QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: jpet.103.051383v1 306/3/889    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Hashimoto, Y. Articles by Nishimoto, I. Search for Related Content PubMed PubMed Citation Articles by Hashimoto, Y. Articles by Nishimoto, I.

CELLULAR AND MOLECULAR

The Cytoplasmic Domain of Alzheimer's Amyloid- Protein Precursor Causes Sustained Apoptosis Signal-Regulating Kinase 1/c-Jun NH₂-Terminal Kinase-Mediated Neurotoxic Signal via Dimerization

Departments of Pharmacology and Anatomy, KEIO University School of Medicine, Medical Research Center, Tokyo, Japan (Y.H., T.N., T.C., E.T., Y.Y., M.I., M.Y., M.N., K.T., S.A., I.N.); and Division of Molecular Signal Transduction Research, Department of Medical and Dental General Research, Tokyo Medical and Dental University (H.K., H.N., H.I.) Tokyo, Japan

Received March 7, 2003; accepted May 27, 2003.

	Abstract

Top Abstract Materials and Methods Results Discussion References

 
The biological function of full-length amyloid- protein precursor (APP), the precursor of A, is not fully understood. Multiple laboratories have reported that antibody binding to cell surface APP causes neuronal cell death. Here we examined whether induced dimerization of the cytoplasmic domain of APP (APP_CD) triggers neuronal cell death. In neurohybrid cells expressing fusion constructs of the epidermal growth factor (EGF) receptor with APP_CD (EGFR/APP hybrids), EGF drastically enhanced neuronal cell death in a manner sensitive to acetyl-L-aspartyl-L-glutamyl-L-valyl-L-aspartyl-aldehyde (Ac-DEVD-CHO; DEVD), GSH-ethyl ester (GEE), and pertussis toxin (PTX). Dominant-negative apoptosis signal-regulating kinase 1 (ASK1) blocked this neuronal cell death, but not -synuclein-induced cell death. Constitutively active ASK1 (caASK1) caused DEVD/GEE-sensitive cell death in a manner resistant to PTX and sensitive to Humanin, which also suppressed neuronal cell death by EGFR/APP hybrid. ASK1 formed a complex with APP_CD via JIP-1b, the c-Jun N-terminal kinase (JNK)-interacting protein. EGFR/APP hybrid-induced and caASK1-induced neuronal cell deaths were specifically blocked by SP600125 (anthra[1,9-cd]pyrazol-6(2H)-one), a specific JNK inhibitor. Combined with our earlier study, these data indicate that dimerization of APP_CD triggers ASK1/JNK-mediated neuronal cell death. We also noticed a potential role of ASK1/JNK in sustaining the activity of this mechanism after initial activation by APP, which allows for the achievement of cell death by short-term anti-APP antibody treatment. Understanding the function of APP_CD and its downstream pathway should lead to effective anti-Alzheimer's disease therapeutics.

Recently, Rohn et al. (2000) and Sudo et al. (2000) independently found that anti-APP antibodies kill primary neurons and immortalized neuronal cells overexpressing APP. Sudo et al. (2000) and Mbebi et al. (2002) found that anti-APP antibodies cause pertussis toxin (PTX)-sensitive neurotoxicity by acting on the cell surface APP. Most recently, Hashimoto et al. (2003) demonstrated that G_o, but not G_i, mediates antibody-stimulated APP neurotoxicity. The cytoplasmic domain (APP_CD)ofAPP [APP_649-695 (numbering by Kang et al., 1987)], especially the middle cytoplasmic region, APP_657-676, has been implicated in this cytotoxic function (Sudo et al., 2001). APP_CD can interact with a number of cytoplasmic adapters (see Kawasumi et al., 2002, for review): 1) G_o through APP_657-676; and 2) phosphotyrosine-interaction domain-containing adapters, such as the Fe65 family, X11, mDab1, and JIP-1b, through APP_677-695. Considering the PTX sensitivity of antibody-stimulated APP neurotoxicity and the involvement of APP_657-676,itis most likely that antibody binding to APP would trigger the toxic function of its cytoplasmic domain that is mediated by G_o.

Based upon circumstantial evidence, Sudo et al. (2000) speculated that antibody-induced oligomerization of cell-surface APP may trigger the toxic function of APP_CD. The present study was conducted to examine whether dimerization of APP_CD induced by an extracellular soluble factor other than antibodies could trigger neuronal cell death.

Based on the well established fact that epidermal growth factor (EGF) causes dimerization of the cell-surface receptor for EGF (EGFR), two fusion constructs of the extracellular domain of EGFR (EGFR_ED) and APP_CD were constructed. One, termed APP_TM+CD hybrid, consisted of EGFR_ED and the transmembrane domain plus the cytoplasmic domain of APP. The other, termed APP_CD hybrid, consisted of the extracellular domain plus the transmembrane domain of EGFR (EGFR_ED+TM) and APP_CD. The results clearly indicate that induced dimerization of APP_CD drastically enhances neuronal cell death via apoptosis signal-regulating kinase (ASK)1. Given that c-Jun N-terminal kinase (JNK) is involved in this function of APP (Hashimoto et al., 2003), the present study indicates that dimerization of APP_CD causes ASK1/JNK-mediated neuronal cell death. We also find that short-term treatment with anti-APP antibody causes death in primary neurons, and that delayed treatment with PTX does not inhibit neuronal death by anti-APP antibody, whereas delayed treatment with a JNK inhibitor suppresses it. This points to the presence of a novel mechanism that sustains the activity of this pathway after initial activation, allowing for the achievement of cell death by short-term antibody treatment. We discuss a potential role of ASK1/JNK in this mechanism.

	Materials and Methods

Top Abstract Materials and Methods Results Discussion References

 
Genes and Materials. The EGFR/APP hybrids were constructed from human EGFR cDNA (kindly provided by Dr. Tadashi Yamamoto, University of Tokyo Institute of Medical Science, Tokyo, Japan), mouse wild-type (wt) APP₆₉₅ (the APP isoform consisting of 695 residues) cDNA (Yamatsuji et al., 1996a,1996b), and the polymerase chain reaction (PCR) with Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA). The sense primer for APP_TM+CD was 5'-TTTTTTGATATCGGCGCCATCATCGGA-3', and the sense primer for APP_CD was 5'-TTTTTTGATATCAAGAAGAAACAGTACACATCC-3'. The antisense primer for both APP_TM+CD and APP_CD was 5'-TTTTTTTCTAGATTAGTTCTGCATTTGCTC-3'. After digestion by EcoRV and XbaI, the DNA fragments were subcloned into corresponding multicloning sites of pcDNA3 vector, termed pcDNA-APP_TM+CD and pcDNA-APP_CD, respectively. EGFR DNA fragments were amplified by PCR with 5'-TTTTTTGGTACCATGCGACCCTCCGGGACG-3' (the sense primer for both EGFR_ED and EGFR_ED+TM) and 5'-TTTTTTGATATCGGCGATGGACGGGATCTTAGG-3' (the antisense primer for EGFR_ED) or 5'-TTTTTTGATATCCATGAAGAGGCCGATCCCCAG-3' (the antisense primer for EGFR_ED+TM). After digestion of the PCR products with KpnI and EcoRV, both EGFR fragments were subcloned into the corresponding sites of pcDNA-APP_TM+CD and pcDNA-APP_CD, respectively. The constructed plasmids were named pcDNA-EGFR_ED/APP_TM+CD for APP_TM+CD hybrid and pcDNA-EGFR_ED+TM/APP_CD for APP_CD hybrid, respectively. We confirmed these hybrids by direct sequencing. The plasmid encoding EGFR(ED+TM) was also constructed from human EGFR cDNA, and the PCR with Pfu Turbo DNA polymerase. The sense and the antisense primers for EGFR(ED+TM) were 5'-TTTTTTGGTACCATGCGACCCTCCGGGACG-3' and 5'-GATATCTCACATGAAGAGGCCGATCCCCAG-3', respectively. PCR products for EGFR(ED+TM) were cloned into pCR-Blunt II TOPO vector using Zero Blunt TOPO Cloning Kit (Invitrogen). After digestion with KpnI and EcoRV, the EGFR(ED+TM) fragments were subcloned into the corresponding sites of pcDNA3 vector. The construction was confirmed by direct sequencing. wtASK1 cDNA, dominant-negative (dn)ASK1 cDNA (ASK1-K709R), and constitutively active (ca)ASK1 cDNA [ASK1(649-1375)], all tagged with HA, were described previously (Saitoh et al., 1998) and used after subcloning into pcDNA. Glutathione-ethyl ester (GEE), L-NMMA (N^G-monomethyl-L-arginine monoacetate salt), PTX, SP600125, PD98059, and SB203580 were from Calbiochem-Novabiochem (San Diego, CA). Apocynin (4-hydroxy-3-methoxyacetophenone; APO) was from Sigma-Aldrich (St. Louis, MO). Ac-DEVD-CHO (acetyl-L-aspartyl-L-glutamyl-L-valyl-L-aspartyl-aldehyde; DEVD) and other tetrapeptide caspase inhibitors were from Peptide Institute (Minoh, Osaka, Japan).

