Modulation of Amyloid Precursor Protein Metabolism by X11α/Mint-1

Modulation of amyloid precursor protein (APP) metabolism plays a pivotal role in the pathogenesis of Alzheimer's disease. The phosphotyrosine-binding/protein interaction (PTB/PI) domain of X11α, a neuronal cytosolic adaptor protein, binds to the YENPTY sequence in the cytoplasmic carboxyl terminus of APP. This interaction prolongs the half-life of APP and inhibits Aβ40 and Aβ42 secretion. X11α/Mint-1 has multiple protein-protein interaction domains, a Munc-18 interaction domain (MID), a Cask/Lin-2 interaction domain (CID), a PTB/PI domain, and two PDZ domains. These X11α protein interaction domains may modulate its effect on APP processing. To test this hypothesis, we performed a deletion analysis of X11α effects on metabolism of APP695 Swedish (K595N/M596L) (APPsw) by transient cotransfection of HEK 293 cells with: 1) X11α (X11α-wt, N-MID-CID-PTB-PDZ-PDZ-C), 2) amino-terminal deletion (X11α-ΔN, PTB-PDZ-PDZ), 3) carboxyl-terminal deletion (X11α-ΔPDZ, MID-CID-PTB), or 4) deletion of both termini (PTB domain only, PTB). The carboxyl terminus of X11α was required for stabilization of APPsw in cells. In contrast, the amino terminus of X11α was required to stimulate APPs secretion. X11α, X11α-ΔN, and X11α-PTB, but not X11α-ΔPDZ, were effective inhibitors of Aβ40 and Aβ42 secretion. These results suggest that additional protein interaction domains of X11α modulate various aspects of APP metabolism.

Alzheimer's disease (AD) 1 is a progressive neurodegenerative disorder that is pathologically defined by the density of neurofibrillary tangles and amyloid plaques in brain. Accumulating biochemical, genetic, and pathologic evidence suggests that amyloidogenesis plays a pivotal and possibly causal role in AD. The major components of amyloid plaque in AD brain are A␤ peptides including A␤40 and A␤42 that are derived by proteolysis of the type I transmembrane glycoprotein amyloid precursor protein (APP). Alternate pathways of APP proteolysis by either ␤-, ␥-secretases or ␣-, ␥-secretases result in the secretion of either APPs␤ and A␤, or APPs␣ and p3, respectively, and the generation of a series of cell-associated carboxylterminal stubs (1).
Mutations identified in rare autosomal dominant pedigrees of AD alter APP processing. For example, the APP sw mutation (K595N/M596L of APP 695 ) was identified in affected subjects of a Swedish kindred of dominantly inherited AD; this mutation is located at the ␤-secretase cleavage site of APP and promotes ␤-secretase cleavage, thus increasing both A␤40 and A␤42 generation and amyloidogenesis in brain. This data strongly suggest that modulation of APP processing and the generation of A␤ peptides in particular play an important role in AD pathogenesis. In support of the amyloid hypothesis, all known familial AD mutations in APP, presenilin-1, and presenilin-2 result in greater generation of the more amyloidogenic A␤42 from APP (1).
Proteins that interact with the highly conserved cytoplasmic carboxyl terminus of APP also modulate its cellular processing. For example, X11␣/Mint-1 and its homologs X11-like (X11␤/ Mint-2) and X11-like2 (X11␥/Mint-3) (2-6), Fe65 and its homologs Fe65-like1 and Fe65-like2 (7)(8)(9)(10)(11)(12)(13), and mDAB1 (mammalian homolog of disabled) (14 -15) interact with the intracellular carboxyl terminus of APP. The X11 and Fe65 gene families and mDAB1 contain phosphotyrosine binding/protein interaction (PTB/PI) domains first described in Shc (16,17). The PTB domain of Shc specifically interacts with ⌿XNPXpY motifs (where ⌿ is a hydrophobic residue, X is any amino acid, N is asparagine, P is proline, and pY is phosphotyrosine) found in tyrosine kinase receptors, thus mediating signal transduction. Similarly, the PTB domains of X11, Fe65, and their homologs bind to the YENPTY motif in the carboxyl terminus of APP, although tyrosine phosphorylation has no discernable effect. The crystal structure of the PTB domain of X11␣ bound to peptides of APP spanning the YENPTY sequence reveals that eight peptide residues make specific contacts with the PTB domain, and they collectively achieve high affinity (K d ϭ 0.32 M) and specificity (18).
