Facilitation of stress-induced phosphorylation of beta-amyloid precursor protein family members by X11-like/Mint2 protein.

Beta-amyloid precursor protein (APP) is the precursor of beta-amyloid (Abeta), which is implicated in Alzheimer's disease pathogenesis. APP complements amyloid precursor-like protein 2 (APLP2), and together they play essential physiological roles. Phosphorylation at the Thr(668) residue of APP (with respect to the numbering conversion for the APP 695 isoform) and the Thr(736) residue of APLP2 (with respect to the numbering conversion for the APLP2 763 isoform) in their cytoplasmic domains acts as a molecular switch for their protein-protein interaction and is implicated in neural function(s) and/or Alzheimer's disease pathogenesis. Here we demonstrate that both APP and APLP2 can be phosphorylated by JNK at the Thr(668) and Thr(736) residues, respectively, in response to cellular stress. X11-like (X11L, also referred to as X11beta and Mint2), which is a member of the mammalian LIN-10 protein family and a possible regulator of Abeta production, elevated APP and APLP2 phosphorylation probably by facilitating JNK-mediated phosphorylation, whereas other members of the family, X11 and X11L2, did not. These observations revealed an involvement of X11L in the phosphorylation of APP family proteins in cellular stress and suggest that X11L protein may be important in the physiology of APP family proteins as well as in the regulation of Abeta production.

␤-Amyloid precursor protein (APP) is the precursor of ␤-amyloid (A␤), which is implicated in Alzheimer's disease pathogenesis. APP complements amyloid precursor-like protein 2 (APLP2), and together they play essential physiological roles. Phosphorylation at the Thr 668 residue of APP (with respect to the numbering conversion for the APP 695 isoform) and the Thr 736 residue of APLP2 (with respect to the numbering conversion for the APLP2 763 isoform) in their cytoplasmic domains acts as a molecular switch for their protein-protein interaction and is implicated in neural function(s) and/or Alzheimer's disease pathogenesis. Here we demonstrate that both APP and APLP2 can be phosphorylated by JNK at the Thr 668 and Thr 736 residues, respectively, in response to cellular stress. X11-like (X11L, also referred to as X11␤ and Mint2), which is a member of the mammalian LIN-10 protein family and a possible regulator of A␤ production, elevated APP and APLP2 phosphorylation probably by facilitating JNK-mediated phosphorylation, whereas other members of the family, X11 and X11L2, did not. These observations revealed an involvement of X11L in the phosphorylation of APP family proteins in cellular stress and suggest that X11L protein may be important in the physiology of APP family proteins as well as in the regulation of A␤ production.
␤-Amyloid precursor protein (APP) 1 is a ubiquitously expressed transmembrane protein with a receptor-like structure and is the precursor of ␤-amyloid (A␤) (1). A␤ is a principal component of senile plaques, a characteristic pathological feature in the Alzheimer's disease (AD) brain. It is widely accepted that production, aggregation, and deposition of A␤ are closely related to AD pathogenesis; however, the detailed molecular mechanisms of this pathogenic process remain to be fully elucidated (2). APP is a member of an evolutionally conserved gene family that in mammals includes amyloid precursor-like protein 1 (APLP1) and 2 (APLP2) (3,4). APP and its family members are physiologically important and functionally redundant because single disruption of each gene causes minor abnormalities, but combined disruption of APP and APLP2 or APLP1 and APLP2 causes lethality in early postnatal mice (5).
The cytoplasmic domain of APP (APPcyt) is composed of 47 amino acid residues, and its phosphorylation at Thr 668 (with respect to the numbering conversion for the APP 695 isoform) is suggested to play an important role in controlling the metabolism and physiological functioning of APP (6 -9). For example, ectopic expression of APP harboring a mutation at Thr 668 affects neurite outgrowth during neuronal differentiation in the PC12 cell line (8). Thr 668 is located in the motif 667 VTPEER 672 , which forms a type I ␤-turn and amino-terminal helix-capping box structure to stabilize its carboxyl-terminal helix structure (10). The phosphorylation of Thr 668 induces significant conformational change in APPcyt and affects its interaction with FE65 (10,11), which is an adaptor protein abundant in neurons that is implicated in APP metabolism, cellular motility, and APP-associated gene transactivation (11)(12)(13). APP-associated gene transactivation by FE65 is also affected by amino acid substitutions at Thr 668 . 2 The Thr 668 residue of APP can be phosphorylated by Cdc2 kinase and cyclindependent protein kinase 5 (Cdk5) in vivo and by glycogen synthase kinase 3␤ in vitro (6,7,14). We and others have also implicated c-Jun NH 2 -terminal kinase (JNK) in the phosphorylation of APP (15)(16)(17); JNK is a major signaling kinase in the cellular stress response and is known to be activated in neurons of AD patients (18 -20).
