Tyrosine Phosphorylation of the β-Amyloid Precursor Protein Cytoplasmic Tail Promotes Interaction with Shc*

β-Amyloid precursor protein (APP) is a widely expressed transmembrane protein of unknown function that is involved in the pathogenesis of Alzheimer's disease. The cytoplasmic tail of APP interacts with phosphotyrosine binding (PTB) domain containing proteins (Fe65, X11, mDab-1, and JIP-1) and may modulate gene expression and apoptosis. We now identify Shc A and Shc C, PTB-containing adapter proteins that signal to cellular differentiation and survival pathways, as novel APP-interacting proteins. The APP cytoplasmic tail contains a PTB-binding motif (Y682ENPTY687) that, when phosphorylated on Tyr682, precipitated the PTB domain of Shc A and Shc C, as well as endogenous full-length Shc A. APP and Shc C were physically associated in adult mouse brain homogenates. Increase in phosphorylation of APP by overexpression of the nerve growth factor receptor Trk A in 293T cells promoted the interaction of transfected APP and endogenous Shc A. Pervanadate treatment of N2a neuroblastoma cells resulted in tyrosine phosphorylation and association of endogenous APP and Shc A. Thus, APP and Shc proteins interact in vitro, in cells, and in the mouse brain. Tyrosine phosphorylation of APP may promote the interaction with Shc proteins.

␤-Amyloid precursor protein (APP) is a widely expressed transmembrane protein of unknown function that is involved in the pathogenesis of Alzheimer's disease. The cytoplasmic tail of APP interacts with phosphotyrosine binding (PTB) domain containing proteins (Fe65, X11, mDab-1, and JIP-1) and may modulate gene expression and apoptosis. We now identify Shc A and Shc C, PTB-containing adapter proteins that signal to cellular differentiation and survival pathways, as novel APP-interacting proteins. The APP cytoplasmic tail contains a PTB-binding motif (Y 682 ENPTY 687 ) that, when phosphorylated on Tyr 682 , precipitated the PTB domain of Shc A and Shc C, as well as endogenous full-length Shc A. APP and Shc C were physically associated in adult mouse brain homogenates. Increase in phosphorylation of APP by overexpression of the nerve growth factor receptor Trk A in 293T cells promoted the interaction of transfected APP and endogenous Shc A. Pervanadate treatment of N2a neuroblastoma cells resulted in tyrosine phosphorylation and association of endogenous APP and Shc A. Thus, APP and Shc proteins interact in vitro, in cells, and in the mouse brain. Tyrosine phosphorylation of APP may promote the interaction with Shc proteins.
The ␤-amyloid precursor protein (APP) 1 is a ubiquitously expressed cell membrane protein that is sequentially cleaved by ␤-secretase and ␥-secretase to release extracellular peptides (including the ␤-amyloid peptides, which are deposited in the brain in Alzheimer's disease) and the APP intracellular domain (AID) (reviewed in Ref. 1).
The biological function of APP and the factor(s) that trigger the APP proteolytic cascade remain unclear. APP and its two closely related homologues, APP-like protein (APLP) -1 and -2, appear to share redundant but essential function(s), because mice with single knockouts of these genes only show mild phenotypic abnormalities, whereas double and triple knockouts are lethal in the prenatal period (2). ␥-Secretase cleavage of APP requires nicastrin and presenilins (reviewed in Ref. 3). Presenilins and APP are the only identified proteins known to be mutated in cases of familial Alzheimer's disease. Presenilins are also required for the ␥-secretase proteolysis of the cellsurface receptor Notch (4), which releases the Notch intracellular domain (NICD) that translocates to the nucleus where it acts as a transcriptional regulator (5). Evidence is accumulating that APP is a receptor-like protein (6) whose function is mediated by AID, analogous to Notch/NICD (7)(8)(9)(10). Peptides consistent with AID have been detected in brain homogenates of Alzheimer's and elderly control subjects (7)(8) and in metabolically labeled cells (9). Overexpression of AID induces apoptosis or may augment apoptosis triggered by other factors (8), offering a potential mechanistic explanation for data functionally linking APP and presenilins to enhanced neuronal apoptosis (11)(12)(13)(14)(15). APP interacts via its cytoplasmic tail with a number of cytoplasmic proteins, including Fe65, X11, and murine DAB 1 (16 -18), all of which share characteristics of adapter proteins that could potentially link APP to intracellular signaling pathways. APP may anchor Fe65 in the cytosol, whereas the release of AID from APP by ␥-cleavage may allow the AID-Fe65 complex to translocate to the nucleus where it acts as a transcription factor (10,19).