The interaction of X11␣ with APP prolongs its half-life within cells and inhibits secretion of A␤40 and A␤42 (4,5) perhaps by retarding APP endocytosis. In contrast, Fe65 and Fe65-like1 promote maturation and translocation of APP to the cell surface and increase APPs␣ and A␤ secretion (9,19). These differential results suggest that additional protein-protein interaction motifs of X11 and Fe65 may mediate these opposing, and perhaps competing, effects on APP metabolism. Fe65 has two PTB domains (only one of which binds to APP) and a WW domain that binds to Mena (mammalian homolog of enabled) (20). The non-APP binding PTB domain of Fe65 binds to the LRP family of lipoprotein receptors (14). X11␣/Mint-1 also has other protein interaction domains, a Munc-18 interaction domain (MID), a Cask/Lin-2 interaction domain (CID) aminoterminal to the PTB/PI domain, and two carboxyl-terminal PDZ (Postsynaptic density-95, Disks-large, epithelial tight junction protein ZO-1) domains. Thus, APP and X11␣ may be components of a multimeric assembly that mediates their normal functions and regulates their trafficking and processing. We hypothesized that the MID, CID, and PDZ domains of X11␣ would modulate its effects on APP processing. To test this hypothesis X11␣, amino-terminal deleted (X11␣-⌬N), carboxylterminal deleted (X11␣-⌬PDZ), and both termini deleted (X11␣-PTB) constructs were generated and transiently co-expressed with APP sw in HEK 293 cells. The X11␣ deletion constructs revealed differential and complex effects on APP metabolism, suggesting a role for the non-PTB interaction domains of X11␣ in its modulation of APP trafficking and processing.
DNA Constructs-Myc-tagged X11␣ and its deletion constructs X11␣-⌬N, X11␣-⌬PDZ, and X11␣-PTB were cloned in pRK5 as described previously (2)(3)(4)21). All constructs were created by a polymerase chain reaction procedure and sequenced using Sequenase Version 2.0 (United States Biochemical). The APP 695 isoform and Swedish mutation (K595N/M596L, APP sw ) in pCD was used for this study. APP sw was used instead of wild-type APP to increase A␤40 and A␤42 levels in conditioned medium, and to more closely mimic neuronal APP metabolism in non-neuronal 293 cells (22).
Cell Transfection and Protein Extraction-Six-cm plates were precoated with poly-D-lysine (10 g/ml) 1 day prior to use. HEK 293 cells were plated (7.5 ϫ 10 5 cells/plate) 24 h prior to transfection with 2 g of DNA using FuGene-6 (Roche Molecular Biochemicals) following the manufacturer's instructions. After 48 h, cells were washed twice with phosphate-buffered saline and lysed in radioimmune precipitation buffer (50 mM HEPES, 10% glycerol, 150 mM NaCl, 1% Triton X-100, 1.5 mM MgCl 2 , 1 mM EGTA, pH 7.5) supplemented with complete protease mixture inhibitor (Roche Molecular Biochemicals). Proteins were collected following centrifugation at 16,000 ϫ g for 15 min at 4°C. Total protein in supernatants was determined by the bicinchoninic acid method (Pierce). Conditioned medium was centrifuged at 16,000 ϫ g for 15 min at 4°C. Subsequently, proteins in cell lysate or conditioned medium supernatants were separated by SDS-PAGE, transferred to nitrocellulose, and detected by immunoblot and enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech) with a PhosphorImager (Molecular Dynamics) and x-ray film (Kodak) as described (4). Protein loading per lane was equalized for cell extracts (or the corresponding cell extracts for conditioned medium). PhosphorImager bands were quantitated using ImageQuant software (Molecular Dynamics).
