Amyloid Precursor Protein Family-induced Neuronal Death Is Mediated by Impairment of the Neuroprotective Calcium/Calmodulin Protein Kinase IV-dependent Signaling Pathway*

The aberrant metabolism of β-amyloid precursor protein (APP) and the progressive deposition of its derived fragment β-amyloid peptide are early and constant pathological hallmarks of Alzheimer's disease. Because APP is able to function as a cell surface receptor, we investigated here whether a disruption of the normal function of APP may contribute to the pathogenic mechanisms in Alzheimer's disease. To this aim, we generated a specific chicken polyclonal antibody directed against the extracellular domain of APP, which is common with the β-amyloid precursor-like protein type 2. Exposure of cultured cortical neurons to this antibody (APP-Ab) induced cell death preceded by neurite degeneration, oxidative stress, and nuclear condensation. Interestingly, caspase-3-like protease was not activated in this neurotoxic action suggesting a different mode of cell death than classical apoptosis. Further analysis of the molecular mechanisms revealed a calpain- and calcineurin-dependent proteolysis of the neuroprotective calcium/calmodulin-dependent protein kinase IV and its nuclear target protein cAMP responsive element binding protein. These effects were abolished by the G protein inhibitor pertussis toxin, strongly suggesting that APP binding operates via a GTPase-dependent pathway to cause neuronal death.

Alzheimer's disease (AD) 1 is a devastating neurodegenerative disorder characterized by deposition of ␤-amyloid (A␤) plaques, accumulation of intracellular neurofibrillary tangles, and neuronal cell loss (1). A␤ peptide is generated by proteolysis of the amyloid precursor protein (APP), which is the product of a gene located on human chromosome 21 and on mouse chromosome 16. APP is expressed in most mammalian tissues (2,3) and is present in at least 10 different spliced variants (4). In the brain, the major isoform generating A␤ is a peptide consisting of 695 residues that contains a single transmembrane domain (5). Although the physiological role of APP is still unclear, several studies have suggested that it is implicated in important physiological functions of neurons, such as neurite outgrowth, synaptogenesis, cell substrate adhesion and neuronal survival (for review see Ref. 6).
Up to now, most of the research has been focused on the toxicity of A␤ peptide related to AD. A␤ has been reported to exert a variety of toxic effects on neurons both in vitro (7,8) and in vivo (9). However, it has been reported that mice overproducing A␤1-42 extracellularly showed no neuronal loss (10), thus suggesting that A␤ peptide seems not to be the only constituent in neurotoxicity associated to AD. Previous studies have demonstrated that overexpression of full-length APP induced degeneration of postmitotic neurons derived from embryonal carcinoma cells (11). Moreover, intracellular accumulation of wild-type APP in the rat hippocampus caused a specific type of neuronal degeneration in vivo in the absence of extracellular A␤ deposition (12), and viral vector-mediated overexpression of wild-type APP induced apoptosis-like death of neurons both in vivo and in vitro (13)(14)(15). Furthermore, it has been found that the familial AD-associated mutant Val-642 of APP caused neuronal DNA fragmentation independently of A␤1-42 production (16,17). These observations confirm the notion of a role for APP in neurotoxicity associated to AD. In this study we explored this hypothesis by analyzing APP signaling mechanisms.
It has been demonstrated that APP possesses a cytoplasmic Go-stimulating domain that associates with GTP-binding proteins, supporting the idea that APP may act as a cell surface receptor (16, 18 -21). Because the nature of the endogenous ligand that binds APP remains unknown, several studies have used antibodies against different extracellular domains of APP to mimic a ligand-receptor interaction. In this study, we analyzed APP signaling mechanisms activated by antibody binding.
It is largely documented that cysteine proteases are crucial executors of apoptosis (for review, see Ref. 22). Among the protease activated during the neuropathological process such as AD, there are cysteinyl proteases-like caspases and calpains (23,24). Caspases are synthesized as inactive precursors that are proteolytically processed to generate active subunits by cleavage at specific aspartic acid residues. Their roles as mediators of apoptosis in a wide range of the cell types are reported (22). Among the identified caspases, caspase-3 is of particular interest as it appears to be a common downstream apoptosis effector and promotes neuronal death during brain development (25). Calpain is an intracellular cysteine protease, constitutively expressed in all vertebrates in which they are highly conserved across species and have been characterized in a wide variety of cell types and tissues. Calpains are considered to participate in various signaling pathways mediated by Ca 2ϩ through modulating the activities and/or functions of other proteins (26). Two ubiquitous isozymes are known,and m-calpains (also called calpains I and calpains II, respectively), which differ in their sensitivity to Ca 2ϩ . Calpains have been reported previously to be activated under oxidative stress conditions in P19, PC12 cells, and in cerebellar granule cells (27)(28)(29). We have previously reported that calcium/calmodulin-dependent protein kinase IV (CaMKIV) plays an anti-apoptotic role in cerebellar granule cells (30). CaMKIV is a serine/threonine kinase that phosphorylates a large variety of substrates and is involved in the regulation of gene transcription (31,32). Its major nuclear target protein is the cAMP responsive element binding protein (CREB). Both CaMKIV activity and CREB phosphorylation on its serine 133 residue promote neuronal survival (30,33,34).