Experiments using F11 Cells. F11 cells were grown in Ham's F-12 plus 18% FBS and antibiotics, as described previously (Yamatsuji et al., 1996a; Hashimoto et al., 2000a, 2001a,b,c). F11 cells, the hybrid cells of a rat embryonic day 13 primary cultured neuron and a mouse neuroblastoma, are one of the best models for primary cultured neurons; they exhibit, without differentiation factor treatment, a number of characteristics for primary neurons, including generation of action potentials (Platika et al., 1985). For the experiments using APP hybrid cDNAs in pcDNA, F11 cells were seeded at 7 x 10⁴ cells/well in a six-well plate and cultured in Ham's F-12 plus 18% FBS for 12 to 16 h, and were transfected with APP hybrid cDNA by lipofection (1 µg of APP hybrid cDNA, 2 µl of LipofectAMINE, and 4 µl of PLUS; Invitrogen) in the absence of serum for 3 h. After subsequent incubation with Ham's F-12 plus 18% FBS for 18 h, cells were cultured with or without EGF in serum-free Ham's F-12, with or without N2 supplement. Unless otherwise specified, N2 supplement was used. For the experiments using various inhibitors, the inhibitor was included in serum-free Ham's F-12 containing EGF. Cell mortality was measured by trypan blue exclusion assay at 48 h after the onset of EGF treatment. For coexpression of ASK1 constructs, F11 cells, similarly seeded at 7 x 10⁴ cells/well in a six-well plate and cultured in Ham's F-12 plus 18% FBS for 12 to 16 h, were transfected with EGFR/APP cDNA with wtASK1 or dnASK1 (1 µg of hybrid cDNA, 1 µg of ASK1 cDNA, 4 µl of LipofectAMINE, and 8 µl of PLUS) in the absence of serum for 3 h. After subsequent incubation with Ham's F-12 plus 18% FBS for 18 h, cells were cultured with or without EGF in serum-free Ham's F-12 containing N2 supplement. Cell mortality was measured by trypan blue exclusion assay at 48 h after the onset of EGF treatment. For constitutively active ASK1 (caASK1) cDNA transfection, F11 cells were similarly transfected with caASK1 cDNA (1 µg of caASK1 cDNA, 2 µl of LipofectAMINE, and 4 µl of PLUS) and cultured in serum-free Ham's F-12 plus N2 supplement. The A53T mutant of -synuclein (SYN) cDNA (in pEF4/Myc/His) was similarly transfected with ASK1 constructs (1 µg of A53T-SYN cDNA, 1 µg of ASK1 cDNA, 4 µl of LipofectAMINE, and 8 µl of PLUS). For EGFR(ED+TM) and APP_CD hybrid transfection, F11 cells were transfected with 1 µg of EGFR(ED+TM) cDNA (or an empty plasmid), 1 µgofAPP_CD hybrid cDNA (or an empty plasmid), 4 µl of LipofectAMINE, and 8 µl of PLUS.

Cytotoxicity and Viability Assays. Cytotoxicity was assessed by trypan blue exclusion assay, as described previously (Hashimoto et al., 2000a,b, 2001a,b,c, 2002a,b). In brief, cells were suspended by pipetting gently, and 50 µl of 0.4% trypan blue solution (Sigma-Aldrich) was mixed with 200 µl of cell suspension (final trypan blue concentration was 0.08%) at room temperature. Stained cells were counted within 3 min after being mixed with the trypan blue solution. The mortality of cells was then determined as a percentage of trypan blue-stained cells in total cells. The cell mortality thus assessed represents the population of dead cells in total cells, including both adhesive and floating cells, at the termination of experiments. Cell viability assay (Cell Counting Kit-8; Wako Pure Chemicals, Osaka, Japan) was performed using WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt], as described in detail previously (Hashimoto et al., 2000a).

Immunoblot Analysis. Immunoblot analysis of expressed EGFR/APP hybrids (50 µg/lane) was performed, using 2 µg/ml mouse anti-EGFR monoclonal antibody (Upstate Biotechnology, Lake Placid, NY), 1/5,000 HRP-labeled anti-mouse IgG antibody, and visualization by ECL (Amersham Biosciences AB, Uppsala, Sweden). Expression of ASK1 constructs (20 µg/lane) were analyzed by 0.1 µg/ml rat anti-HA antibody (Roche Diagnostics, Basel, Switzerland) and 1/2,000 HRP-labeled anti-rat IgG antibody (Wako Pure Chemicals). The bands of APP and JIP-1b were detected by immunoblotting with 2.5 µg/ml anti-APP antibody 22C11 (Chemicon International, Temecula, CA) and 1/1,000 anti-Xpress antibody (Invitrogen), respectively. The secondary antibody used was commonly 1/5,000 HRP-labeled anti-mouse IgG antibody (Bio-Rad, Hercules, CA). Detection of phosphorylated ASK1 was performed as follows. F11 cells were transfected with APP_CD hybrid cDNA (2 µg/well in a six-well plate), and 18 h after transfection, cells were treated with EGF. Cells were then lysed with lysis buffer [20 mM Tris/HCl (pH 7.5), 150 mM NaCl, 10 mM EDTA, 1% deoxycholate, 1% Triton X-100] and submitted to 7% SDS-PAGE. Immunoblot analysis was performed with antibodies for phospho-ASK1 and for ASK1, as described (Tobiume et al., 2002).

