Presenilin-1 Regulates Intracellular Trafficking and Cell Surface Delivery of β-Amyloid Precursor Protein*

Presenilins (PS1/PS2) play a critical role in proteolysis of β-amyloid precursor protein (βAPP) to generate β-amyloid, a peptide important in the pathogenesis of Alzheimer's disease. Nevertheless, several regulatory functions of PS1 have also been reported. Here we demonstrate, in neuroblastoma cells, that PS1 regulates the biogenesis of βAPP-containing vesicles from thetrans-Golgi network and the endoplasmic reticulum. PS1 deficiency or the expression of loss-of-function variants leads to robust vesicle formation, concomitant with increased maturation and/or cell surface accumulation of βAPP. In contrast, release of vesicles containing βAPP is impaired in familial Alzheimer's disease (FAD)-linked PS1 mutant cells, resulting in reduced βAPP delivery to the cell surface. Moreover, diminution of surface βAPP is profound at axonal terminals in neurons expressing a PS1 FAD variant. These results suggest that PS1 regulation of βAPP trafficking may represent an alternative mechanism by which FAD-linked PS1 variants modulate βAPP processing.

Alzheimer's disease (AD) 1 is characterized by the excessive generation and accumulation of ␤-amyloid peptides (A␤). The amyloidogenic A␤ peptide is proteolytically derived from the ␤-amyloid precursor protein (␤APP) within the secretory pathway by distinct enzymatic activities known as ␤and ␥-secretase (1,2). Full-length ␤APP is synthesized in the endoplasmic reticulum (ER) and transported through the Golgi apparatus. The major population of secreted A␤ peptides is generated within the trans-Golgi network (TGN) (3)(4)(5), also the major site of ␤APP residence in neurons at steady state. ␤APP can be transported in TGN-derived secretory vesicles to the cell sur-face if not first proteolyzed to A␤ or an intermediate metabolite.
At the plasma membrane ␤APP is either cleaved to produce a soluble molecule, s␤APP (6) or, alternatively, reinternalized within clathrin-coated vesicles to an endosomal/lysosomal degradation pathway (7,8). Thus, the distribution of ␤APP between the TGN and cell surface has a direct influence upon the relative generation of s␤APP versus A␤. This phenomenon makes delineation of the mechanisms responsible for regulating ␤APP trafficking from the TGN relevant to understanding the pathogenesis of AD.
Although it has generally been accepted that PSs are essential for ␥-cleavage, it has not been firmly established that PSs are the catalytic component of the enzyme complex. For example, recent studies (23) have shown that the production of A␤42 in early compartments of the secretory apparatus is unimpaired in the absence of PSs. Whereas the hypothesis remains attractive that PSs are the ␥-secretase, several reports indicate that PSs mediate additional physiological functions, including roles in calcium homeostasis, neurite outgrowth, apoptosis, and synaptic plasticity (24), and some of these functions are influenced by FAD-linked PS mutations. PS1 has been implicated in regulating intracellular trafficking and maturation of selected transmembrane proteins. PS1 deficiency significantly affects trafficking of the tyrosine kinase receptor TrkB, as well as the PS1-interacting protein ICAM-5/telecephalin (25,26). Indeed evidence has emerged to support the notion that PS1 may facilitate ␥-secretase cleavage of substrates via regulating the maturation and intracellular trafficking of substrates and/or components of the ␥-secretase complex. For example, expression of the PS1 aspartate variants leads to accumulation of ␤APP C-terminal fragments (CTFs) as well as full-length ␤APP at the cell surface (20,27). Several recent studies (28,29) further demonstrate that PS1 regulates the maturation and cell surface accumulation/trafficking of nicastrin.