In this study, we generated and used a chicken polyclonal antibody (APP-Ab) against the extracellular residues 66 -81 localized in the cysteine-enriched, NH 2 -terminal domain of APP (5) to investigate the type of cell death and the intracellular signaling pathways induced by APP-Ab treatment of cortical neurons. We show here that APP-Ab induces neuronal death with neurite disorganization, nuclear condensation, and production of reactive oxygen species (ROS). Both calpain and phosphatase inhibitors promoted cell survival and restored CaMKIV and CREB levels.
Cell Culture-Cultures of cortical neurons were prepared from FVB mice embryos (day 18 of gestation). Briefly, after trypsin/DNase treatment and mechanical dissociation, dispersed cortical neurons were plated onto polyornithine-coated, 96-, 12-, or 6-well plates depending on the experimental protocol. For immunocytochemistry, Hoechst staining, ROS-labeling and transfection experiments, cells were plated on 12-mm glass coverslips treated likewise. Cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 0.5 M insulin, and 50 g/ml gentamicin. After 1 day in culture, neurons were switched to defined medium without serum (Dulbecco's modified Eagle's medium, 0.5 M insulin, 50 g/ml gentamicin, 1 nM T3, 60 M putrescine, 100 M transferrin, and 30 nM sodium selenite), and experiments were performed on day 4. Concentrations of test substances and chemical compounds and the corresponding incubation times are indicated in the figure legends.
Primary Culture from Knock-out Embryos-Embryos were obtained from APP Ϫ/Ϫ mice or intercrosses of APP ϩ/Ϫ APLP2 Ϫ/Ϫ mice as described (35). Each embryo was dissected and plated separately because of their different genotypes, and all the survival test were performed on cultures from individual embryos. Embryos were genotyped by PCR on DNA obtained from tail biopsies of each embryo as previously described (35,36). Wild-type control mice were obtained from 129 Sv(ev) ϫ C57B6 mating and processed in parallel.
Colorimetric MTT Assay-Mitochondrial activity was measured as an indicator of cell viability by a modified procedure of the original method described by Mossmann (37). After treatment with APP-Ab or SSP (in 96-well culture plates), defined medium was changed to a freshly prepared medium containing 0.5 mg/ml MTT. After 1 h of incubation at 37°C, MTT-containing medium was removed, and dark blue crystals formed during reaction in viable cells were dissolved by adding 0.04 N HCl in isopropanol. Plates were read on a Metertech S960 micro-enzyme-linked immunosorbent assay platereader, using a test wavelength of 490 nm and a nonspecific wavelength of 650 nm for background absorbency. Results are presented as a percentage of survival taking control condition as 100% (defined medium without any treatment).
Immunocytochemistry-Cell staining was performed using an indirect immunofluorescence technique. Briefly, cells were fixed with 4% paraformaldehyde for 15 min and permeabilized with 1% Triton X-100 plus 3% bovine serum albumin in phosphate-buffered saline (PBS) for 30 min. A monoclonal antibody against the neuronal cytoskeletal protein MAP2 was diluted 1/500 in 3% bovine serum albumin-PBS. After overnight incubation at 4°C, cells were incubated with carbocyanine-3-coupled donkey anti-mouse-IgG diluted 1/200 in 3% bovine serum albumin-PBS for 1 h at room temperature. After washing, cells were mounted with Fluoromount and stored at 4°C until observation.
Hoechst Staining-Quantification of apoptotic cells was determined by direct visual counting after nuclear staining of 4% paraformaldehyde-fixed cells with the fluorescent probe Hoechst 33342 (1 g/ml) for 30 min at room temperature. Hoechst fluorescence was visualized with an AMCA filter (excitation, 350 mm; emission, 450 mm), and only those neurons containing clearly fragmented nuclei or condensed chromatin were scored as being apoptotic. One hundred neurons are examined per field and five randomly selected fields per experimental condition were analyzed with a 20ϫ objective.
Measurement of Caspase-3 Activity-Caspase-3 activity was determined by measuring the release of AMC from the caspase-3 substrate Ac-DEVD-AMC as previously described by Przywara et al. (38). Briefly, cells cultured in 6-well plates were washed with PBS and lysed on ice with 50 mM Tris, pH 7.5, containing 0.5% igepal NP-4, 0.5 mM EDTA, and 150 mM NaCl. 50 l of lysate was added to a reaction mixture containing reaction buffer (20 mM HEPES, pH 7.5, 50 mM NaCl, and 2.5 mM dithiothreitol) plus 50 M of Ac-DEVD-AMC. After incubation for 1 h at 37°C, the reaction was stopped by adding 10ϫ ice-cold lysis buffer. Fluorescence of free AMC was determined using a PerkinElmer Life Sciences HTS 7000 Bioassay reader (Foster City, CA) at excitation and emission wavelengths of 360 and 465 nm, respectively. One part of the lysates was used for protein determination by the bicinchoninic acid protein assay (Pierce). A standard curve of known AMC concentrations allowed conversion of fluorescence values into picomoles of released AMC/min. All presented results were obtained after 1 h of enzymatic reaction, which is in the linear phase as determined by kinetics measurements (not shown).
