Scholarly article on topic 'Demonstration of BACE ( -secretase) phosphorylation and its interaction with GGA1 in cells by fluorescence-lifetime imaging microscopy'

Demonstration of BACE ( -secretase) phosphorylation and its interaction with GGA1 in cells by fluorescence-lifetime imaging microscopy Academic research paper on "Biological sciences"

Share paper
Academic journal
Journal of Cell Science
OECD Field of science

Academic research paper on topic "Demonstration of BACE ( -secretase) phosphorylation and its interaction with GGA1 in cells by fluorescence-lifetime imaging microscopy"

Research Article

Demonstration of BACE (P -secretase) phosphorylation and its interaction with GGA1 in cells by fluorescence-lifetime imaging microscopy

Christine A. F. von Arnim, Michelle M. Tangredi, Ithan D. Peltan, Bonny M. Lee, Michael C. Irizarry, Ayae Kinoshita and Bradley T. Hyman*

Alzheimer Disease Research Laboratory, Massachusetts General Hospital, Harvard Medical School, 114 16th Street, Charlestown, MA 02129, USA

*Author for correspondence (e-mail: Accepted 28 July 2004

Journal of Cell Science 117, 5437-5445 Published by The Company of Biologists 2004 doi:10.1242/jcs. 01422


P-Secretase (BACE) carries out the first of two proteolysis steps to generate the amyloid-P peptides that accumulate in the senile plaques in Alzheimer's disease (AD). Because most BACE activity occurs in endosomes, signals regulating its trafficking to these compartments are important to an understanding of AD pathogenesis. A DISLL sequence near the BACE C-terminus mediates binding of BACE to the VHS domains of Golgi-localized y-ear-containing ARF-binding (GGA) proteins, which are involved in the sorting of proteins to endosomes. Phosphorylation of the motif's serine residue regulates BACE recycling back to the cell surface from early endosomes and enhances the interaction of BACE with GGA proteins in isolated protein assays. We found that BACE phosphorylation influences BACE-GGA interactions in cells using a new fluorescence-resonance-

energy-transfer-based assay of protein proximity, fluorescence lifetime imaging. Although serine-phosphorylated BACE was distributed throughout the cell, interaction of GGA1 with the wild-type protein occurred in juxtanuclear compartments. Pseudo-phosphorylated and non-phosphorylated BACE mutants remained localized with GGA1 in the Golgi body, but the latter mutation diminished the two proteins' FRET signal. Because BACE phosphorylated at serine residues can be identified in human brain, these data suggest that serine phosphorylation of BACE is a physiologically relevant post-translational modification that regulates trafficking in the juxtanuclear compartment by interaction with GGA1.

Key words: BACE, GGA, Alzheimer's disease, Phosphorylation


Along with intracellular tangles of hyperphosphorylated tau protein, senile plaques composed mainly of aggregated amyloid-P peptides (Ap) are the pathological hallmark of Alzheimer's disease (AD) (Selkoe, 2001). BACE (ß-site of APP-cleaving enzyme) has been identified as the first of two sequential proteases that cleave the Aß precursor protein (APP) to release the 40- or 42-residue Aß peptide and the APP intracellular domain (Hussain et al., 1999; Lin et al., 2000; Sinha et al., 1999; Vassar et al., 1999; Yan et al., 1999). BACE is a type-I membrane-associated aspartic protease. Overexpression of BACE in transgenic mice leads to enhanced generation of Aß (Bodendorf et al., 2002), whereas BACE-null mice produce little or no Aß and yet are phenotypically normal. Additionally, BACE knockout in mice overexpressing human APP rescues behavioral and electrophysiological deficits (Ohno et al., 2004; Roberds et al., 2001). BACE inhibition therefore represents a compelling therapeutic target to prevent Aß deposition, and clarifying the cellular trafficking and activity of this protease is of great importance.

BACE undergoes co-translational ^-glycosylation and subsequent complex glycosylation, as well as proteolytic removal of its prodomain by a furin-like protease (Bennett et al., 2000; Capell et al., 2000; Creemers et al., 2001). Metabolic

pulse-chase experiments reveal that, after glycosylation, BACE is rapidly transported to the Golgi apparatus and distal secretory pathway (Creemers et al., 2001). BACE can undergo phosphorylation at S498 in cell culture, but the role of this post-translational modification is uncertain (Walter et al., 2001). Most BACE protein in the cell is located within Golgi and endosomal compartments, where it localizes with | APP. The acidic pH optimum of BACE (Lin et al., 2000; Sinha et al., 1999; Vassar et al., 1999; Yan et al., 1999) indicates that it is predominantly active within late Golgi compartments, endosomes and/or lysosomes. This is consistent with previous findings that | -secretase cleavage of | APP can occur in all of these acidic compartments (Haass et al., 1992; Koo and Squazzo, 1994; Koo et al., 1996). However, measurable amounts of active BACE protein are also present on the cell surface and in lipid rafts (Huse et al., 2000; Kinoshita et al., 2003; Riddell et al., 2001; Chyung and Selkoe, 2003). From the plasma membrane, BACE is reinternalized to early endosomes and can recycle back to the cell surface. A dileucine motif (residues 499/500) encoded in BACE's intracellular domain and modulated by the phosphorylation state of an adjacent serine residue (S498) might regulate this trafficking and endocytosis. Phosphorylation of S498 does not significantly alter the endocytic pathway of BACE but rather

plays a role in its recycling from early endosome via late endosome and/or trans-Golgi network (TGN) to the cell surface (Walter et al., 2001).

The dileucine motif is similar to the acidic-cluster dileucine (ACDL) sorting signal DXXLL found in the cytosolic tails of mannose-6-phosphate receptors (Puertollano et al., 2001; Takatsu et al., 2001; Zhu et al., 2001) and other proteins (e.g. LRP3, sortilin). The DISLL sequence (residues 496-500) in the BACE C-terminus was shown to mediate binding of BACE to the VHS (Vps-27, Hrs and STAM) domain of the three Golgi-localized y-ear-containing ARF-binding (GGA) proteins (He et al., 2002) by studies of the isolated proteins. GGA proteins are believed to be important sorting adaptors. GGA binding of ACDL-containing proteins might represent the first step in the recruitment of these membrane proteins to coated vesicles on the Golgi membrane (Bonifacino, 2004; Nielsen et al., 2001). There are three GGAs in humans (GGA1, GGA2 and GGA3). Phosphorylation of S498 produced an enhanced binding to VHS of all three GGAs in isolated protein assays (He et al., 2003), with Kd values being ten, four and 14 times lower after BACE phosphorylation for GGA1, GGA2 and GGA3, respectively. A recent study by Shiba et al. (Shiba et al., 2004) confirmed these findings by X-ray crystallography and showed a threefold interaction of BACE and GGA1 with phosphorylation. The increased VHS binding of phosphorylated BACE suggests that GGAs might participate in this recycling process. In addition to the VHS domain, GGA proteins contain three other domains that interact with other proteins participating in vesicular transport, such as clathrin, adaptor protein 1 (AP-1), a-synergin, rabaptin-5 and other potential regulators of vesicle coat assembly (Boman, 2001; Bonifacino, 2004; Hirst et al., 2001; Hirst et al., 2000).

