibodies in PBS-TB for 1 h and then mounted in PermaFlour aqueous mounting medium (Immunon) and viewed with a confocal microscope (Fluoroview, Olympus, Tokyo) as described (43). BAN50 and fluorescein isothiocyanate-conjugated secondary antibody were used for single immunofluorescence detection of C100 derivatives.
Quantitation of A␤ by Two-site ELISAs-Two-site ELISAs that specifically detect the C terminus of A␤ were used as described. BAN50, which was used as a capture antibody, binds only human ␤APP or A␤ and does not cross-react with rodent A␤ or with an N-terminally truncated fragment (e.g. p3). BA27 and BC05 that specifically recognize the C terminus of A␤40 and A␤42, respectively, were conjugated with horseradish peroxidase and used as detector antibodies. Culture medium was collected after an appropriate incubation period (48 h) and subjected to BAN50/BA27 or BAN50/BC05 ELISAs as described (32,43,44). Cell-associated A␤ was quantitated after solubilization in 1% Nonidet P-40. ELISA data were statistically analyzed by analysis of variance using StatView-J.4.11.

RESULTS
Effects of FAD Mutant PS2 on A␤ Production from ␤APP C100 Targeted to ER or TGN-To identify the intracellular compartments where A␤, especially A␤42, is generated and destined to be secreted, we transiently expressed cDNAs coding for the C-terminal 99 amino acids of human ␤APP harboring a signal peptide at the N terminus (C100) or C100 tagged with sorting signals for retention to ER (C100/ER) or for recycling to TGN (C100/TGN) tagged at the C terminus in mouse N2a cells (Fig. 1, A-C). C100, C100/ER, or C100/TGN was expressed as a major ϳ13-kDa and a minor ϳ8-kDa polypeptide on immunoblots, the latter corresponding to fragments cleaved by caspases (see below), and the banding patterns were similar between C100 with and without sorting signals (Fig. 1B). Immunocytochemistry by C4 (against the cytoplasmic tail of ␤APP (48)) combined with anti-BiP antibody that specifically reacts with a KDEL sequence (ER marker) or anti-adaptin-␥ antibody (TGN marker) showed retention of C100/ER in a meshwork like pattern overlapping with an immunolabeling with the ER marker (Fig. 1C, top left), whereas immunoreactive pattern for C100/TGN completely overlapped with that for adaptin-␥ (Fig. 1C, middle right), suggesting proper localization of C100 variants at the intended sites. C100/WT showed a combined ER and TGN localization (Fig. 1C, lower panels). We then quantitated A␤-(1-40) and A␤-(1-42) secreted from N2a cells expressing C100 variants by two-site ELISAs using BAN50 as a capture antibody, which specifically detects human A␤ but not endogenous murine A␤ (Fig. 1D). Cells transfected with C100/TGN secreted ϳ2000 pM A␤-(1-40) and ϳ200 pM A␤- , which were at comparable levels with those secreted from cells expressing C100/WT. In contrast, C100/ER did not secrete detectable levels of A␤-(1-40) or A␤-(1-42) as in N2a cells transfected with an empty vector. These results suggested that TGN would be the major intracellular site in which ␥-cleavage to yield A␤-(1-40) as well as A␤-(1-42) that are destined for secretion takes place.
We then expressed C100/ER, C100/TGN, or C100/WT in N2a cells that stably express WT or N141I or M239V FAD mt PS2 and examined the production of secreted or intracellular A␤. N2a cells expressing WT PS2 transiently transfected with C100/WT or C100/TGN secreted similar levels of A␤-(1-40) and A␤-(1-42), whereas those with C100/ER did not secrete detectable levels of A␤, in almost identical patterns to those observed in cells without exogenous PS2. In contrast, cells expressing mt PS2 transiently transfected with C100/WT or C100/TGN secreted larger amounts of A␤-(1-42) compared with A␤-(1-40), whereas those with C100/ER did not secrete detectable A␤ ( Fig.  2A). These data suggest that TGN or later intracellular compartments, but not ER, are the intracellular site where mt PS2 affects ␥-cleavage of ␤APP to promote secretion of A␤42 in N2a cells.