DNA Laddering-Cortical neurons cultured in 10-cm-diameter plates were treated either with 5 g/ml APP-Ab or 10 M SSP. After 24 h of treatment, neurons were washed with PBS and lysed with 20 mM Tris-HCl, pH 7.5, containing 20 mM EDTA and 0.4% Triton X-100 for 20 min on ice. Centrifugation at 15,000 rpm eliminated cell debris. The supernatant was incubated with a buffer containing 150 mM NaCl, 100 mg/ml proteinase K, 10 mM Tris-HCl, pH 8, 40 mM EDTA, and 1% SDS for 1 h at 45°C. After a phenol/chloroform (v/v) treatment, the upper phase was precipitated with sodium acetate and washed with ethanol. Lysates were treated with 1 mg/ml RNase for 30 min at 37°C. DNA samples were resuspended in TE and electrophoresed on a 2% agarose gel.
Measurement of ROS Production-ROS were detected with H2DCF-DA, which produces a green fluorescence when oxidized (39). After the indicated incubation times, cells were loaded with 10 M H2DCF-DA for 30 min at 37°C and immediately observed under a microscope with a 20ϫ objective. The number of both fluorescent and total cells was determined in five randomly selected fields per experimental condition.
Western Blot Analysis-Cells from six-well culture plates were washed three times with ice-cold PBS and lysed in 20 mM Tris-HCl containing 150 mM NaCl, 2 mM EDTA, 1% Triton-X100, and protease inhibitors (10 g/ml leupeptin, 10 g/ml pepstatin, and 1 mM phenylmethylsulfonyl fluoride). To standardize gel loading, protein concentrations were determined using the bicinchoninic acid assay. Samples were then supplemented with ␤-2-mercaptoethanol and denatured for 3 min at 100°C. Equal amounts of protein (20 g) were separated with SDS-PAGE and transferred to nitrocellulose membranes. To ensure a correct protein transfer, membranes were colored with Ponceau Red. After washing, nonspecific labeling was blocked with 10% blotto (10% non-fat dry milk in 150 mM Tris-HCl, pH 7.5, 50 mM NaCl, and 0.05% Tween 20) for 1 h at room temperature. Membranes were then incubated overnight at 4°C with appropriate primary antibodies diluted in 3% blotto. After washing, membranes were incubated for 2 h at room temperature with horseradish peroxidase-conjugated secondary antibodies diluted 1/2000 for sheep anti-mouse IgG or 1/5000 for goat anti-rabbit IgG. Blots were revealed by ECL (Amersham Biosciences). Experiments were performed at least three times, and representative blots are shown. The relative optical density of the immunoreactive bands was quantified using the Bio-Rad analysis software Multi-Analyst.
Gene Transfer-The expression vector used was pSV2-CREB-VP16 (a generous gift from Dr. C. Bancroft, Mount Sinai School of Medicine, New York, NY), which encodes a chimera protein where CREB-(1-341) is fused to the transcription activator region of VP16, thus generating a constitutively active CREB protein (40). Neuronal gene transfer was performed as reported previously, using polyethyleneimine 25 kDa (PEI 25K, Sigma) as DNA carrier (41). After 4 days in vitro, cell cultures on glass coverslips placed in 12-well plates, were transfected with 1.5 g/ml plasmid. DNA plasmid containing the expression construct was mixed with the EGFP-expression vector (CLONTECH) in a ratio of 3/1 by weight and diluted in 150 mM NaCl. This mixture was then mixed with the PEI, and after 10 min of incubation, the DNA/PEI mixture was added to the cells for 30 min. One day after transfection and recovery in fresh medium, cells were treated with APP-Ab. After 24 h of APP-Ab treatment, cells were fixed (4% paraformaldehyde), and Hoechst 33342 was added (1 g/ml). The GFP-positive cells were inspected by an observer who was unaware of the treatment condition being evaluated. The "blind" observer scored each fluorescent cell as apoptotic or not based on nuclear morphology (visualized by Hoechst fluorescence). We estimate that ϳ0.1-1% cortical cells were GFP-positive.
Statistical Analysis-Data are expressed as means Ϯ S.E. values. Statistical significance was assessed by means of one-way analysis of variance followed by the Newman-Keuls multiple comparison test. Differences were considered significant at p Ͻ 0.05.

APP-Ab Mediated Neuronal
Death-In the present study, biological effects evoked by application of a chicken polyclonal anti-APP were investigated in primary cultures of cortical neurons. The APP-Ab was generated in chicken against residues 66 -81 of APP localized in the cysteine-enriched, NH 2 -terminal domain (5). This antibody revealed an immunoreactive band migrating on SDS-PAGE at 120 kDa, which corresponds to APP, at the same position as recognized by the 22C11 antibody, a commercially available monoclonal anti-APP antibody (Fig.  1a). Cell viability studies showed a toxic effect induced by this antibody in dissociated cortical neurons. Fig. 1b illustrates a dose-dependent effect on cell death after 24 h of treatment, as determined by MTT assay. Five and 10 g/ml provoked a significant increase in cell death (50 and 55%, respectively), whereas lower doses exhibited little effect. The 5 g/ml dose induced a time-dependent cell death that was detected early at 12 h and became significant at 24 and 48 h of culture (50 and 81%, respectively) (Fig. 1c). Based on these results, and unless otherwise indicated, neurons were treated with 5 g/ml APP-Ab for 24 h in further experiments. To determine the specificity of APP-Ab-induced cell death, its effect was com-pared with different positive and negative controls (Fig. 1d). Preincubation of APP-Ab with the synthetic peptide containing its antigenic epitope significantly attenuated neuronal damage. As positive control, we used the 22C11 monoclonal antibody. It provoked 52% of cell death under these experimental conditions. A negative control, the mouse monoclonal antibody Jonas, which reacts with the cytoplasmic carboxyl-terminal end of intact APP, had no effect on neuronal survival. These data suggest that the action of APP-Ab on APP triggers neuronal cell death through a specific interaction with its extracellular domain. SSP, a protein kinase inhibitor known to induce neuronal apoptosis (42), caused 70% of cell death in our experimental conditions. This control is included in all further experiments.
