Amyloid (cid:1) Protein Precursor Is Phosphorylated by JNK-1 Independent of, yet Facilitated by, JNK-Interacting Protein (JIP)-1*

Alzheimer’s disease (AD) is genetically linked to the processing of amyloid (cid:1) protein precursor (A (cid:1) PP). Aside from being the precursor of the amyloid (cid:1) (A (cid:1) ) found in plaques in the brains of patients with AD, little is known regarding the functional role of A (cid:1) PP. We have recently reported biochemical evidence linking A (cid:1) PP to the JNK signaling cascade by finding that JNK-interacting pro-tein-1 (JIP-1) binds A (cid:1) PP. In order to study the functional implications of this interaction we assayed the carboxyl-terminal of A (cid:1) PP for phosphorylation. We found that the threonine 668 within the A (cid:1) PP intracellular domain (AID or elsewhere AICD) is indeed phosphorylated by JNK1. We surprisingly found that although JIP-1 can facilitate this phosphorylation, it is not required for this process. We also found that JIP-1 only facilitated phosphorylation of A (cid:1) PP but not of the two other family members APLP1 (amyloid precursor-like protein 1) and APLP2. Understanding the connection between A (cid:1) PP phosphorylation and the JNK signaling pathway, which mediates cell response to stress may have important implications in understanding the pathogenesis of Alzheimer’s disease.

Alzheimer's disease (AD) is genetically linked to the processing of amyloid ␤ protein precursor (A␤PP). Aside from being the precursor of the amyloid ␤ (A␤) found in plaques in the brains of patients with AD, little is known regarding the functional role of A␤PP. We have recently reported biochemical evidence linking A␤PP to the JNK signaling cascade by finding that JNK-interacting protein-1 (JIP-1) binds A␤PP. In order to study the functional implications of this interaction we assayed the carboxyl-terminal of A␤PP for phosphorylation. We found that the threonine 668 within the A␤PP intracellular domain (AID or elsewhere AICD) is indeed phosphorylated by JNK1. We surprisingly found that although JIP-1 can facilitate this phosphorylation, it is not required for this process. We also found that JIP-1 only facilitated phosphorylation of A␤PP but not of the two other family members APLP1 (amyloid precursorlike protein 1) and APLP2. Understanding the connection between A␤PP phosphorylation and the JNK signaling pathway, which mediates cell response to stress may have important implications in understanding the pathogenesis of Alzheimer's disease.
Alzheimer's disease (AD) 1 is the most common neurodegenerative disease constituting approximately two thirds of all cases of dementia (1). AD is genetically linked to a few molecules, one being A␤PP. A␤PP is a type I transmembrane protein, which undergoes processing by the secretases to produce various fragments. Following processing by the ␤and ␥-secretases, the A␤ fragment (from the ␤ to ␥ sites) and AID (from the ␥ site to the carboxyl-terminal) are produced (2,3). Recently, another ␥-secretase-dependent cleavage has been described to occur at the "epsilon" site, which lies within AID (4 -7). This would cause shorter AID fragments of either 49 or 50 amino acids. The pathologic cascade, which leads to clinical manifestations of AD, has not been identified fully; however, the "Amyloid Hypothesis" has been used to explain certain aspects of AD pathology. According to this hypothesis, the accumulation of A␤ is the primary event that leads to all subsequent events in the pathology of AD (8).
However, considering that production of both A␤ and AID are dependent on the ␥-secretase, we and others have attempted to understand the cellular effects of AID production. Indeed it has been found that AID is able to trigger apoptosis or lower the cell's threshold to other apoptotic stimuli (9). Furthermore, many proteins have been found to interact with AID such as X11, Fe65, mDab, Shc, Numb, and Numb-like (10 -14). Use of "guilty by association" has been used to speculate possible roles for AID in the cell. One interacting protein, which has attracted much interest recently, is JNK-interacting protein (JIP)-1 (15). JIP-1 is a cytoplasmic protein that binds members of the JNK signaling cascade and scaffolds these proteins to allow for efficient activation of the JNK pathway (16). We and others (17)(18)(19) have shown that APP binds JIP-1.