Assessment of JNK Activation by 22C11. Primary neurons were seeded in poly-L-lysine-coated six-well plates (Sumitomo Bakelite, Akita, Japan) at 1 x 10⁶ cells/well in Neuron Medium (Sumitomo Bakelite), as previously described (Hashimoto et al., 2003). The purity of neurons by this method was >98%. On day in vitro 3, the cultured medium was changed to Dulbecco's modified Eagle's medium containing N2 supplement (DMEM-N2). On day in vitro 6, neurons were treated with 2 µg/ml 22C11 with or without 100 ng/ml PTX, 300 µM APO, or 1 mM L-NMMA. At the termination of the reaction, cells were lysed in ice-cold lysis buffer [50 mM Tris/HCl (pH 8.0), 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, and 1 tablet/50 ml of protease inhibitor cocktail (Roche Diagnostics)]. After centrifugation at 10,000g, cell lysates were submitted to SDS-PAGE and electroblotted onto polyvinylidene difluoride membranes. After blocking by 1% bovine serum albumin-Tris-buffered saline/Tween 20 for 3 h at room temperature, the membranes were probed with 1 µg/ml anti-phosphorylated JNK antibody (Cell Signaling Technology Inc., Beverly, MA) and 1/5,000 HRP-labeled anti-mouse IgG antibody. The membranes were reprobed with 1/500 anti-JNK antibody (Santa Cruz Biotechnology) and 1/5,000 HRP-labeled anti-rabbit IgG antibody (Bio-Rad).

Assessment of Cell Death by Short-Term 22C11 Treatment. Prepared neurons (seeded in 96-well plates; 2.5 x 10⁴, 100 µl of culture medium/well) were treated with 5 µg of 22C11 for 6 or 12 h, and the cultured medium was changed to DMEM-N2 with 100 ng/ml PTX or 100 nM SP600125 for 66 or 60 h, respectively. Calcein assay was performed by treating neurons with 1 µl of 600 µM Calcein-AM (Dojindo, Kumamoto, Japan). After 2 h of incubation in a CO₂ incubator at 37°C, the cultured medium was changed to PBS to lower the background, and calcein-specific fluorescence (excitation, 485 nm; emission, 535 nm) was measured by a spectrofluorometer (Wallac 1420 ARVOsx Multi Label Counter; PerkinElmer Life Sciences, Boston, MA).

Statistical Analysis. All of the experiments described in this study were repeated at least three times with independent transfections and treatments, each of which yielded essentially the same result. Statistical analysis was performed with one-way ANOVA followed by the post hoc test, in which p < 0.05 was assessed as significant.

	Results

Top Abstract Materials and Methods Results Discussion References

 
Toxic Effect of EGF in Neuronal Hybrid Cells Expressing EGFR/APP Hybrids. To examine whether dimerization of APP_CD triggers the mechanism for neuronal cell death, we created two fusion constructs of EGFR_ED and APP_CD: APP_TM+CD hybrid and APP_CD hybrid. We tested them in F11 neuronal cells, which are the hybrid cells of a rat embryonic day 13 primary cultured neuron and a mouse neuroblastoma NTG18 (Platika et al., 1985). They exhibit, without differentiation factor treatment, a number of characteristics for primary neurons, including generation of action potentials, and have been used in various studies of neuronal functions as one of the best immortalized models for primary neurons (Yamatsuji et al., 1996a; Storms and Rutishauser, 1998; Ghil et al., 2000; Hagiwara et al., 2000; Hashimoto et al., 2000a,b, 2001a,b,c, 2003; Huang et al., 2000; Niikura et al., 2000, 2001; Sudo et al., 2000, 2001). F11 cells were thought to be most appropriate for the investigation of APP_CD functions by its transfection because treatment of F11 cells with anti-APP antibody robustly causes cell death only when F11 cells are transfected with APP cDNA (Sudo et al., 2000, 2001).

First, simple overexpression of APP_TM+CD hybrid or APP_CD hybrid in F11 cells resulted in a significant increase in neurotoxicity from the basal 10% mortality level to 30% mortality level (Fig. 1A, upper panel, for APP_TM+CD hybrid; Fig. 1A, lower panel, for APP_CD hybrid: the hybrid structures in the inset of Fig. 1B). This was quite reasonable; our earlier studies (Hashimoto et al., 2000b; Sudo et al., 2001) had found that simple overexpression of full-length APP in F11 cells causes 30 to 40% mortality via the cytoplasmic C-terminal APP_677-695 region in the absence of N2 supplement. It was thus highly likely that overexpression of APP_CD would induce cell death in F11 neuronal cells to a certain degree. Notably, this neurotoxicity was suppressed by N2 supplement (Fig. 1A). We also found that in F11 cells expressing either EGFR/APP hybrid, 100 nM EGF greatly enhanced neuronal cell death to 60 to 70% mortality levels, whether N2 supplement was present or not (Fig. 1A). The presence of N2 supplement little affected the transfection efficiency or expression levels of the EGFR/APP hybrids (insets of Fig. 1A). These data suggest that dimerization of overexpressed APP_CD causes neuronal cell death in an N2 supplement-resistant manner, which was different from N2 supplement-sensitive neurotoxicity by simple overexpression of APP_CD. To focus on dimerization-induced APP_CD neurotoxicity, we used N2 supplement in subsequent experiments.

View larger version (76K): [in this window] [in a new window]   Fig. 1. Effect of EGF on neuronal cell death by EGFR/APP hybrids. A, after F11 neurohybrid cells were transfected with or without APPTM+ hybrid cDNA (upper panel) or APPCD hybrid cDNA (lower panel), cells were treated with or without 100 nM EGF in the absence (none) or (N2) of N2 supplement. Cell mortality was measured by trypan blue exclusion assay 48 h after the onset of EGF treatment. *, significantly cytotoxicity relative to corresponding APP hybrid cytotoxicity in the absence of EGF and N2 supplement. **, significantly increased relative to corresponding APP hybrid cytotoxicity in the absence of EGF and the presence of N2 supplement. ***, significantly decreased relative to corresponding APP hybrid cytotoxicity in the absence of EGF and N2 supplement. All values presented in this study indicate means S.D. of three to four independent experiments. Insets: immunoblot analysis of each APP hybrid. F11 cells were transfected with or APPTM+CD hybrid cDNA (upper panel) or APPCD hybrid cDNA (lower panel) and treated with or without 100 nM EGF in the presence or of N2 supplement. The cell lysates were submitted to SDS-PAGE and immunoblot analysis with anti-EGFR antibody. Arrowheads indicate A hybrids. The numbers correspond to those of the experiments performed in the same panels. B, dose effects of EGF on APP hybrid-mediated cell death. F11 cells were transfected without (no T) or with empty pcDNA (pcDNA), APPTM+CD hybrid cDNA (APPTM+CD hybrid), or APPCD cDNA (APPCD hybrid) and treated with increasing concentrations of EGF in the presence of N2 supplement. Forty-eight hours after the onset of treatment, cell mortality was measured by trypan blue exclusion assay. *, significant versus corresponding APP hybrid in the absence of EGF. illustration of two APP hybrids. ED, TM, and CD, extracellular domain, transmembrane domain, and cytoplasmic domain, respectively. C, effects EGFR(ED+TM). F11 cells were transfected with or without EGFR(ED+TM) cDNA with or without APPCD hybrid cDNA and treated with or 100 nM EGF in the presence of N2 supplement. Forty-eight hours after the onset of EGF treatment, cell mortality was measured by trypan exclusion assay. n.s., not significant. *, significant versus corresponding APPCD hybrid in the absence of EGF. **, significant versus APPCD alone or in the presence of EGF. Inset: expression of EGFR(ED+TM) or APPCD hybrid in the experiments shown in C. The cell lysate samples immunoblotted by anti-EGFR antibody. The arrowhead indicates EGFR(ED+TM) or APPCD hybrid. The numbers correspond to those of experiments performed in C. D, effects of DEVD, GEE, or PTX on neuronal cell by APP hybrids. F11 cells were transfected without (no T) or empty pcDNA (pcDNA), APPTM+CD hybrid cDNA (APPTM+CD hybrid), or APPCD hybrid cDNA (APPCD hybrid) and treated with 100 nM EGF the presence of N2 supplement without (-) or with 100 µM DEVD (D), 1 mM GEE (G), or 1 µg/ml PTX (P). Forty-eight hours after the onset of treatment, cell mortality was measured by trypan blue exclusion assay. *, significant versus corresponding APP hybrid alone in the presence of E, effect of APO or L-NMMA on EGF-dependent neuronal cell death by APP hybrids. F11 cells were transfected without (no T) or with empty (pcDNA), APPTM + CD hybrid cDNA (APPTM+CD hybrid), or APPCD hybrid cDNA (APPCD hybrid) and treated with 100 nM EGF in the of N2 supplement without (-) or with 300 µM APO (A) or 1 mM L-NMMA (N). Forty-eight hours after transfection, cell mortality was measured trypan blue exclusion assay. * and #, significant and not significant versus corresponding APP hybrid alone in the presence of EGF.