Based on these observations, we investigated the potential role of PS1 in regulating intracellular trafficking of full-length ␤APP through the secretory pathway. We previously established a cell-free system in which we can reconstitute both the formation of A␤ and the trafficking of ␤APP/A␤ from the ER or the TGN (3, 5, 30 -32). By utilizing this system, we directly examined individual cellular processes involved in PS1 regulation of ␤APP trafficking. We report here that PS1 deficiency or expression of loss-of-function variants led to robust formation of ␤APP-containing vesicles, concomitant with increased maturation and/or cell surface accumulation of ␤APP. In contrast, vesicle formation from the TGN and the ER was impaired in cells expressing FAD-linked PS1 mutants, resulting in a reduction of ␤APP delivery to the cell surface. We also observed a profound reduction of surface ␤APP at axonal terminals in neurons harboring an FAD PS1 mutation (A246E), compared with neurons expressing wild type (wt) PS1. Taken together, these results suggest that PS1 can modulate metabolism of ␤APP via regulating ␤APP trafficking within the secretory pathway and thus affect A␤ generation by controlling the availability of substrate ␤APP to appropriate secretases.
Preparation of Permeabilized N2a Cells-It has been well established that incubation of cells at 15 (36) or 20°C (30) leads to an accumulation of membrane and secretory proteins in the ER and TGN, respectively. To assay ␤APP trafficking from TGN, cells were pulse-labeled with [ 35 S]methionine (500 Ci/ml) for 15 min at 37°C, washed with phosphate-buffered saline (prewarmed to 20°C), and chased for 2 h at 20°C in prewarmed complete media. To assay ␤APP trafficking from the ER, cells were pulse-labeled for 4 h at 15°C. For both types of preparations, cells were permeabilized at the termination of incubation as follows. Cells were first incubated at 4°C in "swelling buffer" (10 mM KCl, 10 mM HEPES, pH 7.2) for 10 min. The buffer was discarded and replaced with 1 ml of "breaking buffer" (90 mM KCl, 10 mM HEPES, pH 7.2), after which the cells were broken by scraping with a rubber policeman. Cells were centrifuged at 800 ϫ g for 5 min, washed in 5 ml of breaking buffer, and resuspended in 5 volumes of breaking buffer. This results in Ͼ95% cell breakage as evaluated by trypan blue staining. Broken cells (cell-free system) were incubated in a final volume of 300 l containing 2.5 mM MgCl 2 , 0.5 mM CaCl 2 , 110 mM KCl, cytosol (30 g protein) prepared from N2a cells (32,37), and an energy-regenerating system consisting of 1 mM ATP, 0.02 mM GTP, 10 mM creatine phosphate, 80 g/ml creatine phosphokinase, and a protease inhibitor mixture. Incubations were carried out at 37°C for various times (15-120 min) to observe the kinetics of protein trafficking.
Formation of Nascent Secretory Vesicles in Permeabilized Cells and Immunoprecipitation-Following incubation of cell-free systems, vesicle and membrane fractions were separated by centrifugation at 11,000 rpm for 30 s at 4°C in a Brinkmann centrifuge (Brinkmann Instruments). Vesicle (supernatant) and membrane (pellet) fractions were diluted with IP buffer (50 mM Tris-HCl, pH 8.8, 150 mM NaCl, 6 mM EDTA, 2.5% Triton X-100, 5 mM methionine and cysteine, and 1 mg/ml bovine serum albumin), immunoprecipitated using ␤APP C-terminal antibody 369 (7,38) or anti-NCAM and anti-FGFR1 antibodies (Santa Cruz Biotechnology), and analyzed by SDS-PAGE. Each experiment was performed at least three times. Band intensities were analyzed and quantified using NIH ImageQuant software, version 1.52.
Biotinylation and Detection of Cell Surface ␤APP-Stably transfected N2a cells were labeled with [ 35 S]methionine (500 Ci/ml) for 10 min at 37°C and chased for 2 h at 20°C in complete medium to accumulate labeled proteins in the TGN. To restore vesicle trafficking from TGN, cells were transferred to 37°C for various time intervals. Cells were then incubated at 4°C with 0.5 mg ml Ϫ1 sulfo-N-hydroxysuccinimide biotin (Pierce) to biotinylate cell surface proteins. Biotinylated and non-biotinylated proteins were first separated into two fractions by binding to streptavidin-agarose beads (Pierce), and ␤APP from each fraction was immunoprecipitated with antibody 369 (7,38) and analyzed by SDS-PAGE (39).