This biochemical connection between A␤PP (and therefore AD), and JIP-1 (and therefore the JNK signaling cascade) provided an intriguing connection that could be important in understanding the pathology of AD. Pathological hallmarks of AD include amyloid plaques, neurofibrillary tangles, and neurodegeneration, with the last seeming to be most directly related to the clinical manifestations of AD (20). The JNK signaling cascade, which is responsible for responding to cell stress and mediating apoptosis (or survival) in different contexts, would seem likely to be involved in the neurodegeneration seen in AD (21). In fact, there have already been studies implicating the JNK signaling pathway in various aspects of AD pathology including both amyloid plaques and neurofibrillary tangles (22)(23)(24)(25)(26). While investigating the functional manifestations of this A␤PP-JIP-1 interaction we have found that AID in combination with JIP-1 is able to cause transcriptional activation of a reporter gene (27).
In this study we present data supporting the possibility that there are two pathways for JNK1-dependent phosphorylation of threonine 668 on the cytoplasmic tail of A␤PP. In one pathway, JIP-1 is able to scaffold the components necessary for JNK1 activation and A␤PP phosphorylation, while in the second A␤PP is phosphorylated independently of JIP-1. We further show that in the JIP-1-dependent pathway JNK1 is not capable of phosphorylating the cytoplasmic tails of APLP1 and APLP2, indicating that the functional role of JIP-1 is asymmetric with respect to action on A␤PP and APLPs.

MATERIALS AND METHODS
DNA Constructs-The full-length JIP-1 construct used was the human JIP-1e described previously and is referred to as JIP-1 throughout this study. JIP-1␦JBD is also described elsewhere. Expression constructs for MLK3, JNK1, and MKK7 were obtained from Dr. C. W. Cell Culture-HEK293T cells were grown in RPMI 1640 media with 10% heat-inactivated fetal calf serum. HEK293 cells overexpressing A␤PP were grown in the same medium supplemented with 5 g/ml of puromycin to maintain stable expression.
Immunoprecipitations, JNK Activation, Kinase Assays, and Western Blotting-Immunoprecipitations were performed as described elsewhere with the addition of the phosphatase inhibitors sodium fluoride (50 mM) and sodium vanadate (0.2 mM) to all buffers. To immunoprecipitate A␤PP, antibody P2-1 (BIOSOURCE) was used. HEK293 cells were treated with 60 mJ/cm 2 of UVB using a Bio-Rad cross-linker to activate JNK. SP600125 (Biomol) was added at the indicated concentrations 10 min before UVB treatement. Kinase assays were performed using lysis buffer and kinase buffer from Cell Signaling Technology as follows. For experiments in Figs. 3, 4, and 5, cells were lysed by sonicating for 10 s. After spinning down, the supernatant was added to GST fusion protein and incubated overnight at 4°C. The beads were washed twice with lysis buffer and twice with kinase buffer followed by the addition of 50 l of kinase buffer supplemented with 100 M ATP. Following incubation at 30°C for 30 min, the reaction was halted by boiling in loading buffer. For experiments in Figs. 2 and 5b, proteins were purified by immunoprecipitation for 4.5 h at 4°C and washed twice with lysis buffer and twice with kinase buffer. Appropriate purified proteins were then added to GST-AID. The kinase assay was then performed as described above. Proteins were separated by SDS-PAGE and Western blotting was carried out with the following antibodies: A␤PP, (22C11, Chemicon or APP-C-terminal, Zymed Laboratories Inc.); FLAG, (Sigma); MLK3, JNK1, Phospho-JNK1, Jun, phospho-Jun, panphospho-threonine (Cell Signaling Technology), monoclonal A␤PP-Phospho-T668 (P. Davies). (17,18) have previously identified an interaction between A␤PP and JIP-1. Considering that JIP-1 is a scaffold protein which is able to bind components of the JNK signaling cascade and A␤PP contains a consensus sequence for JNK phosphorylation (Thr-Pro or TP) in the intracellular domain, we questioned whether A␤PP is phosphorylated by JNK1 in a JIP-1-dependent manner. HEK293 cells expressing A␤PP (but not JIP-1, data not shown) were transfected with JIP-1, JNK1, and MLK3 in order to activate JNK1 in the context of JIP-1 and compared with control-transfected cells. A␤PP was immunoprecipitated, and proteins were resolved with SDS-PAGE. Because previous reports have indicated