To identify cellular pathways to which the AID peptide may signal, we used the yeast 2-hybrid system to identify proteins with which AID may interact. As expected, our yeast 2-hybrid screen identified known APP-interacting proteins, including Fe65 and X11. We (20) and others (21) recently identified human c-JUN N-terminal kinase (JNK) interacting protein (JIP)-1 as a novel APP-binding protein. X11, Fe65, mDAb-1, and JIP-1 share as a common feature the presence of phosphotyrosine binding (PTB) domains. PTB domains are protein interaction domains that bind tyrosine residues located within the cytoplasmic tails of activated cellular receptors (22)(23). To investigate further the interaction between APP and the proteins identified in the yeast 2-hybrid screen, we performed glutathione S-transferase (GST) binding and co-immunoprecipitation experiments. As a negative control for these assays, we used the PTB domain of Shc A, which has been reported not to interact with APP (18). Shc A is a member of a family of cytoplasmic adapter proteins (Shc A, Shc B, and Shc C) that interacts with its PTB and Src homology-2 (SH2) domains with receptor tyrosine kinases and activated growth factor receptors (24 -26). Surprisingly, we found that AID and the Shc A PTB domain interact in GST pulldown and immunoprecipitation experiments. Tyrosine phosphorylation of Tyr 682 of the amino acid sequence YENPTY (amino acids 682-687 of the APP 695 isoform), the PTB-binding motif that is contained in APP, was required for the APP-Shc A interaction. Tyrosine-phosphorylated AID also interacted with Shc C, an Shc isoform expressed in the brain. Finally, we show that endogenous APP and Shc C are physically associated in the adult mouse brain.
Cell Culture and Transfection-293T cells, COS7 cells, and N2a cells were grown in RPMI 1640 media (Invitrogen) supplemented with 10% fetal calf serum and penicillin/streptomycin. Transient transfections were performed using FuGENE 6 (Roche Molecular Biochemicals) at 3 l per 1 g of DNA.
In Vitro Transcription and Translation-The TnT-coupled in vitro transcription/translation system (Promega) was used as described previously (12).
GST Fusion Proteins-Recombinant GST fusion proteins were made using the pGEX vector system (Amersham Biosciences) and expressed in Escherichia coli strain BL21 (Invitrogen) or TKB1 bacteria (Stratagene). Phosphorylation of fusion proteins was confirmed by immunoblotting. Proteins were purified using glutathione-Sepharose beads. The following recombinant proteins were expressed in pGEX: vector alone (GST), Ccas, AID, AID Y682G , AID T668A , AID T668E , as well as APP residues 680 -689 and APP residues 682-687.
Immunoprecipitations, Immunoblotting, and Antibodies-For GST pulldown of in vitro translated Shc proteins, a standard protocol was used as described previously (12). For immunoprecipitation of proteins expressed in vivo, cells were harvested 24 -48 h after transfection, washed in PBS, and lysed in lysis buffer containing 1% Triton X-100, sodium fluoride (50 mM), sodium vanadate (1 mM), and protease inhibitors (Mini Tablets, Roche Molecular Biochemicals). The protein concentration was determined using the Bradford method (Protein Assay, Bio-Rad). Equal amounts of protein were immunoprecipitated for 2 h at 4°C.
Antibodies used for immunoprecipitation included APP monoclonal antibody (mAb) (6E10; Signet; used at a concentration of 1 l per 50 l of lysate) or APP polyclonal antibody (R1155, C7, or C8; these are rabbit polyclonal antisera raised to a synthetic polypeptide of APP residues 676 -695; see Refs. 28 and 32; kindly provided by Dr. Dennis Selkoe; each used at a concentration of 1:200), followed by binding to protein A-agarose beads for 1 h (ImmunoPure, Pierce; 25 l per IP), or anti-FLAG mAb bound to agarose beads (M2; Sigma; 20 l per IP). Samples were washed 3 times with lysis buffer and then boiled with SDS loading buffer and DTT. Immunoprecipitates and total lysates were separated by SDS-PAGE and transferred to nitrocellulose membranes.