Using a fluorescence-resonance-energy transfer (FRET)-based assay of protein proximity, we now explore the subcellular localization of phosphorylated BACE and of the BACE-GGA1 interaction. We find that BACE binds GGA1 in cells in the TGN and that this interaction is sensitive to BACE phosphorylation. The presence of serine-phosphorylated BACE in human brain demonstrates that this interaction might have in vivo relevance.

Materials and Methods

Generation of expression constructs of GGA1 and BACE

The GGA1 construct that encodes GGA1 with Myc at its C-terminus was subcloned from cDNA of human GGA1 (generous gift of M. S. Robinson, University of Cambridge, Cambridge, UK) by polymerase chain reaction (PCR) using the primers 5'-TATGCTAGCCACCATG-GAGCCCGCGAT-3' and 5'-AATCTCGAGAGAGGCTACCCCAG-GTTTC-3'. The PCR product was inserted into the Nhel/XhoI restriction sites of pCDNA3.1-Myc (Invitrogen, Carlsbad, CA).

The details of the BACE constructs have been described elsewhere (Kinoshita et al., 2003). In brief, BACE was cloned by PCR from a whole human brain cDNA library (Quick Clone cDNA; Clontech, Pal Alto, CA) using the primers 5'-AGCCACCAGCACCACCA-GACTTG-3' and 5'-ACTGGTTGGTAACCTCACCCATTA-3'. The PCR product was inserted into pcDNA3.1/V5/His-Topo vector (Invitrogen) (BACE-V5). The phosphorylation-site mutants of BACE (Walter et al., 2001) were generated by substitution of the serine residue with alanine (mimicking a non-phosphorylated form) or to an aspartate at residue 498 of BACE-V5 or BACE-GFP constructs as well as the catalytically inactive BACE construct by substitution of the aspartates at residues 93 and 289 to alanine using the QuickChange

site-directed mutagenesis kit (Stratagene, La Jolla, CA). BACE was cut out from the BACE-V5 at HindIII and SacII sites, and inserted into the pEGFP-N1 vector (Clontech) to make the BACE/green-fluorescent-protein (BACE-GFP) construct. The APP695 fused N-terminally to secreted alkaline phosphatase (SEAP-APP) construct was a generous gift of S. F. Lichtenthaler, LMU Munich, Germany). Authenticity of the PCR-generated constructs was confirmed by DNA sequencing.

Cell culture conditions and transient transfection

Mouse neuroblastoma N2a and HEK293 cells are used in this study. N2a cells were cultured in OPTI-MEM®I (Grand Island, NY) and HEK cells in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). Transient transfection of the cells was performed using a liposome-mediated method (FuGene 6; Roche Molecular Biochemicals, Indianapolis, IN). For immunocytochemistry, cells were split into four-well chambers 1 day before the transfection. First, a mixture of 1 |ig plasmid DNA and 3 |il FuGene6 was made in 100 l DMEM and left for 30 minutes at room temperature. Then, 25 |il of this mixture was added to the medium in each well. Double transfection of GGA1 and BACE constructs was performed in the same way.


Immunostaining was done on the cells 24-48 hours after transfection. Cells were fixed in 4% paraformaldehyde for 10 minutes, washed in TBS (pH 7.3) and permeabilized with 0.5% Triton X-100 for 20 minutes, then blocked with 1.5% normal goat serum for 1 hour. To detect the localization of GGA1 and BACE, cells transfected with GGA1-Myc and BACE-V5 were immunostained by rabbit anti-Myc antibody (Ab) (1:1000; Upstate Biotechnology, Lake Placid, NY) and anti-V5 monoclonal Ab (mAb) (1:500; Sigma, St Louis, MO) or rabbit anti-BACE Ab against the C-terminus of BACE (BACE-CT) (1:500; Calbiochem, San Diego, CA); anti-GGA1 Abs are not available. Cells were then washed three times in TBS and labeled with FITC-conjugated anti-rabbit Ab and Cy3-conjugated anti-mouse Ab (10 g ml-1; Jackson Immunoresearch, West Grove, PA) for 1 hour at room temperature. Alternatively, N2a cells co-transfected with GGA1-Myc and BACE-V5, and immunostained with the same primary antibodies were labeled with FITC-conjugated anti-mouse Ab and Cy3-conjugated anti-rabbit Ab. Immunostained cells were covered with a coverslip and mounted for confocal or two-photon microscopic imaging. The immunostained cells were observed with the appropriate filters using a BioRad 1024 confocal microscope.

FRET measurements using fluorescence-lifetime imaging microscopy

FRET is observed when two fluorophores are in very close proximity (<10 nm). FRET measurements using fluorescence-lifetime imaging microscopy (FLIM) rely on the observation that fluorescence lifetime (the time of fluorophore emission after brief excitation, measured in picoseconds) is shorter in the presence of a FRET acceptor. We applied a validated FLIM technique that can quantify protein-protein interactions using multiphoton microscopy (Bacskai et al., 2003; Berezovska et al., 2003). A mode-locked Ti-sapphire laser (Spectra Physics, Fremont, CA) sends a femtosecond pulse every 12 nanoseconds to excite the fluorophore. Images were acquired using a BioRad Radiance 2000 multiphoton microscope (BioRad, Hercules, CA). We used a high-speed Hamamatsu MCP detector (Hamamatsu, Bridgewater, NJ) and hardware and software from Becker and Hickl (Berlin, Germany) to measure fluorescence lifetimes on a pixel-by-pixel basis. Excitation at 800 nm was empirically determined to excite FITC but not Cy3. Donor fluorophore (FITC) lifetimes were fitted to two exponential decay curves to calculate the proportion of

fluorophores within each pixel that interacts with an acceptor. As a negative control FITC lifetime was measured in the absence of acceptor (Cy3), which showed lifetimes equivalent to FITC-IgG alone or in solution. Another negative control was measuring the lifetime of BACE labeled at the extracellular N-terminus [rb-anti-BACE (46-65); Calbiochem] in the presence of GGA1-Myc labeled at the intracellular C-terminus with Cy3. Despite colocalization, the FITC lifetime is the same as in the negative control, confirming that GGA1-Myc labeled with Cy3 does not act as an acceptor to FITC on BACE if the labeled epitopes are too far apart to support FRET. This control demonstrates that a change in the fluorescence lifetime, indicating FRET, is a more sensitive measure of proximity than simple co-localization.