Antibodies and ELISA-The anti-Myc antibody 9E10 (Santa Cruz Biotechnology) was used for immunoblot detection of expressed Myctagged X11␣ constructs. Anti-tubulin antibody (Santa Cruz Biotechnology) was used to detect tubulin. The monoclonal antibody 22C11 (Roche Molecular Biochemicals) raised to an amino-terminal region of APP (residues 66 -81) detects APP in cell lysates and total APPs in medium. An antibody raised to the carboxyl terminus of APP (Chemicon) was also used to detect APP in cell lysates. A sensitive and specific sandwich ELISA for A␤40 and A␤42 was performed as described (4,23). BAN50 was used as the capture antibody and either horseradish peroxidase-coupled BA-27 or BC-05 as the detection antibody for A␤40 or A␤42, respectively. BAN-50 is a monoclonal antibody specific for A␤1-10.
Statistical Analysis-Significant differences between means were determined by multiple analyses of variance using a two-tailed t test.

RESULTS
To examine the potential effects of additional protein-protein interaction domains of X11␣ on APP processing, we deleted the amino terminus (X11␣-⌬N), carboxyl terminus (X11␣-⌬PDZ), or both (PTB domain only). A schematic representation of the X11␣ protein-protein interaction domains, and the deletion constructs used in these experiments are shown in Fig. 1A. Immunoblot of all expressed Myc-tagged X11␣ constructs revealed similar protein levels of the expected M r (Fig. 1B). PhosphorImager quantitation of expression levels of the four X11␣ constructs revealed no significant differences (n ϭ 12 separate experiments).
Coexpression of APP sw with X11␣ or X11␣-⌬N, but not vector, X11␣-⌬PDZ, or X11␣-PTB significantly increased APP sw in cell lysates (Fig. 2). Immunoblot results were identical with an antibody that recognized either the amino terminus ( Fig. 2A) or the carboxyl terminus (Fig. 2B) of APP, thus demonstrating these effects on full-length APP. Quantitation of immunoblots as shown in Fig. 2A revealed that X11␣ and X11␣-⌬N, but not X11␣-⌬PDZ and X11␣-PTB, increased cellular APP 2-fold (p Ͻ 0.01, n ϭ 12 separate experiments) (Fig. 2C). These results suggest that the carboxyl-terminal PDZ domains of X11␣ mediate relative increases in cellular APP. Reprobing the blot with an anti-tubulin antibody revealed equivalent levels, suggesting that the effects were specific to APP and not tubulin and that equal levels of protein were loaded per lane (data not shown).