The cytoplasmic domains of APP family proteins are highly homologous, and the phosphorylation site corresponding to Thr 668 of APP is conserved in APLP2 as Thr 736 (with respect to the numbering conversion for the APLP2 763 isoform) (4,37). Indeed, similar to Thr 668 of APP, Thr 736 of APLP2 can be phosphorylated by Cdc2 kinase (37). Moreover, also similar to APP, phosphorylation of APLP2 is suggested to act as a molecular switch for binding to cytosolic proteins such as FE65 (11). However, whether the molecular mechanisms of APLP2 phosphorylation and its regulation are identical to that of APP has not been fully elucidated.
In this study, therefore, we investigated the phosphorylation of APP and APLP2 by JNK and the regulation of these phosphorylation events by their binding proteins. APP and APLP2 were phosphorylated by JNK1, JNK2, and JNK3 in vitro and in cells in response to cellular stress. Furthermore, among several known binding proteins, X11L, but not X11 or X11L2, was found to markedly enhance phosphorylation of APP and APLP2. These findings further elucidate the molecular mechanisms of phosphorylation and physiological function(s) of APP and APLP2 as well as differential functions among X11 family proteins.
Expression and Analysis of Proteins in Cultured Cell Lines-Human embryonic kidney 293 (HEK293) cells and HEK293 cells stably expressing human APP 695 were cultured as described previously (22,36). For transient protein expression, HEK293 cells were transfected with plasmid expression vectors using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's protocol. The cells were treated as indicated, collected, lysed in radioimmunoprecipitation buffer (50 mM Tris-HCl, 0.1% (w/v) SDS, 0.5% (w/v) sodium deoxycholate, 1% (v/v) Nonidet P-40, 0.15 M NaCl) containing 5 g/ml chymostatin, 5 g/ml leupeptin, 5 g/ml pepstatin A, and 1 M microcystin-LR, and centrifuged at 12,000 ϫ g for 10 min at 4°C. Protein kinase inhibitors SP600125 and SB203580 (BIOMOL Research Laboratories Inc.) were added 30 min prior to hyperosmotic treatment of cells. The resulting supernatants were subjected to immunoprecipitation with G369, UT-424, or M2 antibody as described previously (36). Cell lysates or immunoprecipitated proteins were analyzed by immunoblotting with indicated antibodies using an ECL detection kit (Amersham Biosciences). For quantification, immunoblot analysis was also performed using 125 I-protein A (Amersham Biosciences) with radioactivity being quantified using a Fuji BAS 1800.
Preparation of Proteins-Production and purification of recombinant GST fusion proteins were as described previously (16,22). Briefly GST fusion proteins were generated in Escherichia coli BL21 containing pGEX-4T-1 cDNA constructs and purified with glutathione-Sepharose 4B (Amersham Biosciences). To prepare FLAG-tagged X11 and X11L proteins, HEK293 cells transiently expressing FLAG-X11 and FLAG-X11L were lysed and subjected to immunoprecipitation with anti-FLAG M2 affinity gel (Sigma). These proteins were eluted with elution buffer (100 g/ml FLAG peptide (Sigma), 50 mM HEPES (pH 7.4), 100 mM NaCl). Proteins were concentrated with Microcon (Millipore) as required, and protein purity was examined by staining of gels with Coomassie Brilliant Blue R-250 following SDS-PAGE.

