α-Secretase-derived Fragment of Cellular Prion, N1, Protects against Monomeric and Oligomeric Amyloid β (Aβ)-associated Cell Death*

Background: Cellular prion undergoes α-secretase cleavage, yielding N1. We examined whether N1 protects against Aβ monomers and oligomers. Results: N1 protects against Aβ monomers and oligomers prepared from APP-London-expressing human cells and Alzheimer disease-affected brains. Conclusion: N1 could protect from Aβ-associated toxicity at the early asymptomatic phase of Alzheimer disease. Significance: These data emphasize the cross-talk between PrPc and βAPP catabolites. In physiological conditions, both β-amyloid precursor protein (βAPP) and cellular prion (PrPc) undergo similar disintegrin-mediated α-secretase cleavage yielding N-terminal secreted products referred to as soluble amyloid precursor protein-α (sAPPα) and N1, respectively. We recently demonstrated that N1 displays neuroprotective properties by reducing p53-dependent cell death both in vitro and in vivo. In this study, we examined the potential of N1 as a neuroprotector against amyloid β (Aβ)-mediated toxicity. We first show that both recombinant sAPPα and N1, but not its inactive parent fragment N2, reduce staurosporine-stimulated caspase-3 activation and TUNEL-positive cell death by lowering p53 promoter transactivation and activity in human cells. We demonstrate that N1 also lowers toxicity, cell death, and p53 pathway exacerbation triggered by Swedish mutated βAPP overexpression in human cells. We designed a CHO cell line overexpressing the London mutated βAPP (APPLDN) that yields Aβ oligomers. N1 protected primary cultured neurons against toxicity and cell death triggered by oligomer-enriched APPLDN-derived conditioned medium. Finally, we establish that N1 also protects neurons against oligomers extracted from Alzheimer disease-affected brain tissues. Overall, our data indicate that a cellular prion catabolite could interfere with Aβ-associated toxicity and that its production could be seen as a cellular protective mechanism aimed at compensating for an sAPPα deficit taking place at the early asymptomatic phase of Alzheimer disease.

The ␤-amyloid precursor protein (␤APP) 5 and cellular prion are central to two neurodegenerative diseases, namely Alzheimer and prion diseases (1,2). In Alzheimer disease (AD), anatomical and genetic clues directed research toward the investigation of ␤APP proteolysis, and it emerged relatively rapidly that this protein could undergo various physiological or potentially pathological alternative cleavages via distinct proteolytic entities. Most of the normal cleavage of ␤APP occurs in a constitutive or regulated manner by two disintegrins belonging to the a disintegrin and metalloprotease (ADAM) family, ADAM10 and ADAM17, respectively (3). This ␣-cleavage has been proved to be very important for at least two main reasons. First, the cleavage occurs within the A␤ domain of ␤APP and thereby prevents the production, accumulation, and aggregation of A␤. Second, the ␣-secretase-mediated breakdown of ␤APP gives rise to a secreted product referred to as sAPP␣ that is cytotrophic and neuroprotectant and has the remarkable property to protect against A␤-related toxicity both in vitro and in vivo (4 -6). Because, ␣and ␤-secretases apparently compete for ␤APP substrate (7,8), it has been speculated that part of Alzheimer disease pathology could be linked to a deficiency in sAPP␣-associated protection against A␤-related toxicity.
The cellular prion is a glycosylphosphatidylinositol-anchored protein (9). It is known that glycosylphosphatidylinositol-linked protein is mainly released from cellular membranes by the action of the well described phosphatidylinositol-specific phospholipase C (10). Unexpectedly, as is the case for ␤APP, we have established that PrP c can undergo both constitutive and regulated proteolysis by ADAM10 and ADAM17, respectively (11,12), although the ADAM family of enzymes was thought to target only transmembrane proteins (13,14). The striking similarity between the enzymatic machineries responsible for the physiological processing of ␤APP and PrP c led us to pursue our investigation of PrP c cleavage regulation and the putative biological function harbored by disintegrinmediated PrP c catabolites. Data concerning the muscarinic control of the regulated processing of PrP c (15) and the nature of the protein kinase C and downstream kinases involved delineate striking similarities, although a few differences remain (13,16,17).