In F11 cells expressing APP_TM+CD hybrid, EGF greatly enhanced neuronal cell death from the basal levels to 60 to 70% mortality levels with the EC₅₀ value at 1 to 10 nM (Fig. 1B). This was also the case with APP_CD hybrid. In cells transfected with APP_CD hybrid, EGF dose dependently increased neurotoxicity from the basal levels to 60 to 70% mortality levels with the EC₅₀ value at 1 to 10 nM (Fig. 1B). The level of APP_TM+CD hybrid expression was similar to that of APP_CD hybrid expression under the same condition (Fig. 1A, insets). Because it has been demonstrated that EGF dimerizes EGFR at 1 to 10 nM, these data indicate that induced dimerization of APP_CD hybrid causes neuronal cell death.

To confirm that APP_CD mediates the toxic function of APP_CD hybrid, we examined EGFR(ED+TM), the EGFR construct consisting of the extracellular domain and the transmembrane domain of EGFR. When F11 cells were transfected with EGFR(ED+TM) cDNA under the same conditions as APP_CD hybrid transfection, 100 nM EGF could not significantly exert neurotoxicity (Fig. 1C). Although the expression level of EGFR(ED+TM) was somewhat lower than that of APP_CD hybrid, the expression of EGFR(ED+TM) was significant (Fig. 1C, inset). When cells were transfected with APP_CD hybrid in the presence of EGFR(ED+TM) cotransfection, EGF-induced neurotoxicity was drastically suppressed (Fig. 1C), suggesting that EGFR(ED+TM) acts as a dominantly interfering construct of APP_CD hybrid. At this moment, it remains unknown why at expression levels considerably lower than that of APP_CD hybrid, EGFR(ED+TM) strongly suppressed the EGF-dependent toxicity of APP_CD hybrid. However, it is most likely that in addition to the sequestration of EGF, EGFR(ED+TM) suppressed the toxic function of APP_CD hybrid by forming an inactive heterodimer with APP_CD hybrid. Alternatively, EGFR(ED+TM) might be targeted to the cell surface, where EGF acts on APP_CD hybrid, more efficiently than APP_CD hybrid. We were unable to examine the cellular localization of EGFR(ED+TM) and APP_CD hybrid. These data strengthen the notion that APP_CD mediates the toxic function of APP_CD hybrid.

Effects of Pharmacological Inhibitors on Neuronal Cell Death by EGFR/APP Hybrids. We next investigated the sensitivity of APP_CD-induced neurotoxicity to DEVD, GEE, and PTX. It has been shown that neuronal cell deaths by both antibody-bound APP and by V642I-APP, which causes familial AD, are mediated by the cytoplasmic region APP_657-676 (Yamatsuji et al., 1996a,b; Hashimoto et al., 2000b; Sudo et al., 2001) and that neuronal cell death by V642I-APP is sensitive to DEVD (an established caspase inhibitor), GEE (an established antioxidant), and PTX (a specific G_i/o inhibitor) (Yamatsuji et al., 1996a,b; Giambarella et al., 1997; Hashimoto et al., 2000b, 2002b). We therefore expected that EGFR/APP hybrid-induced neuronal cell death would also be sensitive to DEVD, GEE, and PTX. As expected, in F11 neuronal cells expressing either APP_TM+CD hybrid or APP_CD hybrid, cell death enhanced by EGF was suppressed by DEVD, by GEE, and by PTX (Fig. 1D). APP hybrid expression was not affected by any of these inhibitors (data not shown).

Because neuronal cell death by antibody-bound wtAPP as well as by V642I-APP is mediated by NADPH oxidase (Hashimoto et al., 2002b, 2003), the GEE-sensitive neurotoxicity by APP_CD dimerization was assumed to be mediated by NADPH oxidase. Thus, we next examined whether this could be the case. APO is the established NADPH oxidase inhibitor, and NADPH oxidase is the only reactive oxygen species-generating enzyme that is inhibited by 300 µM APO (Hart and Simons, 1992). It has also been noted that F11 cells express the subunits of NADPH oxidase (Hashimoto et al., 2002b). In F11 cells expressing either APP_TM+CD hybrid or APP_CD hybrid, EGF-induced cytotoxicity was drastically inhibited by 300 µM APO (Fig. 1E). This inhibition by APO was specific, because 1 mM L-NMMA had no effect. It was therefore likely that neuronal cell death by APP_CD dimerization is mediated by NADPH oxidase.

Effect of dnASK1 on APP_CD Hybrid-Induced Neuronal Cell Death. Since ASK1 is one of the established targets of reactive oxygen species for cell death (Saitoh et al., 1998), we examined whether ASK1 is involved in DEVD-sensitive neuronal cell death by APP_CD dimerization. For this purpose, F11 cells were transfected with APP_CD hybrid with ASK1-K709R, the dominant negative mutant of ASK1 (dnASK1), and treated with or without EGF (Fig. 2A). As shown in Fig. 2B, the EGF-induced increase in neuronal cell mortality was drastically inhibited by coexpressed dnASK1. In contrast, EGF-induced neurotoxicity was little inhibited by similarly coexpressed wtASK1 (Fig. 2B). Under the same conditions, expression of APP_CD hybrid was not inhibited by coexpression of dnASK1 or wtASK1 (Fig. 2A). Drastic inhibition of APP_CD hybrid-induced cell death by dnASK1, but not by wtASK1, was also the case, when cell viability was assessed by WST-8 assay. The viability of cells expressing APP_CD hybrid was drastically impaired by EGF treatment. This impairment was recovered by coexpressed dnASK1, but not by coexpressed wtASK1 (Fig. 2C).

