| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-secretase activity regulates protein clearance from the endocytic recycling compartment
Neurobiology of Disease Laboratory, Genetics and Aging Research Unit, Department of Neurology/MIND, Massachusetts General Hospital, Harvard Medical School, Charlestown, Massachusetts, USA
3Correspondence: Neurobiology of Disease Laboratory, Genetics and Aging Research Unit, Department of Neurology/MIND, Massachusetts General Hospital, Harvard Medical School, 114 16th Street, Charlestown, Massachusetts, 02129, USA. E-mail: dora_kovacs{at}hms.harvard.edu
ABSTRACT
The presenilin (PS)/
-secretase complex proteolytically cleaves more than 20 different proteins in addition to the amyloid precursor protein (APP). These substrates are almost exclusively type I membrane proteins. Many undergo internalization from the cell surface followed by degradation or recycling back to the plasma membrane through the endocytic recycling compartment (ERC). Evidence shows that the PSs also regulate intracellular trafficking of APP and its C-terminal fragments (CTFs). To investigate whether PS/-secretase activity is required for normal endosomal recycling, we performed live cell imaging experiments with fluorescently labeled transferrin, reported to specifically traffic through the ERC. By using pharmacological -secretase inhibitors or cell lines lacking functional PS/-secretase, here we show that PS/-secretase activity is required for clearance of transferrin from the ERC. Interestingly, lack of PS/-secretase function also resulted in the accumulation of APP and APP-CTFs in the ERC in addition to the cell surface. Familial Alzheimer’s disease mutations in APP-CTFs did not affect endocytic recycling of these proteins. Our results suggest that PS/-secretase activity is required for normal endosomal recycling of soluble and membrane-associated proteins through the ERC and propose a new mechanism by which impaired PS/-secretase function may eventually contribute to neurodegeneration.—Zhang, M., Haapasalo, A., Kim, D. Y., MacKenzie Ingano, L. A., Pettingell, W. H., and Kovacs, D. M. Presenilin/-secretase activity regulates protein clearance from the endocytic recycling compartment.
Key Words: amyloid precursor protein • transferrin • trafficking • Alzheimer’s disease • neurodegeneration
THE TWO MAIN pathological hallmarks in Alzheimer’s disease (AD) are the extracellular, amyloid-ß (Aß) containing plaques and intracellular neurofibrillary tangles resulting in a diffuse loss of neurons and synapses in different cortical regions of the brain (1
, 2)
. Even though the pathogenic events underlining the Aß accumulation and subsequent neurodegeneration are poorly understood, mutations in the genes encoding the amyloid precursor protein (APP) and presenilin-1 and -2 (PS1 and PS2) have been identified as causing early-onset AD. The more common form of the disease, late-onset AD, is associated with genetic polymorphisms operating as risk factors and/or genetic modifiers (3)
.
Aß is proteolytically produced from the type I transmembrane protein APP (4)
. At the cell surface, the extracellular domain of APP is mainly cleaved by the metalloproteases of the ADAM family (
-secretase) to release a soluble, secreted APP (5
, 6)
. Alternatively, APP undergoes a different N-terminal cleavage by BACE (ß-site APP cleaving enzyme or ß-secretase) (7
8
9
10)
on the cell surface and endosomes. Both cleavages are followed by restricted intramembraneous processing (RIP) of APP in its transmembrane domain by -secretase at two adjacent sites (- and 
-cleavages) (11)
. These cleavages produce a large N-terminal ectodomain (sAPP), Aß or p3, and a small, soluble APP intracellular domain (AICD) that has been recently shown to directly or indirectly activate transcription (12
13
14)
. -secretase is a large multiprotein complex consisting of at least four proteins, PS, nicastrin, Aph-1 and Pen-2, which are all necessary and sufficient for -secretase activity (15
16
17
18
19
20)
. Recently, in addition to APP, -secretase has been shown to cleave over 20 different proteins, most of which are type I membrane proteins displaying a variety of functions from cell-type specification, cell adhesion, and axonal pathfinding to synaptic activity, such as Notch, cadherins, nectin-1, and voltage-gated sodium channel ß2 subunit (11
, 21
22
23
24
25)
.