Pervanadate Treatment of Cells-Pervanadate was prepared by mixing sodium orthovanadate and H 2 O 2 and adding the mixture (or PBS for the control wells) to the culture medium for 10 min at 37°C (final concentrations, of 0.1 and 0.3 mM, respectively). The reaction was stopped by aspiration of the medium, the addition of ice-cold PBS containing 2 mM Na 3 VO 4 , and lysis of the cells as outlined above.
Mice-Mouse experiments were performed in accordance with the ethical standards of the Helsinki Declaration of 1976 (revised in 1983) and the animal care regulations of the Albert Einstein College of Medicine. Adult BALB/c mice (age 3 months) were euthanized, and brains were removed and homogenized in lysis buffer as described above. One mg of protein was used for immunoprecipitations with ␣APP antibody (C7; 1:200), Shc C polyclonal antibody (Ref. 26; 5 l), or Shc A polyclonal antibody (Transduction Laboratories; 100 l). IPs were performed over-night at 4°C, followed by incubation of immunoprecipitates with Sepharose beads (25 l per IP), washing, and immunoblots as described above.

AID Interacts with the Shc A PTB Domain in Vitro-
The experimental details of the yeast 2-hybrid screen by means of which we identified X11 and Fe65 and the novel APP-interacting molecule, JIP-1, have been reported (20). Shc A, which contains both a PTB and an SH2 domain (Fig. 1B), has been reported not to interact with APP (17). Therefore, we cloned the Shc A PTB domain into pcDNA3.1-FLAG to use as a negative control for the investigation of the interaction between APP and X11, Fe65, or JIP-1. The Shc A PTB domain was produced in vitro and incubated with glutathione-Sepharose beads bound to either GST alone, GST-AID, or GST APP-Ccas (the latter consists of APP residues 665-695, the C-terminal caspase cleavage product of APP; see Fig. 1A) (29 -32). Both AID and Ccas contain the Y 682 ENPTY 687 PTB-binding motif ( Fig. 1A) that is fully conserved from Drosophila to mammalian APP, suggesting that it has an important biological role. As shown in Fig. 1C, the Shc A PTB domain interacted with GST-AID and, to a reduced degree, with GST-Ccas but not with GST alone. In addition, the YENPTY hexapeptide alone, with or without the 2 flanking up-and downstream amino acids, was able to pull down Shc PTB, even though the interaction was weaker (Fig. 1C). As reported elsewhere (20), an unrelated protein, ALG-2-interacting protein (AIP)-1 (27), was not pulled down by GST-AID (data not shown). In addition, neither the Shc A CH1 nor the SH2 domain nor the CH2 domain of the p66 isoform of Shc A interacted with any of the GST-AID proteins (Fig. 1D). These in vitro results suggest that AID binds the Shc A PTB domain and that although the YENPTY sequence of APP is sufficient to bind the Shc A PTB domain, the interaction is augmented by AID residues outside the YENPTY motif.
Phosphorylation of AID Increases the Interaction with Shc A PTB in Vitro-For the interaction with the PTB domains of X11, mDAB-1, Fe65, and JIP-1, tyrosine phosphorylation of the cytoplasmic tail of APP is not essential (16 -18, 20 -21) and may even reduce the interaction (16). In contrast, the interaction between Shc A and both the epidermal and nerve growth factor receptors is promoted by tyrosine phosphorylation of the receptor's NPXY consensus PTB binding sequence (34). To determine whether the binding of Shc A PTB to the GST-AID proteins was influenced by phosphorylation of the AID YENPTY motif, we expressed GST-AID in the E. coli strain TKB1 and immunoblotted the bead-bound proteins with phosphotyrosine antibodies. Tyrosine phosphorylation was present in the GST-AID beads when expressed in TKB1 E. coli but not in BL21 E. coli (Fig. 1D, right panel). Compared with GST-AID, phosphorylated GST-AID bound the in vitro translated Shc A PTB to a greater degree (Fig. 1C, right panel).