APP-Ab Mediated Neuronal Death in Knock-out Mice-APP belongs to a multigene family that also includes two other homologs named amyloid precursor-like proteins 1 and 2 (APLP1 and APLP2) (4,(43)(44)(45). Because APP and APLP2 present a high sequence homology (46), we asked whether APP-Ab is able to induce programmed cell death through both proteins. To this aim, we tested our antibody in cortical neuron cultures derived from APP and/or APLP2 knock-out embryos. Microscopic observation of these cultures revealed that neurons from knock-out mice were viable and presented a normal morphology with an extended neurite network and a non-condensed soma as compared with wild-type cultures (not shown). Fig. 1e shows the analysis of cell survival in the different knock-out embryos cultured individually. APP-Ab treatment in APP Ϫ/Ϫ -APLP2 ϩ/ϩ neurons induced a degree of cell death similar to wild-type neurons. This suggests that the presence of APLP2 is sufficient to mediate the antibody-induced toxicity. In contrast, a gradual restoration of cell survival was observed to be proportional to the loss of APP, APLP2, or both, being completely restored in double knock-out APP Ϫ/Ϫ -APLP2 Ϫ/Ϫ neurons. Conversely, the survival response of APP Ϫ/Ϫ -APLP1 Ϫ/Ϫ neurons was not different from APP Ϫ/Ϫ alone (data not shown), suggesting that APLP1 is not recognized by the APP-Ab. These findings indicate that APP-Ab induces cell death by activating independently APP and APLP2 signaling pathways.
APP-Ab Induces Neurite Degeneration-To determine whether APP-Ab causes dystrophic changes in cortical neurons, we used a specific monoclonal antibody against the neuronal cytoskeletal protein MAP-2. Significant differences in MAP-2 staining were found between control and APP-Ab-exposed cells (Fig. 2). The neuritic network extended by cortical neurons showed progressive signs of degeneration in the presence of APP-Ab, and many neurons presented fragmented neurite after 12 h of treatment as compared with the healthy morphology exhibited by control neurons. SSP, used as a positive control, induced a severe degeneration of the neuritic network. Preincubation of APP-Ab with its epitope peptide reduced signs of neurite dystrophy, corroborating the specificity of the toxic effect of APP-Ab. Therefore, exposure of cortical neurons to APP-Ab induces significant morphological changes that precede neuronal death.
APP-Ab-induced Programmed Cell Death (PCD) Is Distinct from Apoptosis-To investigate the type of cell death induced by APP-Ab treatment, we examined several characteristics of PCD. First, we monitored nuclear morphology by Hoechst staining. As illustrated in Fig. 3a, nuclei of untreated neurons are large and round exhibiting an uniform staining, whereas an increasing number of nuclei with strong chromatin condensation or clearly fragmented appeared after APP-Ab treatment (Fig. 3b). Preincubation of APP-Ab with its epitope peptide prevented nuclei condensation (Fig. 3c). As expected, SSP treatment also produced nuclear features of PCD (Fig. 3d). Quantitative analysis of these observations is shown in Fig. 3e. In addition, the proportion of condensed and fragmented nuclei in the presence of APP-Ab was time-dependent (Fig. 3f). However, when fragmented and condensed nuclei where scored separately, APP-Ab and SSP showed quantitatively distinct effects. Indeed, APP-Ab induced a large range of condensed nuclei (88%) compared with SSP (67%). Conversely, SSP induced more fragmentation (33%) than APP-Ab (12%) (Fig. 3g). This suggests that APP-Ab-induced cell death is different from the classical apoptosis induced by SSP.
Apoptosis is a cell death mechanism that requires a specialized cellular machinery, including the activation of a family of cysteine proteases called caspases (for review see Ref. 47). Among these, caspase-3 has been proposed as a key mediator of mammalian apoptosis in particular in the central nervous system (48). To determine more precisely the type of neuronal death induced by APP-Ab, we measured caspase-3-like protease activity in our culture system using the fluorogenic sub-strate Z-DEVD-AMC. As depicted in Fig. 4a, caspase-3-like activity, which was detected in control neurons at relatively high levels, did not increase after APP-Ab treatment for 6, 18, or 24 h. In contrast, SSP induced a marked increase in the enzymatic activity as soon as after 6 h of treatment. To obtain independent evidence of these findings, we analyzed by Western blot the levels of pro-caspase-3 protein (CPP32) and the active fragment of caspase-3 (p20) (Fig. 4b). No variation in CPP32 or p20 generation was detected whatever the time of APP-Ab treatment as compared with control. In contrast, in SSP-treated cells, a significant decrease of the full-length form of caspase-3 (32 kDa) and a marked increase of the active p20 fragment was found, confirming the measurements of enzymatic activity. ␣-actin was used as internal control of protein levels. These results argue against a major role of caspase-3 in APP-Ab-induced neuronal death. Finally, we evaluated internucleosomal DNA fragmentation, a major hallmark of apoptosis. Fig. 4c shows that APP-Ab failed to induce any laddering typical of apoptosis at 24 h after treatment, even though obvi- ous cell death was observable at this point and only an accumulation of large DNA fragments was detected. In contrast, treatment with SSP induced a clear DNA ladder, a typical pattern of apoptosis. These data, together with the observation that APP-Ab does not induce the same nuclear modifications (condensed versus fragmented nuclei) than SSP, suggests that APP-Ab induces a type of PCD different from classical apoptosis.