Converging phenotypes conveyed by sAPP␣ and N1, the ␣-secretase-derived product yielded by ADAM-dependent PrP c cleavage, also exist. Thus, as had been documented for sAPP␣ (4 -6), we established that N1 protects various cells, including primary cultured neurons, from various proapoptotic challenges (18). Most interesting is that N1 also protects neurons in vivo in a pressure-induced ischemia model of rat retina (18). In both in vitro and in vivo approaches, N1 triggers its protective phenotype by down-regulating the p53-dependent pathway (18). Interestingly, a previous study suggested that A␤ peptides could elicit cell death by exacerbating the p53 pathway (19). Altogether, this led us to postulate that N1 could potentially protect cells from A␤-induced toxicity. Here we show that recombinant N1 and sAPP␣ similarly protect human cells from staurosporine (STS)-induced cell death by reducing p53 pathway activation. Interestingly, N1 reduces the toxicity and p53 pathway activation in cells expressing familial AD-linked mutations in ␤APP and PS2 (i.e. engineered to overproduce A␤). Finally, we establish that N1 protects cells from A␤ oligomers recovered in the secretions of cells expressing ␤APP bearing the London mutation or prepared from pathogenic AD-affected brain extracts. Altogether, our study is the first demonstration that a PrP c catabolite could interfere with A␤ toxicity. We speculate on the possibility that such a mechanism could be part of the compensatory mechanisms likely taking place during the early asymptomatic phase of AD pathology.
Mock-transfected and APP695WT-or APP695 LDN -expressing CHO cells were obtained by stable transfection of pcDNA 4 empty vector and hAPP695WT or hAPP695 LDN cDNA subcloned in pcDNA 4 vector. Cells were maintained in DMEM containing 10% FBS, sodium hypoxanthine-thymidine supplement, and 300 M proline. cDNA encoding APP695 LDN was obtained by site-directed mutagenesis of APP695WT cDNA as described below. Cells were stably transfected with 2 g of cDNA constructs according to Lipofectamine protocols reported previously (26). Clones were selected with 250 g/ml Zeocin (Invitrogen).
Conditioned Media from CHO Cell Lines-Mock-transfected, APP695WT, or APP695 LDN CHO cells were grown in 150-mm-diameter dishes until reaching 80% confluence, then washed with PBS, and allowed to secrete for 24 h into 15 ml of Neurobasal medium (Invitrogen). Secretions were centrifuged (1000 ϫ g for 10 min) and then concentrated into Amicon Ultra-15 3000 filters (4000 ϫ g for 30 min). One-milliliter aliquots of concentrates were stored at Ϫ80°C until use (27).
Recombinant Fragments-N1 and N2 recombinant fragments were produced as described previously (18,29). Briefly, the pGEX-KG glutathione S-transferase-N1-or -N2-expressing vectors were transformed into the BL21-Gold strain of Escherichia coli. Bacteria were grown in Luria broth medium, and then the fusion protein was induced with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside for 4 h at 37°C. Cells were pelleted, resuspended in PBS with Complete protease inhibitor mixture and lysozyme, incubated on ice for 30 min, and then solubilized by 1% Triton X-100, 10 mM MgCl 2 , and 5 g/ml DNase I. Debris was pelleted, and then glutathione-Sepharose beads were added to the lysate and swirled for 1 h at 4°C. Peptides were cleaved with thrombin. Thrombin was removed using Sepharose-benzamidine beads. sAPP␣695 was purified as described (29). A strain of Pichia pastoris expressing sAPP␣695 (kindly provided by Dr. R. Cappai, University of Melbourne, Melbourne, Australia) was grown at 30°C in 1% yeast extract (Invitrogen), 2% peptone (Invitrogen), 2% D-glucose (Sigma). Protein expression was induced during 48 h in BMMY (1% yeast extract, 2% peptone, 1.34% yeast nitrogen base without amino acids (Sigma), 4 ϫ 10 Ϫ5 % biotin (Sigma), 2% methanol (Merck)). Purification was carried out on ice using a modification of the method of Henry et al. (30). Yeast cultures (0.5-2 liters) were centrifuged at 16,000 ϫ g for 10 min at 4°C, and supernatants were filtered (0.45 m; Whatman). The supernatant was diluted to ionic strength 0.2 in 20 mM imidazole, 5 mM EDTA, 10 mg/liter phenylmethylsulfonyl fluoride (PMSF), pH 5.5 and applied at a flow rate of 7 ml/min onto a Q-Sepharose column (HR 26-10, GE Health-care) pre-equilibrated with 6 volumes of the same buffer. The column was washed with 20 volumes of the same buffer to which 250 mM NaCl was added, and the protein was eluted in 20 mM Tris-HCl, 1 M NaCl, 5 mM EDTA, 10 mg/liter PMSF, pH 7.4. Protein-containing fractions were pooled and desalted in prepacked Sephadex G-25 columns (GE Healthcare). The desalted pool was applied onto a 5-ml HiTrap heparin-Sepharose column (GE Healthcare) connected to an HPLC system and preequilibrated with 50 mM Tris-HCl, pH 7.4. The column was washed until the A 280 had reached base line, and the protein was eluted with a linear gradient up to 1 M NaCl. Eluted peaks were monitored by the absorbance at 280 nm. Fractions in the sAPP␣695 peak were pooled and desalted in 50 mM Tris-HCl, pH 7.4; concentrated on a Centricon-30 device (Amicon, Bedford, MA); and stored at 4°C. The identity of sAPP␣695 was verified by SDS-PAGE and Western blots using anti-APP monoclonal antibody 22C11 (Roche Applied Science).