Immunofluorescence Confocal Microscopy-For staining of fulllength ␤APP, N2a cells grown in chamber slides were incubated at 4°C with primary antibody 6E10 (diluted 1:100 in growth medium, Senetek PLC) for 1 h without fixation or permeabilization. Following incubation with secondary antibodies and FITC-conjugated Vicia villosa agglutinin (1:100, Vector Laboratories Inc.), cells were fixed with 4% formaldehyde at room temperature for 15 min. Cultured neurons were fixed twice with 4% formaldehyde for 15 min twice. For surface staining of full-length ␤APP, cells were directly incubated with primary antibody mAb348 (1:100 dilution, Roche Molecular Biochemicals) at 4°C overnight. In some experiments, neurons were permeabilized by ice-cold methanol for 2 min prior to overnight incubation with mAb348 or anti-GAP43 antibody (1:4000). Immunofluorescence staining was examined by confocal microscopy (LSM510, Zeiss).

RESULTS
PS1 Deficiency Leads to Increased ␤APP Transport from TGN to Plasma Membrane and from ER to Golgi-It has been well established that PS1-deficient neurons fail to secrete A␤ but accumulate intracellular ␤APP C-terminal fragments (25,40). In addition to affecting ␥-secretase activity, recent observations (25,28,29) that PS1 deficiency or loss-of-function alters the maturation and cell surface accumulation of certain membrane proteins, such as ␤APP, nicastrin, and Notch-1, suggest a potential role for PS1 in intracellular protein trafficking. To support further the notion that wt PS1 may have a direct regulatory effect on ␤APP trafficking through secretory compartments, we assessed the formation of ␤APP-containing vesicles from the TGN and from the ER in PS1 Ϫ/Ϫ fibroblasts using a cell-free reconstitution system.
The formation of ␤APP-containing vesicles from the TGN was examined using a cell-free system that has been used extensively to study ␤APP trafficking and A␤ generation (see "Experimental Procedures" and Refs. 3, 5, and 32). This cellfree system has been used to investigate nascent secretory vesicle budding from a variety of cells (30,31), and the integrity of TGN stacks and derived vesicles has been demonstrated by electron microscopy (37). To study trafficking of ␤APP-containing vesicles from the TGN, we labeled wt and PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ fibroblasts with [ 35 S]methionine and then incubated the cells at 20°C, a temperature at which transport of proteins, including ␤APP, from the ER to the TGN is unimpaired. However, under these conditions, the egress of secretory vesicles from the TGN is blocked, thus allowing labeled ␤APP (and other membrane proteins) to accumulate in the TGN (3,5,32). This experimental design ensures that TGN-specific vesicle biogenesis is measured. In our cell-free trafficking assays, following permeabilization and incubation at 37°C, budding of ␤APPcontaining vesicles from the TGN was greatly increased at all time points examined in preparations that lacked PS1 when compared with preparations from cells that express wt PS1 (Fig. 1, a and b). After 2 h of incubation, the amount of ␤APP transported out of the TGN was 54.2% higher in PS1 Ϫ/Ϫ cell preparations compared with preparations from PS1 wt cells (i.e. 21.9 versus 14.2% of the total labeled ␤APP in the TGN).
We next tested whether PS1 deficiency might affect ␤APP vesicle transport from the ER to Golgi utilizing a modified cell-free system in which ␤APP-containing post-ER vesicles were reconstituted. In this case, cells were labeled with [ 35 S]methionine at 15°C to accumulate ␤APP within the ER (see "Experimental Procedures" and Ref. 5), followed by permeabilization and incubation at 37°C to initiate vesicle release. The budding of ␤APP-containing ER vesicles was accelerated (ϳ2-fold) in PS1 Ϫ/Ϫ cells compared with wt cells (Fig. 1, c and

d).