APP-Ab Induces Oxidative Stress and a Loss of Neuroprotective Proteins-ROS are considered as important effectors of PCD (49). Increased oxidative damage to macromolecules has been reported in brain tissue of AD patients (50). Moreover, APP has been shown to modulate oxidative stress in primary neuronal cultures (51). Here, we examined whether APP-Ab induces ROS generation by quantifying the number of cells labeled with the fluorescent dye H2DCF-DA. Fig. 5a shows the increase in the amount of ROS-positive cells when treated with either APP-Ab or SSP in a time-dependent manner. A maximum was observed after 18 h of APP-Ab treatment, a time point that precedes cell death. ROS have been reported to induce a loss of two important proteins involved in cell survival: CaMKIV and its nuclear substrate CREB, both of which play an important role in neuroprotection (29,30). To determine whether the levels of CaMKIV are modulated after APP-Ab treatment, we measured the amount of CaMKIV protein present in neurons at different time points. As shown in Fig. 5b, the anti-CaMKIV antibody revealed a band at 66 kDa that decreased with time of treatment (see inset). Quantitative analysis showed that the progressive decline of CaMKIV levels was visible after 12 h of treatment and became dramatic at 24 and 48 h. This decrease in the amount of the neuroprotective protein may account for the mechanism of APP-Ab-induced cell death. The functional consequence of CaMKIV degradation was evaluated on the phosphorylation state of CREB, which is specifically phosphorylated by CaMKIV on the serine 133 residue (52,53). Quantitative analysis by Western blot with an antibody directed against the phosphorylated form of CREB (p-CREB) showed that the levels of p-CREB decreased in a time dependent manner in response to APP-Ab treatment with a kinetic comparable with those observed for CaMKIV degradation (Fig. 5c). We further measured the levels of the total amount of native CREB (N-CREB) regardless of its phosphorylation state. APP-Ab induced a loss of N-CREB in a time dependent manner. The time course analysis showed that the loss of p-CREB precedes the loss of total N-CREB protein levels. Interestingly, a decrease in p-CREB levels was observed 18 h after treatment, whereas the decrease in N-CREB was significant only at 24 h, suggesting that independent mechanisms are involved in the loss of N-CREB and the dephosphorylation process. The fact that p-CREB levels declined before N-CREB suggests a specific dephosphorylation mechanism that precedes the loss of total CREB. To test the potential role of activated CREB in neuronal survival, in our experimental model, we performed transient co-transfection experiments with a plasmid containing a dominant active mutant of CREB and the EGFP expression plasmid. To generate a constitutively active CREB protein, CREB is fused to the transcription region of VP16. EGFP-expressing vector was used to identify transfected cells. The effects of CREB mutant overexpression were monitored by scoring the alteration of nuclear morphology in the transfected cell population (EGFP positive). Neurons cotransfected with the EGFP vector and an expression vector without any insert served as controls in the evaluation of cells displaying condensed nuclei. This activated CREB mutant prevented APP-Ab-induced chromatin condensation (Fig. 6, e and f) as compared with APP-Ab-treated cells co-transfected with the empty vector (Fig. 6, c and d) in which nuclei are condensed. Non-treated cells presented a normal nucleus morphology (Fig. 6, a and b). Quantitative analysis showed that the presence of the CREB mutant reduced significantly the number of condensed nuclei (Fig. 6g), indicating that transcription-ally active CREB (p-CREB) exerts neuroprotective effects against APP-initiated PCD.
APP-Ab-dependent PCD Recruits Calpain and Calcineurin-Because caspase-3 does not appear to be involved in APP-Abinduced proteolytic events and subsequent cell death, we investigated the possible involvement of calpain, another member of the cysteine protease superfamily that has been shown to participate in PCD (54). Inhibition of calpain activity using the specific synthetic inhibitor CII partially prevented both cell death and ROS production induced by APP-Ab (Fig. 7,  a and b, respectively). These data suggest that calpains operate upstream of ROS production and that they inactivate an as yet unidentified protein implicated in the control of ROS production. We checked CII effects on CaMKIV and CREB degradation. As seen in Fig. 7c, treatment with CII also reduces CaMKIV and CREB loss. One of the phosphatase implicated in CREB dephosphorylation is the Ca 2ϩ /calmodulin-dependent serine/threonine phosphatase 2B (calcineurin) (55), which can also mediate apoptosis (56). Treatment of cortical neurons with the calcineurin inhibitor cyclosporin A protected against APP-Ab-induced cell death (Fig. 7a) and inhibited ROS generation (Fig. 7b). Interestingly, this treatment inhibited the proteolytic degradation of N-CREB and was sufficient to restore the levels of transcriptionally active CREB (p-CREB) (Fig. 7c). ␣-actin was used as an internal control of gel loading. To further investigate the contribution of calcineurin in the control of APP-Ab-dependent proteolytic events, we analyzed the expression of CaMKIV under the same experimental conditions. As shown in Fig. 7c, the decrease in CaMKIV immunoreactivity induced by APP-Ab was restored by addition of cyclosporin A. Taken together, these data suggest that CaMKIV and CREB play an important role in promoting survival of cortical neurons and that the loss of its activity, likely through calpain-and calcineurin-dependent mechanisms, is involved in APP-Abinduced cell death.