Phosphorylation assay

24 hours after transfection, N2a cells were serum starved (0.1% FBS in OPTI-MEM®I) for 18 hours followed by a 1-hour treatment with 20 nM okadaic acid (OA) (Sigma), a phosphatase inhibitor. The cells were then immunostained and fixed in 4% paraformaldehyde with 20 nM OA for 10 minutes, washed in TBS (pH7.3) containing 20 nM OA, permeabilized with 0.5% Triton X-100 containing 20 nM OA for 20 minutes, and blocked with 1.5% normal goat serum for 1 hour. To detect the localization and amount of phosphorylated wild-type BACE, cells transfected with BACE-GFP were immunostained with mouse anti-phosphoserine Ab (1:500; Sigma). FRET would be expected to occur only where BACE contained a phosphoserine epitope. Rabbit anti-BACE CT Ab (1:500; Calbiochem) served as a positive control. Cells were then washed three times in TBS and labeled with Cy3-conjugated anti-mouse Ab or anti-rabbit Ab (10 |ig ml-1; Jackson Immunoresearch, West Grove, PA) for 1 hour at room temperature. Immunostained cells were covered with a coverslip and mounted for confocal or two-photon microscopic imaging. The immunostained cells were observed with the appropriate filters using confocal microscopy.

Tissue homogenization under denaturing conditions

The Massachusetts Alzheimer Disease Research Center Brain Bank provided temporal cortex tissue. A total of 18 temporal cortices were used, nine from control brains and nine from AD brains. The mean ages of control and AD cases were 82±6 years and 81 ±7 years, respectively. Control cases had an average post-mortem interval (PMI) of 17±6 hours and AD cases had an average PMI of 8±6 hours. Our protein-solubilization procedure was adapted from previously reported studies (Orlando et al., 2002) with minor modifications. The tissue was homogenized in ice-cold TEVP + sucrose buffer (10 mM Tris, pH 7.4, 5 mM NaF, 1 mM Na3VO4, 1 mM EDTA, 1 mM EGTA, 320 mM sucrose) at 1 ml per 100 mg tissue. The homogenates were centrifuged at 20,000 g for 16 minutes at 4°C and the supernatants were removed. The pellets were resuspended in 800 n l TEVP + 1% sodium dodecyl sulfate (SDS) (10 mM Tris, pH 7.4, 5 mM NaF, 1 mM Na3VO4, 1 mM EDTA, 1 mM EGTA, 1% SDS). Next, the resuspended pellets were sonicated for 10 seconds then boiled for 5 minutes. The samples were centrifuged and the supernatant was collected for immunoprecipitation reactions after the protein concentration was determined by protein assay (BioRad).


Immunoprecipitation experiments were carried out with BioMag goat anti-mouse IgG (PerSeptive Biosystems, Framingham, MA). The magnetic beads were precoupled separately to mouse anti-BACE-CT antibody (1:500; Chemicon) and mouse anti-phosphoserine (1:500; Sigma) in preparation for BACE and serine-phosphorylated protein capture, respectively. The incubation took place overnight at 4°C on a rocking platform. Excess antibodies were removed and brain-tissue lysates were added with lysis buffer to the bead-antibody complex for

3 hours at 4°C. As a negative control, we used beads that were precoupled to an irrelevant antibody, mouse anti-phosphotyrosine, (1:500; Cell Signaling Technology, Beverly, MA). As another negative control, we incubated the beads in the presence of anti-BACE-CT antibody but lysates were not applied. After the supernatants were collected, the beads were washed four times in lysis buffer. Next, loading buffer was added to the magnet beads of each condition and the bound protein was denatured at 90°C for 510 minutes, and centrifuged for 1 minute. The supernatants and recombinant BACE (R&D Systems, Minneapolis, MN) were loaded onto 8% Tris-glycine polyacrylamide gels (Novex, San Diego, CA) under denaturing and reducing conditions. The proteins were transferred to Immobilon-P membranes (Millipore, Bedford, MA) and blocked in 2.5% non-fat dried milk. BACE was detected by rabbit anti-BACE Ab against the N-terminus of BACE (BACE-NT) (1:10,000; Calbiochem) and phosphorylated serine residues were detected by mouse anti-phosphoserine antibody at 1:1000. Secondary antibodies conjugated to HRP were applied and visualized by chemiluminescence.

Phosphatase assay

HEK293 cells passaged into 12-well plates were transfected with a fS-galactosidase (fS-gal) reporter, APP695 fused N-terminally to secreted alkaline phosphatase and empty vector, BACE or a BACE mutant. Each condition was transfected in triplicate. Medium was changed 24 hours later, and then collected after another 24 hours. Measurement of SEAP activity in the conditioned media was carried out in triplicate by chemiluminescent assay (Roche, Mannheim, Germany) according to the manufacturer's instructions. SEAP activity was normalized to P-gal activity, which was measured by hydrolysis of o-nitrophenyl-fS -D-galactopyranoside in cells lysed with reporter lysis buffer (Promega, Madison, WI).


Co-localization of GGA1 and BACE

To assess the impact of phosphorylation of the serine residue in the C-terminal DISLL motif of BACE on interactions with GGA1, we mutated BACE to create pseudophosphorylated and non-phosphorylated forms. The negative charge of an aspartic acid residue at position 498 (S498D) mimicked the negative charge on phosphorylated serine and replacing the serine with an alanine (S498A) created a non-phosphorylatable BACE (Walter et al., 2001). Double immunostaining was performed and analysed by confocal microscopy using a BioRad 1024 confocal three-channel instrument. Mutation of S498 altered BACE localization by increasing (S498D) or decreasing (S498A) the concentration within perinuclear compartments, consistent with previous observations (Walter et al., 2001). GGA1-positive structures nonetheless largely overlapped with BACE-positive structures in both BACE-mutants when they were co-expressed.