To assess whether PS1 deficiency selectively affects ␤APP trafficking, budding of fibroblast growth factor receptor (FGFR 1 )-containing vesicles from the ER was determined in parallel and remained unchanged by PS1 deficiency (Fig. 1e). Collectively these results suggest that PS1 selectively regulates ␤APP trafficking from ER and TGN compartments.
To support further the notion that ␤APP is transported more efficiently out of the ER and Golgi compartments in the absence of PS, we examined the patterns of ␤APP maturation in detergent lysates prepared from PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ cells and wt fibroblasts. As expected from the PS1 Ϫ/Ϫ -permeabilized cell studies, the amount of mature (glycosylated/sialylated) ␤APP in PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ cells was significantly higher than that in PS1 wt cells (Fig. 1f), indicative of increased residence and transit of ␤APP through the late secretory compartments. However, alternation in ␤APP glycosylation/maturation was much less dramatic in PS1 Ϫ/Ϫ cells than in PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ cells (data not shown). Although the underlying mechanism is not clear, this may be attributed to differential roles of PS1 and PS2 in APP trafficking versus maturation.
Loss-of-Function PS1 Mutants Accelerate ␤APP Trafficking from the ER without Altering TGN Vesicle Budding-To confirm further that loss of PS1 activities results in accelerated ␤APP trafficking through the secretory pathway, N2a cells expressing loss-of-function PS1 variants were examined to assess their effects on intracellular transport of ␤APP. Previous studies demonstrated that a single amino acid substitution in PS1 transmembrane domain 7 (D385A) (19,20) or deletion of the first two PS1 transmembranes (⌬1,2) (41) lowers A␤ secretion and leads to accumulation of ␤APP ␤CTFs. Unexpectedly, ␤APP-containing vesicle budding from the TGN was not changed by the expression of PS1 loss of function mutations (D385A or ⌬1,2) (ϳ35% of maximal ␤APP vesicle budding in both D385A and ⌬1,2 versus ϳ38% in PS1 wt cells) (Fig. 2, a  and b). However, the trafficking of ␤APP from the ER to Golgi was significantly increased in these loss-of-function PS1 mutants (Fig. 2, c and d). The maximal level of vesicle budding was increased by ϳ1.5-(for D385A) to 2-fold (for ⌬1,2) when compared with PS1 wt; this increase in budding is similar to that observed in PS1-deficient cells. As a result of accelerated ER trafficking, the total amount of ␤APP residing in TGN membrane in loss-of-function PS1 mutants was increased compared with that in wt TGN membrane (Fig. 2a), although the rate of budding of ␤APP-containing vesicles from the TGN was unchanged (Fig. 2b).
Together, these results suggest that the loss of function for PS1 in ␥-cleavage of ␤APP and the complete absence of PS1 protein differentially regulate vesicle biogenesis from the TGN, although the efflux of ␤APP molecules from the ER is enhanced under both conditions.
Loss-of-Function PS1 Mutations Increase the Amount of Fulllength ␤APP Delivered to the Plasma Membrane-Despite the lack of marked differences in TGN vesicle biogenesis, increased ER to Golgi trafficking of ␤APP observed in the loss-of-function mutant cells may be sufficient to elevate the steady-state levels of ␤APP at the cell surface in intact cells. In addition, as recently reported (42), a delay in the internalization of surfacebound ␤APP may further facilitate increased surface accumulation of ␤APP. Indeed, live staining of ⌬1,2 cells using monoclonal antibody 6E10 revealed an obvious increase in the amount of surface-bound ␤APP compared with PS1 wt cells. The amount of total surface glycoproteins is identical in the two types of cells as judged by staining for V. villosa agglutinin, a lectin that binds to N-acetyl-D-galactosamine linked to serine or threonine residues in glycoproteins (43) (Fig. 3a). The amount of newly synthesized ␤APP delivered to the cell surface was measured quantitatively in these cells by pulse-chase labeling in combination with cell-surface biotinylation (39) in intact cells. Cells were pulse-labeled for 10 min with [ 35 S]methionine at 37°C and chased for 2 h at 20°C to accumulate labeled ␤APP in the TGN. Subsequent incubation at 37°C allows trafficking of ␤APP from TGN to cell surface in a synchronized fashion (39). Newly arrived [ 35 S]methionine-labeled cell surface ␤APP molecules were then labeled with biotin at 4°C for 15 min and separated from TGN-associated ␤APP by streptavidin bead precipitation. Up to 14.3% of nascent ␤APP was transported to the plasma membrane after 120 min of chase, a value that is 48.9% greater than that in PS1 wt cells (Fig. 3, b and c). The findings from loss-of-function PS1 mutants together with studies using PS1-deficient cells strongly suggest that PS1 plays a role in regulating the delivery of nascent ␤APP molecules to the cell surface.