C100 recombinant substrate of ␥-secretase was produced as described previously (31). Briefly, cDNA encoding the C-terminal 99 amino acids of human ␤APP was subcloned into a pet29c vector fused to FLAG tag at the C terminus and harboring an additional methionine at the N terminus. Recombinant protein was expressed in E. coli, and then the cell pellet was solubilized with 20 mM Tris lysis buffer, pH 7.5 containing 0.1 mM EDTA, 1 mg/ml lysozyme, and 1% N-lauroylsarcosine (v/v). After 3 h under agitation at 4°C, the lysate was spun at 5000 ϫ g for 2 h at 4°C. The supernatant was then spun for 75 min at 20,000 ϫ g at 4°C, and the resulting supernatant was subjected to filtration on a 30-kDa-cutoff membrane.
Human Brain Tissue Homogenate Preparation-Frontal cortex samples of human brains from an Alzheimer disease-affected patient diagnosed in Braak stage VI (91-year-old female) and a control patient (82-year-old female) or control patients provided by the Groupement d'Intérêt Économique de Collections d'Échantillons Biologiques (Hôpital de la Pitié-Salpêtrière, Paris, France) were obtained by adaptation of the protocol used by Shankar et al. (32). Briefly, frozen human brain tissues were diced and homogenized using a MagNA Lyser instrument (3000 rpm for 30 s) and related MagNA Lyser green beads (Roche Applied Science) in 5 volumes of TBS containing protease and phosphatase inhibitor mixtures I and II (Sigma). Homogenates were centrifuged at 175,000 ϫ g for 30 min, and then supernatants were either immunoprecipitated and analyzed by Western blotting or used for treatment of mouse primary cultured neurons (Invitrogen).
A␤ Recovery and Analysis-HEK293 cells were grown in 6-well dishes and allowed to secrete for 8 h in Opti-MEM (1 ml; Invitrogen) containing 10 M phosphoramidon (Sigma) to prevent A␤ degradation by neprilysin. Media were collected, onetenth volume of 10ϫ immunoprecipitation assay buffer (100 mM Tris-HCl, pH 8.0) containing 1.5 M NaCl and 50 mM EDTA was added, and the samples were incubated overnight with a 100-fold dilution of anti-A␤ antibody FCA18, FCA3340, or FCA3542 (33) and protein A-agarose beads (VWR, France). Beads were washed twice with 1ϫ radioimmunoprecipitation assay buffer and subjected to Tris-Tricine 16.5% polyacrylamide gels.
Immunoprecipitation of A␤40 and A␤42 species or total A␤ from human brain tissue homogenates was performed according to Shankar et al. (32) using FCA3340, FCA3542, or 6E10 antibody, respectively. Immunoprecipitates were then subjected to Tris-Tricine 16.5% polyacrylamide gels. The same gel analysis conditions were used for CHO conditioned medium concentrates.
Proteins were transferred onto nitrocellulose and incubated overnight with the 2H3 or 6E10 monoclonal antibody at a 1:1000 dilution. Immunological complexes were detected with a goat anti-mouse peroxidase-conjugated antibody (1:2000 dilution). Chemiluminescence was recorded using a LAS-3000 Luminescence Image Analyzer (Raytest, Courbevoie, France), and quantifications were performed using the AIDA analyzer software.