APP-Ab-inducedCellDeathInvolvestheActivationofaGTPasedependent Pathway-Our present findings strongly suggest that APP-Ab acts as a ligand of APP/APLP2 to induce neuronal cell death through a cell surface receptor-initiated transduction mechanism. To support this hypothesis, we examined the effect of PTX, a specific inhibitor of heterotrimeric G proteins, on APP-Ab-induced cell death. PTX restored cell survival in the presence of APP-Ab (Fig. 8a), and prevented CaMKIV and CREB down-regulation (Fig. 8b), which further reinforces the notion that APP binding and the subsequent activation of a PTX-sensitive, GTPase-dependent pathway affects specific CaMKIV/CREB-regulated transcriptional events that ultimately lead to cell death.
The cAMP-dependent Signaling Pathway Decreases APP-Abinduced Cell Death-The cAMP-dependent kinase signaling transduction pathway plays an important role in the regulation of neuronal development, survival, and neuroprotection (57)(58)(59). We examine whether elevating cAMP and activating PKA could counteract APP-Ab-induced cell death. In our model, increasing intracellular cAMP levels by the adenylate cyclase activator forskolin or the cAMP analogue 8-bromo-cAMP inhibited neuronal death provoked by APP-Ab treatment (Fig. 9a). This was also true for ROS production because forskolin or 8-bromo-cAMP was able to strongly attenuate the amount of ROS-positive cells induced by APP-Ab treatment (Fig. 9b). We further analyzed the effect of elevated cAMP levels on CaMKIV and CREB levels. Both stimulators of the cAMP/PKA-dependent pathway prevented APP-Ab-induced CaMKIV and CREB down-regulation (Fig. 9c), therefore indicating the protective role of such pathway against the neurotoxic APP binding. FIG. 4. APP-Ab does not activate caspase-3. a, caspase-3-like activity was measured by degradation of the fluorogenic substrate Z-DEVD-AMC in extracts of cortical neurons cultured in the absence or presence of 5 g/ml APP-Ab, 10 M SSP, or the caspase-3 inhibitor 10 M Z-DEVD-fmk at the indicated times. Data are means Ϯ S.E. from three independent experiments. ***, p Ͻ 0.001 versus control. b, cortical neuron extracts were submitted to Western blot analysis with an antibody directed against the caspase-3 full-length protein (32 kDa) and with an antibody specifically directed against the active p20 fragment of caspase-3. Cells were treated with either 5 g/ml APP-Ab for the indicated times or 10 M SSP for 24 h. Blots were reprobed with an anti-␣-actin antibody to ensure equal protein loading. c, neurons were collected and prepared for a DNA fragmentation assay as described under "Experimental Procedures." A 100-bp DNA ladder is used as molecular weight marker. Clear areas are the migration fronts of the respective dyes.

APP-Ab-dependent PCD Is Distinct from Apoptosis-This
study demonstrates that a polyclonal antibody directed against the extracellular domain of APP/APLP2 induces cortical neuron death. APP belongs to a multigene family that includes the two amyloid precursor-like proteins 1 and 2 (APLP1 and APLP2) (4,(43)(44)(45). Most of the functional domains and structural motifs found in APP are also present in APLPs (particularly APLP2) (60), suggesting that they may share common biological functions in neurons. Using the monoclonal antibody 22C11, several studies have reported that this antibody cannot discriminate between APP and APLP2 (45,60,61). The antibody that we generated is also likely to recognize both APP and APLP2 (but not APLP1) because its death-promoting effect is only suppressed in neurons obtained from mice where both genes have been deleted. Because the antibody induces PCD in cells where either APP or APLP2 is deleted our data clearly show that both APP and APLP2 are able to induce independ-ently programmed neuronal death (Fig. 1e).
The antibody-induced cell death is preceded by neurite degeneration because many of the MAP-2 immunoreactive neurons exposed to APP-Ab showed signs of neurite damage after 12 h of treatment, when 88% of the cells were still alive as judged by their nuclear morphology. Moreover, labeling of living cells with the fluorescent vital dye calcein-AM revealed in our culture system the early presence of neurite disorganization, while at the same time nuclei appeared normal (data not shown). Whether neurites degenerate as a result of a general process affecting the whole cell or because APP-Ab induces a local, deleterious effect on the neurite network remains to be determined. Previous studies have shown that beading and degeneration of neurites are early events occurring in response to a variety of death insults including A␤ (62). In addition, it has been demonstrated that neurite beading by mechanical stress or other insults involves changes in the organization of the neurite cytoskeleton, likely through activation of local pro- or presence of 5 g/ml APP-Ab (close circles) or 10 M SSP (triangles) for the indicated times. b, quantitative analysis of the CaMKIV immunoreactivity as determined by Western blot after treatment of cortical neurons with 5 g/ml APP-Ab for the indicated times. A typical autoradiogram is shown (inset). c, Western blot analysis of p-CREB and N-CREB protein levels after treatment of cortical neurons with 5 g/ml APP-Ab for the indicated times. Blots were reprobed with an anti-␤-actin antibody to ensure equal protein loading. The graph shows the quantitative analysis of the decrease in CREB immunoreactivity. Data represent means Ϯ S.E. of three to four independent experiments. *, p Ͻ 0.05; ***, p Ͻ 0.001 versus corresponding control. teolytic mechanisms (63). It is thus possible that such processes occur in cortical neurons as we show here that calpains are involved in APP-Ab-induced cell death.