Tyrosine phosphorylation of C-terminal fragments of APP has been detected in brain homogenates (35), and a recent report (36) describes tyrosine phosphorylation of the cytoplasmic tail of APP by a constitutively active form of the non-receptor tyrosine kinase Abl. In the Y 682 ENPTY 687 motif of APP, Tyr 682 (not Tyr 687 , as might be expected from the sequence of the NPXY consensus PTB-binding motif) was phosphorylated by constitutively active Abl (36). To investigate further the importance of tyrosine phosphorylation of the YENPTY motif for the interaction with Shc A PTB, we mutated Tyr 682 of AID to glycine and expressed this mutated AID Y682G in TKB1 bacteria. Tyrosine phosphorylation was not detected, suggesting that AID is phosphorylated on Tyr 682 (Fig. 1D). Compared with wild type AID, binding of Shc A PTB was reduced to background levels when Shc A PTB was incubated with AID Y682G (Fig. 1C). These data suggest that ty-rosine phosphorylation of AID Tyr 682 is important for the interaction with the Shc A PTB domain.
APP may also be phosphorylated on Thr 668 (37), which may result in alteration of the conformational structure of AID and reduced interaction with at least one cytoplasmic binding partner, Fe65 (38,39). To investigate the role of Thr 668 for the interaction of AID with Shc A PTB, we generated two AID mutants in which Thr 668 was mutated to either Ala or Glu and expressed these constructs as GST fusion proteins. Interestingly, when in vitro translated Shc A PTB was incubated either with AID T668A or with AID T668E , the AID-Shc A PTB interaction was reduced (Fig. 1C). These findings indicate, as has recently been shown for Fe65 (38), that the interaction of AID with the PTB domain of Shc A is mediated by the AID Y 682 ENPTY 687 motif, whereas Thr 668 , a residue N-terminal to this motif, may modulate the interaction.
AID Interacts with Shc A in Vivo-To reconcile our finding that Shc PTB interacts with AID in vitro with the negative results of Margolis and co-workers (18) who did not find the PTB domain of Shc A to interact with APP, we expressed FLAG-tagged Shc A PTB in 293T cells, and we repeated the pulldown experiments with GST-AID proteins, followed by SDS-PAGE and anti-FLAG immunoblotting. These experimental conditions are analogous to those used by Margolis and co-workers (18). As shown in Fig. 2A, Shc A PTB interacted with AID only when phosphorylated on tyrosine, because GST-AID expressed in E. coli strain BL21 as well as AID Y682G was unable to pull down Shc PTB. In addition, the YENPTY motif alone with or without the flanking two N-and C-terminal amino acids was sufficient to precipitate Shc PTB but only when phosphorylated on tyrosine (Fig. 2B). In contrast, neither the CH2-, CH1-, nor the SH2 domain, a phosphotyrosine interaction domain, interacted with any of the GST-AID proteins ( Fig. 2A). As expected, overexpression of full-length Shc A p66 yielded the same results, i.e. precipitation by AID or YENPTY only when phosphorylated on Tyr 682 (Fig. 2B). These findings suggest that the Shc A PTB domain interacts with tyrosinephosphorylated AID in mammalian cells. It is unclear why the in vitro produced Shc A PTB domain interacts with unphosphorylated AID, whereas Shc A expressed in mammalian cells interacts only with phosphorylated AID. It is possible that modifications of Shc A or interactions with other proteins in vivo modulate its interaction with AID.
Next we investigated whether GST-AID could interact with endogenous Shc A in 293T cells. Phosphorylated GST-AID precipitated both the p52 and the p46 isoform of endogenous Shc A from untransfected 293T cell lysates (Fig. 2C), and the phosphorylated YENPTY sequence was sufficient to precipitate endogenous Shc A (Fig. 2D).
We then overexpressed an APP construct comprising the 99 C-terminal amino acids that is a cleavage product of ␤-secretase (C99; see Fig. 1A), along with FLAG-tagged Shc PTB. As shown in Fig. 3A, anti-FLAG antibodies immunoprecipitated C99 in these lysates. C99 was not precipitated in lysates where only C99 or C99 together with FLAG-tagged AIP-1 was expressed. These data suggest that Shc A interacts via its PTB domain with both AID and a longer C-terminal fragment of APP (C99) in mammalian cells. Under these experimental conditions that include overexpression of both C99 and Shc A PTB, the interaction occurred even without evident exogenous tyrosine phosphorylation, suggesting that C99 was basally phosphorylated to a certain extent.
Overexpression of Trk A Promotes the APP-Shc A Interaction-To investigate whether full-length APP interacts with endogenous Shc A in mammalian cells, we attempted unsuccessfully to immunoprecipitate Shc A with anti-APP antibodies and vice versa. One possible explanation was that adequate tyrosine phosphorylation of the APP YENPTY motif was not

FIG. 2. Interaction of AID and Shc A in 293 T cells and adult mouse brain.