Phosphorylation of APP and APLP2 by JNK Family
Proteins-We initially examined the ability of JNK to phosphorylate APP at Thr 668 and APLP2 at Thr 736 (their phosphorylation sites are schematically represented in Fig. 1A). JNK proteins are expressed from three genes, Jnk1, Jnk2, and Jnk3, as low and high molecular weight splice variants (18). We performed in vitro protein phosphorylation of purified APPcyt protein ( Fig. 1B) or APLP2cyt protein ( Fig. 1C) with activated JNK isoforms including JNK1 (␣1 and ␣2), JNK2 (␣1 and ␣2), and JNK3. Phosphorylation was examined by immunoblotting with antibodies that specifically recognize Thr 668 -phosphorylated APP or Thr 736 -phosphorylated APLP2. Phosphorylation of APP at Thr 668 was induced by all of these JNK isoforms (Fig. 1B,  upper panel). In addition, phosphorylation of APLP2 at Thr 736 was also mediated by all of the JNK kinases that we examined (Fig. 1C, upper panel). These results indicate that all JNK family proteins are capable of phosphorylating both APP at Thr 668 and APLP2 at Thr 736 .
Stress-induced Phosphorylation of APP and APLP2 by JNK-JNK is a stress-activated protein kinase that is critical for signaling cell death and/or cell survival in response to various cellular stresses (18). We therefore investigated whether phosphorylation of APP could be induced in response to stress ( Fig. 2A). HEK293 cells stably expressing APP were treated with various stimuli including a protein synthesis inhibitor (anisomycin), radiation (UV), and hyperosmolarity (high concentration of sorbitol). In cells that had been subjected to these stressful stimuli, activation of endogenous JNK proteins was detected as evidenced by the appearance of dually phosphorylated forms of JNK in immunoblot ( Fig. 2A, lower   panel). Phosphorylation of APP was detected in response to each of the stressful stimuli and was most pronounced in cells subjected to hyperosmolarity but was hardly observed in nontreated cells ( Fig. 2A, upper panel). This indicated that phos- phorylation of APP at Thr 668 was induced by these stress stimuli. Such stress-induced phosphorylation of APP at Thr 668 was also observed in other cell lines including mouse neuroblastoma neuro-2a cells (data not shown).
To determine whether JNK could be involved in stress-induced phosphorylation of APP, we transiently overexpressed JNK isoforms and APP. Phosphorylation of APP was induced by hyperosmotic stress in cells transiently expressing APP (Fig. 2B), as it was in cells stably expressing APP, except that phosphorylation of the mature form of APP (Fig. 2B, second  panel, upper band) was barely detectable. Overexpression of JNK strongly enhanced the hyperosmotic stress-induced phosphorylation of APP (Fig. 2B, first panel). In addition, a small degree of APP phosphorylation was observed in the absence of stressful stimuli, and this appeared to correlate with spontaneous activations of overexpressed JNK in transfected cells (Fig. 2B, first and third panels). This correlation between the amount of phosphorylated APP and that of phosphorylated JNK suggests that JNK may be involved in the phosphorylation of APP.
Therefore, we next examined the involvement of endogenous JNK in stress-induced APP phosphorylation using selective kinase inhibitors in HEK293 cells stably expressing APP. An appropriate concentration of SP600125, a recently described selective chemical inhibitor of JNK (40), significantly inhibited APP phosphorylation induced by hyperosmolarity (Fig. 2C). On the contrary, SB203580, a specific inhibitor of another major stress-responsive kinase, p38, did not affect APP phosphorylation. In addition, we observed that co-expression of JNK-binding domain (JBD), which is a short polypeptide within JIP-1 (amino acids 127-281) and inhibits the activity of JNK (41), significantly inhibited APP phosphorylation induced by hyperosmolarity (Fig. 2D, left panel). Phosphorylation of APLP2 at Thr 736 was also examined in cells. Phosphorylation of APLP2 at Thr 736 was observed in sorbitol-treated cells as evidenced by immunoblotting with anti-phospho-APLP2 but was barely detectable in non-treated cells. When JBD was co-expressed, phosphorylation was apparently reduced (Fig. 2D, right panel). Taken together, these observations suggest that APP and APLP2 are phosphorylated at Thr 668 and Thr 736 , respectively, by JNK in response to cellular stress.