Interaction of GGA1 and BACE by FLIM analysis

Although double immunostaining showed subcellular compartment colocalization of BACE and GGA1 predominantly in juxtanuclear compartments (Fig. 1), this does not necessarily imply a close interaction. We therefore used an alternative technique to probe the proximity of GGA1 and BACE in order to test the idea that the GGA 1-BACE interaction depends on BACE phosphorylation and to evaluate further the localization of this interaction. We used FLIM, a

Fig. 1. Phosphorylation-state-dependent localization of BACE and GGA1. BACE-V5 [wild-type (WT), S498D and S498A] and GGA1-Myc co-transfected N2a cells grown on glass coverslips were immunostained with anti-V5 mAb (visualized using Cy3) and rabbit anti-Myc Ab (visualized using FITC). Notice that the subcellular localization of BACE in juxtanuclear compartments and the partial colocalization with GGA1 is the same in WT, S498D and S498A transfected cells, whereas distribution in the periphery of the cell predominates in the non-phosphorylated BACE mutant (S498A).

morphology-based FRET technique that can reveal proteinprotein interactions in intact cells. The fluorescence lifetime of a donor fluorophore is influenced by the surrounding microenvironment and is shortened in the immediate vicinity of a FRET acceptor molecule. The degree of lifetime shortening can be displayed with very high spatial resolution

in a pseudo-color-coded image. The lifetime is concentration independent. Therefore, the lifetimes of fluorophores attached to even diminutive amounts of protein expressed can be determined.

We measured changes in the lifetime of the donor fluorophore (FITC) under different experimental conditions. In the absence of an acceptor fluorophore, the lifetime of FITC conjugated to IgG (hereafter referred to simply as FITC) is 2250-2400 picoseconds. If an acceptor fluorophore is present but remains too distant from the donor (i.e. there is no interaction), donor lifetimes remain in this range. The lifetime of FITC attached to the C-terminus of wild-type BACE alone (2269±56 picoseconds) was significantly shorter when co-expressed GGA1 was labeled C-terminally with Cy3 (1970±83 picoseconds, P<0.001), indicating FRET between the two fluorophores (Table 1). This was confirmed when the acceptor and donor fluorophores were exchanged (Table 1). Plotting the mean lifetimes calculated for each pixel shows that FITC exhibits a single population of lifetimes in the absence of an acceptor fluorophore, but the proximity of an acceptor creates a second population with a much shorter lifetime (Fig. 2). By creating an image that assigns different colors to pixels belonging to each of these two populations, we were able to determine the subcellular localization of the GGA 1-BACE interaction. Although BACE is distributed throughout the cell (Fig. 2A,C), the interaction with GGA1 occurs only in juxtanuclear compartments (Fig. 2B,D). In order to test the idea that the decrease in lifetime observed in the BACE-V5/GGA1 FLIM assay was caused by FRET, we performed an additional control. In this experiment, BACE was labeled by FITC at its N-terminus, across the membrane from the cytoplasmic Cy3 label on GGA1 and therefore too distant to support FRET. Although there was still striking colocalization of the two proteins, FITC underwent no significant lifetime change (Table 1). This experiment demonstrates the specificity of the proximity assay in this FLIM-based method of measuring FRET.

We next tested the localization of GGA1 in N2a cells co-transfected with wild-type BACE-V5, the pseudophosphorylated BACE-V5 S498D or the non-phosphorylated BACE-V5 S498A. We then co-transfected cells with these BACE mutants and wild-type GGA1, and labeled the C-termini with FITC and Cy3, respectively. The lifetimes of FITC tagged to GGA1 were statistically shorter in the presence of wild-type BACE (Table 1) or

Table 1. Summary data of the FLIM assay for BACE and GGA1 proximity in transfected N2a cells


Lifetime [picoseconds] (mean±s.d.)

n (cells)

Significance (compare to control)


none GGA 1-myc (Cy3) none

BACE-wt-V5 (Cy3)

none GGA 1-myc (Cy3)







10 10 10 10 10 10

PC0.001 PC0.001 n.s.

If there is no interaction, specific lifetimes of the fluorophore are observed, as seen in the negative controls in the absence of the acceptor fluorophore. Statistically shorter lifetimes between BACE and GGA1 suggest that there is FRET between them. However, if the fluorophore is located at the N-terminus of BACE, no lifetime change is observed, because the distance across the membrane (the N-terminus compared with the C-terminus) is too large to support a FRET interaction, despite complete colocalization at the light level. To confirm the data, acceptor and donor fluorophores were exchanged. Statistically shorter lifetimes between GGA1 and BACE suggest that FRET occurs if the fluorophores are exchanged.

Fig. 2. FLIM analysis of the proximity of wild-type BACE and GGA1 within cells. N2a cells were transfected with BACE-V5 (labeled with FITC) and GGA1-Myc (labeled with Cy3). Only the donor fluorophore (FITC) was applied for the negative control (A,B). The intensity image shows the standard immunostaining pattern for BACE (A,C). The color-coded FLIM image shows the lifetimes (in picoseconds) of FITC in the absence (B) or presence (D) of the acceptor Cy3. The shorter FITC lifetimes reflecting proximity between GGA1 and BACE appear only in juxtanuclear compartments (TGN) (D). The negative control exhibits a single population of lifetimes for FITC in the absence of an acceptor, represented by a homogenous blue-green pseudocolor image.

pseudophosphorylated BACE-S498D than non-phosphorylated BACE-S498A or FITC alone (Table 2, P<0.001). The interaction of GGA1 with wild-type or BACE-S498D generated a second population with shorter FITC lifetimes that was restricted to the juxtanuclear compartment and was absent when Cy3 was absent or attached to non-phosphorylated BACE-S498A (Fig. 3). Because this FLIM analysis method is concentration independent, the data indicate that interaction occurs only in a relatively small area of the total cell but, as indicated by the intensity images, this area is also the area of high expression of both BACE and GGA1.