A, the interaction of AID and Shc A is mediated by the Shc A PTB domain. Immunoblot analysis with anti-FLAG mAb of the indicated FLAG-tagged Shc A domains or GST alone (GSTϪ lane) were expressed in 293 T cells, and the total cell lysate (T.L.) is also shown. Tyrosine phosphorylation of AID is required for the interaction. B, the phosphorylated Y 682 ENPTY 687 motif of APP is sufficient for the interaction with the Shc A PTB domain (upper panel) and full-length Shc A p66 (lower panel). Cell lysates were pulled down with GST fusion proteins and analyzed by immunoblotting with an anti-FLAG mAb. Refer to Fig. 1A for a representation of the APP residues expressed as GST fusion peptides. C, tyrosine-phosphorylated AID interacts with endogenous Shc A p52/46. After GST pulldown of cell lysates samples were analyzed by Western blot with anti-Shc A antibodies. T., transfected Shc A p52/46, used as a marker; e., endogenous Shc A, detected in the total lysate. D, the phosphorylated 682 YENPTY 687 motif of APP is sufficient for the interaction with endogenous Shc A. GST pulldown and immunoblot with anti-Shc A antibodies. The total cell lysate (T.L.) of untransfected cells was used for the GST pulldown and is shown in the leftmost lane. e., endogenous Shc A detected in the total lysate. available for the interaction with Shc A. To increase tyrosine phosphorylation of cellular proteins, including APP, the following approach was taken. We have recently shown that tyrosine phosphorylation of APP is increased in cells in which the nerve growth factor receptor Trk A is overexpressed. 2 The mechanism by which Trk A expression results in phosphorylation of APP as well as the physiological relevance of this finding are currently not clear, and experiments to evaluate APP phosphorylation in cells expressing Trk A by its physiological ligand nerve growth factor (NGF) are ongoing in our laboratory. Nevertheless, APP phosphorylation via overexpression of Trk A requires intact tyrosine kinase activity of Trk A because a mutated tyrosine kinase-inactive Trk A (where Lys 538 is mutated to Asn) failed to increase tyrosine phosphorylation of APP, in our experiments using phosphotyrosine immunoprecipitation and APP immunoblot and vice versa. 2 We therefore transfected 293T cells with either wild type Trk A or mutated Trk A K538N and immunoprecipitated APP from lysates of these cells. Because the basal expression of APP in COS7 and 293T cells is low (data not shown), we also transfected APP, which was detected in the lysates and immunoprecipitates (Fig. 3B). We then immunoblotted the APP immunoprecipitates with Shc A antibodies, and we found that the p52 isoform of endogenous Shc A was precipitated only in lysates expressing wild type Trk A but not in lysates in which mutated Trk A K538N was expressed (Fig. 3B). This result was confirmed by the converse immunoprecipitation with Shc A antibodies and immunoblot with APP antibodies, which showed that APP was only precipitated in lysates expressing wild type Trk A but not Trk A K538N (Fig. 3B). Thus, and within the limitations of our experimental design, it appears that by increasing tyrosine phosphorylation of APP by overexpression of Trk A APP is promoted to interact with Shc A. The possibility, however, that Trk A induces alternative modifications of Shc A or APP that promote the APP-Shc A interaction cannot be ruled out at present.