Endocytosis is a universal cellular function, by which cell surface proteins are internalized along with extracellular material and plasma membrane, and subsequently trafficked through a series of vesicular compartments (26
27
28
29
30)
. The internalization of proteins takes place most commonly in clathrin-coated vesicles but also by caveolae or macropinocytosis (31)
. After shedding of the clathrin coat, the vesicles fuse with early endosomes (32
, 33)
. Early endosomes receive cargoes also from the Golgi and late endosomes, some of which are recycled back to these compartments. Many of the cell surface proteins or lipids are continuously internalized and recycled back to the plasma membrane in recycling endosomes that are regulated by Rab4, Rab11, and the SNARE protein cellubrevin (34)
. Some of the proteins, in turn, are directed to the late endosomes and lysosomes for degradation (35
, 36)
. The ß-cleavage of APP most likely takes place at the cell surface and in early endosomes after APP internalization (37)
, whereas components of the -secretase complex have been identified in many cellular compartments such as plasma membrane, both early and late endosomes, autophagic vacuoles, lysosomes, and ER (19
, 38
39
40
41
42)
. Interestingly, Aß generation has been demonstrated to markedly slow down when APP internalization and endocytosis are inhibited (43
, 44)
and, in contrast, accelerate when endocytosis is stimulated (45
, 46)
. Altered neuronal endocytosis has been shown to be one of the earliest known neuropathological changes in AD (45
, 47
, 48)
. In late-onset AD, endosome pathology is indicative of the initial abnormal increase in Aß levels (48
, 49)
. Furthermore, in individuals with Down syndrome, morphologically abnormal endosomes can be detected already decades before the onset of AD (48
49
50
51)
.
By using pharmacological -secretase inhibitors or cell lines lacking the functional -secretase complex, we show here that -secretase activity is required for normal endosomal recycling. This is indicated by a delay in trafficking of two different proteins, transferrin and APP, through the endocytic recycling compartment (ERC). Our results suggest a new mechanism by which impaired -secretase function may eventually result in dysfunction or degeneration of neurons.
MATERIALS AND METHODS
DNA constructs and reagents
The GFP-Rab7, GFP-Rab9, and GFP-Rab11 mammalian expression constructs were gifts from Dr. R. E. Pagano (Mayo Clinic and Foundation, Rochester, MN; (52)
). APP CTF-GFP (C99-GFP) mammalian expression construct was a gift from Dr. B. Hyman (Massachusetts General Hospital, Charlestown, MA). AD causing mutations were introduced into C99-GFP vector by using QuickChange site-directed mutagenesis kit (Stratagene). APP 695-yellow fluorescent protein mammalian expression construct was a gift from Drs. Mandelkow and Biernat (MPG-ASMB, Max-Planck-Gesellschaft, Hamburg, Germany). DAPT and L685,458 were obtained from Calbiochem and used to inhibit -secretase activity in cells (800 nM and 1 µM, respectively). Fluorescent transferrin (Alexa 633-, Alexa 568-, or Alexa 488-transferrin) and fluorescent dextran (Alexa 488-dextran; Mr 70,000, lysine-fixable) were purchased from Invitrogen/Molecular Probes (Eugene, OR).
Cell culture and transfection
Wild-type (WT) or stably transfected Chinese hamster ovary (CHO) cells were maintained in Ham’s F-12 medium containing 10% FBS. Cells were grown for 20 h to achieve 
60% confluency before transfection or other treatments. To generate stable cell lines, individual zeocin-resistant colonies were isolated and screened for expression of the transfected proteins by Western blot analysis. Clones with similar expression levels were maintained in selection medium. WT PS1/2 or PS1/2 double knockout (PS1/2 dKO) embryonic stem (ES) cells were cultured in knockout-Dulbecco’s modified Eagle medium (Life Technologies, Carlsbad, CA) containing 10% ES cell-quality FBS, L-glutamine (2 mM), penicillin-streptomycin (100 U/ml, and 100 µg/ml, respectively), ß-mercaptoethanol (110 µM), nonessential amino acids (100 µM) and ESGRO (1000 U; Invitrogen).