BACE1 and ␥-Secretase Assays-For BACE1, cells were lysed with 10 mM Tris-HCl, pH 7.5, and then homogenates were monitored for their BACE1 activity by means of a fluorometric assay described previously (35,36).
The in vitro ␥-secretase assay was performed according to Sevalle et al. (31). Briefly, "solubilized membranes" were obtained from HEK293 cells resuspended in solubilization buffer and diluted to yield a 1 mg/ml final protein concentration; then diluted with an equal volume of sodium citrate buffer, 1 mg/ml egg phosphatidylcholine, and 50 g/ml recombinant C100-FLAG; and assayed as described.
Membrane integrity was evaluated by measuring the lactate dehydrogenase activity in the culture medium using a Cyto-Tox-ONE TM kit (Promega Corp., Madison, WI). Cells were grown in 96-well plates and allowed to reach 80% confluence. Assays were performed according to the manufacturer's recommendations for 40 l of medium transferred in 96-well dark plates, and fluorescence was recorded with excitation and emission wavelengths of 560 and 590 nm, respectively.
TUNEL Analysis-Cells were fixed for 30 min with 4% paraformaldehyde, rinsed in phosphate-buffered saline, permeabilized for 10 min in 0.01% Triton X-100, and then processed for the dUTP nick-end labeling TUNEL technique according to the manufacturer's recommendations (Roche Applied Science). Fragmented DNA labeling corresponds to green spots. A second labeling with DAPI (1:20,000 in PBS) was carried out to visualize the total number of nuclei.
Caspase-3-like Activity Measurements-Cells were grown in 6-well plates and incubated with STS (2 M for 16 h for HEK293 cells and primary cortical neurons, 1 M for 2 h for fibroblasts, and 1 M for 16 h for CHO cells) after they reached confluence. Samples were processed for caspase-3-like activity assay as described previously (37). Caspase-3-like activity is calculated from the linear part of fluorimetry recorded and expressed in units/h/mg of proteins (established by the Bio-Rad procedure). One unit corresponds to 4 nmol of 7-amino-4-methylcoumarin released.
Statistical Analysis-Statistical analysis was performed with PRISM software (GraphPad Software, San Diego, CA) using the Newman-Keuls multiple comparison tests for one-way analysis of variance and t test.
To confirm our data in another cell system, we used a stably transfected HEK293 cell line that overexpresses N141I mutated presenilin-2 (PS2 N141I ; see Fig. 2A). This cell system was chosen first because it leads to increased production of A␤ (mainly A␤42 (42)) and second because we previously established that this mutation triggers a drastic exacerbation of both STS-stimulated caspase-3 activation and p53-dependent cell death (38). Fig. 2B confirms an enhanced basal and STS-stimulated caspase-3 activity in N141I-PS2-expressing HEK293 cells (48.2 Ϯ 7.6% increase when compared with mock-transfected cells, n ϭ 4, p Ͻ 0.05) and further documents a protective effect of N1, but not N2, that reduces STS-induced caspase-3 activity (Ϫ30.5 Ϯ 3.3%, n ϭ 4, p Ͻ 0.05; Fig. 2C). This set of data indicates that N1 could protect against cell death triggered by various familial AD mutations responsible for enhanced formations of either total A␤ (␤APP Swedish mutation) or more selectively A␤42 (N141I-PS2 mutation).
We also examined the protective phenotype of N1 in fibroblasts devoid of ␤APP and its family members APLP1 and APLP2 (triple KO). This cell system was used to abolish a putative protective effect of endogenous sAPP␣ (absent in cells lacking ␤APP) competing with that of N1. Furthermore, this cell system allows the examination of a putative cell-specific dependent phenotype associated with N1. N1, but not N2, similarly protected wild-type and triple KO fibroblasts from STSinduced caspase-3 activation (supplemental Fig. 3).

JOURNAL OF BIOLOGICAL CHEMISTRY 5025
caspase-3 activation by reducing p53 activity and promoter transactivation.