We show in this study that APP-Ab triggers a programmed cell death mechanistically different from classical apoptosis. SSP-induced death, used as a control for this study, occurred with typical chromatin condensation and fragmentation associated with early caspase-3 activation and DNA cleavage at nucleosome size. In APP-Ab-treated cells, we also found chro-matin condensation but little fragmentation, and no nucleosomal DNA cleavage was observed (Figs. 3 and 4). In addition, no increase of caspase-3 activity was detected, despite the presence of a basal activity that could be completely blocked by the specific caspase-3 inhibitor Z-DEVD-fmk (data not shown). However, these findings do not exclude that other caspases might be involved. For example, it has been shown that neuronal apoptosis induced by A␤ in hippocampal neurons is mediated by caspase-8 (64). Also, caspase-6 is implicated in apoptosis of serum-deprived primary cultures of human neurons (65). In our model, preliminary experiments did not reveal any caspase-6 activity (data not shown). Recent reports show that many populations of neurons are able to exhibit normal amounts of PCD in the absence of caspase-3 and that the morphology of these degenerating neurons differs from the more typical apoptotic type of cell death (66,67). Studies using caspase inhibitors and genetically modified mice demonstrated that caspases have diverse functions that may overlap in some cell types. The tissue-specific phenotypes of mice lacking individual caspases suggest tissue-and cell type-specific functions for caspases (68). Other recent reports indicate that caspasedependent and caspase-independent pathways may mediate neuronal death in specific brain regions and cell types at specific ages (69). Some authors have suggested the existence of different types of caspase-independent programmed cell death. This type of cell death seems to be physiologically relevant because it has been observed during development (70) and in AD (71). This large body of data may account for apparently conflicting data dealing with the APP-dependent death. Rohn et al. (72) have clearly shown the neuroprotective effect of caspase inhibition (Z-VAD) against APP-mediated neuronal death. Because we show here a basal caspase-3 activity but no further increase in the presence of APP-Ab, it is conceivable that this basal activity, insufficient by its own to trigger cell death, participates to the PCD execution phase when additional APP-dependent signaling mechanisms come into play. Such a situation is not without precedent because recent findings indicate clearly that caspase activation can also occurs in neurons under conditions in which cells do not die (73).
APP Signal Transduction-To date, little is known about the normal functions of APP. Nevertheless, studies on the effects of anti-APP antibodies in primary neurons suggest that APP has a physiological relevance in normal conditions. Recently, several groups have independently reported that APP may act as a cell surface receptor and that APP actions could be mediated through an interaction with the GTP-binding protein G 0 (19,21,74). Interestingly, it has been shown that the G proteincoupling function of wild-type APP can be stimulated by the three Val-642 mutations linked to familial AD (75). Moreover, expression of familial AD mutants of APP causes apoptosis through PTX-sensitive G proteins. It was shown that these mutants induce nuclear fragmentation, which is independent of A␤ production and mediated by the G␤ 2 ␥ 2 complex of G 0 (16,20). These findings suggest that APP has an intrinsic signaling function in the cell. Accordingly, we found that in cortical neurons, the G 0 /G i inhibitor PTX was able to protect cells against APP-Ab treatment. Moreover, PTX was also able to counteract APP-induced signaling. It inhibited the decrease in CaMKIV and CREB protein levels induced by the antibody. It is noteworthy that these results are in good agreement with those obtained by Sudo et al. (76), showing that the death of F11/APP-transfected cells triggered by the antibodies against APP ectodomain is also mediated by a PTX-sensitive GTPase.
APP Recruitment and Oxidative Stress-Oxidative stress has been implicated in a number of age-associated disorders and neurodegenerative diseases. Numerous studies have clearly pointed out the importance of an oxidative imbalance in these diseases including AD. This oxidative hypothesis is supported not only by the presence of oxidative markers in AD brain (77,78) but also by studies demonstrating that A␤ toxicity is associated with oxidative stress (79,80). In our study, APP-Ab treatment increased ROS production, supporting a role for oxidative stress in mediating the activation of APP. This result is in agreement with a recent study reporting the ability of an antioxidant to prevent the morphological changes associated with the cell death process induced by the APPdirected 22C11 antibody (72). In addition, in our model APP-Ab-induced ROS production was also inhibited by the ROS scavenger lipoic acid, though neuroprotection was not complete due to an intrinsic toxicity of this compound. 2 Extending the observation of Rohn et al. (72), we show by a detailed kinetic analysis that ROS production is not an early signaling mechanism of APP. It immediately precedes changes in nuclear morphology characteristic of PCD but follows the earliest detectable signs of proteolytic as dephosphorylation activity. These temporally organized events suggest that ROS are primarily recruited for the execution phase of PCD and may be controlled by rapidly activated signals, including proteases and phosphatases. In line with this interpretation, we show that both calpain and calcineurin inhibitors (that protect against PCD) strongly reduce oxidative stress. Taken together, our findings show that PCD induced by APP-Ab binding is mediated, at least in part, by ROS production in cortical neurons.