As expected we saw no significant difference between pseudophosphorylated and wild-type BACE by FLIM; the

Fig. 3. Phosphorylation-dependent interaction of GGA1 with BACE. Intensity images of GGA1 (A,C,E,G) and pseudocolored FLIM images when co-transfected with wild-type BACE (WT-BACE; D) or the BACE-S498D (F) and BACE-S498A (H) mutants in N2a cells. In the absence of an acceptor or Cy3-tagged, non-phosphorylated BACE mutants (S498A), no change in fluorescence lifetime was detected. WT-BACE and pseudo-phosphorylated BACE mutants (S498D) showed a significant decrease of fluorescence lifetime in the juxtanuclear compartment.

pseudophosphorylated construct acts like wild-type BACE under the conditions used, as shown by Walter et al. (Walter et al., 2001). As we show in this paper, wild-type BACE is phosphorylated in cells. Using the S498D mutant, we show that the effect on the BACE-GGA1 interaction is dependent on the type of mutation at S498 rather than the mere fact of mutation.

Table 2. Summary data of the FLIM assay for GGA1 and BACE mutants


Lifetime [picoseconds] (mean±s.d.)

Significance n (cells) (compared to control)

GGA 1-myc (FITC) GGA 1-myc (FITC) GGA 1-myc (FITC)

BACE-S498D-V5 (Cy3) BACE-S498A-V5 (Cy3)

2383±64 2195±67 2336±52

10 10 10


N2a cells were stained for GGA1-Myc with rabbit-anti-Myc Ab (labeled with FITC) and for BACE mutants with mouse-anti-V5 Ab (labeled with Cy3). For the negative control, the primary antibody for the acceptor fluorophore was not applied. Statistically shorter lifetimes between GGA1 and wild-type and pseudophosphorylated BACE (S498D) show that proximity between GGA1 and BACE is phosphorylation dependent. Statistical testing was performed using Student's t test.

Localization and amount of phosphorylation of wild-type BACE

We next investigated the proportion of BACE that is phosphorylated at Ser498 in N2a cells as well as its cellular distribution. To do so, we developed a FRET-based assay that allows us to observe the phosphorylation of a specific protein without undertaking the arduous task of generating an antibody to the phosphorylated form of that protein. Our method involves attaching a donor fluorophore to the protein of interest and using an anti-phosphoserine antibody to carry the acceptor fluorophore using methods analogous to those of Ng et al. (Ng et al., 1999). The donor fluorophore should be attached to the protein or an epitope of the protein close to - but not blocking - the phosphorylation site. When the protein is phosphorylated, we observed a shorter lifetime for the donor fluorophore owing to quenching by the acceptor attached to the phosphorylation site. In this case, we transfected cells with BACE fused at the C-terminus to GFP and stained the fixed cells with a Cy3-labeled anti-phophoserine antibody. BACE-GFP was chosen because antibodies to V5 at the C-terminus of BACE empirically blocked binding of the phosphorylation-specific antibody (which is only two amino acids from the C-terminus). Before fixation, the cells were treated for 1 hour with okadaic acid, a serine-phosphatase inhibitor, to increase the proportion of BACE that was serine phosphorylated. Immunostaining showed a typical localization of BACE and a very high level of phosphoserine staining widely distributed around the cell,

Fig. 4. Localization of BACE-GFP and endogenous phosphoserine residues. N2a cells were transfected with BACE-GFP, treated with OA and stained with an antibody specific to phosphorylated serine residues, labeled by Cy3. BACE-GFP shows the standard immunostaining pattern for BACE. Phosphoserine residues show a homogenous staining throughout the cell.

as expected (Fig. 4). The fluorescence lifetime of GFP alone ranges from 2200-2350 picoseconds, and we measured the lifetime of BACE-GFP in the absence of Cy3 as 2270±56 picoseconds. When Cy3 was attached to phosphorylated serine residues, we observed a significant decrease in the lifetime of BACE-GFP to 2115±57 (P<0.001, Table 3). In pseudocolored images, BACE-GFP with shorter lifetimes representing phosphorylated BACE appears to occur in distal cell compartments, at or near the cell membrane (Fig. 5).

To show that this interaction is unique to BACE's S498 phosphorylation site and does not indicate the proximity of a phosphoserine on some other protein interacting closely with BACE, we created a non-phosphorylatable BACE-GFP mutant (S498A) and repeated the assay. No significant FRET was observed despite OA treatment (Table 3).

Immunoprecipitation of phosphorylated BACE

We immunoprecipitated phosphorylated BACE from homogenized human brain tissue to determine whether BACE phosphorylation occurs under physiological conditions in brain where BACE function is presumed to be important in the pathogenesis of AD. Phosphorylated BACE was captured under two conditions: (1) immunoprecipitation with anti-BACE-CT antibody followed by probing with anti-phosphoserine; (2) immunoprecipitation with anti-phosphoserine antibody followed by probing with anti-BACE-NT antibody. Both conditions produced strong immunoreactive bands at ~64 kDa, the expected size of endogenous, glycosylated BACE (Fig. 6, lanes 2,3). BACE that is not glycosylated (~55 kDa) does not appear in the phosphoserine pull-down lanes (Fig. 6, lane 1) and, to a much lesser extent, in the BACE pull-down probed with anti-phosphoserine antibody compared with probing with anti-BACE antibody. This indicates that serine phosphorylation occurs preferentially in glycosylated BACE. In the BACE pull-down, we saw additional bands at ~50 kDa and ~90 kDa, which might be due to pull-down of BACE complexes with other phosphoserinylated proteins. The capture of phosphorylated BACE did not appear to be dependent on the diagnosis, age or post-mortem index of the cases (Fig. 6, lanes 1,2). The recombinant BACE control as well as the homogenates that were captured and probed for BACE confirmed the presence of glycosylated BACe at ~64 kDa and unglycosylated BACE (Capell et al., 2000) at ~55 kDa (Fig. 6, lanes 4,5) in human brain. Recombinant BACE migrated at the same size as phosphorylated BACE, as shown by Walter et al. (Walter et al.,

Table 3. Summary data of the FLIM assay for BACE phosphorylation


Lifetime [picoseconds] (mean±s.d.)

Significance n (cells) (compared to control)






Anti-phosphoserine Ab (Cy3) Anti BACE-CT Ab (Cy3) = pos. control none

Anti-phosphoserine Ab (Cy3)






10 10 10 15 15

P<0.001 P<0.001

N2a cells were transfected with BACE-GFP, treated with OA and stained with an antibody specific to phosphorylated endogenous serine residues for the analysis (labeled by Cy3). For the negative control, the primary antibody for the acceptor fluorophore was not applied. For the positive control a Cy3-labeled antibody to the BACE C-terminus was applied. Statistically shorter lifetimes between BACE-GFP and anti-phosphoserine Cy3-labelled antibody, but not between the non-phosphorylated BACE-S498A-GFP and anti-phosphoserine Cy3-labelled antibody show a specific measurement of BACE phosphorylation. Statistical testing was performed using Student's t test.