To verify the intracellular production of A␤-(1-40) and A␤-(1-42) from C100 with or without sorting signal motifs, we analyzed RIPA-extracted lysates of N2a cells expressing C100/ WT, C100/ER, or C100/TGN together with WT or N141I mt PS2 by Western blotting with antibodies to ␤APP (i.e. BAN50 against the A␤ N terminus) or with those against C termini of A␤ after immunoprecipitation with BAN50 (Fig. 2B). C100/WT, C100/ER, or C100/TGN were expressed as ϳ13-kDa as well as ϳ8-kDa polypeptides as observed without co-expression of PS2. In N2a cells co-expressing WT PS2, C100/TGN yielded a ϳ4-kDa polypeptide positive for A␤40 as well as an equally intense A␤42-positive ϳ4-kDa polypeptide, in a similar pattern to those observed with C100/WT. In N2a cells expressing N141I or M239V mt PS2, however, C100/WT and C100/TGN yielded comparable levels of A␤42-positive 4-kDa bands, whereas only trace amounts of A␤40-positive bands were detected in cell lysates, despite robust expression of C100 and its derivatives. In contrast, detectable levels of A␤40 or A␤42-positive polypeptides were not observed in Western blots of cell lysates expressing C100/ER together with WT or N141I or M239V mt PS2. To confirm the lack of detectable cell-associated A␤ in cells expressing C100/ER, we solubilized N2a cells expressing C100/ER or C100/WT in 70% formic acid and analyzed the extracted proteins by Western blotting with BA27 and BC05. Comparable levels of A␤40-positive and A␤42-positive ϳ4-kDa proteins were detected in cells expressing C100/WT, whereas no A␤-positive bands were detectable in cells expressing C100/ER despite robust expression of C100/ER holoprotein (Fig. 2C), suggesting that the levels of formic acid-extractable ER-associated A␤ are very low, if any, in our N2a cells overexpressing C100 derivatives.
We next analyzed the relationship between the subcellular localization of PS2 and the intracellular production site of A␤42. To this end, we expressed C100/WT together with WT or N141I mt PS2, separated the cells by iodixanol density fractionation, and analyzed the fractions by Western blotting (Fig.  2D). N-terminal fragments of PS2 were chiefly distributed in fractions 4 -10, which overlapped with those positive for a TGN marker (adaptin-␥; fractions 3-10). In contrast, the distribution of full-length PS2 (fractions 11 and 12) was limited to those positive for an ER marker (i.e. BiP; fractions 10 -12). The distribution patterns of PS2 and its derivatives were similar between WT and N141I mt PS2. ELISA quantitation of cell fraction-associated A␤42 showed that cells expressing N141I mt PS2 harbored elevated levels of A␤-(1-42) in fractions 8 and 9, which corresponded to those positive for PS2 N-terminal fragment as well as for a TGN marker, supporting the notion that TGN is the site in which mt PS2 affects ␥-cleavage of C100 to promote A␤42 production (Fig. 2E).
To examine if the conclusions drawn from experiments using tagged C100 above are applicable to full-length ␤APP, we transiently co-expressed human ␤APP tagged with KKLN (␤APP/ ER) or SDYQRL (␤APP/TGN) or without the tags (␤APP/WT) in N2a cells that stably express WT or N141I or M239V FAD mt PS2 and examined the production of secreted A␤. N2a cells expressing WT PS2 transiently transfected with ␤APP/WT or ␤APP/TGN secreted similar levels of A␤-(1-40) and A␤-(1-42), whereas those with ␤APP/ER did not secrete detectable levels of A␤ (Fig. 3A, left), and the results were similar to those in cells expressing C100 and its derivatives. In contrast, cells expressing N141I or M239V mt PS2 transiently transfected with ␤APP/WT or ␤APP/TGN secreted increased levels of A␤-(1-42), whereas those with ␤APP/ER again did not secrete detectable A␤ (Fig. 3A, middle and right). To examine whether the lack of ER-associated A␤ observed in experiments based on C100 is reproducible with full-length ␤APP in our N2a system, we solubilized N2a cells expressing ␤APP/ER or ␤APP/WT in 70% formic acid and analyzed the extracted proteins by Western blotting with BA27 and BC05. Comparable levels of A␤40positive and A␤42-positive ϳ4-kDa polypeptides were detected in cells expressing ␤APP/WT, whereas no A␤-positive bands were detectable in cells expressing ␤APP/ER despite comparable levels of expression of ␤APP/ER and ␤APP/WT (Fig. 3B). Taken together, it was strongly suggested that late intracellular compartments including TGN, but not ER, are the major intracellular site of A␤ production and of mt PS2 effects on ␥-cleavage of ␤APP to promote secretion of A␤42 in N2a cells.
Expression of C-terminally Truncated C100 and Effects of Co-expression of mt PS2 on A␤ Secretion-Since C100 was fully susceptible to ␥-cleavage as well as to the A␤42-promoting effect of mt PS2, we next examined if the C-terminal cytoplasmic domain of ␤APP, which has been implicated in a number of functions including intermolecular association, caspase cleavage, and endocytosis, is required for the abnormal function of mt PS2. For this purpose, we constructed following C100 derivatives truncated at various positions within the cytoplasmic domain: C100/stop68 truncated at Asp 68 (numbering starting from residue Asp 1 of A␤) documented as a caspase-3 cleavage site (49), C100/stop56 retaining the KKKQ sequence flanking the membrane that is attributed to membrane anchoring (50), and C100/stop52 that lacks the entire cytoplasmic domain (Fig. 4A).