FAD-linked PS1 Mutants Delay ␤APP Transport from the TGN and the ER-We next investigated whether FAD-linked PS1 variants might influence ␤APP trafficking. The kinetics of ␤APP-containing vesicle budding from the TGN of N2a cells expressing wt PS1 was compared with that of cells expressing FAD-linked PS1 variants. The budding of TGN-derived ␤APP vesicles was dramatically impaired in FAD-linked PS1 mutants (⌬E9, A246E, and M146L) at all time points examined, compared with PS1 wt cells. Among the three FAD-linked PS1 mutations examined, PS1 A246E demonstrated the highest degree of impairment in ␤APP vesicle budding from the TGN, with maximal ␤APP vesicle budding of 21.8% at 90 min compared with 38% in PS1 wt cells (Fig. 4, a and b).
We further tested whether FAD-linked PS1 mutations affect ␤APP vesicle transport from the ER to Golgi. Similar to the results obtained for TGN budding, ␤APP trafficking from the ER to Golgi, as judged by the release of ␤APP-containing vesicles from the ER, was significantly reduced in FAD-linked mutants, with a 50.9 and 57.9% decrease in the maximal levels of vesicle budding in PS1 ⌬E9 and A246E cells, respectively, compared with PS1 wt cells. The impairment in ER vesicle biogenesis was relatively mild in the PS1 M146L mutation, with an ϳ30% reduction compared with PS wt (Fig. 4, c and d).
To ascertain the selectivity of impaired ␤APP-containing vesicle budding in cells expressing FAD-linked PS1 variants, we examined the trafficking of neural cell adhesion molecule (NCAM) from the TGN using the permeabilized cell system. In contrast to ␤APP, the rate of TGN budding of vesicles containing NCAM isoforms was almost identical in PS1 wt and mutant cells (Fig. 4e), suggesting that FAD-linked PS1 variants specifically delay ␤APP transport from the TGN to the cell surface.
FAD-linked PS1 Mutants Lower the Amount of Full-length ␤APP Delivered to the Plasma Membrane-We reasoned that the alteration in ␤APP trafficking by PS1 mutations seen in the permeabilized cell system should affect the steady-state level of ␤APP on the plasma membrane. To test this hypothesis we stained live N2a cells with monoclonal antibody 6E10 to visualize surface-bound ␤APP and ␤APP C-terminal fragments and FITC-conjugated V. villosa agglutinin to label all glycoproteins. As shown in Fig. 5a, the intensity of ␤APP immunofluorescence on the cell surface of mutant cells was markedly reduced as compared with PS1 wt cells. V. villosa agglutinin staining of surface glycoproteins was comparable between PS1 wt and mutant cells (Fig. 5a). These findings are consistent with the results from permeabilized cell experiments described above and suggest that FAD-linked PS1 mutants impair the trafficking of ␤APP to the cell surface.