N1 Modifies Neither ␤and ␥-Secretases Expression nor Activities in Human Cells-The exacerbation of both caspase-3 activation and p53 pathway in HEK293 cells engineered to overexpress a familial mutation in ␤APP or PS2 strongly suggest that cell death is triggered by overload of A␤. Therefore, the N1 protective effect could indeed account for a late effect on A␤-associated toxicity but also theoretically for an early upstream interference with A␤ production. In this context, we

N1 Protects Cells against A␤-induced Apoptosis
examined the putative effect of N1 on the expression of secretases in both APPWT-and APPswe-expressing cells. Fig. 4A first shows that N1 treatment did not influence the expression of APPWT and APPswe or the recovery of A␤ in culture media (compare Ct and N1 lanes). Furthermore, N1 did not modify the endogenous expressions of PS1 or PS2, Aph1, Pen2, and nicastrin (Fig. 4B), the four members of the biologically active ␥-secretase complex (46). Finally, N1 did not affect the endogenous expression of the ␤-secretase BACE1 (Fig. 4C) (7). This set of data was corroborated by the demonstration that ␥-secretase activity in reconstituted membranes (47) prepared from APPWT-and APPswe-expressing HEK293 cells and BACE1 activity were not altered by N1 (Fig. 4, D and E, respectively). Thus, this set of data indicates that N1 does not interfere with the proteolytic machineries responsible for the production of A␤ in APPWT and APPswe cells and therefore that the N1 protective effect is not due to its propensity to reduce the load of A␤ production.

N1 Protects Primary Cultured Mouse Cortical Neurons from Apoptosis Induced by A␤ Oligomer-enriched Conditioned
Medium Prepared from APP LDN CHO Cells-Several lines of evidence recently suggested that at least part of the AD-related neurodegenerative process could be due to A␤ multimers of high molecular weights (48 -50). More recently, it was suggested that A␤ dimers and trimers could account for most of the A␤-related toxicity (51). To examine whether N1 could protect against A␤ oligomer-related toxicity, we developed CHO cell lines stably expressing either wild-type ␤APP695 or its homolog APP LDN harboring the London mutation (28) that is known to yield and secrete high levels of A␤ oligomers (27). Fig. 5A shows that cells expressing similar levels of APPWT and APP LDN both produced higher levels of A␤ when compared with mock-transfected CHO cells (Fig. 5A) but that the London mutation selectively exacerbated A␤42 production, yielding an enhanced A␤42/A␤40 ratio (2.2 Ϯ 0.5 and 0.3 Ϯ 0.06 for APP LDN and APPWT, respectively, n ϭ 3, p Ͻ 0.05; Fig. 5A). Furthermore, although both APPWT-and APP LDN -expressing cells secreted high molecular weight oligomers (although to a much higher amount in the latter cells), only APP LDN cells produced detectable A␤ dimers and trimers in our experimental conditions (Fig. 5B). Interestingly, conditioned medium from APP LDN -expressing cells time-dependently triggered cellular toxicity in primary cultured neurons (28.4 Ϯ 0.9 versus 54.2 Ϯ 7% cell death, n ϭ 4, p Ͻ 0.01 after 24-h incubation for treatment with mock versus APP LDN secretions, respectively; 35.9 Ϯ 3.7 versus 67.2 Ϯ 7.5%, n ϭ 4, p Ͻ 0.01 after 48-h incubation; FIGURE 5. Conditioned media from APP LDN -expressing CHO cells contain A␤ oligomers and are toxic for primary cultured neurons. A, cell extracts and secretion medium from APP LDN -expressing CHO cell lines were analyzed for ␤APP expression and monomeric A␤ species, respectively. ␤APP and tubulin expressions were monitored by Western blot with 22C11 antibody, and A␤40 and A␤42 were immunoprecipitated using FCA3340 or FCA3542 antibody, respectively, as described under "Materials and Methods." B, 20 l of conditioned medium were loaded on a Tris-Tricine 16.5% acrylamide gel, then transferred onto nitrocellulose membrane, and blotted with 6E10 antibody as described under "Materials and Methods." C, primary cultured neurons from mouse cortices were prepared and maintained in culture as described under "Materials and Methods" and exposed to A␤ oligomer-enriched conditioned medium for the indicated time periods, and then cellular toxicity was assessed using an lactate dehydrogenase assay as described under "Materials and Methods." Bars are the means of four independent determinations. Error bars ϭ S.E. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. We subsequently examined the potential of N1 to protect primary cultured mouse cortical neurons from A␤ oligomerassociated toxicity. Fig. 6 shows that A␤ oligomer-enriched conditioned medium from APP LDN , but not APPWT, cells reduced neuronal viability (Fig. 6A) and increased the number of TUNEL-positive apoptotic neurons by about 2-fold (Fig. 6, B and C). Interestingly, N1, but not N2, abolished A␤ oligomerassociated toxicity (Fig. 6A) and cell death (14.7 Ϯ 1.8% (n ϭ 6) versus 6.7 Ϯ 2.1 (n ϭ 6) of apoptotic nuclei per optical field in N1-treated and control neurons, respectively; Fig. 6, B and C). Therefore, our data demonstrate for the first time that N1 protects neurons from A␤ oligomer-induced cell death.