All cell lines were transfected by using FuGENE6 transfection reagent (Roche, Munich, Germany) according to kit instructions. Transfected cells were either imaged as live cells at different times of expression of the transfected cDNAs or as fixed cells (4% paraformaldehyde (PFA) in PBS for 30 min).
Labeling the endocytic pathway in living cells
The endocytic recycling compartment (ERC) was labeled by loading cells with fluorescently labeled transferrin (250 µg/ml in cell culture medium; ((53)
) for 8 minutes, followed by fixation immediately or after chasing. Endocytosed transferrin was chased after washing the cells three times with PBS in prewarmed + 37°C medium for indicated times. Late endocytic compartments (late endosomes and lysosomes) were labeled at + 37°C with endocytosed fluorescent dextran (1 mg/ml Alexa 488-dextran in cell culture medium) for 17 h followed by 1-hour chase before live cell imaging.
Western blot analysis
Total cell lysates were prepared in a lysis buffer containing 100 mM Tris, pH 7.6, 2 mM EDTA, 150 mM NaCl, 1% Triton X-100, 0.25% Nonident P-40 and protease inhibitor cocktail (Roche), run on 4–12% Bis-Tris SDS-PAGE gels, and blotted on PVDF membranes. To detect yellow fluorescent protein (YFP)-tagged full-length (FL) APP or GFP-tagged APP CTF (C99), anti-GFP (1:1000, Roche) or APP C-terminal 6E10 antibody (Ab) (1:2000, Signet) was used. The anti-GFP Ab recognizes both GFP and YFP. The blots were reprobed with actin Ab (1:1000, Neomarkers) to verify equal loading.
Microscopy
Confocal images were collected from cells cultured on LabTek chambered coverslips using a Zeiss LSM Pascal 5 confocal microscope system equipped with C-apochromat x63/1.4 numerical aperture (NA), or x100/1.4 NA oil immersion objective, and RGB Laser Module (Arg 458/488/514 nM, HeNe 543, and HeNe 633 nM for FITC/GFP/yellow fluorescent protein, TX Red/Alexa 568, and Alexa 633, respectively; Carl Zeiss, Thornwood, NY). For each set of experiments, identical parameters were used in confocal imaging. The images in each experiment were processed identically using Zeiss LSM software and Adobe Photoshop software. Metamorph software (Molecular Devices) with a region analysis module was used to quantify the amount of fluorescent transferrin in ERC from the original LSM images. Intensity of transferrin fluorescence in the ERC within each cell was measured within an area of identical size. The background fluorescence in 10 different locations outside of cells was measured within areas of identical sizes to those of the ERCs and averaged. The average background intensity was then subtracted from the average intensity of transferrin fluorescence in the ERC in each image. Average of 15 cells in each microscopic field was measured. Data shown are expressed as the average intensity of transferrin fluorescence in the ERC as relative units ± SEM The statistical analyses were performed by Independent Samples Test (Figures 1C
and 2B)
or one-way ANOVA with Bonferroni correction as a post hoc test (Figure 4B
). Conventional fluorescence microscopy and image processing were carried out with Nikon TE300 fluorescence microscope equipped with a color Spot camera and Spot image-processing software.