X11L Regulates the Phosphorylation of APP Family by JNK
tion was examined following treatment or non-treatment of the cells with sorbitol (Fig. 3A). Phosphorylation of APP at Thr 668 was observed in all treated cells. Most of the APP-binding proteins did not significantly affect the level of phosphorylation; however, in the presence of X11L, the level of phosphorylated APP was greatly increased. Consistent with the previous observation that the expression of X11L stabilizes intracellular APP metabolism (30), we also observed an increase in APP in cells expressing X11L (Fig. 3A, left middle  panel). However, the increase of the phosphorylated form of APP was too large to be accounted for by accumulation of intracellular APP. It was found that the ratio of phosphorylated to total APP was increased by ϳ3-fold by co-expression of X11L when an equal amount of APP was analyzed (Fig. 3A, right panels). In the cells co-expressing exogenous JNK, APP phosphorylation was also elevated corresponding to an increase of phosphorylated JNK. On the other hand, no significant increase of phosphorylated JNK was observed in the cells expressing X11L (Fig. 3A, left lower panel). Overexpression of X11L facilitates the activation of neither endogenous nor exogenous JNK induced by hyperosmolarity (Supplemental Fig. 1). Interestingly we also detected phosphorylated APP in cells expressing X11L that were not exposed to stress, although it was much smaller than in cells exposed to hyperosmotic stress (Fig. 3A). This phosphorylation of APP may reflect basal cellular JNK activity because it was suppressed by inhibiting JNK with JBD protein (data not shown). These results suggest that X11L expression could increase both the amount and the ratio of phosphorylated APP in cells without further activation of JNK.
Several researchers have reported that all three mammalian LIN-10 family proteins, X11, X11L, and X11L2, commonly interact with the cytoplasmic domain of APP and modulate APP metabolism (21)(22)(23)(27)(28)(29)(30). Thus, we examined whether X11 and X11L2 could elevate the phosphorylation level of APP in a manner similar to X11L (Fig. 3B). Similar to X11L, co-expression of either X11 or X11L2 with APP in HEK293 cells increased cellular APP, suggesting that they may play a role in stabilizing cellular APP (27,28,30). When APP phosphorylation was induced by high osmotic stress, the level of phosphorylated APP was increased slightly in cells expressing X11 or X11L2 compared with cells expressing APP alone, probably as a consequence of increasing total cellular APP. However, these increases in the cells expressing X11 and X11L2 were much lower than that observed in cells expressing X11L, and the ratio of phosphorylated APP to cellular APP was not significantly elevated. We also investigated the effect of X11L and its family proteins on APP phosphorylation induced by DLK overexpression (16) as this protein activates JNK kinase in cells. In cells not expressing any X11 family proteins, DLK-induced phosphorylation was weak, and this was slightly enhanced by expression of X11 or X11L2, probably reflecting increased cellular APP. On the contrary, in cells expressing X11L a high level of phosphorylated APP was detected. Taken together, these results suggest that the ability to elevate APP phosphorylation is a unique function of X11L among the X11 family and is probably independent of the effects of this family on cellular APP accumulation.
Unlike APP, the interaction of APLP2 with intracellular proteins has not been well characterized. However, similar to APP, APLP2 is known to bind X11L in its cytoplasmic domain (22). Therefore, it is likely that the phosphorylation of APLP2 at Thr 736 is elevated in the presence of X11L. Thus, APLP2 was co-expressed with JNK1 or X11L, and the level of phosphorylated APLP2 was examined after hyperosmotic stress and DLK overexpression (Fig. 3C). Both hyperosmotic stress and DLK overexpression induced the phosphorylation of APLP2 at Thr 736 , and co-expression of JNK1, which could act as the responsible kinase, increased the amount of phosphorylated APLP2. Co-expression of X11L apparently increased the level of phosphorylated APLP2 as did co-expression of JNK1. These results suggest that, similar to APP, phosphorylation of APLP2 at Thr 736 is likely to be enhanced in the presence of X11L.
Interaction of X11L with APP Is Required to Elevate the Phosphorylation Level of APP-To address the detailed mechanism of X11L on APP phosphorylation, we examined whether a direct interaction between X11L and APP was necessary to increase phosphorylated APP. The interaction involves the PI domain of X11L and the 681 GYENPTY 687 motif of APP. APP lacking amino acid residues 681-690 (APP⌬681-690) failed to bind X11L in yeast two-hybrid binding analysis (22). Therefore, we first examined the binding abilities of APP and APP⌬681-690 to X11L in cells by co-immunoprecipitation analysis (Fig.  4A). X11L was co-precipitated with APP but not with APP⌬681-690, indicating that APP⌬681-690 is indeed unable to interact with X11L in cells. Then we examined the effect of X11L on phosphorylation of APP⌬681-690 by JNK (Fig. 4B). In the absence of X11L, APP⌬681-690 was phosphorylated as efficiently as wild-type APP after high osmotic stimulation. This indicates that the deletion does not by itself affect the phosphorylation at Thr 668 by JNK. However, when X11L was co-expressed, no elevation of phosphorylated APP⌬681-690 could be observed, while the level of phosphorylated wild-type APP was markedly increased. Similarly X11L failed to increase the phosphorylated form of APP⌬681-690 when its phosphorylation was induced by overexpression of DLK. This suggests that the interaction of X11L with APP is necessary for its increase in APP phosphorylation.