that glycogen synthase kinase (GSK)-3␤, CDC2, and JNK3 (28 -30) are able to phosphorylate A␤PP at threonine 668, we also began our analysis at this location. Western blotting using a monoclonal antibody specific for A␤PP phosphorylation on threonine 668 (using A␤PP 695 numbering) showed (Fig. 1a) that A␤PP is phosphorylated in the presence of JIP- 1/JNK1/MLK3 but not in the control cells. In order to determine which components of the signaling cascade are required for this phosphorylation, we transfected the various combinations of JIP-1, JNK1, and MLK3. Surprisingly, we found (Fig.  1b) that the combination of JNK1 and MLK3 was sufficient to cause phosphorylation, in fact to an even greater extent than in combination with JIP-1. A similar experiment was carried out using the MAP kinase kinase MKK7 instead of MLK3 to activate JNK1. Once again we found that phosphorylation was stronger in the absence of JIP-1 (Fig. 1c). Although these data could be explained by stating that JIP-1 acts as a JNK inhibitor and therefore there is less phospho-JNK1 in the sample, Fig. 1, b and c indicates that there is equal if not more phospho-JNK1 in the JIP-1-supplemented samples. A more likely explanation would be that JIP-1 could play a direct role in facilitating A␤PP phosphorylation but due to JIP-1 overexpression there is quenching due to distribution of A␤PP and active kinases to different JIP-1 complexes. Nevertheless, these data indicate that active JNK1, even not in the context of JIP-1, is able to cause A␤PP phosphorylation.

JNK1 Activation Results in a JIP-1-independent A␤PP Phosphorylation in Vivo-We and others
To determine whether activation of endogenous JNK results in A␤PP phosphorylation, we treated HEK293 cells with UVB to activate JNK. Thirty minutes after UVB exposure JNK was activated (data not shown), and A␤PP was phosphorylated on threonine 668 (Fig. 1d). Moreover, the specific JNK inhibitor SP600125 blocked A␤PP phosphorylation on threonine 668 in a dose-response manner (Fig. 1d). Thus, A␤PP is a JNK substrate.
Active JNK1 Directly Phosphorylates AID in Vitro Independent of JIP-1-The above results suggest that active JNK1 can phosphorylate A␤PP at threonine 668 in vivo. However, it is also possible that A␤PP could be phosphorylated by other endogenous kinases activated by JNK1. To determine whether JNK1 is directly responsible for this phosphorylation, we used an in vitro system. Cells were transfected individually with JIP-1 or JNK1, with JNK1 either being activated with anisomycin or not. Cells were lysed and proteins were purified by immunoprecipitation (Fig. 2a). Different combinations of these proteins were incubated with GST-AID in the presence of ATP and phosphorylation was analyzed by Western blotting. Fig. 2b shows that active JNK1 is sufficient to phosphorylate AID, even in the absence of JIP-1. These data indicate that JNK1 can directly phosphorylate A␤PP.
A␤PP Is Phosphorylated by JNK1 in Vitro in a JIP-1-dependent Manner-Although the above data indicated that JIP-1 was not necessary for A␤PP phosphorylation, we speculated that JIP-1 might function under certain circumstances to bring JNK1 and A␤PP together to allow efficient phosphorylation. In order to test for this, we overexpressed combinations of JIP-1, JNK1, and MLK3 (Fig. 3a) and analyzed whether GST-AID could pull-down JIP-1 already loaded with active JNK1 that could phosphorylate AID (Fig. 3b). Fig. 3c shows using a panphospho-threonine antibody that JIP-1 is able to serve as the link between GST-AID and active JNK1 to facilitate AID phosphorylation in vitro. Using an antibody specific for A␤PP Thr-668 phosphorylation we determined that threonine 668 was being phosphorylated (Fig. 3d). In order to determine whether this was the only location being phosphorylated, we incubated the JIP-1, JNK1, and MLK3 combination along with GST-AID-T668A and compared it to GST-AID. Fig. 3d shows that although binding of the JIP-1/JNK1/MLK3 complex is not decreased by this mutation, no threonine phosphorylation is detected using a pan-phospho-threonine antibody or an antibody recognizing phosphorylation at Thr-668. Importantly, use of both a GST-AID-YENPTY motif mutant (GST-AID-⌬) and a GST-AID-Y682G point mutant, (both of which do not bind JIP-1,) was used to confirm that binding of JIP-1 to AID is required for in vitro association and phosphorylation in this system (Fig. 3, c and d).