|
|
|
RESULTS
PS/-secretase activity is required for normal clearance of transferrin from the ERC
PS/-secretase substrates are almost exclusively type I membrane proteins (21)
. Many of these proteins are known to undergo continuous internalization from the cell surface, followed by degradation or recycling back to the plasma membrane. To investigate whether PS/-secretase activity is required for normal endosomal recycling, we performed live imaging experiments with fluorescently labeled transferrin (A568-transferrin) in several cell model systems with decreased -secretase activity. Transferrin receptors are reportedly associated with the endocytic recycling compartment (ERC) together with Rab11 (54)
. In a preliminary experiment, we confirmed that transferrin colocalized with the ERC marker Rab11. CHO cells transiently overexpressing the GFP-tagged Rab11 (Rab11-GFP), or GFP-tagged late endosomal or lysosomal markers Rab7 or Rab9 (Rab7-GFP and Rab9-GFP, respectively) were labeled with transferrin for 8 minutes. As expected, transferrin colocalized with Rab11, but not with the endosomal or lysosomal markers Rab7 or Rab9, demonstrating that transferrin localization was restricted to the endocytic recycling pathway (Figure 1
A). This experiment confirmed that transferrin specifically labeled the ERC.
To begin assessing the role of PS/-secretase in the recycling of transferrin, CHO cells were treated with 800 nM DAPT, a specific -secretase inhibitor, for 17 h and labeled with transferrin for 8 min. Cells treated with DMSO vehicle served as a control. Localization of transferrin was studied immediately after the labeling (Figure 1A
; 0 h) or after 1-hour chase (Figure 1B
). In control cells, transferrin partially localized to the ERC immediately after labeling (Figure 1Ab
), and it almost completely cleared out from the cells by 1 hour (Figure 1Bb
). In contrast, in DAPT-treated cells transferrin was mainly present in the ERC immediately after labeling (Figure 1Ag
), and it was still retained in the ERC after a 1-hour chase (Figure 1Bg
). Quantitation of the data indicated that there was no difference in the intensity of transferrin labeling in the ERC between control and DAPT-treated cells immediately after labeling (Figure 1C
). However, there was about six times more transferrin label left in the DAPT-treated cells (relative intensity 32.3) after 1 hour compared to the control cells (relative intensity 5.2; Figure 1C
). Similar to the control cells, in DAPT-treated cells, transferrin did colocalize with Rab11, but not with Rab7 or Rab9, at either time points after -secretase inhibition, further implying that transferrin was withheld in the ERC (Figure 1A
and B
). These experiments suggested that PS/-secretase activity is required for normal clearance of transferrin from the ERC.
To further investigate whether transferrin clearance from the ERC requires functional PS/-secretase, we labeled CHO cells overexpressing either WT PS1 or the dominant negative mutant of PS1, PS1/D385A, with transferrin for 8 min. Transferrin-labeled cells were imaged immediately after 8 min of incubation (Figure 2
Aa and c; 0 min) or after a 30-min chase following labeling (Figure 2Ab
and d
). Incubation times were reduced here and in subsequent experiments from the previous 1 hour to allow for detection of transferrin in cells harboring functional -secretase. Immediately after labeling of both WT and PS1/D385A CHO cells, transferrin was present in the ERC (Figure 2Aa
and c
). However, after 30 min, most of the protein was cleared out from the ERC in WT PS1 overexpressing cells (Figure 2Ab
). In contrast, expression of the PS1/D385A mutant resulted in a delayed transferrin clearance from the ERC (Figure 2Ad
). Quantitation of the data showed that there was about almost twice as much transferrin present in the PS1/D385A cells after 30 min as compared to the WT PS1 overexpressing cells (relative intensities 88 vs. 46.5, respectively; Figure 2B
). Different incubation times, CHO cell clones, and possibly subtle temperature changes among experiments do not allow for direct comparison between the magnitude of the effects of -secretase inhibitors (Fig. 1)
and the PS1/D385A mutation (Fig. 2)
. In both cases, enzyme activity of -secretase clearly regulates transferrin recycling from the ERC.