View larger version (90K): [in this window] [in a new window]   Fig. 2. Effect of dnASK1 on EGF-induced neuronal cell death by APPCD hybrid. A, immunoblot analysis of APPCD hybrid and ASK1 constructs. F11 neurohybrid cells were transfected without (-) or with empty pcDNA (pcDNA), APPCD hybrid cDNA (APPCD hybrid) with empty pcDNA (vec), dnASK1 cDNA (dnASK1), or wtASK1 cDNA (wtASK1) and treated with or without 100 nM EGF in the presence of N2 supplement. Forty-eight hours after the onset of EGF treatment, the cell lysates were submitted to immunoblot analysis with anti-EGFR antibody (for APPCD hybrid) or anti-HA antibody (for ASK1 constructs). The upper panel illustrates wtASK1, dnASK1, and caASK1. Numbers indicate the position of amino acids. The shaded region (amino acid 678-936) corresponds to the kinase domain. The lower panel indicates expressed ASK1 constructs (bottom subfigure) and APPCD hybrid (middle subfigure). The numbers on the left sides of the gels indicate molecular mass in kilodaltons. B, inhibition by dnASK1 of EGF-induced neuronal cell death by APPCD hybrid. F11 cells were transfected without (no T) or with empty pcDNA (pcDNA) or APPCD hybrid cDNA (APPCD hybrid) without (none) or with empty pcDNA (pcDNA), ASK1-K709R cDNA (dnASK1), or wtASK1 cDNA (wtASK1), and treated with or without 100 nM EGF in the presence of N2 supplement. Forty-eight hours after the onset of EGF treatment, cell mortality was measured by trypan blue exclusion assay. *, not significant versus APPCD hybrid alone in the presence of EGF. **, significant versus APPCD hybrid alone and APPCD hybrid with wtASK1 in the presence of EGF. ***, significantly increased cytotoxicity as compared with cytotoxicity in the absence of EGF or EGF-induced cytotoxicity in cells not expressing APPCD hybrid. C, cell viability assay results. Experiments were performed in the same way as in B. Forty-eight hours after the onset of EGF treatment, cell viability was measured by WST-8 assay. These experiments were performed independently from the experiments shown in B. *, not significant versus APPCD hybrid alone in the presence of EGF. **, significantly higher cell viability as compared with that of cells transfected with APPCD hybrid alone as well as that of cells transfected with APPCD hybrid and wtASK1. ***, significant versus cell viability in the absence of EGF or viability of cells not expressing APPCD hybrid in the presence of EGF. D, inhibition by dnASK1 of anti-APP antibody-induced neuronal cell death by full-length wtAPP. F11 cells were transfected without (no T) or with empty pcDNA (pcDNA) or full-length wtAPP cDNA (wtAPP) with empty pcDNA (pcDNA) or dnASK1 cDNA (dnASK1), and treated with or without 2 µg/ml 22C11 in the presence of N2 supplement. Forty-eight hours after the onset of 22C11 treatment, cell mortality was measured by trypan blue exclusion assay. *, significant versus wtAPP alone in the absence of 22C11. **, not significant versus wtAPP plus dnASK1 in the absence of 22C11. Upper panels: expression of full-length APP (topmost subfigure) ASK1 constructs (middle subfigure) in the experiment corresponding to that shown in the bottom panel. E, effect of dnASK1 on mutant SYN-induced neuronal cell death. F11 cells were transfected without (no T) or with empty vector (vec) or mutant SYN cDNA (A30P SYN or A53T SYN) with empty vector (vec) or dnASK1 cDNA (dnASK1) in the presence of N2 supplement. Forty-eight hours after the onset of transfection, cell mortality was measured by trypan blue exclusion assay. Insets: immunoblot analysis of SYN constructs (upper) and dnASK1 (lower) in the experiment shown in the same panel. Arrowheads indicate the corresponding constructs. The numbers correspond to those of the experiments performed in the same panels. F, induction of neuronal cell death by caASK1 and its sensitivity to various inhibitors. F11 cells were transfected without (no T) or with empty pcDNA (vec), wtASK1 cDNA (wtASK1), or caASK1 cDNA (caASK1) in the presence of N2 supplement without (-) or with 100 µM DEVD (D), 1 mM GEE (G), or 1 µg/ml PTX (P). Forty-eight hours after transfection, cell mortality was measured by trypan blue exclusion assay. * and **, significant and not significant versus caASK1 alone, respectively. Inset: effect of APO or L-NMMA on neuronal cell death by caASK1 (caASK1). F11 cells were transfected without (no T) or with empty pcDNA (vec), caASK1 cDNA (caASK1) in the presence of N2 supplement without (none), or with 300 µM APO (APO) or 1 mM L-NMMA (L-NMMA). Forty-eight hours after the onset of transfection, cell mortality was measured by trypan blue exclusion assay. * and **, significant and not significant versus caASK1 alone, respectively. The upper panel indicates expressed HA-tagged caASK1 in the corresponding cell lysates by anti-HA antibody.

Effect of dnASK1 on Neuronal Cell Death by Full-Length APP. We next examined whether expression of dnASK1 also suppresses neuronal cell death caused by full-length wtAPP. As previously reported (Sudo et al., 2000, 2001; Hashimoto et al., 2003), 22C11 (monoclonal anti-APP antibody) could cause cell death in F11 neurohybrid cells, when cells had been transfected with wtAPP (Fig. 2D). Under the same condition, when dnASK1 was cotransfected instead of an empty vector, 22C11 could not stimulate cell death in F11 cells similarly overexpressing wtAPP (Fig. 2D). These data suggest that dnASK1 inhibition of APP_CD neurotoxicity reflects the involvement of ASK1 in neurotoxicity caused by full-length APP.

We also examined whether expression of dnASK1 suppresses neuronal cell death caused by mutant -synuclein (A30PSYN and A53TSYN). A30PSYN and A53TSYN are the genes that cause dominantly inherited Parkinsonism with dementia, and they are known to cause nonapoptotic neuronal cell death in the F11 neurohybrid cell system (Hashimoto et al., 2003). As shown in Fig. 2E, overexpression of dnASK1 could not affect neuronal cell death caused by these mutant SYNs in the same system. Therefore, the inhibitory effect of dnASK1 is considered to be specific.

Neuronal Cell Death by caASK1 Is Sensitive to DEVD and GEE/APO but Resistant to PTX. These data provide strong evidence that ASK1 plays a key role in mediating DEVD-sensitive neuronal cell death by APP_CD dimerization. If so, simple expression of caASK1 must also cause DEVD/APO-sensitive neuronal cell death. N-terminally deleted ASK1 [ASK1(649-1375)] has been demonstrated to be caASK1 (Saitoh et al., 1998; Kanamoto et al., 2000). Using this caASK1, we found that expression of caASK1 caused neuronal cell death sensitive to DEVD (Fig. 2F). caASK1-induced cell death was sensitive to GEE and APO, but resistant to L-NMMA, indicating that caASK1 triggers DEVD/APO-sensitive cell death similar to that caused by APP_CD dimerization (Fig. 2F and inset). Neither APO nor L-NMMA altered the expression levels of caASK1 (inset of Fig. 2F). These data provide evidence that 1) activation of ASK1 can cause neuronal cell death, and that 2) caASK1 activates a neurotoxic pathway with properties quite close to those of the APP_CD-activated pathway.

In contrast to the effectiveness of DEVD and GEE, PTX had no effect on caASK1-induced neuronal cell death (Fig. 2F), under the same conditions in which PTX was very effective in inhibiting neuronal cell death by APP_CD (Fig. 1D), as well as by wtAPP (Hashimoto et al., 2003). Combined with the drastic inhibition by dnASK1 of APP_CD-induced neuronal cell death, these data not only provide evidence that inhibition by DEVD and GEE was not an artifact but also indicate that the hierarchy is the PTX target, ASK1, and the GEE/APO target in this toxic pathway. Since it has been demonstrated, using the same system, that G_o, JNK, and NADPH oxidase, in this order, are involved in 22C11-stimulated APP neurotoxicity (Hashimoto et al., 2003), it is most likely that DEVD-sensitive neuronal cell death by APP_CD dimerization is mediated by ASK1, by G_o as the upstream signal transducer of ASK1, and by NADPH oxidase as the downstream transducer of ASK1.