FIG. 2. Active JNK phosphorylates AID in vitro. a, purified JIP-1, JNK1, and active JNK1 were purified from cells which were overexpressing these proteins. Western blotting was used to confirm the purification. b, combinations of JIP-1, JNK1, active JNK1or control material were added to GST-AID beads, and a kinase assay was performed to test for AID phosphorylation. Western blotting with an antibody specific for A␤PP phosphorylation at Thr-668 was used to determine that only active JNK1 and not the addition of JIP-1 is necessary for in vitro phosphorylation. PR was used to detect total GST-AID protein.
FIG . 3. JIP-1 facilitates AID phosphorylation by binding both AID and JNK-1. a, various combinations of JIP-1, JNK, and MLK3 were transfected into HEK 293 cells, and Western blotting was used to confirm expression. b, protein complexes were pulled down from the cell lysate with either GST-AID (AID) or with GST-AID lacking the YENPTY motif necessary for JIP-1 to A␤PP binding (⌬). c, kinase assays were performed with the material from b and only when JIP-1, JNK, and MLK3 were present in the total lysate (and pull-down) was AID phosphorylated. AID lacking the YENPTY motif (GST-AID⌬) was never phosphorylated. d, kinase assays were done as in c; however, AID Y682G or AID T668A were also used as substrates to test directly the requirement for JIP-1 binding and whether Thr-668 was the only phosphorylation site.
There are two forms of JIP-1, which have been identified in both mouse and rat (31), JIP1b contains a complete PTB domain that interacts with A␤PP while JIP1a is missing 47 amino acids at the beginning of the PTB domain and is therefore unable to bind A␤PP. In order to determine whether JIP1a and JIP1b may mediate A␤PP phosphorylation in vitro, we performed a kinase assay as above however this time using either JIP1a or JIP1b. Fig. 4a shows that only JIP1b is able to bind GST-AID and mediate phosphorylation of AID in vitro.
These data suggest that there may be two pools of JIP-1, one that is involved in A␤PP function and one that functions independently of A␤PP pathways. We next wanted to clarify the role of JNK1 along this JIP-1-mediated pathway. A JIP-1 mutant lacking the JNK binding domain (JIP-1␦JBD) was used to perform the kinase assay such that JNK1 would not be recruited to A␤PP. Fig. 4b shows that although both the wildtype JIP-1 and the JIP-1␦JBD mutant associate with AID, JNK1 (and therefore phospho-JNK1) is not recruited by JIP-1␦JBD and AID is not phosphorylated. These data indicate that JNK1 recruitment plays an essential role in the JIP-1-dependent A␤PP Thr-668 phosphorylation in vitro.
JIP-1 Facilitates Phosphorylation of A␤PP but Not the APLPs-A␤PP is part of a larger gene family that is comprised of A␤PP, APLP1, and APLP2; however, most research on the family has been focused on A␤PP because only it has been directly implicated in AD (2,33). Work using mice with either single or compound knockouts of the different family members indicate that although there are unique roles for each of the different A␤PP family members during development, there is also functional redundancy among them that allows for compensation between the family members (34,35). APLP1 and APLP2 are also able to bind JIP-1 although they bind JIP-1 more weakly than does A␤PP (27). Sequence alignments between the three A␤PP family members show that all three family members contain the YENPTY motif, which is essential for JIP-1 binding (Fig. 5a). Furthermore, both A␤PP and APLP2 contain the TP motif, which may be phosphorylated by JNK1, whereas APLP1 does not contain this motif (Fig. 5a).
We first wanted to determine whether active JNK can phosphorylate APLP1 and APLP2. Fig. 5b shows that as predicted from the absence or presence of the TP motif APLP1 cannot be phosphorylated by JNK while APLP2 can. This also would indicate that JIP-1 is not essential for APLP2 phosphorylation. We therefore wanted to determine whether JIP1 can facilitate phosphorylation of APLP2 similar to A␤PP. To do this, we transfected the JIP-1/JNK1/MLK3 complex into 293 cells and performed a kinase assay as before using either GST-AID, GST-ALID1, or GST-ALID2. Western blotting with a pan-phospho-threonine (Fig. 5c) antibody demonstrated that only AID associates with and is phosphorylated by the complex. APLP2, which binds JIP-1 weakly, is not phosphorylated. Attempts to reproduce this data in vivo were inconclusive due to poor antibody performance and cross-reactivity. FIG. 5. JIP-1 facilitates phosphorylation of APP but not the APLPs. a, alignment of A␤PP, APLP1, and APLP2. The conserved YENPTY motif responsible for A␤PP to JIP-1 binding, as well as the JNK consensus phosphorylation motif (TP) is indicated. Note that APLP1 does not contain the TP motif. b, inactive (JNK) or active JNK ( p JNK) was added to GST fusion proteins (AID, AID T668A, ALID1, ALID2) and incubated to allow for phosphorylation. Only AID and ALID2 were phosphorylated by active JNK. c, GST fusion proteins were used to pull-down JIP1 and other components that might be bound. Kinase assays were performed, and Western blotting was carried out. Both pan-phospho-threonine antibody and phospho-T668 antibody reveal that only AID is phosphorylated.