Inactivation of the CaMKIV/CREB-dependent Neuroprotective Pathway-Several studies document that CaMKIV promotes the survival of a variety of cell types including T cells (81), Purkinje cells (82), and cerebellar granule cells (30). CaMKIV mediates Ca 2ϩ -dependent stimulation of gene transcription through phosphorylation of the transcription factor CREB at serine 133 (52,83). Here, APP-Ab treatment induced a decrease of CaMKIV levels, which progressively declined in parallel with the loss of CREB. The early loss of this neuroprotective signaling pathway, before the appearance of a significant cell death, reinforces the idea that it is involved in the maintenance of survival-promoting mechanisms. In support of this hypothesis, overexpression of a dominant active form of CREB, which is constitutively active, reversed APP-Ab-induced PCD. Dismantling of this neuroprotective pathway through proteolytic degradation of CaMKIV and CREB seems to be a major mechanism that significantly contributes to the PCD signaling pathway activated by APP-Ab. Impairment of CREdependent transcription may directly induce PCD through yet unknown mechanisms. This PCD-inducing mechanism may also be indirect by impairing of the transcription of neuroprotective genes. Such neuroprotective genes might include the gene encoding brain-derived neurotrophic factor (84,85) or the antiapoptotic protein Bcl-2 (86), that are both positively regulated by CREB.
The inactivation of the CaMKIV/CREB signaling pathway in response to APP-Ab rests on two distinct signaling mechanisms. First, APP-Ab alters the state of phosphorylation of the cell by recruiting calcineurin, a Ca 2ϩ -dependent phosphatase that is abundantly expressed in the central nervous system especially in neurons vulnerable to ischemic and traumatic insults (for review see Ref. 87). Through this rapid mechanism, APP-Ab dephosphorylates p-CREB, thus converting it into a transcriptionally inactive state prior to its proteolytic degradation. Second, permanent inactivation is achieved through proteolytic events that are sensitive to calpain inhibition. Interestingly, these two events, dephosphorylation and proteolysis, must be activated in concert to induce PCD. Indeed, inhibition of one or the other (by either calpain or calcineurin inhibitors) is sufficient to block oxidative stress and PCD. The molecular link between the initial transduction step of APP binding and protease/phosphatase recruitment has not been studied here. However, a possible common mechanism might be an increase of intracellular free Ca 2ϩ levels because both calpain and calcineurin are activated by this second messenger and increased Ca 2ϩ levels are associated with PCD in a variety of cell types and experimental conditions (88,89).
The cAMP/PKA-dependent pathway has been proposed to promote neuronal survival (57,90). We show here that stimulation of this pathway reverses indeed the death-promoting effects of APP-Ab, confirming the neuroprotective role of the cAMP/PKA-dependent pathway in cortical neurons, but also showing that APP signaling can be modulated by other intracellular pathways. This functional cross-talk between APP and the neuroprotective cAMP/PKA pathway may have important physiological implications. One may speculate that within a functional neuronal network, trans-synaptic inputs up-regulating cAMP signaling inhibit APP-induced PCD. Conversely, the lack of these inputs might be sufficient to liberate the deleterious potential of APP and thereby contribute to AD pathogenesis.
The present study shows the cell death-promoting effect of an anti-APP/APLP2-directed antibody through impairment of specific CREB-dependent neuroprotective pathways. Fig. 10 represents schematically the different events that lead to PCD, which is preceded by neurite degeneration. Binding of the FIG. 10. APP and APLP2 recruit multiple death promoting pathways in cortical neurons. a, in normal conditions, electrical activity and neurotrophic inputs support neuronal survival through activation of CaMKIV. This, in turn, maintains CREB in a phosphorylated state that induces the transcription of neuroprotective genes. Similarly, the cAMP/PKA-dependent pathway also contributes to this neuronal survival. b, APP binding by interacting with the signaling molecule Go, recruits different death effectors ( proteases, phosphatases, and ROS) that impair the stability of these protective pathways thus leading to cell death. Within a given normal network, the cAMP/ PKA-dependent pathway, by inhibiting ROS generation and restoring the levels of phosphorylated CREB, may counterbalance this APP-Abinduced cell death. antibody to APP, or to its very related homolog APLP2, stimulates a GTPase-dependent mechanism that activates at least two death effectors acting in a concerted way. First, the induction of a calpain-dependent protease activity is responsible for the degradation of important neuroprotective molecules, particularly CaMKIV and its target CREB. Second, activation of specific phosphatases, such as calcineurin, affects the phosphorylation state of CREB, thus contributing to reduce the protective potential of CREB. In parallel, protease activity may operate to produce ROS, which constitute by themselves a highly powerful signal of neuronal death. Within a neuronal network, neurons may recruit other survival mechanisms, such as the cAMP/PKA-dependent pathway, to inactivate the death effectors and restore the normal levels of phosphorylated CREB. Taken together, the findings presented here strongly suggest that CREB is a key player in neuronal survival and that a disruption of the normal signaling of APP may represent an additional cause of AD neurodegeneration.