N1 Protects Primary Cultured Mouse Cortical Neurons from Apoptosis Induced by Brain Extracts from Patient with AD-
We prepared extracts from cortical tissues of AD-affected patient diagnosed with Braak stage VI. We characterized the 〈␤ species that were immunoprecipitable from extracts of control and AD samples. The AD sample was characterized by an increase in both A␤40 and A␤42 accounting for an overall augmentation of A␤ as well as increased high molecular weight oligomers (Fig. 7A). AD extracts increased the number of TUNEL-positive primary cultured neurons (Fig. 7B) by about 50% (28.70 Ϯ 1.60% (n ϭ 6) versus 20.8 Ϯ 1.3% (n ϭ 6) of apoptotic nuclei in AD-treated and control cells, respectively). Interestingly, apoptosis associated with AD extract was abolished by N1 treatment (28.70 Ϯ 1.6%, n ϭ 6 versus 20.75 Ϯ 1.7%, n ϭ 8) but not by N2 (27.4 Ϯ 1.9%, n ϭ 8). This study demonstrates that N1 can protect primary cultured neurons from A␤-induced apoptosis triggered by pathogenic brain extracts.

DISCUSSION
We show here that the ␣-secretase-derived fragments of ␤APP and PrP c , namely sAPP␣ and N1, respectively, similarly protect human cells from STS-induced caspase-3 activation by interfering with the p53-dependent pathway. Furthermore, we establish that N1 protects cells from toxicity associated with the overload of A␤ derived from the overexpression of Swedish mutated ␤APP and N141I-PS2, two proteins involved in familial cases of Alzheimer disease. We show that these mutations exacerbate STS-induced caspase-3 activation and p53, the extent of which is lowered by both recombinant N1 and sAPP␣.
Interestingly, several studies have demonstrated that intracellular A␤42 can trigger toxicity via p53 and Bax in primary cultured human neurons (52) likely by directly activating the transactivation of p53 promoter (19). Later, the tight link between A␤-associated toxicity and the p53 pathway was reinforced by the identification of the late effectors of A␤ toxicity that corresponded to the p53 downstream targets RNA-dependent-protein kinase and mammalian target of rapamycin (53). Furthermore, similar to the present study, it was reported that wild-type ␤APP confers resistance to p53-induced apoptosis and that this phenotype was abolished by familial AD mutated ␤APP (41). Finally, mutated PS1 and PS2 trigger p53-dependent apoptosis (26,38,54) that could be prevented by ␥-secretase inhibitors (54). Combined with our present results, we propose that N1 protects cells expressing familial AD-linked APP or PS2 by down-regulating A␤-associated p53-dependent cell death.