Pervanadate Promotes Tyrosine Phosphorylation and Association of Endogenous APP and Shc A-To confirm these findings, and to determine whether phosphorylation leads to association of endogenous APP and Shc A, we treated N2a neuroblastoma cells with pervanadate or PBS before cell lysis and immunoprecipitation. As shown in Fig. 3C (top panel), analysis of total N2a lysates showed that tyrosine phosphorylation was greatly increased in the cells treated with pervanadate. Phosphotyrosine immunoblot revealed bands in both the pervanadate-treated APP and Shc A immunoprecipitates that are consistent in size with APP (and only in the APP immunoprecipitates and APP C-terminal fragments). Conversely, a band consistent with Shc A was detected in pervanadatetreated APP and Shc A immunoprecipitates by phosphotyrosine immunoblot. These findings suggest tyrosine phosphorylation as well as association of APP and Shc A in pervanadate-treated cells. To confirm this, we performed immunoblots with APP and Shc A antibodies on the APP and Shc A immunoprecipitates. APP, including fragments consistent with C-terminal APP fragments generated by ␤and/or ␣-secretase cleavage, was detected in the APP immunoprecipitates from both treated and untreated cells, whereas these bands (W.B.␣APPCT), visualizing both full-length and C-terminal fragments of APP (APPCT). Note that the exposure of the image of APPCT is longer than of full-length APP. Shc A immunoblot was with polyclonal anti-Shc A antibody. C, co-immunoprecipitation of endogenous APP and Shc A in cells treated with pervanadate. N2a neuroblastoma cells were incubated for 10 min with either pervanadate or PBS as described under "Experimental Procedures." Cells were lysed, and lysates (T.L.) were analyzed by SDS-PAGE and phosphotyrosine immunoblot (W.B.␣PY), or immunoprecipitated with anti-APP antibodies (a mixture of R1155 and C8), or Shc A polyclonal antibodies and analyzed by SDS-PAGE and immunoblot. Phosphotyrosine immunoblot reveals bands potentially consistent with Shc A (*) and APP C-terminal fragments (**) in the APP immunoprecipitate, as well as bands possibly consistent with fulllength APP (***) in the Shc A immunoprecipitate. APP immunoblot was with CT695, an antibody recognizing amino acids 676 -695 of APP were only seen in Shc A immunoprecipitates from pervanadatetreated cells (Fig. 3C, middle panel). Conversely, Shc A was detected in both treated and untreated Shc A immunoprecipitates, whereas Shc A was immunoprecipitated by APP antibodies only from cells treated with pervanadate (Fig. 3C, lower  panels). These findings indicate that endogenous APP associates with endogenous Shc A when phosphorylation, including tyrosine phosphorylation, of cellular proteins is increased by pervanadate treatment of N2a neuroblastoma cells.
APP Interacts with Shc C in Vitro, in 293T Cells, and in the Adult Mouse Brain-The level of Shc A in the brain declines during embryogenesis and is low during postnatal life. Concomitant with this down-regulation of Shc A in the brain there is induction of expression of Shc C, a highly homologous Shc isoform that is expressed predominantly in mature neural tissues (26,40). Shc C shares the same modular architecture with PTB, CH1, and SH2 domains. We therefore sought to determine whether APP also interacts with Shc C. To this end, we expressed full-length Shc C in 293T cells and performed GST-AID pulldown experiments as were done for Shc A. As shown in Fig. 4A, and analogous to Shc A, Shc C was pulled down by phosphorylated but not unphosphorylated GST-AID fusion proteins. Similarly, the YENPTY motif of APP with or without the two flanking up-and downstream amino acids was sufficient to precipitate Shc C. As for Shc A, the PTB domain of Shc C, which we transcribed/translated in vitro or expressed in 293T cells, was sufficient for the interaction with phosphorylated GST-AID proteins, in contrast to the CH1 and SH2 domains of Shc C which were not pulled down by GST-AID proteins (Fig.  4B). Whereas the in vitro translated PTB domain of Shc A interacted to a reduced degree also with unphosphorylated AID (Fig. 1C), this was not seen with the PTB domain of Shc C under identical experimental conditions.
To verify if endogenous APP and Shc C interact in the adult mouse brain, we made homogenates of mouse brains and performed immunoprecipitations with APP, Shc A, and Shc C antibodies. As shown in Fig. 4C, APP was precipitated with both APP and Shc C antibodies, whereas APP was not precipitated with Shc A antibodies, consistent with low levels of Shc A expression in the postnatal brain. Similar findings were obtained by the reverse experiment, that is Shc C was precipitated by both Shc C and APP antibodies (Fig. 4C). These experiments indicate that endogenous APP and Shc C associate in the adult mouse brain. DISCUSSION In this report, we describe Shc A and Shc C as novel proteins that interact with the cytoplasmic tail of APP in GST pulldown and immunoprecipitation experiments. The relevance of these findings is underscored by the physical association of endogenous APP and Shc C in adult mouse brain lysates. These findings provide a potential functional link between APP and the modulation of cell survival, because Shc proteins modulate cellular and neuronal differentiation and proliferation (24 -26, 40, 41), and the p66 isoform of Shc A has been linked to apoptosis in response to extracellular stress and the longevity of the organism (42).