We next examined the effects of X11␣ and its deletion constructs on secreted APPs in conditioned medium. After transient transfection of HEK 293 cells with APP sw and either vector, X11␣, X11␣-⌬N, X11␣-⌬PDZ, or X11␣-PTB, proteins in conditioned medium were separated by SDS-PAGE followed by immunoblot detection of total APPs with 22C11. X11␣ and X11␣-⌬PDZ, but not X11␣-⌬N or X11␣-PTB, increased total APPs in medium compared with cells transfected with APP sw alone (Fig. 3A). PhosphorImager quantitation of APPs from immunoblots as shown in Fig. 3A revealed that X11␣ and X11␣-⌬PDZ, but not X11␣-⌬N and X11␣-PTB, increased APPs approximately 4-and 8-fold, respectively (p Ͻ 0.01, n ϭ 12 separate experiments) (Fig. 3B). These results suggest that the amino-terminal domains of X11␣ promote APPs secretion. The downward shift in M r suggests that APPs␤ is responsible for the bulk of the relative increase in total APPs. Finally, we measured A␤40 and A␤42 levels in conditioned medium by ELISA (4,23). Cotransfection of APP sw with X11␣, X11␣-⌬N, or X11␣-PTB, but not X11␣-⌬PDZ, decreased levels of both A␤40 and A␤42 in parallel ( Fig. 4; p Ͻ 0.01, n ϭ 8 separate experiments). In contrast, X11␣-⌬PDZ increased secreted A␤40, although this did not reach statistical significance (Fig. 4). These results suggest that the interaction of the PTB domain of X11␣ with the YENPTY motif of APP inhibits A␤ secretion; however, this inhibitory effect is reversed by the deletion of PDZ domains. DISCUSSION The effect of X11␣ on modulation of APP processing is complex and may be influenced by additional protein-protein interaction domains of X11␣. The results may be interpreted by a model of alternate and competing pathways of APP metabolism, with the shunting of APP among these pathways influenced by these additional X11␣ domains. The results best fit a model including: 1) inhibition of endocytic APP metabolism by the PTB domain interaction, 2) inhibition of all APP metabolism by carboxyl-terminal PDZ domain interactions, and 3) promotion of APPs and A␤ secretion by amino-terminal domain interactions. Although this model may be useful in the interpretation of results (Figs. 2-4) in relation to the literature on APP trafficking and processing, further experiments are required to probe its accuracy. Inhibition of endocytic processing is plausible because clathrin-mediated endocytosis also requires the binding of adaptor proteins to the cytoplasmic YENPTY sequence of APP. Also supportive of this model, a yeast two-hybrid screen identified self-interaction of Lin-10, the Caenorhabditis elegans homolog of X11␣ (24), suggesting that X11␣ may dimerize at PDZ domains to retard APP catabolism.
Whereas stabilization of APP in cell lysates and inhibition of A␤ secretion by X11␣ are invariate, results with APPs are more complex. Cotransfection of wt APP with X11␣ decreases APPs␣ and total APPs secretion in conditioned medium. Similarly, cotransfection of APP sw with X11␣ decreases APPs␣ (4, 5). However, cotransfection of APP sw with X11␣ increases total APPs (Fig. 3) and may represent an increase in APPs␤. This FIG. 4. The isolated PTB domain of X11␣ is as effective as X11␣ in inhibition of A␤40 and A␤42 secretion. HEK 293 cells were transiently transfected with empty vector or APP sw alone or in combination with X11␣-wt, X11␣-⌬N, X11␣-⌬PDZ, or X11␣-PTB. Conditioned medium were collected 48 h after transfection and A␤40 and A␤42 were detected by ELISA. The data represent absorbance minus background converted to picograms using standard curves. Results were corrected for protein levels in cell lysates (pg/g protein). Bars indicate mean Ϯ S.E. *, significantly different from APP sw alone; p Ͻ 0.01, n ϭ 8. may be explained by the fact that the APP sw mutation promotes ␤-secretase cleavage at the expense of ␣-secretase cleavage, and that amino-terminal domains of X11␣ may further promote ␤-secretase cleavage of APP sw in the secretory pathway (Fig. 3). Furthermore, the effects of X11␣ on APPs may have biphasic kinetics consisting of an initial inhibition followed by a relative increase in APPs secretion as bulk flow of nascent APP through the secretory pathway resumes whereas endocytic processing is inhibited. Analysis with APPs␤-specific probes and kinetic experiments will shed light on these possibilities. These results are consistent with prior studies demonstrating that generation of A␤ peptides from wt APP requires endocytosis from the cell surface, whereas generation of APPs and A␤ from APP sw occurs in parallel and independent of endocytosis in the exocytic pathway (25)(26)(27)(28)(29).