APP Forms a Complex with ASK1 via JIP-1b, the JNK-Interacting Protein. Because ASK1 belongs to the mitogen-activated protein kinase kinase kinase-class kinases of the JNK pathway, there was a possibility that ASK1 would form a direct complex with JIP-1b, which is the established scaffold protein of the kinases in the JNK pathway. However, since it had not been shown that JIP-1b forms a direct complex with ASK1, we examined this issue. When F11 cells were transfected with JIP-1b and ASK1 cDNAs, immunoprecipitation of JIP-1b coprecipitated ASK1 (Fig. 3A). Coprecipitation with JIP-1b was also the case with dnASK1 and caASK1, indicating that the interaction with JIP-1b does not depend on the activation status of ASK1 (Fig. 3A).

View larger version (51K): [in this window] [in a new window]   Fig. 3. APP forms a complex with ASK1 via JIP-1b and HN inhibits APP hybrid/caASK1-induced neuronal cell death. A, after F11 cells were transfected with or without JIP-1b and with or without wtASK1, dnASK1, or caASK1, JIP-1b in the cell lysates was immunoprecipitated by anti-(His)6 antibody, and the precipitate was analyzed with 0.1 µg/ml anti-HA antibody (Roche Diagnostics) for ASK1. The five lanes in the right half of the figure indicate immunoblot analysis of the precipitated samples, and the five lanes on the left indicate the inputs for immunoprecipitation. B, after F11 cells were transfected with wtAPP, JIP-1b, and wtASK1 (lane 3) or APP19, JIP-1b, and wtASK1 (lane 4), cell lysates were immunoprecipitated by 22C11 for APP (leftmost panels) or anti-(His)6 antibody for JIP-1b (middle panels). The samples were submitted to immunoblot analysis with anti-Xpress antibody (for JIP-1b blot), anti-HA antibody (for ASK1 blot), or 22C11 (for APP blot). Lanes 1 and 2 indicate no transfection and vector transfection cases, respectively. The rightmost panels indicate the inputs for immunoprecipitation. C, activation of endogenous ASK1 by APPCD hybrid in response to EGF. F11 cells were transfected without (-) or with APPCD hybrid. Eighteen hours after transfection, cells were stimulated with 10 ng/ml (1.7 nM) EGF and lysed at each indicated time point (hours) after the onset of EGF treatment. Cell lysates were analyzed by immunoblotting (WB:p-ASK1) with anti-phospho-ASK1 antibody. Cell lysates were also analyzed with anti-ASK1 antibody as loading controls (WB:ASK1). As another loading control, cells were transfected with HA-tagged ASK1 for 18 h, and cell lysates were analyzed by anti-HA antibody, anti-phospho-ASK1 antibody, or anti-ASK1 antibody (WB:HA). As a positive control, F11 cells were treated with 1 mM H2O2 for 30 min (H2O2). The arrowheads denote the specific bands of activated ASK1 (upper panel) and total ASK1 (middle panel), respectively. D, effects of HN, HNG, and HNA on APPCD hybrid-mediated neuronal cell death. F11 cells were transfected with empty pcDNA (pcDNA) or APPCD hybrid cDNA (APPCD hybrid) and cultured in the presence or absence of increasing concentrations (0, 10 nM, 100 nM, 1 µM, or 10 µM, from left to right) of HN, increasing concentrations (10 pM, 100 pM, 1 nM, or 10 nM, from left to right) of HNG, or 10 µM HNA for 48 h. Cell mortality was then measured by trypan blue exclusion assay. no T represents the results of cell mortality without transfection. * and **, significant and not significant versus APP hybrid plus EGF. Right panel, immunoblot analysis of expressed APPCD hybrid in the experiments shown in the left panel. An arrowhead indicates the expressed APPCD hybrid protein. The employed concentrations of HN, HNG, and HNA were 10 µM, 10 nM, and 10 µM, respectively. E, effects of HN, HNG, and HNA on caASK1-mediated neuronal cell death. F11 cells were transfected with empty vector (vec) or caASK1 cDNA (caASK1) and cultured in the presence or absence of increasing concentrations [0 (lane 1), 10 nM (lane 2), 100 nM (lane 3), 1 µM (lane 4), or 10 µM (lane 5), from left to right] of HN, increasing concentrations [10 pM (lane 6), 100 pM (lane 7), 1 nM (lane 8), or 10 nM (lane 9), from left to right] of HNG, or 10 µM HNA (lane 10) for 48 h. Cell mortality was then measured by trypan blue exclusion assay. no T represents the results of cell mortality without transfection. * and **, significant and not significant versus caASK1 alone. Upper panel: immunoblot analysis of expressed caASK1 in the experiments shown in the lower panel.

Since it has been shown that JIP-1b interacts with APP and that the interaction domain of APP is the G⁶⁸¹YENPTY⁶⁸⁷ region contained in APP_CD (Matsuda et al., 2001; Scheinfeld et al., 2002), we examined whether APP forms a complex with ASK1 via JIP-1b. When F11 cells were transfected with JIP-1b, ASK1, and full-length APP cDNAs, immunoprecipitation of APP coprecipitated JIP-1b and ASK1 (Fig. 3B). When F11 cells were transfected with JIP-1b, ASK1, and APP lacking the Met⁶⁷⁷-Asn⁶⁹⁵ region (APP19; the Met⁶⁷⁷-Asn⁶⁹⁵ region includes the JIP-1b-interacting G⁶⁸¹YENPTY⁶⁸⁷ region), immunoprecipitation of the APP mutant hardly coprecipitated JIP-1b and ASK1 (Fig. 3B). Very faint bands of JIP-1b and ASK1 observed in the APP mutant precipitation would most likely be the JIP-1b and ASK1 coprecipitated by endogenous APP. The same findings were observed when JIP-1b was immunoprecipitated. Immunoprecipitation of JIP-1b coprecipitated ASK1 as well as full-length APP, but not APP19 (Fig. 3B). These findings provide definite evidence that APP forms a complex with ASK1 via JIP-1b at APP_CD.

Activation of ASK1 by EGFR/APP Hybrid in Response to EGF. In an attempt to establish the intermediary role of ASK1, we examined whether ASK1 was activated by the EGFR/APP hybrid in response to EGF. Activation of ASK1 was assessed by specific anti-phospho-ASK1 antibody, as described previously (Tobiume et al., 2002). Whereas endogenous ASK1 was detected in F11 cells by specific anti-ASK1 antibody (Fig. 3C), little phosphorylation occurred in ASK1 by EGF treatment of F11 cells. In contrast, when F11 cells had been transfected with APP_CD hybrid, treatment of cells with nanomolar levels of EGF caused significant activation of endogenous ASK1 without alterations in its expression. These data concur with the notion that F11 cells endogenously express ASK1 and that the APP_CD hybrid induces ASK1 activation in response to EGF.

Effect of HN on APP_CD-Induced and caASK1-Induced Neurotoxicity. Humanin (HN) is a recently identified 24-residue peptide factor that suppresses neuronal cell death not only by various familial AD mutants [APP mutants (V642I, K595N/M596L, A617G, L648P), presenilin (PS)1 mutants (M146L, H163R, A246E, L286Y, C410Y), and N141I-PS2], and A peptides (A1-42/43, and A25-35) (Hashimoto et al., 2001a,b,c), but also by anti-APP antibody (Hashimoto et al., 2001b). HN also suppresses A toxicity in cerebrovascular smooth muscle cells (Jung and Van Nostrand, 2003). Antibody-stimulated APP neurotoxicity is suppressed by 1 to 10 µM HN and 1 to 10 nM (S14G)HN (HNG), but not by (C8A)HN (HNA). Caricasole et al. (2002) identified rat HN, termed rattin, whose biochemical characteristics are very close to those of HN. Understanding the mechanism for the rescue function of HN would provide an important clue to develop novel anti-AD therapeutics. We thus examined whether HN protects neuronal cells from APP_CD hybrid-induced toxicity.