In the third set of experiments, we used embryonic stem cells (ES) lacking both PS1 and PS2 (PS1/2 dKO) to confirm that PS/-secretase activity regulates transferrin recycling through the ERC. WT ES cells were used as a control. Imaging of WT ES cells transfected with Rab11 and labeled with transferrin verified that ERC was properly formed in these cells (Figure 3
Aa–c). However, similar to the -secretase inhibitor-treated CHO cells or the cells expressing the PS1/D385A mutant, delayed transferrin clearance was observed in PS1/2 dKO ES cells after a 20-min chase when compared to the WT PS1/2 expressing ES cells (Figure 3B
). Colabeling of the cells with FITC-dextran, a marker for the late endocytic pathway, including late endosomes and lysosomes, indicated that the majority of transferrin remained in the ERC and was not directed to the late endocytic compartments within the PS1/2-deficient cells (Figure 3Bb
). Treatment of WT PS1/2 ES cells with two different -secretase inhibitors, DAPT (800 nM, 17 h; Figure 4
Ad) and L685,458 (1 µM, 17 h; 4Af), also revealed a similar delay in transferrin clearance, when examined at 40 min after its uptake. Compared to control cells (relative intensity 19.4), at 40 min DAPT-treated cells contained 2.2 times more (relative intensity 42.7) and L685,458 treated cells 2.6 times more (relative intensity 51) transferrin (Figure 4B
). Lack of transferrin colocalization with FITC-dextran further corroborated that transferrin clearance from the ERC was delayed on inhibited -secretase function and that transferrin did not localize to the late endocytic compartments. Taken together, these results imply that normal PS-dependent -secretase function is required for transferrin recycling from the ERC back to the cell surface and out of the cells.
|
PS/-secretase activity also regulates clearance of APP-CTFs from the ERC
APP is first cleaved by ß- or -secretase, both resulting in ectodomain shedding and generation of membrane-anchored APP C-terminal fragments (CTFs). The subsequent -secretase-mediated processing of the remaining CTFs generates Aß or p3, and the cytoplasmic AICD fragment. Lack of PS or PS loss-of-function have previously been reported to result in increased intracellular transport and accumulation of APP on the plasma membrane (39
, 55
56
57
58
59)
, increased levels of APP CTFs (60
61
62
63)
, and decreased internalization (57)
. We next studied the role of -secretase function in the trafficking of APP through the ERC. CHO cells were transiently transfected with YFP-tagged full-length APP (APP-yellow fluorescent protein), treated with the -secretase inhibitor DAPT (800 nM, 17 h) or L-685,458 (1 µM, 17 h), and labeled with transferrin. As shown in Figure 5
A, APP-yellow fluorescent protein in control cells was mainly localized inside the cells, and it did not colocalize with transferrin to the ERC (Figure 5Aa
). However, on treatment of the cells with either DAPT or L-685,458, APP-yellow fluorescent protein accumulated on the cell surface and in the ERC, as indicated by its colocalization with transferrin in this compartment (Figure 5Ab
and c
). Western blotting confirmed that APP-yellow fluorescent protein is correctly processed by -secretase in CHO cells (Figure 5B
). Interestingly, -secretase inhibitor-treated cells showed an increase only in APP-CTF, but not full-length APP, levels (Figure 5B
). This suggests that the newly accumulated CTFs are most likely to accumulate on the cell surface and the ERC subsequent to the -secretase inhibitor treatment.
|
Finally, we directly assessed the effect of -secretase inhibition of APP-CTF localization by stably transfecting CHO cells with a recombinant APP-C99-GFP construct lacking the whole ectodomain of APP. In control cells, C99-GFP was undetectable because of its constitutive -secretase-mediated processing (Figure 6
Aa). Similar to CTFs deriving from full-length APP-yellow fluorescent protein, treatment with the -secretase inhibitors DAPT or L-685,458 resulted in the stabilization and accumulation of C99-GFP on the cell surface and in the ERC, as indicated by the colocalization of C99-GFP with transferrin (Figure 6Af
and i
). Western blot analysis further showed that the C99-GFP levels increased in the DAPT- and L685,458-treated cells as compared to the control cells, where C99-GFP levels remained undetectable (Figure 6B
). In addition, we observed that APP-C99-GFP bearing different flavin adenine dinucleotide (FAD)-linked mutations (EGg, V717F, and V717I) stabilize and accumulate in the ERC in a similar manner to the WT C99 on -secretase inhibition with DAPT or L685,458 (Supplemental Fig. 1
). Thus, FAD mutations in APP-CTFs do not visibly alter their trafficking through the ERC, in cells with normal or reduced -secretase function. In summary, our data suggest that the normal function of the PS/-secretase is required for proper trafficking of at least transferrin and APP through the ERC, and further imply that PS/-secretase may function as an important regulator of protein trafficking through the ERC in general.