We then transiently transfected cDNAs coding for C-terminally deleted C100 in N2a cells stably expressing WT or N141I or M239V mt PS2 and quantitated A␤-(1-40) and A␤-(1-42) secreted into culture media. In cells expressing WT PS2, levels of secreted A␤-(1-40) were similar among all C100 derivatives, and C100/stop52 yielded significantly reduced levels or ratios (ϳ3%) of A␤-(1-42) compared with those from other C100 derivatives, as observed in cells without transfection of PS2 ( Fig. 5A; *, p ϭ 0.011 in ratios by analysis of variance). In contrast, co-transfection of FAD mt PS2 increased the percentage of secreted A␤-(1-42) as a fraction of total A␤ to ϳ60% (N141I) or ϳ50% (M239V) with all types of C-terminally deleted C100, except that co-expression of C100/stop52 and M239V mt PS2 yielded ϳ20% of A␤42, although this was still significantly higher than that with WT PS2 (Fig. 5A). To confirm that the cytoplasmic domain of ␤APP is dispensable for ␥-cleavage and mt PS2 effect on A␤42 generation on a full-length ␤APP basis, we next co-expressed C-terminally deleted full-length ␤APP in N2a cells stably expressing WT or N141I or M239V mt PS2 and quantitated A␤-(1-40) and A␤-(1-42) in culture media (Fig. 5B). Upon cotransfection with WT PS2, the levels of total secreted A␤ were significantly lower in cells expressing C-terminally truncated ␤APP compared with that with fulllength ␤APP presumably due to lack of endocytosis as previously reported (51), whereas there was a uniform increase in the secretion of A␤42 in cells stably expressing N141I or M239V mt PS2. These data suggest that the anterior half of the C100 (i.e. A␤ sequence plus the following intramembranous portion of ␤APP) is sufficient for the full effect of mt PS2 to increase secretion of A␤- . DISCUSSION In this study, we showed that (i) TGN is one of the major intracellular sites in which a secretable pool of A␤42 is produced and up-regulated by the abnormal function of FAD mt A, schematic depiction of C100 truncated at the cytoplasmic domain. C100/stop68 is truncated at Asp 68 (starting from Asp 1 of the A␤ sequence), which is inferred as the caspase-3 cleavage site; C100/stop56 retains the membrane-flanking four amino acid residues KKKQ; and C100/stop52 lacks the entire cytoplasmic domain. B, Western blot analysis of C-terminally truncated C100 transiently expressed in N2a cells with BAN50. An asterisk shows the co-migration of an ϳ8-kDa polypeptide derived from C100/WT with C100/stop68. Molecular mass markers are shown in kilodaltons, and the names of transfected cDNAs are indicated above each lane. C, inhibition of the generation of ϳ8-kDa band (*) from C100/WT by a caspase inhibitor zVAD-fmk. Ϫ, N2a cells transfected with C100/WT without zVAD-fmk treatment. zVAD, N2a cells transfected with C100/WT treated with 100 mM zVAD-fmk for 24 h. D, differential extraction of C-terminally truncated C100 by Na 2 CO 3 or Triton X-100. Microsomal fractions of N2a cells transiently expressed C100/WT (wt), C100/stop68 (stop68), C100/stop56 (stop56), C100/stop52 (stop52) were extracted by 0.5 M Na 2 CO 3 (pH 11.0) or 1% Triton X-100, and solubilized proteins (S) and insoluble pellets (P) were analyzed by Western blotting with BAN50. E, immunofluorescence localization of C-terminally truncated C100 in N2a cells revealed by BAN50. Scale bar, 10 m. PS2, (ii) lack of the entire cytoplasmic domain of C100 selectively decreases the production of A␤42, and (iii) the anterior half of C100 lacking the entire cytoplasmic domain is sufficient for the abnormal function of mt PS2 to increase production of A␤42.