In F11 neurohybrid cells transfected with APP_CD hybrid, EGF strongly enhanced cell death, and the EGF-stimulated toxicity was inhibited by 1 to 10 µM HN and 1 to 10 nM HNG, but not by 10 µM HNA (Fig. 3D). Expression of the APP_CD hybrid was not affected by HN and HN derivatives (Fig. 3D, right panel). These data are consistent both qualitatively and quantitatively with the known functions of HN and HN derivatives against APP-mediated cell death (Hashimoto et al., 2001b).

We next examined the effects of HN on caASK1-induced cell death. Figure 3E indicates that HN and its potent derivative HNG dose dependently inhibited caASK1-induced neuronal cell death. The IC₅₀ values and the concentrations allowing complete suppression were 100 nM and 10 µM for HN and 10 to 100 pM and 10 nM for HNG, respectively. Under the same conditions, HNA had no rescue effect on caASK1-induced cell death even at 10 µM. Any concentration of HN, HNG, or HNA used in this experiment did not inhibit caASK1 expression (Fig. 3E, upper panel). The IC₅₀ values and the doses allowing complete suppression by HN and HNG were virtually identical on 22C11-stimulated APP neurotoxicity, EGF-stimulated APP_CD hybrid neurotoxicity, and caASK1-induced neurotoxicity. These data indicate that ASK1 neurotoxicity is inhibited by HN and its potent derivative HNG but not by its inactive derivative HNA, in the same dose dependencies as those of HN, HNG, and HNA on APP-induced neuronal cell death. Therefore, HN would suppress APP neurotoxicity by inhibiting the downstream pathway of ASK1. This notion concurs with a recent study (Kariya et al., 2002) that HN inhibits NGF deprivation-induced neuronal cell death, as ASK1 has been implicated in NGF deprivation-induced neuronal cell death (Kanamoto et al., 2000). It is therefore likely that HN would be able to inhibit a variety of types of neuronal cell death in which ASK1 is involved.

Effects of a Specific JNK Inhibitor, SP600125, a Specific MEK/MAPK Inhibitor, PD98059, and a Specific p38MAPK Inhibitor, SB203580, on APP_CD Hybrid-Induced and caASK1-Induced Neuronal Cell Death. Although it has been demonstrated that antibody-dependent full-length APP-induced neuronal cell death is inhibited by a specific JNK inhibitor, SP600125, but not by a specific MEK/MAPK inhibitor, PD98059, nor by a specific p38MAPK inhibitor, SB203580 (Hashimoto et al., 2003), it remained unknown whether APP_CD-induced neuronal cell death, if any, is similarly sensitive to SP600125 and resistant to PD98059 and SB203580. As shown in Fig. 4A (bottom panel), EGF-stimulated APP_CD-mediated neuronal cell death was almost completely blocked by 100 nM SP600125, but resistant to 50 µM PD98059 and 20 µM SB203580. Whereas 50 µM PD98059 and 20 µM SB203580 appeared to somewhat augment the expression of APP_CD hybrid (Fig. 4A, middle panel), 100 nM SP600125 did not inhibit its expression under the same condition in which 100 nM SP600125 suppressed toxicity by APP_CD hybrid (Fig. 4A, middle panel). These results revealed that APP_CD-induced neuronal cell death would be mediated by JNK, but not by MEK/MAPK or p38MAPK, as is the case with full-length APP-induced neuronal cell death.

View larger version (47K): [in this window] [in a new window]   Fig. 4. Effect of a specific JNK inhibitor, SP600125, on APPCD-induced and caASK1-induced neuronal cell death and unique effects of delayed treatment with inhibitors of antibody-triggered APP-mediated neuronal death. A, inhibition by SP600125, but not PD98059 or SB203580, of both APPCD-induced and caASK1-induced neuronal cell death. F11 neurohybrid cells were transfected without (no T) or with an empty vector (pcDNA) or caASK1 cDNA, and 24 h after transfection, cells were cultured in the presence of N2 supplement with or without 100 nM SP600125, 50 µM PD98059, or 20 µM SB203580 for 48 h. Forty-eight hours after the onset of transfection, cell mortality was measured by trypan blue exclusion assay. For the APPCD hybrid experiment, F11 cells were transfected with APPCD hybrid cDNA. Twenty-four hours after the onset of transfection, cells were treated with or without 100 nM EGF for 48 h in the presence of N2 supplement with or without 100 nM SP600125, 50 µM PD98059, or 20 µM SB203580. **, significant versus pcDNA alone or APPCD hybrid alone. *, significant versus caASK1 alone or APPCD hybrid plus EGF treatment. Upper photos: immunoblot analysis of caASK1 (upper) and APPCD hybrid (lower) in the presence or absence of SP600125 (SP), PD98059 (PD), or SB203580 (SB) under the same experiments shown in the bottom panel. Arrowheads indicate expressed caASK1 and APPCD hybrid proteins. B, primary neurons (2.5 x 104 cells/well) were treated with 5 µg/ml 22C11 for 6 or 12 h, and then the medium was changed to DMEM-N2 plus 100 ng/ml PTX or 100 nM SP600125 without 22C11. Seventy-two hours after the onset of 22C11 treatment, neuronal viability was measured by calcein fluorescence assay and indicated as the percentage of the viability of each no-treatment case. The viabilities of neurons without treatment were 242,376 ± 60,760 in 6-h delayed medium change, 328,607 ± 58,960 in 12-h delayed medium change, and 560,649 ± 63,622 in 0-h delayed medium change. The viabilities of neurons treated with mouse nonspecific IgG for 72 h were 260,906 ± 106,970 in 5 µg/ml IgG treatment, 180,748 ± 36,183 in 5 µg/ml IgG plus 100 ng/ml PTX, and 232,434 ± 39,915 in 5 µg/ml IgG plus 100 nM SP600125. Therefore, the increase in neuronal viability in the case of 72-h 22C11 treatment was attributed to a putative neurotrophic effect of 22C11, which was not observed in short-term 22C11 treatment. *, significant reduction versus no-treatment case; **, not significant reduction versus no-treatment case. C, primary neurons were treated with 2 µg of 22C11 at time 0, and after various periods of incubation, cell lysates were submitted to the analysis by JNK phosphorylation, as described under Materials and Methods. In an experiment, neurons were pretreated with 100 ng/ml PTX 12 h before the onset of 22C11 treatment, and the medium was changed to DMEM-N2 plus 2 µg of 22C11. In some experiments, 100 ng/ml PTX was added 1 h after the onset of 22C11 treatment. In some experiments, neurons were treated with 300 µM APO simultaneously with 2 µg of 22C11. In an experiment, neurons were treated with 1 mM L-NMMA simultaneously with 2 µg of 22C11 as a control of APO. Each diagram corresponds to each panel of JNK activation. The lower panels indicate total amounts of JNK corresponding to each time point. Arrowheads show phosphorylated JNK. Similar experiments were performed three times for each, and similar results were obtained. D, a potential mechanism for dimerization-induced APPCD neurotoxicity. Anti-APP antibody or EGF binds to cell-surface APP or APPCD hybrid, respectively, and induces APPCD dimerization, which triggers neurotoxicity via Go, ASK1, and JNK. Go binds to the middle cytoplasmic His657-Lys676 region of APP (Nishimoto et al., 1993; Brouillet et al., 1999), and the G of Go activates JNK (Coso et al., 1996). ASK1 and JNK binds to the extreme C-terminal Met677-Asn695 region of APP via JIP (Matsuda et al., 2001; Scheinfeld et al., 2002; this study), and JIP stabilizes the activity of JNK (Dickens et al., 1997; Hashimoto et al., 2003). There is a positive feedback mechanism around the ASK1/JNK system. Both constitutively active ASK1 and constitutively active JNK exert neurotoxicity via NADPH oxidase (Hashimoto et al., 2003: this study). Superoxide generated by NADPH oxidase in turn activates ASK1 (Saitoh et al., 1998). Therefore, a simple interpretation is that the activity of the ASK1/JNK/NADPH oxidase system could be sustained without upstream APP stimulation, once this system is sufficiently activated. However, it should be noted that such a feedback loop as shown in this figure has not been conclusively demonstrated and is not the only possibility. See the text for details.