DISCUSSION
In this report we demonstrate that A␤PP can be phosphorylated on threonine 668 by JNK-1. A study published during preparation of this article has come to a similar conclusion (19). Our data differ importantly from that study in that we show that although JIP-1 may play a role in assembling the components needed for A␤PP phosphorylation by JNK1, JIP-1 is not necessary for this process. Phosphorylation at Thr-668 is particularly interesting considering the work of Ramelot and Nicholson (36) who studied phosphorylation at this site by NMR spectroscopy. They found that phosphorylation induces changes in backbone dihedral angles and shifts in the cis/trans equilibrium of the prolyl bond between threonine 668 and proline 669. This may increase or decrease the affinity of A␤PPbinding proteins as has been observed for Fe65 in one report (32), or affect other aspects of A␤PP biology such as trafficking and processing. Further work will be needed to identify the functional consequences of Thr-668 phosphorylation by JNK1 both in the normal and the AD neuron.
In this report we have found that active JNK1 alone is sufficient to phosphorylate A␤PP. For a kinase to phosphorylate a substrate, there must be contact between the two, made possible by either direct affinity between the kinase and the substrate or their both having affinity for a common protein or complex. Based on our data, the most likely conclusion is that the affinity between active JNK1 and A␤PP is great enough to allow phosphorylation (Fig. 6b). We have further found that JIP-1 is capable of scaffolding active JNK pathway components to A␤PP and facilitating A␤PP phosphorylation. This may indicate that although JIP-1 appears unnecessary, it may facilitate phosphorylation in the cellular context (Fig. 6a). JIP-1 can play an essential role of isolating, transporting, and restricting phospho-JNK and A␤PP to allow for a spatially controlled and efficient phosphorylation process. Still it is important to note two other possibilities. First, there may be another scaffold or adapter protein that remains bound to active JNK1 when it is purified from cell lysates and this unidentified scaffold tethers JNK1 and A␤PP together (Fig. 6c). A possible candidate for this is JIP-2, which like JIP-1 binds JNK1 and A␤PP. However, we did not find JIP-2 expressed in the cells that were used (data not shown). Second, JNK1 may bind and activate another kinase, remaining bound during protein purification, and it is this second kinase that is phosphorylating A␤PP (Fig. 6c).
Finally, We have also found that that in vitro only AID is phosphorylated in a JIP-1-dependent manner, while the ALIDs are not. We have also previously found that only AID and not the ALIDs are capable of combining with JIP-1 to form a transcriptionally active complex. AD is genetically linked to A␤PP but not the APLPs, presumably because only A␤PP yields A␤ fragment following processing by the secretases. Our data now suggest a new JIP-1-related difference between A␤PP and the APLPs. Phosphorylated AID fragments may mediate the toxicity found in AD while ALID fragments, which are not phosphorylated in a JIP-1-facilitated manner, cannot mediate these functions. Further work will be necessary to identify the functional consequences of JNK1-mediated phosphorylation at Thr-668 particularly if it is involved in the pathogenesis of Alzheimer's Disease.
FIG. 6. Models illustrating mechanisms for A␤PP phosphorylation. a, JIP-1 plays the role of facilitating JNK1 dependent phosphorylation of A␤PP by assembling active kinases and bringing them to A␤PP (or AID). b, JNK1 phosphorylates A␤PP independent of JIP-1 or any other scaffold. c, JNK1 phosphorylates A␤PP dependent upon an unidentified scaffold (Scaffold X), which is required to hold JNK1 and A␤PP together to allow phosphorylation. d, JNK1 activates another kinase (Kinase X), which in turn phosphorylates A␤PP.