To examine directly the delivery of newly synthesized fulllength ␤APP trafficking from the TGN to the plasma membrane, we carried out pulse-chase labeling in combination with cell-surface biotinylation. As shown in Fig. 5, b and c, in PS1 wt cells, 9.6% of nascent ␤APP that accumulated in the TGN during incubation at 20°C left the TGN and traveled to the cell surface after a 2-h chase at 37°C. However, the amount of newly synthesized ␤APP that accumulated on the cell surface at various chase time periods was much lower in PS1 (⌬E9) mutant cells, with only 3-5% of nascent ␤APP molecules delivered to the plasma membrane even after 2 h of chase. Therefore, both immunofluorescence and biochemical studies indi- FIG. 3. Delivery of full-length ␤APP to the plasma membrane in cells harboring loss-of-function PS1 mutations. a, live N2a cells were incubated with primary antibody 6E10 (1:100) at 4°C for 1 h to label cell surface ␤APP (red) and FITC-conjugated V. villosa agglutinin to stain all surface glycoproteins (green). Cells were then fixed and visualized by confocal microscopy. b and c, cells were labeled with [ 35 S]methionine at 37°C for 10 min and chased at 20°C for 2 h to accumulate labeled ␤APP in the TGN (total cell ␤APP). Cells were then incubated at 37°C for various times to allow transport of ␤APP to the plasma membrane. Cell surface proteins were then biotinylated at 4°C for 15 min. Biotinylated and non-biotinylated proteins were separated into two fractions by binding to streptavidin beads. ␤APP was immunoprecipitated from each fraction and analyzed. Quantitative data represent mean Ϯ S.E. from three independent experiments. cate that the levels of full-length ␤APP delivered from TGN to plasma membrane are diminished in cells expressing FADlinked PS1 variants, most likely by delaying the budding of ␤APP-containing vesicles from the TGN and the ER. (44 -48) have demonstrated that ␤APP is axonally transported by the kinesin-mediated, fast anterograde component. It has also been reported (49) that both full-length and processed derivatives of ␤APP accumulate at presynaptic terminals of cortical neurons. We further assessed whether FAD-linked mutations affect the distribution of full-length ␤APP on the surface of primary neurons.

An FAD-linked PS1 Variant Causes a Profound Reduction of Cell Surface-bound ␤APP at Axons and Axonal Terminals in Primary Neurons-Previous studies
Embryonic cortical neurons from PS1 knockout mouse embryos rescued with comparable levels of expression of either wt human PS1 (line 17-3) or FAD-linked PS1 A246E mutant (line 16-4) (34) were grown in culture, and ␤APP molecules on the cell surface at the axonal terminals and in the cell body were visualized by immunofluorescence staining using monoclonal antibody mAb348, which recognizes the ectodomain of ␤APP. As shown in Fig. 6a, immunofluorescence of full-length ␤APP on the surface of live (non-permeabilized) wt PS1 rescued neurons was evident both in the cell body and at the axonal terminals, whereas the immunofluorescence of surface-bound ␤APP in PS1 A246E mutant rescued neurons was limited to the cell body, being absent from axons and axonal terminals (Fig. 6a, top panels). As a control, live cells were also doubly stained with V. villosa agglutinin to confirm that the staining of surface-bound glycoproteins was comparable in wt and mutant PS1 neurons (Fig. 6a, bottom panels). However, after permeabilization, the distribution pattern of intracellular full- length ␤APP was identical in PS1 wt and A246E neurons (Fig.  6b, top panels). In addition, intracellular localization of growthassociated protein 43 (GAP43), a protein associated with growth cone membranes, was similar in wt and mutant PS1 neurons (Fig. 6b, bottom panels). These data suggest that the FAD-linked mutants selectively impair the targeting or fusion of ␤APP-containing vesicles to the plasma membrane, especially at axons and axonal terminals of neurons.