It has been shown that ASK1 activates JNK and that caASK1 induces DEVD-sensitive cell death (Ichijo et al., 1997). It is therefore reasonable to assume that ASK1-induced cell death is simply mediated by JNK. However, there is a possibility that p38MAPK and MEK/MAPK, which are simultaneously activated by ASK1, could mediate, to some extent, ASK1-induced cell death. Thus, we also examined whether caASK1-induced neuronal cell death is blocked by SP600125 and affected by PD98059 and SB203580. As was the case with APP_CD-induced cell death, caASK1-induced neuronal cell death was almost completely blocked by 100 nM SP600125, but resistant to 50 µM PD98059 and 20 µM SB203580 (Fig. 4A, bottom panel). Whereas 50 µM PD98059 and 20 µM SB203580 appeared to decrease the expression of caASK1 (Fig. 4A, upper panel), 100 nM SP600125 did not inhibit its expression under the same condition in which 100 nM SP600125 suppressed toxicity by caASK1 (Fig. 4A, middle panel). These results indicate that caASK1-induced neuronal cell death is specifically mediated by JNK, but not by MEK/MAPK or p38MAPK.

A Potential Role of the ASK1/JNK System in Sustaining the APP-Triggered Neurotoxic Mechanism after Short-Term Stimulation of APP. Hashimoto et al. (2003) showed that NADPH oxidase is the major downstream mediator of JNK in APP neurotoxicity. Given that JNK is the established downstream target of ASK1 (Ichijo et al., 1997), this concurs with the present data that NADPH oxidase is the downstream mediator of ASK1 in the APP-triggered neurotoxic pathway, suggesting that the ASK1/JNK system is the upstream regulator of NADPH oxidase in this mechanism.

On the other hand, intracellular superoxide activates ASK1 (Saitoh et al., 1998). Therefore, it is conceivable that superoxide generated by NADPH oxidase can up-regulate the activity of upstream ASK1 in a positive feedback manner, once NADPH oxidase is activated in the APP-triggered neurotoxic pathway. If so, relatively short-term treatment with anti-APP antibody would be sufficient to cause neuronal cell death. This idea was consistent with our earlier study (Sudo et al., 2000) showing that 12-h treatment of neurohybrid cells with anti-APP antibody is sufficient to cause cell death, which occurs for 72 h following treatment. In the present study, we confirmed that this was also the case with primary neurons. Six- to 12-h treatment of primary cortical neurons with anti-APP antibody 22C11 was sufficient to cause neuronal death for the next 72 h (Fig. 4B). Therefore, it was possible that there would be a certain positive feedback mechanism in which the ASK1/JNK/NADPH oxidase system played a central role.

We performed a series of experiments to examine whether this idea was consistent with the facts. If such a positive feedback mechanism around ASK1/JNK were present, delayed PTX treatment would no longer block 22C11-stimulated neuronal death. We thus examined whether death of primary neurons by 22C11, treated for only 6 to 12 h, was inhibited by delayed treatment with PTX. The results, shown in Fig. 4B, revealed that delayed PTX little inhibited neuronal death (which occurred for the next 72 h) caused by both short-term 6-h and 12-h treatment with 22C11. When PTX was simultaneously used with 22C11, PTX blocked 22C11 neurotoxicity, even when neurons were exposed to long-term (72 h) treatment with 22C11 (Fig. 4B), as reported previously using the same system (Hashimoto et al., 2003). Since it takes only 60 to 80 min for PTX to exert its blocking effect on G proteins in cultured systems (Nishimoto et al., 1987), this result suggested that a 60- to 80-min delay of the PTX effect does not affect the blocking effect of PTX on neurotoxicity by anti-APP antibody.

In clear contrast, delayed treatment with a specific JNK inhibitor, SP600125, greatly inhibited subsequent 72-h neurotoxicity caused by both 6-h and 12-h treatment with 22C11. Taken together, these data indicate that continuous JNK activation is required for APP to accomplish neurotoxicity, whereas upstream APP stimulation is no longer necessary once JNK is activated.

Our earlier study (Hashimoto et al., 2003) has shown that 22C11 treatment activates JNK in primary neurons and that simultaneous PTX treatment suppresses 22C11-induced JNK activation. If a positive feedback mechanism is present, we might be able to observe its presence by assessing JNK activation. Using primary neurons, we thus further examined whether sustained activation of this pathway could be observed, as assessed with JNK assay (Fig. 4C). When neurons were continuously treated with 22C11, JNK was continuously activated (Fig. 4C, left upmost panel), as reported previously (Hashimoto et al., 2003). Whereas 12-h prior PTX treatment completely blocked subsequent activation of JNK induced by 22C11 (Fig. 4C, left middle panel), 1-h delayed PTX treatment could no longer affect JNK activation by 22C11 (Fig. 4C, right upmost panel). Because it takes only 1 to 2 h for PTX to exert its effect and the effect of PTX is irreversible, this result suggested that 1) JNK cannot be activated when G_o is inhibited before APP stimulation; and 2) activation of JNK triggered by APP stimulation is sustained, even when G_o is inhibited after initial APP activation. Therefore, to activate JNK, triggering by APP stimulation is essential, but once JNK is activated, JNK activation is sustained without continuous APP stimulation.

In contrast, when neurons were treated with 300 µM APO, 22C11 could only activate JNK transiently (Fig. 4C, left bottom panel). NADPH oxidase is the only enzyme that is inhibited by 300 µM APO, which cannot affect any of the other superoxide-generating enzymes (Hart and Simons, 1992). We confirmed the functional specificity of APO in this effect, by examining the action of 1 mM L-NMMA, a specific NOS inhibitor, which had no effect (Fig. 4C, right middle panel). Therefore, this result suggests that in order for APP to cause sustained activation of JNK, NADPH oxidase would play an essential role. In addition, delayed treatment with PTX could not affect the transient activation of JNK by 22C11 in the presence of 300 µM APO (Fig. 4C, right bottom panel). Combined with the result that prior PTX treatment abolished 22C11-induced JNK activation (Fig. 4C, left middle panel), it is conceivable that G_o activation is essential for transient activation of JNK by APP.

Taken together, these data indicate that JNK activation by APP is sustained even after upstream signal blockade and that the sustained activation of JNK by APP no longer occurs when the downstream NADPH oxidase is inhibited. This result was consistent with the idea that there is a positive feedback mechanism after G_o before NADPH oxidase; that is, in the ASK1/JNK system.