Next we further characterized the effect of X11L using several truncated X11L proteins. X11L is composed of a unique amino-terminal region, a central PI domain, and a carboxylterminal region containing two PDZ domains. We utilized truncated X11L including NϩPI (containing the amino-terminal region and attached PI domain), PIϩC (containing the carboxyl-terminal half of X11L including PI and PDZ domains), and N (the amino-terminal region alone) (Fig. 5A). APP lacking a part of the extracellular region (APP⌬124 -303), which behaves in the same manner as full-length APP in cells, 3 was also utilized instead of full-length APP for precise determination of APP phosphorylation since phosphorylated full-length APP could overlap with NϩPI, which was nonspecifically recognized with anti-phospho-APP antibody in immunoblots. APP phosphorylation induced by hyperosmotic stress was examined in cells co-expressing full-length X11L or these truncated proteins, and the ratio of the phosphorylated form of APP to total APP was compared. Expression of NϩPI markedly elevated the ratio of APP phosphorylation as did full-length X11L, whereas expression of PIϩC and N was without significant effect (Fig. 5B). When we induced APP phosphorylation by DLK overexpression instead of high osmotic stress, we obtained similar results. The ratio of APP phosphorylation was increased by co-expression of full-length X11L and NϩPI by ϳ8-fold but was not influenced by PIϩC and N (Fig. 5C). Expression of the PI domain alone failed to elevate phosphorylated APP, although it is not possible to conclude that this domain alone does not influence phosphorylation because its cellular expression level was much lower than the other domains (data not shown). These results suggest that the interaction of X11L with APP via its PI domain is responsible, and the amino-terminal region of X11L is also important for its effect on APP phosphorylation. The ability of X11L to stabilize cellular APP is likely insufficient to elevate APP phosphorylation, and the effects of X11L on the stability and phosphorylation of APP might be constituted of independent mechanisms since PIϩC could stabilize APP but could not enhance the phosphorylation. This conclusion seems to be consistent with the observations that X11 and X11L2, which resemble X11L in their PI and carboxyl-terminal regions and stabilize cellular APP (21-23), did not elevate APP phosphorylation level as X11L did (Fig. 3B).
X11L Facilitates the Phosphorylation of APP by JNK in Vitro-Association of X11L with APP was required to elevate the phosphorylation level (Fig. 4). However, expression of X11L did not affect JNK activation (Fig. 3 and Supplemental Fig. 1). In addition, insofar as we examined, there was no interaction of X11L with JNK protein, whereas another APP-binding protein, JIP-1, can bind JNK (Supplemental Fig. 2). Therefore, a more likely mechanism of increased APP phosphorylation by X11L is that X11L renders APP more susceptible to phosphorylation by JNK through its direct interaction rather than facilitating the activation of JNK or scaffolds between JNK and APP as is observed for JIP-1 (16,17,33). Thus, we tested whether X11L could enhance the phosphorylation of APP by JNK in vitro using purified proteins. APPcyt protein was subjected to phosphorylation with activated JNK in the presence or absence of either purified X11L or X11 protein, and the phosphorylation at Thr 668 was examined (Fig. 6A, left panels). APP was similarly phosphorylated in the absence and presence of X11. However, in the presence of X11L, the level of APP phosphorylation was higher than in the absence of X11L. This is consistent with the observation in cultured cells that X11L, but not X11, increased APP phosphorylation (Fig. 3B). On the other hand, X11L did not affect JNK-mediated phosphorylation of GST-c-Jun (Fig.  6A, right panels), which is a widely used substrate to analyze catalytic activity of JNK (42), suggesting that X11L selectively facilitates the phosphorylation of APP rather than directly elevates catalytic activity of JNK. We also examined the effect of truncated X11L proteins purified as GST fusion proteins on APP phosphorylation (Fig. 6B). Addition of GST-X11L (full length) and GST-NϩPI to the reaction increased APP phosphorylation, while GST protein (Fig. 6B, ctrl) and GST-N were 3 K. Iijima and T. Suzuki, unpublished observation. without effect. The effect of NϩPI on APPcyt phosphorylation was more marked than that with the full-length protein, probably reflecting the higher affinity of NϩPI for APP in vitro as we have reported previously (22). These results suggest that X11L can directly facilitate the phosphorylation at Thr 668 by JNK through its association with APPcyt, whereas other X11 family proteins cannot. DISCUSSION Phosphorylation of APP at Thr 668 is proposed to act as a conformational switch for protein-protein interactions and is involved in the regulation of its metabolism and putative function(s) (6 -11). In this study, we demonstrated that phosphorylation of APP and APLP2 is induced by JNK family proteins in response to cellular stress and is regulated by one of their binding proteins, X11L.