|
DISCUSSION
Here, we show that -secretase activity is required for normal clearance of transferrin and APP-CTFs from the ERC. Lack of -secretase activity resulted in the accumulation of soluble transferrin and membrane-bound APP-CTFs in the ERC, indicating that protein clearance from this early endosomal compartment is regulated by PS/-secretase. Cell surface levels and signaling of a variety of proteins are regulated by endocytosis. Receptor tyrosine kinases and G-protein coupled receptors are rapidly endocytosed after ligand-induced activation and are often recycled back to the plasma membrane via ERC (30)
. Transferrin receptor, essential for iron uptake, has been widely used to label and characterize ERC function (64)
. Evidence shows that small GTP-binding proteins Rab11 (65)
and Rab15 (66)
, the SNARE protein cellubrevin (67)
, and syntaxin 13 (68)
localize to and regulate transport via the ERC. RME-1/EHD family proteins have been proven important for protein exit from the ERC and trafficking back to cell surface (69
, 70)
. To study the role of PS/-secretase activity in protein recycling through ERC, we utilized fluorescently labeled transferrin in different cell models deficient of PS/-secretase function. We observed that PS/-secretase inhibition results in delayed clearance of transferrin from the ERC in CHO cells, suggesting that lack of PS/-secretase activity results in a functional defect of this particular endocytic compartment.
Presenilins regulate intracellular trafficking of APP, APP CTFs, and -secretase complex components Nicastrin and Pen-2 (55, 57–59, 71). PS/-secretase proteolytically cleaves over 20 different type I membrane proteins in addition to APP and Notch (21)
, such as synaptic junction protein nectin-1 (23)
, voltage-gated sodium channel ß2 (24
, 25)
, p75 neurotrophin receptor (p75; (72)
), and cell adhesion molecules N- and E-cadherin (22
, 73)
. Interestingly, PS1 regulates trafficking of N-cadherin from endoplasmic reticulum to the plasma membrane, thus influencing cell-cell adhesion (74)
. Many of the PS/-secretase substrates undergo endocytosis from the cell surface. Notch and p75 are endocytosed and accumulate to recycling endosomes, and endocytosis directs -secretase-mediated cleavage of Notch and subsequent signaling of its intracellular domain (ICD; (75
, 76)
). Also, E-cadherin is endocytosed and colocalizes with Rab11 to recycling endosomes (77
, 78)
. E-cadherin recycling has been suggested to regulate its availability for cellular junction formation and tissue remodeling (79)
. The PS/-secretase mediated cleavage, on the other hand, releases the E-cadherin ICD, resulting in cell-cell dissociation (22)
. Our data together with these reports suggest an important role for PS/-secretase in the regulation of both degradation and recycling of proteins, even though the relationship between these two functions remains yet to be elucidated.