We show that increase in phosphorylation of APP in cells by overexpression of Trk A or pervanadate treatment was required for the interaction with Shc A. Site-directed mutagenesis and GST pulldown experiments identified Tyr 682 in the PTB-binding motif Y 682 ENPTY 687 of the APP cytoplasmic tail as the target of tyrosine phosphorylation which promoted the interaction with Shc proteins. Interestingly, tyrosine phosphorylation of C-terminal fragments of APP has recently been detected in the brains of elderly subjects with or without Alzheimer's disease, suggesting a functional role for this process (35). The factor(s) that promote tyrosine phosphorylation of APP in vivo, the signaling pathways that may be modulated by the APP-Shc interaction, and indeed the physiological function of APP are incompletely understood at present.
The emerging model suggests that APP is a signaling receptor whose function is regulated by the cleavage of membranebound APP by ␥-secretase and/or caspases (29 -32). APP-interacting proteins are retained at cell membranes by uncleaved APP, whereas ␥-secretase cleavage of APP releases AID into the cytoplasm, together with the interacting protein. AID complexes may then translocate to other cellular locations includ- FIG. 4. A, phosphorylated AID interacts with full-length Shc C in 293T cells. The phosphorylated Y 682 ENPTY 687 motif of APP is sufficient for the interaction. Cell lysates were pulled down with GST fusion proteins and analyzed by Western blot with a Shc C mAb. B, phosphorylated AID directly interacts with the Shc C PTB domain. In vitro transcription/translation of Shc C PTB domain and GST pulldown as above. C, endogenous APP and Shc C interact in the adult mouse brain. Immunoprecipitation (I.P.) and Western blot (W.B.) analyses of total brain homogenates with the indicated antibodies. HC indicates the heavy chain of the antibodies used for IP. All IPs were done with rabbit polyclonal antibodies; immunoblot was with Shc C mAb and goat anti-mouse secondary antibodies. The visualization of the rabbit HC by the goat anti-mouse secondary antibodies may reflect either a high concentration of antibodies used for IP and/or low level cross-reactivity between the secondary antibody and the rabbit antibodies used for IP. The middle panel shows the portion of the anti-APP immunoblot that includes the heavy chain (HC). Note that although the HC is visualized by both the APP and Shc C antibodies, the Shc C band above the heavy chain, which is seen in the anti-Shc C immunoblot, is not seen in the anti-APP immunoblot (middle panel).
ing the nucleus and axons (10,19,43). This mechanism of regulation of the function of APP is analogous to the function of the Notch receptor. Membrane-bound Notch is sequentially cleaved in response to its physiological ligands to release NICD which translocates to the nucleus where it acts as a transcription factor.
In addition to cleavage of APP by secretases or caspases, tyrosine phosphorylation of APP on Tyr 682 may represent a second mechanism that regulates the putative function of APP as a signaling molecule. Tyrosine phosphorylation of APP may regulate the selection of a binding partner for the APP cytodomain and differentially couple APP to apoptotic or proliferation/differentiation pathways. It is important to note that phosphorylation of APP is not required for the interaction with any of the APP-interacting proteins that have been identified to date (Fe65, mDAB-1, X11, JIP-1, and the recently identified kinesin) (16 -18, 20 -21, 44). Indeed, the interaction of at least one APP-interacting protein (Fe65) may be reduced by APP phosphorylation (38). Thus, as suggested by recent reports by our laboratory (20) and others (21), APP could be linked to the pro-apoptotic JNK pathway via JIP-1, in the absence of tyrosine phosphorylation. In contrast, we show here that tyrosine phosphorylation of APP may promote the docking of Shc proteins via the PTB domain, with the potential to couple APP to cellular pathways generally associated with proliferation and survival, including the Ras/mitogen-activated protein kinase (MAPK) pathway and/or the phosphatidylinositol 3-kinase (PI3K)/Akt pathway (24,25,45). Neuronal differentiation is regulated by signals from receptor tyrosine kinases and the kinetics of subsequent MAPK activation (46) which is modulated by the availability and/or activity of Shc C (40). In addition, activation of MAP kinases (which include p38, p44/42 ERK, and JNK) is well documented in the brain of AD and is associated with the presence of amyloid plaques and neuronal degeneration. MAP kinases may also mediate the hyperphosphorylation of tau protein, the major constituent of intracellular neurofibrillary tangles (33,(47)(48)(49)(50)(51)(52). Elucidation of how APP and/or AID might modulate the function of Shc and MAPK pathways in vivo is currently under investigation in our laboratory and may have implications for the normal biological function of APP and for its pathological function in the pathogenesis of Alzheimer's disease (AD).