JNK is implicated in the phosphorylation of APP at Thr 668 by our group and others (15,16). Here we found that JNK-mediated phosphorylation of APP at Thr 668 was induced in response to cellular stress (Fig. 2). Well known targets of activated JNK include transcription factors such as c-Jun and ATF2; however, it is also known that activated JNK localizes to the cytoplasm and contributes to the phosphorylation of several cytosolic proteins (18). APP is a possible cytosolic substrate of JNK. During the preparation of this report, Inomata et al. (17) and Scheinfeld et al. (43) also reported APP phosphorylation to be induced by stress. However, our analysis extended these findings, particularly with regard to answering two previously unresolved questions. First, what kind of JNK protein kinases can phosphorylate APP? JNK protein kinases are encoded by three genes, Jnk1, Jnk2, and Jnk3, which are highly homologous and are involved in signaling apoptosis, stress responses, and proliferation (18). However, physiological differences among the three JNK protein kinases have also been reported. JNK1 and JNK2 are expressed ubiquitously, whereas JNK3 is expressed most abundantly in brain (18). JNK1 and JNK2 also exhibit differences in substrate specificities, in signal-specific induction of apoptosis in fibroblast, and in response to withdrawal of tropic support in cerebellar neurons (18,44,45). In the case of APP, we found that JNK1, JNK2, and JNK3 can all phosphorylate the Thr 668 residue to a comparable extent in vitro (Fig.  1B) and that JNK1 and JNK2 isoforms can commensurately contribute to stress-induced APP phosphorylation in cells (Fig.  2B). The ability of all JNK protein kinases to phosphorylate APP is consistent with the observation that APP phosphorylation was induced not only in neuronal cells but also in nonneuronal cells. The second question was whether another APP family protein, APLP2, was phosphorylated by JNK in response to cellular stress. We demonstrated that APLP2 was phosphorylated at Thr 736 , which corresponds to Thr 668 of APP, by JNK1, JNK2, and JNK3 in vitro (Fig. 1C) and that this phosphorylation event was induced by cellular stress (Fig. 2D). Therefore, we conclude that APP and APLP2 can be phosphorylated by JNK protein kinases JNK1, JNK2, and JNK3 in a similar manner.
Structural analyses using NMR and CD spectroscopy indicate that phosphorylation causes prolyl cis/trans isomerization of the Thr 668 -Pro 669 peptide bond of the VTPEER motif and alters the overall structure of the cytoplasmic domain of APP (10,11). The conformational changes induced by phosphorylation affect the interaction of APP with binding proteins such as FE65 (11). The VTPEER motif of APP is highly conserved in APLP2 as LTPEER, and a mutation at Thr 736 of APLP2 affects its interaction with FE65 protein as does mutation of Thr 668 of APP (5,11). Thus, phosphorylation of APP and APLP2 probably acts as a conformational switch for the complementary physiological functions of these proteins.
Recently, Lee et al. (46) reported that Thr 668 -phosphorylated APP, especially phosphorylated carboxyl-terminal fragments, was elevated and accumulated in human AD brain and raised the possibility that the phosphorylation may increase A␤ generation. In the pathogenesis of AD, activated JNK is significantly increased and is localized in the cytoplasm of neurons in a manner that correlates with the progression of the disease (19,20). Activation of JNK is also observed in a mouse model of AD (47). Therefore, it is possible that APP and APLP2 could serve as substrates of cytoplasmic JNK in AD and that this results in neural dysfunction. Detailed immunohistochemical analysis with a highly specific antibody aimed at examining whether abnormal phosphorylation of APP and APLP2 correlates with JNK activation may contribute to elucidating the neuronal degeneration process of AD. However, the antibodies we used here are not suitable for immunohistochemical analysis on fixed brain sections.