APP is internalized from the plasma membrane and proteolytic processing of APP, and Aß production may take place in different cellular compartments (43
, 80
81
82
83
84)
. Whereas -secretase-mediated cleavage of APP mostly occurs on the plasma membrane (5
, 6)
, BACE and components of the PS/-secretase complex are localized along the secretory and endosomal pathways, in addition to the plasma membrane (41
, 58
, 85
86
87
88
89
90
91
92
93
94)
. Endosomal sorting increases BACE-mediated cleavage of APP (95)
, and overexpression of Rab5, a positive regulator of endocytosis, results in increased Aß and APP ßCTF levels in early endosomes and trans-Golgi network (46)
. In contrast, inhibition of APP endocytosis increases its - and ß-secretase-mediated processing on the plasma membrane (43)
. Furthermore, -secretase was recently shown to associate with APP CTFs and cleave APP and Notch on the plasma membrane (84)
. In PS1-deficient neurons or cells overexpressing the PS1/D385A mutant, APP CTFs accumulate on the plasma membrane, endosomes, endoplasmic reticulum, Golgi, and lysosomes (56
, 62)
. Our imaging studies of DAPT- and L685,458-treated C99-GFP expressing cells suggest that the APP CTFs, in particular, accumulate in the ERC. This is indicated by the fact that most C99-GFP appears to colocalize with transferrin in the ERC, although cell surface C99-GFP is clearly present in these cells. Furthermore, our Western blot analysis of full-length APP expressing cells showed that PS/-secretase inhibitors specifically increased the levels of APP CTFs, and not full-length APP. Our results most probably indicate a problem in the clearance of endocytosed APP from the ERC when PS/-secretase activity is inhibited. Therefore, our data suggest that -secretase may not only cleave APP-CTFs, but also regulate the delicate balance of APP and/or APP-CTF pools available for Aß production.
Strong evidence indicates that the endosomal system is greatly affected in AD. Altered neuronal endocytosis is one of the earliest known neuropathological changes in AD (45
, 47
, 48)
. At early stages of AD, concurrently with the increasing soluble Aß levels, many neurons exhibit enlarged Rab5-positive endosomes containing Aß. At the same time, expression of Rab4, a marker for endosomal recycling, is increased, indicating an up-regulation of the endosomal system (45
, 48
, 49)
. The 4 allele of apolipoprotein E (ApoE) promotes endosome pathology along with earlier onset of AD and a more severe amyloid pathology (45)
. Furthermore, in individuals with Down syndrome (DS), enlarged neuronal endosomes can be detected already before birth (48
49
50)
. These data suggest that disturbances in normal endosomal function may result in neurodegeneration in AD. Aging conditional PS1/2 double knockout mice display impairments in hippocampal memory and synaptic plasticity associated with increasing neurodegeneration. These impairments are accompanied with reduced NMDA receptor (NMDAR)-mediated responses and synaptic levels of NMDAR subunits, recently also shown to be regulated by synaptic Aß levels (96
, 97)
. Reduction of active NMDAR levels on the cell surface could lead to neurodegeneration. In our study, lack of PS/-secretase activity in PS1/2 dKO cells leads to impairment of ERC function. It is thus possible that impaired NMDAR recycling back to cell surface could also play a part in neurodegeneration.
To our knowledge, this is the first report that PS/-secretase activity directly or indirectly, perhaps through proteolytic cleavage of a type I membrane protein or accumulation of various membrane protein CTFs, is involved in the regulation of protein trafficking through the ERC. This is supported by the notion that trafficking of two different proteins, transferrin and APP, is delayed through the ERC in cells lacking functional PS/-secretase. These results not only suggest that PS/-secretase activity is required for normal endosomal recycling, but further indicate that lack of PS/-secretase function results in impairment of a specific endocytic compartment. Our results suggest a new mechanism by which impaired PS/-secretase function may eventually result in dysfunction or degeneration of neurons.
ACKNOWLEDGMENTS
The authors wish to thank Dr. R. E. Pagano (Mayo Clinic and Foundation, Rochester, MN) for the GFP-Rab7, GFP-Rab9, and GFP-Rab11 expression constructs, Dr. B. Hyman (Massachusetts General Hospital, Charlestown, MA) for the APP CTF-GFP (C99-GFP) construct, and Drs. Mandelkow and Biernat (MPG-ASMB, Max-Planck-Gesellschaft, Hamburg, Germany) for the APP 695-yellow fluorescent protein construct. This work was supported by NIH/NIA grants (D.M.K) and the Academy of Finland (A.H.).
FOOTNOTES
1 Current address: Synta Pharmaceuticals Corp., 45 Hartwell Ave., Lexington, MA 02421, USA. 
2 These authors contributed equally to this work.
Received for publication December 16, 2005. Accepted for publication January 26, 2006.
REFERENCES
0.0001).