We cannot exclude the possibility that some of the effects of the X11␣ deletion constructs on APP metabolism may be mediated in part by differences in their binding affinity to APP. However, X11␣, X11␣-⌬N, X11␣-⌬PDZ, and the isolated PTB domain of X11␣ are all capable of binding APP (2,3,18). Similarly, APP binds to the isolated PTB domain of X11␤ as well as full-length X11␤; however, deletion of the carboxyl terminus of X11␤ enhances its binding to APP whereas deletion of its amino terminus suppresses binding. Apparently independent of binding affinities, only full-length X11␤ inhibits A␤40 but not A␤42 secretion (30). This effect is blocked by XB51, a protein that binds to an amino-terminal region of X11␤ (31). Whereas the carboxyl-terminal protein interaction domains in the X11 gene family are highly conserved, the aminoterminal sequences are divergent and thus may modulate APP metabolism differently.
Munc-18 interacts with a domain in the amino-terminal region (MID) of X11␣ and X11␤ (32), but not X11␥. Munc-18 plays an essential role in the docking of vesicles during exocytosis (33). Cask also interacts with an amino-terminal domain (CID) of X11␣, but not X11␤ or X11␥. Cask is a member of the evolutionarily conserved heterotrimeric complex Mint/Cask/ Veli and its C. elegans homologs Lin-10/Lin-2/Lin-7. In C. elegans this modular adaptor protein complex is essential in targeting the epidermal growth factor receptor Let-23 to the basolateral side of vulval epithelial cells and the GLR-1 glutamate receptor to the postsynaptic region of neurons (34 -36). The Mint/Cask/Veli complex is also present in mammalian brain (3,37). Based on their location and other binding partners, functions in both presynaptic and postsynaptic complexes, and synaptogenesis, have been proposed (3, 21, 38 -42). Consistent with this notion, data from knock-out and overexpression models of the Drosophila melanogaster APP homolog suggest that it regulates synaptic structure and number, and this effect is mediated by cytoplasmic domains of APP (43). Furthermore, neurons from APP-deficient mice display deficient survival and neurite outgrowth (44).
X11␣ also has several potential binding partners of its PDZ domains. A presynaptic voltage-gated calcium channel (45) or the dendritic kinesin KIF-17 binds to the first PDZ domain of X11␣. The complex of KIF-17 and Mint-1/Cask/Veli may play a role in transport of N-methyl-D-aspartate receptor containing vesicles to the postsynaptic density (46). Spinophilin/neurabin II, a protein with putative roles in presynaptic formation and function, binds to the second PDZ domain of X11␣ (47,48). Although the role of these potential X11␣ binding partners in APP metabolism in transfected 293 cells is unclear, these protein interactions may modulate APP trafficking and processing in vivo. The APP/Mint/Cask complex is not only localized in neuronal processes suggestive of synaptic localization and function, but also in the Golgi of NT2 neurons, thus suggesting APP tethering and targeting roles for the Mint-1/Cask/Veli complex (21).
The ␤and ␥-secretase activities responsible for APP metabolism have recently been identified. X11, Fe65, and other APPbinding proteins may influence its processing by multiple mechanisms including steric hindrance of proteases or by trafficking APP to alternate cellular compartments containing different complements of protease activities. Recent therapeutic strategies for the treatment of AD have included identifying specific inhibitors of ␤or ␥-secretases. However, it is unlikely that these proteases are specific to APP, and thus, inhibitors may have untoward effects. For example, presenilin-1 promotes intramembranous Notch cleavage as well as APP cleavage (1). Notch signaling is vital for normal mammalian development and has important roles in the adult. The interaction of X11␣ and other binding partners with APP suggests novel therapeutic strategies based on substrate protection rather than protease inhibition. Alternatively, X11␣ has unknown effects on potentially deleterious carboxyl-terminal stubs of APP and intracellular A␤. Therefore, whether this mechanism proves to be a viable therapeutic strategy for AD depends on further in vitro and in vivo studies of the consequences of their interaction.