Fig. 5. FLIM analysis of BACE phosphorylation. N2a cells were transfected with BACE-GFP, treated with 20 nM OA and stained with an antibody specific to phosphorylated serine residues for the analysis. Negative controls were only transfected with BACE-GFP and treated with OA (A,B). The intensity image shows the standard immunostaining pattern for BACE (A,C). The color-coded FLIM image shows the lifetimes (in picoseconds) of GFP in the presence of acceptor Cy3. The lifetime reflects the proximity between BACE-C-terminus and phosphoserine residues, demonstrated in pseudocolor. The FLIM image suggests that there is a close proximity between the BACE C-terminus and phosphoserine residues not only in juxtanuclear compartments (TGN), but throughout the cell (D). The chart shows mean changes in lifetime and statistically shorter lifetimes in the presence of Cy3-labeled endogenous phosphoserine (P<0.001).

Fig. 6. Immunoprecipitation of BACE phosphorylated at its Serine residue in human brain tissue. (Lane 1) AD human brain. Immunoprecipitation (IP) with ms anti-phosphoserine, probe with rb anti-BACE (N-terminus). (Lane 2) Normal human brain. IP with ms anti-phosphoserine, probe with rb anti-BACE (N-terminus). (Lane 3) Normal human brain. IP with rb anti-BACE (C-terminus), probe with ms anti-phosphoserine. (Lane 4) Normal human brain. IP with ms anti-BACE (C-terminus), probe with rb anti-BACE (N-terminus) (positive control). (Lane 5) BACE standard. Probe with rb anti-BACE (N-terminus) (positive control). (Lane 6) Buffer. IP with ms anti-phosphoserine, probe with rb anti-BACE (N-terminus) (negative control). (Lane 7) Normal human brain. IP with ms anti-phosphotyrosine, probe with rb anti-BACE (N-terminus) (negative control).

2001), with only an undetectable change in size through phosphorylation being observed, as one would expect from phosphorylation at only one site. Immunoprecipitation reactions using anti-phosphotyrosine antibodies and reactions using lysis buffer in place of brain homogenates were free of this band (Fig. 6, lanes 6,7).

APP shedding

In order to assess the effect of BACE phosphorylation and interaction with GGA1 via its dileucine motif on APP processing, we measured shedding of the ectodomain of APP with the cDNA of SEAP fused to its N-terminus (Lichtenthaler et al., 2003) as a very sensitive indicator of BACE cleavage. After co-transfection of the SEAP-APP construct with a fi-gal reporter construct and empty vector, wild-type BACE, BACED93/289A, BACE-S498A, S498D or LL499/500AA, SEAP activity was measured in the medium and normalized to f -gal activity. Wild-type BACE produced a significant increase in APP cleavage compared with the baseline, whereas catalytically inactive BACED93/289A exhibited no effect, as expected. Mutating BACE at S498 to simulate or prevent phosphorylation or at LL499/500 to prevent interaction with GGA1 entirely did not significantly alter BACE activity from the wild-type level (Fig. 7). This is in accordance with previous reports showing no effect of BACE S498 or dileucine mutations on Af generation (Pastorino et al., 2002; Walter et al., 2001).


Our data address the subcellular localization of BACE phosphorylation and BACE-GGA interactions. Endogenous

700000 600000 500000 400000 300000 200000 100000 0


D93/289A S498A S498D LL499/500AA

Fig. 7. APP shedding assay. HEK293 cells were transfected with SEAP-APP, P-gal and empty vector (Topo) or with a plasmid encoding wild-type BACE, a catalytically inactive BACE mutant (BACE D93/289A), non-phosphorylated BACE mutant (BACE S498A), pseudophosphorylated BACE mutant (BACE S498D) or BACE dileucine mutant (LL 499/500 AA). The alkaline phosphatase activity in the conditioned medium is shown, normalized to f -gal activity, with the means and SD after both transfection and measurement were carried out in triplicate, representing two independent assays. **, P<0.001 (ANOVA, Fisher's PLSD post hoc test).



GGA1 localizes predominantly to the TGN but is also present in early endosomes (Takatsu et al., 2001). BACE resides predominantly in the Golgi and endosomal compartments, but can also be observed on the cell surface (Kinoshita et al., 2003). The colocalization in the juxtanuclear compartments and, to a lesser extent, the endosomes reported here and elsewhere is, however, not sufficient to demonstrate protein-protein interaction in these compartments. To detect interaction, we measured FRET between fluorophores attached to the C-termini of each protein by means of FLIM. FLIM is a versatile, concentration-independent measure of proteinprotein interactions in whole cells (Berezovska et al., 2003). Our FLIM data suggest that BACE and GGA1 come into close proximity primarily in the TGN in N2a cells, indicating that the interaction occurs primarily in juxtanuclear compartments.

In addition, we showed in cells that this interaction is dependent on the S498 phosphorylation state of BACE, consistent with results from binding constants with isolated proteins (He et al., 2002; He et al., 2003). Shiba et al. (Shiba et al., 2004) recently described the X-ray crystal structure of a heterodimeric crystal of the BACE C-terminus and GGA1 VHS domain at 1.9 Ä resolution. Increased GGA affinity for phosphorylated BACE did not arise from interaction of negatively charged atoms in the VHS domain with the positively charged phosphate but rather increased hydrogen bonding and electrostatic interactions between the two proteins. These data further support our findings in cells. The VHS domain of GGA1 binds more efficiently to the BACE ACDL motif when this serine residue is phosphorylated by casein kinase 1. GGA function itself has been shown to be regulated by cycles of phosphorylation and dephosphorylation of a serine residue upstream of the DXXLL sequence of GGA by casein kinase 2 (Ghosh et al., 2003).

Although BACE and its unprocessed form pro-BACE have been identified in human brain, it is unclear whether the extensive post-translational modifications identified in cell culture also occur in vivo. The identification of phosphorylated BACE in AD and control brain indicate that this modification of BACE is physiologically relevant. In N2a cells challenged with a serine-phosphatase inhibitor, FLIM reveals that BACE phosphorylation at S498 occurs in multiple cell compartments - surprisingly, not only in the Golgi but also at or near the cell membrane. Thus, this post-translational modification probably represents a regulatory mechanism because, in pull-down assays of endogenous BACE from human brain, only the mature, glycosylated form of BACE appears to undergo phosphorylation at S498.