DISCUSSION
To date, the mechanisms by which PSs exert their effects on ␤APP metabolism are not fully understood, although multiple lines of evidence support a direct role of PS1 in facilitating ␥-cleavage of ␤APP, Notch, and other substrates (1,50). On the other hand, it has been reported (24) that PS1s may play multiple physiological roles such as those in calcium homeostasis, neuronal development, neurite outgrowth, apoptosis, synaptic plasticity, and tumorigenesis. Recent evidence indicates a novel function of PS1 in regulating intracellular trafficking of a selected set of proteins including those associated with PS1 and ␤APP metabolites (25,26,42). In the present report, we demonstrate the following: 1) PS1 deficiency leads to the accelerated trafficking of ␤APP from both the TGN and the ER; 2) loss-of-function PS1 mutants with impaired ␥-secretase function increase ER biogenesis of ␤APP-containing vesicles without affecting TGN budding and eventually elevate the amount of ␤APP transported to the plasma membrane; 3) FAD-linked PS1 mutants impair ␤APP trafficking from the TGN to the plasma membrane, as well as from the ER to Golgi, resulting in delayed delivery of ␤APP to the cell surface; 4) a profound reduction of surface ␤APP at axonal terminals of neurons that express FAD-linked PS1 mutants. Taken together, these results indicate that the role of PS1 in facilitating ␥-secretase processing of APP extends beyond its putative function in the catalytic process.
The above findings are consistent with a model in which PS1 might regulate ␤APP metabolism by altering ␤APP trafficking (Fig. 7). Our model proposes that PS1 provides a retention FIG. 5. FAD-linked PS1 mutations reduce the amount of fulllength ␤APP delivered to the plasma membrane. Surface staining and budding assays were performed as described in Fig. 3, except that N2a cells coexpressing ␤APPswe and PS1 harboring one of the indicated FAD-linked mutations were used. These assays were done in parallel to the experiments described in Fig. 3, and the WT data are the same as in Fig. 3. Quantitative data represent mean Ϯ S.E. from three independent experiments.
FIG. 6. FAD-linked PS1 mutations cause more profound reduction of surface-bound ␤APP in axons and axonal terminals. Primary neurons that express comparable levels of either wt human PS1 or FAD-linked PS1 A246E mutant in PS1 Ϫ/Ϫ background were cultured for 7 days. a, after 15 min of 4% formaldehyde fixation, nonpermeabilized cells were incubated with APP N-terminal antibody mAb348 and FITC-conjugated V. villosa agglutinin. b, in some experiments neurons were permeabilized with ice-cold methanol for 2 min and then incubated with mAb348 (red) and anti-GAP43 (green) antibodies. Immunofluorescence staining was visualized by confocal microscopy.
signal, which guides ␤APP delivery to appropriate compartments where ␥-secretase processing occurs (Fig. 7a). Gain-offunction FAD mutants (Fig. 7c) may direct sustained retention of ␤APP-containing vesicles and consequently increase the availability of ␤APP to cleavage enzymes resident in the TGN and/or ER, the major sites of intracellular A␤ generation. In contrast, PS1 knockout (Fig. 7b) or loss-of-function mutants (Fig. 7bЈ) abolish or reduce, respectively, the retention property and thereby allow accelerated ␤APP trafficking to the cell surface. In addition, based upon our observations and another report (42), it is conceivable that PS1 regulates ␤APP trafficking in both the secretory as well as the endocytic pathway where some A␤ may also be generated (51).
Previously it was shown that PS1 deficiency causes missorting of select type I membrane proteins. For example, PS1 is required for the maturation and intracellular trafficking of nicastrin, an integral component of the ␥-secretase complex (14,28,29,52). Moreover, in PS1-deficient neurons, telencephalin/ICAM is translocated from the plasma membrane to large intracellular clusters (26). It is interesting to note that PS1 deficiency causes enhanced localization of ␤APP at the cell surface but has the opposite effect on surface accumulation of nicastrin and telencephalin/ICAM.