The intracellular site of A␤42 generation in relation to PS function has been a matter of controversy (38). Here we showed that C100 targeted to TGN (C100/TGN) yielded similar levels of intracellular as well as secreted forms of A␤-(1-42) and A␤-(1-40), compared with those derived from C100 without sorting signals, whereas C100 targeted to ER (C100/ER) did not produce detectable levels of intracellular as well as secreted A␤. Moreover, FAD-linked N141I and M239V mt PS2 fully increased the production of A␤-(1-42) from C100/TGN at an extent comparable with that for C100 without signals, whereas production of intracellular as well as secreted A␤-(1-42) was not up-regulated by co-expression of C100/ER. Use of C100 that does not require ␤-cleavage, which is presumed to occur in the post-Golgi compartments (52), to trigger ␥-cleavage enabled us to directly address the intracellular site where ␥-cleavage takes place and is affected by the abnormal effects of mt PS2. It has been reported that cultured neurons (33,34) as well as human embryonic kidney 293 cells (35) produce A␤-  in ER upon overexpression of ␤APP. These findings were in good agreement with the predominant ER localization of PS, which has been implicated in ␥-cleavage. However, it was subsequently shown that the ER-associated A␤42 was not directly secreted and was considered to comprise a distinct pool from the secreted A␤ (53). The reason for our failure to detect ER-associated A␤ in our cell system is not clear at present; however, the following possibilities could be considered to explain these discrepancies: (i) full-length A␤ may be detected in ER only by extremely high level expression of APP (e.g. by Semliki Forest virus infection) (33), and modest levels of overexpression of ␤APP or C100 fail to produce detectable levels of A␤; (ii) in N2a cells, small amounts of A␤42 truncated at the N terminus, but not full-length species, have been detected in ER (36,47), which escaped our detection system specific for human full-length A␤; and (iii) ER-derived A␤ is present at a relatively small amount compared with those in late compartments, the former being lower than the detection limit of our highly sensitive immunoblot assay.
Our data indicating that TGN harbored elevated levels of "secretable" A␤42 upon co-expression of mt PS2 strongly support the view that the active form of PS (i.e. endoproteolytic fragments that are stabilized (44) and form a high molecular weight complex (54)) resides in Golgi/TGN as well as in additional late intracellular compartments and that mt PS1 upregulates production and secretion of A␤42 in these compartments (39,40). In this case, a relatively small amount of "active" presenilin complex may be sufficient for the generation of A␤ in the late compartments. Notwithstanding the present data, the problem of the "spatial paradox" (38) between the localization of PS and ␥-secretase activities has not been completely clarified. Further careful studies on the intracellular distribution of presenilin complex and ␥-cleavage activities for the processing of ␤APP as well as Notch in different types of cells will be needed. We and others have shown that co-expression of C100 that lacks the majority of the extracellular domain of ␤APP with FAD-associated mt PS1 (40) or PS2 (32) is sufficient to induce overproduction of A␤42. To examine whether the cytoplasmic domain of ␤APP is required for the abnormal effect of mt PS, we co-expressed C-terminally truncated forms of C100 and evaluated the secretion of A␤. Unexpectedly, we found that the expression of C100 lacking the entire cytoplasmic domain (C100/stop52), with or without co-expression of WT PS2, dramatically reduced the secretion of A␤42, although the total levels of A␤ secretion were not significantly altered. Similar results were obtained also in COS cells (data not shown). The mechanism whereby ␥-secretase differentially cleaves A␤40 and A␤42 within the transmembrane segment of ␤APP is not well understood. However, accumulating data suggest that ␥-cleavage occurs in a position-dependent manner within the membranous portion, irrespective of the amino acid sequences (54,55). The lack of the cytoplasmic domain including the KKKQ motif at the membrane-flanking portion, which is presumed to work as a membrane anchor (50), may destabilize the positioning of the transmembrane domain of ␤APP, thereby leading to predominant cleavage at the A␤40 position. Moreover, C100 ending at the putative caspase-3 cleavage site (C100/stop68) did not change the level or proportion of secreted A␤. It has been shown that ␤APP truncated at the caspase-3 cleavage site increased the secretion of A␤40 (49). The reason for this discrepancy is unknown, but it is possible that the caspase-cleaved ␤APP may promote ␤-cleavage (49), thereby increasing A␤40 secretion.
Finally, we have shown that the cytoplasmic domain of C100 is dispensable for the abnormal effects of mt PS2 to increase production of A␤42. This domain is implicated in a number of ␤APP functions including interaction with a number of binding proteins (i.e. FE65 (57) or X11 (58,59)) as well as endocytosis (51), all of which are known to alter A␤ production (60). However, co-transfection of mt PS2 fully increased the secretion of A␤42 from all types of C100 truncated at the cytoplasmic domain. It has recently been suggested that PS serves as a ␥-secretase harboring two intramembranous aspartates in TM6 and TM7 domains as a catalytic center (61). Taken together with our present data, shift of ␥-cleavage from the predominant A␤40 position to a more pathogenic A␤42 position caused by the abnormal gain-of-function of FAD mt PS may require solely the intramembranous interaction between the TM domains of ␤APP and PS. Further analysis on the molecular mechanism whereby mt PS leads to increased production of A␤42 should facilitate the understanding of the pathogenesis of AD as well as of the unusual but important proteolytic mechanism recently referred to as regulated intramembrane endoproteolysis (62).