We propose that, after endocytosis, phosphorylated BACE interacts in the Golgi with GGA1 for recycling of BACE to cell surface and eventually to cell compartments in which BACE can interact with and cleave APR Proteins interacting with other GGA1 domains affect Aß production. Rabaptin5 interacts with the GGA1 GAT domain, and upregulation of its effector (Rab5) leads to increased Aß-40 and -42 production (Grbovic et al., 2003). Similarly, upregulation of cation-independent mannose-6-phosphatase receptors (MPR), which interacts with the VHS-domain of GGAs, also increases Aß generation (Mathews et al., 2002). These data suggest a role for GGA in regulating APP cleavage. However, this effect might be indirect, because neither BACE phosphorylation (Walter et al., 2001) nor mutation of the BACE dileucine motif

(Pastorino et al., 2002) affects Aß production or BACE-dependent APP shedding. Further studies will be required to determine whether BACE or GGA1 phosphorylation alters APP metabolism in a more subtle fashion. Taken together, these data suggest that serine phosphorylation of BACE is a physiologically relevant post-translational modification that influences BACE trafficking and might thereby indirectly affect BACE interaction with APP.

We thank M. S. Robinson for the GGA construct and S. F. Lichtenthaler for the SEAP-APP construct. Our work was supported by NIH AG 12406, a grant from the DFG (Deutsche Forschungsgemeinschaft) to CAFvA (AR 379/1-1) and from the American Federation for Aging Research Beeson-Award to MCI and BML.


Bacskai, B. J., Skoch, J., Hickey, G. A., Allen, R. and Hyman, B. T. (2003). Fluorescence resonance energy transfer determinations using multiphoton fluorescence lifetime imaging microscopy to characterize amyloid-beta plaques. J. Biomed. Opt. 8, 368-375.

Bennett, B. D., Denis, P., Haniu, M., Teplow, D. B., Kahn, S., Louis, J. C., Citron, M. and Vassar, R. (2000). A furin-like convertase mediates propeptide cleavage of BACE, the Alzheimer's beta-secretase. J. Biol. Chem. 275, 37712-37717.

Berezovska, O., Ramdya, P., Skoch, J., Wolfe, M. S., Bacskai, B. J. and Hyman, B. T. (2003). Amyloid precursor protein associates with a nicastrin-dependent docking site on the presenilin 1-gamma-secretase complex in cells demonstrated by fluorescence lifetime imaging. J. Neurosci. 23, 45604566.

Bodendorf, U., Danner, S., Fischer, F., Stefani, M., Sturchler-Pierrat, C., Wiederhold, K. H., Staufenbiel, M. and Paganetti, P. (2002). Expression of human beta-secretase in the mouse brain increases the steady-state level of beta-amyloid. J. Neurochem. 80, 799-806.

Boman, A. L. (2001). GGA proteins: new players in the sorting game. J. Cell Sci. 114, 3413-3418.

Bonifacino, J. S. (2004). The GGA proteins: adaptors on the move. Nat. Rev. Mol. Cell Biol. 5, 23-32.

Capell, A., Steiner, H., Willem, M., Kaiser, H., Meyer, C., Walter, J., Lammich, S., Multhaup, G. and Haass, C. (2000). Maturation and propeptide cleavage of beta-secretase. J. Biol. Chem. 275, 30849-30854.

Chyung, J. H. and Selkoe, D. J. (2003). Inhibition of receptor-mediated endocytosis demonstrates generation of amyloid beta-protein at the cell surface. J. Biol. Chem. 278, 51035-51043.

Creemers, J. W., Ines Dominguez, D., Plets, E., Serneels, L., Taylor, N. A., Multhaup, G., Craessaerts, K., Annaert, W. and de Strooper, B. (2001). Processing of beta-secretase by furin and other members of the proprotein convertase family. J. Biol. Chem. 276, 4211-4217.

Ghosh, P., Griffith, J., Geuze, H. J. and Kornfeld, S. (2003). Mammalian GGAs act together to sort mannose 6-phosphate receptors. J. Cell Biol. 163, 755-766.

Grbovic, O. M., Mathews, P. M., Jiang, Y., Schmidt, S. D., Dinakar, R., Summers-Terio, N. B., Ceresa, B. P., Nixon, R. A. and Cataldo, A. M.

(2003). Rab5-stimulated up-regulation of the endocytic pathway increases intracellular beta-cleaved amyloid precursor protein carboxyl-terminal fragment levels and Abeta production. J. Biol. Chem. 278, 31261-31268.

Haass, C., Koo, E. H., Mellon, A., Hung, A. Y. and Selkoe, D. J. (1992). Targeting of cell-surface beta-amyloid precursor protein to lysosomes: alternative processing into amyloid-bearing fragments. Nature 357, 500503.

He, X., Chang, W. P., Koelsch, G. and Tang, J. (2002). Memapsin 2 (beta-secretase) cytosolic domain binds to the VHS domains of GGA1 and GGA2: implications on the endocytosis mechanism of memapsin 2. FEBS Lett. 524, 183-187.

He, X., Zhu, G., Koelsch, G., Rodgers, K. K., Zhang, X. C. and Tang, J.

(2003). Biochemical and structural characterization of the interaction of memapsin 2 (beta-secretase) cytosolic domain with the VHS domain of GGA proteins. Biochemistry 42, 12174-12180.

Hirst, J., Lui, W. W., Bright, N. A., Totty, N., Seaman, M. N. and Robinson, M. S. (2000). A family of proteins with gamma-adaptin and VHS domains

that facilitate trafficking between the trans-Golgi network and the vacuole/lysosome. J. Cell Biol. 149, 67-80.

Hirst, J., Lindsay, M. R. and Robinson, M. S. (2001). GGAs: roles of the different domains and comparison with AP-1 and clathrin. Mol. Biol. Cell 12, 3573-3588.

Huse, J. T., Pijak, D. S., Leslie, G. J., Lee, V. M. and Doms, R. W. (2000). Maturation and endosomal targeting of beta-site amyloid precursor protein-cleaving enzyme. The Alzheimer's disease beta-secretase. J. Biol. Chem. 275, 33729-33737.

Hussain, I., Powell, D., Howlett, D. R., Tew, D. G., Meek, T. D., Chapman, C., Gloger, I. S., Murphy, K. E., Southan, C. D., Ryan, D. M. et al.

(1999). Identification of a novel aspartic protease (Asp 2) as beta-secretase. Mol. Cell. Neurosci. 14, 419-427.