To fully understand the regulation of ␤APP metabolism by PS1, it would be important to determine whether PS1 exerts an effect on trafficking of ␤APP CTFs similar to that on full-length ␤APP. It has been shown that loss of PS1 results in the accumulation of the ␣-/␤-CTF (25,40). However, further studies on the subcellular distribution of CTFs in PS1-deficient cells indicate the complexity of intracellular trafficking of CTFs. For example, the generation of CTFs mostly occurs in the late secretory compartments, whereas ␤APP CTFs accumulate in the ER, Golgi, and lysosomes (53). It has also been reported (53) that CTFs may accumulate in restricted and unpredicted intracellular compartments in PS1-deficient cells. Although our preliminary data indicate that PS1 may regulate intracellular trafficking of CTFs in a similar manner as its regulation of full-length ␤APP (data not shown), vesicle budding assays (pulse-chase and low temperature incubations) have not yet distinguished between trafficking and production of CTFs in the ER and Golgi compartments. More rigorous studies are underway to establish appropriate cell lines (e.g. those overexpressing ␤CTFs) and optimal experimental conditions. Much remains to be learned about the mechanisms by which PS1 regulates intracellular trafficking of ␤APP. Based on our permeabilized cell data, budding of vesicles containing FGFR 1 and NCAM is not influenced by PS1 function, indicating that PS1 exerts its regulation on select membrane proteins. One attractive model is that PS1 regulates the recruitment or the association of trafficking factors with cytoplasmic sorting signals within ␤APP, thereby selectively regulating the sorting of ␤APP to the surface. In the absence of PS function, ␤APP containing vesicles may be transported to the cell surface via the default constitutive pathway. On the other hand, FADlinked mutations modify the interaction between ␤APP and trafficking factors in a manner that interferes with efficient ␤APP trafficking. In this regard, it has been reported that PS1 and PS2 associate with Rab11, a member of the GTP-binding protein family of membrane trafficking regulators implicated in protein transport along the biosynthetic and endocytic pathways (54). In addition, PS1 binds to Rab GDP dissociation inhibitor (RabGDI), a protein that functions in vesicular membrane transport to recycle Rab GTPases, and PS1 deficiency leads to impaired Rab-GDI membrane association (55). Furthermore, our preliminary data suggest that addition of phospholipase D1 to the cell-free budding assays prevents impaired ␤APP trafficking from the TGN in cells expressing FAD PS1 mutations, whereas inhibition of phospholipase D1 activity by primary butanol diminishes the accelerated ␤APP trafficking from the TGN in cells expressing loss-of-function PS1 mutations.
Finally, our studies demonstrate that FAD-linked PS1 mutants result in a decreased distribution of surface ␤APP in axons and axonal terminals of neurons. This finding suggests that PS1 may specifically affect targeting to and/or fusion of ␤APP-containing vesicles at the nerve terminus. As reported previously, full-length ␤APP has been implicated in a number of physiological functions such as synapse formation, growth cone outgrowth, and axon guidance (55,56). Furthermore, based on the demonstration that ␤APP plays an essential role in axonal trafficking (47,48), our findings that PS1 may regulate ␤APP sorting along the axon have much broader implications as well. For example, impaired delivery of full-length ␤APP to the cell surface at axonal terminals by FAD-linked PS1 variants may interfere with neurite initiation, elongation and branching, and synaptic plasticity. It is important to note that in AD presynaptic pathology is more severe than neuronal loss (57). By affecting vesicle transport and surface delivery of full-length ␤APP, pathogenic PS1 mutants might directly modulate ␤APP metabolism and, in addition, indirectly contribute to the pathogenesis and progression of AD. FIG. 7. Model proposing that PS1 may regulate ␤APP metabolism by altering ␤APP trafficking through the secretory pathway. a, it is proposed that PS1 provides a retention signal to guide ␤APP delivery to appropriate compartments where ␥-secretase processing occurs. b, in the absence of PS1 expression ␤APP trafficking is accelerated. As a consequence, A␤ generation is reduced. bЈ, PS1 lossof-function mutants accelerate ␤APP transport from the ER without affecting budding from the TGN. The amount of ␤APP molecules in each vesicle derived from TGN membrane is increased without altering the rate of vesicle budding. Therefore, ␤APP delivery to cell surface is increased. c, in contrast, FAD-linked PS1 mutants may direct sustained retention of ␤APP, and consequently increase the availability of ␤APP to cleavage enzymes resident in the TGN and/or ER, the major intracellular sites for A␤ generation. However, the possibilities that PS1 may recruit certain cytosolic trafficking factors (such as phospholipase D1 and/or Rab11) to TGN and/or ER membrane and thereby regulate ␤APP transport cannot be excluded.