The influence of interactions between APP and cytosolic proteins on the phosphorylation at Thr 668 has not been well understood. Among major APP-binding proteins, we found that only X11L significantly increases the phosphorylation of APP at Thr 668 by JNK. In our study, much higher levels of phosphorylated APP were observed in cells expressing X11L with APP than in the cells expressing APP alone or APP with other APP-binding proteins (Fig. 3). In the in vitro phosphorylation assay in which dephosphorylation of APP was inhibited, X11L also increased the level of phosphorylated APP (Fig. 6). Therefore, increased phosphorylation in the presence of X11L probably reflects an enhancement of phosphorylation by X11L rather than an inhibition of dephosphorylation or of phosphorylated APP degradation, although we could not thoroughly rule out these latter possibilities.
Previously we have found that expression of JIP-1 protein increases APP phosphorylation induced by activating the JNK pathway by DLK overexpression (16). JIP-1 is known as a scaffold protein within the JNK signaling cascade that interacts with components of the JNK cascade to organize and facilitate signal transduction (32). Recent reports suggest that JIP-1 simultaneously interacts with both APP and JNK to enhance the phosphorylation of APP by JNK in vitro (17,43). In our analysis, however, JIP-1 had no significant effect on the level of phosphorylated APP in cells during a stress response (Fig. 3A). Taken together, these results suggest that JIP-1 possibly controls APP phosphorylation by regulating JNK activation and by scaffolding APP with activated JNK to form a substrate-enzyme complex (Fig. 7, middle) and, therefore, in certain circumstances may elevate APP phosphorylation. On the contrary, X11L influenced APP phosphorylation much more profoundly than JIP-1. Binding of X11L to APP is necessary for it to influence APP phosphorylation because mutated forms of X11L or APP that are unable to bind each other could not enhance or be enhanced with respect to phosphorylation both in cells and in vitro (Figs. 4 and 6). In contrast to JIP-1, X11L did not bind JNK or facilitate the activation of JNK at least insofar as we examined (Fig. 3A and Supplemental Figs. 1 and 2). In addition, X11L was suggested to selectively elevate APP phosphorylation rather than generally enhance JNK-mediated phosphorylation (Fig. 6A). Thus, it is conceivable that X11L binds APP to facilitate its phosphorylation by JNK without scaffolding them (Fig. 7, right). Although Cdk5 also executes APP phosphorylation (7,46), Cdk5-mediated APP phosphorylation was not enhanced in the presence of X11L as JNKmediated APP phosphorylation was at least in our in vitro analysis (Supplemental Fig. 3). Therefore, X11L possibly acts as a specific regulator of JNK-mediated APP phosphorylation.
X11L plays a role in stabilizing cellular APP and suppressing A␤ generation (22,29,30); however, the effect of X11L on APP phosphorylation may be independent from these previously proposed roles because the enhancement also occurs in vitro (Fig. 6), and PIϩC, a region sufficient for stabilization (29), cannot enhance phosphorylation (Fig. 5). Moreover the other two mammalian X11 family members, X11 and X11L2, which also stabilize cellular APP and suppress A␤ generation (27)(28)(29)(30), had no significant effects on APP phosphorylation (Fig. 3B), suggesting that X11 family proteins may have differential roles in APP metabolism. X11L is possibly involved in the metabolism and functions of APP in multiple ways: not only by stabilizing cellular APP and suppressing A␤ generation directly but also by regulating the interactions of APP with other proteins by enhancing APP phosphorylation. The observations that the interaction of X11L with APP can be positively and negatively regulated by Alcadein and XB51 protein, respectively, suggest the presence of a cellular regulatory system for APP that involves X11L (30,48,49).
X11L interacts with both APP and APLP2 (22), whereas JIP-1 prefers APP over APLP2 as a binding partner (50). In the presence of X11L, phosphorylation of APLP2 at Thr 736 was enhanced as efficiently as that of APP (Fig. 3C). Therefore, X11L is a common regulator for APP family proteins, which may be important not only in AD but also in physiological conditions.