Kinoshita, A., Fukumoto, H., Shah, T., Whelan, C. M., Irizarry, M. C. and Hyman, B. T. (2003). Demonstration by FRET of BACE interaction with the amyloid precursor protein at the cell surface and in early endosomes. J. Cell Sci. 116, 3339-3346.

Koo, E. H. and Squazzo, S. L. (1994). Evidence that production and release of amyloid beta-protein involves the endocytic pathway. J. Biol. Chem. 269, 17386-17389.

Koo, E. H., Squazzo, S. L., Selkoe, D. J. and Koo, C. H. (1996). Trafficking of cell-surface amyloid beta-protein precursor. I. Secretion, endocytosis and recycling as detected by labeled monoclonal antibody. J. Cell Sci. 109, 991998.

Lichtenthaler, S. F., Dominguez, D. I., Westmeyer, G. G., Reiss, K., Haass, C., Saftig, P., de Strooper, B. and Seed, B. (2003). The cell adhesion protein P-selectin glycoprotein ligand-1 is a substrate for the aspartyl protease BACE1. J. Biol. Chem. 278, 48713-48719.

Lin, X., Koelsch, G., Wu, S., Downs, D., Dashti, A. and Tang, J. (2000). Human aspartic protease memapsin 2 cleaves the beta-secretase site of beta-amyloid precursor protein. Proc. Natl. Acad. Sci. USA 97, 1456-1460.

Mathews, P. M., Guerra, C. B., Jiang, Y., Grbovic, O. M., Kao, B. H., Schmidt, S. D., Dinakar, R., Mercken, M., Hille-Rehfeld, A., Rohrer, J. et al. (2002). Alzheimer's disease-related overexpression of the cation-dependent mannose 6-phosphate receptor increases Abeta secretion: role for altered lysosomal hydrolase distribution in beta-amyloidogenesis. J. Biol. Chem. 277, 5299-5307.

Ng, T., Squire, A., Hansra, G., Bornancin, F., Prevostel, C., Hanby, A., Harris, W., Barnes, D., Schmidt, S., Mellor, H. et al. (1999). Imaging protein kinase Calpha activation in cells. Science 283, 2085-2089.

Nielsen, M. S., Madsen, P., Christensen, E. I., Nykjaer, A., Gliemann, J., Kasper, D., Pohlmann, R. and Petersen, C. M. (2001). The sortilin cytoplasmic tail conveys Golgi-endosome transport and binds the VHS domain of the GGA2 sorting protein. EMBO J. 20, 2180-2190.

Ohno, M., Sametsky, E. A., Younkin, L. H., Oakley, H., Younkin, S. G., Citron, M., Vassar, R. and Disterhoft, J. F. (2004). BACE1 deficiency rescues memory deficits and cholinergic dysfunction in a mouse model of Alzheimer's disease. Neuron 41, 27-33.

Orlando, L. R., Dunah, A. W., Standaert, D. G. and Young, A. B. (2002). Tyrosine phosphorylation of the metabotropic glutamate receptor mGluR5 in striatal neurons. Neuropharmacology 43, 161-173.

Pastorino, L., Ikin, A. F., Nairn, A. C., Pursnani, A. and Buxbaum, J. D. (2002). The carboxyl-terminus of BACE contains a sorting signal that regulates BACE trafficking but not the formation of total A(beta). Mol. Cell. Neurosci. 19, 175-185.

Puertollano, R., Aguilar, R. C., Gorshkova, I., Crouch, R. J. and Bonifacino, J. S. (2001). Sorting of mannose 6-phosphate receptors mediated by the GGAs. Science 292, 1712-1716.

Riddell, D. R., Christie, G., Hussain, I. and Dingwall, C. (2001). Compartmentalization of beta-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts. Curr. Biol. 11, 1288-1293.

Roberds, S. L., Anderson, J., Basi, G., Bienkowski, M. J., Branstetter, D. G., Chen, K. S., Freedman, S. B., Frigon, N. L., Games, D., Hu, K. et al. (2001). BACE knockout mice are healthy despite lacking the primary beta-secretase activity in brain: implications for Alzheimer's disease therapeutics. Hum. Mol. Genet. 10, 1317-1324.

Selkoe, D. J. (2001). Alzheimer's disease: genes, proteins, and therapy. Physiol. Rev. 81, 741-766.

Shiba, T., Kametaka, S., Kawasaki, M., Shibata, M., Waguri, S., Uchiyama, Y. and Wakatsuki, S. (2004). Insights into the phosphoregulation of beta-secretase sorting signal by the VHS domain of GGA1. Traffic 5, 437-448.

Sinha, S., Anderson, J. P., Barbour, R., Basi, G. S., Caccavello, R., Davis, D., Doan, M., Dovey, H. F., Frigon, N., Hong, J. et al. (1999). Purification and cloning of amyloid precursor protein beta-secretase from human brain. Nature 402, 537-540.

Takatsu, H., Katoh, Y., Shiba, Y. and Nakayama, K. (2001). Golgi-localizing, gamma-adaptin ear homology domain, ADP-ribosylation factor-binding (GGA) proteins interact with acidic dileucine sequences within the cytoplasmic domains of sorting receptors through their Vps27p/Hrs/STAM (VHS) domains. J. Biol. Chem. 276, 28541-28545.

Vassar, R., Bennett, B. D., Babu-Khan, S., Kahn, S., Mendiaz, E. A., Denis, P., Teplow, D. B., Ross, S., Amarante, P., Loeloff, R. et al. (1999). Beta-secretase cleavage of Alzheimer's amyloid precursor protein by the transmembrane aspartic protease BACE. Science 286, 735-741.

Walter, J., Fluhrer, R., Hartung, B., Willem, M., Kaether, C., Capell, A., Lammich, S., Multhaup, G. and Haass, C. (2001). Phosphorylation regulates intracellular trafficking of beta-secretase. J. Biol. Chem. 276, 14634-14641.

Yan, R., Bienkowski, M. J., Shuck, M. E., Miao, H., Tory, M. C., Pauley, A. M., Brashier, J. R., Stratman, N. C., Mathews, W. R., Buhl, A. E. et

al. (1999). Membrane-anchored aspartyl protease with Alzheimer's disease beta-secretase activity. Nature 402, 533-537.

Zhu, Y., Doray, B., Poussu, A., Lehto, V. P. and Kornfeld, S. (2001). Binding of GGA2 to the lysosomal enzyme sorting motif of the mannose 6-phosphate receptor. Science 292, 1716-1718.