The similar protective effect of sAPP␣ and N1 fragments does not per se prove that they lower A␤-related toxicity via identical cellular pathways. A line of data suggests that it could be indeed the case. Thus, both sAPP␣ and N1 signaling could involve the phosphatidylinositol 3-kinase/Akt cellular pathway (18,55,56). However, close examination of the literature and our present data could counter such a hypothesis. It was reported that sAPP␣ interacts with the LRP1 and that this interaction triggers sAPP␣ endocytosis and subsequent degradation (57). Therefore, sAPP-LRP1 interaction ultimately leads to the inactivation of the sAPP␣-associated phenotype. At first sight, this concurs with the proposal of LRP1 as a pathogenic trigger in AD. However, in the case of N1, it appears that LRP1 deficiency abolishes the N1-mediated protective function and therefore that LRP1 could overall trigger a protective function. This agrees fairly well with another study showing that LRP1 FIGURE 6. N1 protects primary cultured mouse cortical neurons from A␤ oligomer-induced toxicity and cell death. A, primary cultured mouse cortical neurons (seeded on 96-well plates) were treated (24 h) with recombinant N1, N2 (1 M), or an equivalent volume of control buffer (Ct) with 〈␤ oligomer-enriched conditioned media (4-fold dilution). Twenty-four hours after incubation, cells were treated as above for an additional 24-h period, and then neuronal viability was determined by 2,3-bis[2-methyloxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide assay as described under "Materials and Methods." Values are the means of three independent experiments (carried out in triplicates) and are expressed as percentage of control untreated cells (taken as 100). Error bars ϭ S.E. B, representative pictures of mouse primary cultured neurons seeded on glass coverslips treated as in A and then processed for TUNEL as describe under "Materials and Methods." C, TUNELpositive nuclei were counted in six independent optical fields. Bars correspond to the percentage of labeled nuclei of total DAPI-stained nuclei. *, p Ͻ 0.05; **, p Ͻ 0.01.; ns, not statistically significant.

N1 Protects Cells against A␤-induced Apoptosis
triggers a neuronal antiapoptotic phenotype by promoting the Akt survival pathway as N1 does (18,45). However, it is unlikely that N1 signaling needs LRP1-mediated endocytosis because we previously showed that the N1 protective function did not require internalization/endocytosis (18). This set of data indicates that although sAPP␣ and N1 could display similar phenotypes and both protect against p53-dependent cell death, the two fragments could elicit their effects via distinct cellular pathways.
Recent advances led to the proposal that the A␤-associated toxicity may due to soluble A␤ oligomers instead of A␤ monomers that could display a protective function (47). Thus, Shankar et al. (58) showed that oligomers prepared from human brain pathogenic extracts indeed alter long term potentiation and therefore affect cognitive functions linked to memory and learning. This prompted us to assess whether N1 also protected against A␤ oligomer-associated toxicity from two sources, i.e. conditioned medium from cells engineered to produce oligomers and pathogenic human brain extracts. Our data show that in both cases N1 protects primary cultured neurons against A␤ oligomer-associated cell death.
How N1 can protect against A␤ oligomer toxicity can be deduced from recent works showing that PrP c could be necessary for A␤ oligomer-associated deleterious effects. Thus, it has been suggested that PrP c could act as a potential receptor for A␤ oligomers (59,60). The functional consequence of the physical interaction between A␤ oligomers and PrP c is still under extensive debate (61)(62)(63)(64), but the validity of the protein-protein interaction is less controversial. Two PrP c regions bind to soluble A␤42 oligomers: the 95-110 sequence and the 23 KKRPK 27 N-terminal basic residues of PrP c (59, 65). Indeed, a partial loss of A␤ binding was observed for ⌬101-110 deleted huPrP fragment, whereas removal of the 23-90 segment resulted in a complete loss of interaction between PrP c and A␤42 oligomers (65). Strikingly, both domains implicated in PrP c -A␤ oligomer binding are included in the N1 sequence. Interestingly, we previously demonstrated that the mutation of the N-terminal KKRPK sequence or its deletion led to the loss of the neuroprotective function of N1 (18). Whether N1 could directly bind A␤ peptide oligomers remains to be established, but one can envision that the activation of ␣-secretase cleavage and subsequent production of N1 may be a potential way to deplete extracellular media of A␤ toxic oligomeric species. In this context, interestingly, a recent study reports that A␤ oligomers recruit PrP c to the neuronal cell surface (66). These results could be seen as an attempt for neurons to enhance N1 availability to protect themselves from A␤ toxicity.
The various possible cellular mechanisms underlying the N1 protective function toward A␤ or A␤ oligomer-associated neurotoxicity , i.e. 1) the activation of the intracellular cell survival pathway by modulating the Akt/p53 pathway, 2) the depletion of toxic A␤ species by direct binding to N1, 3) the obstruction by direct competition of the N-terminal PrP c -dependent binding site necessary for A␤-mediated cell death, and 4) a combination of the three previous possibilities, reinforce the view that a strong and intimate functional dialogue between PrP c and ␤-amyloid precursor protein and their catabolites exists. If one considers that in AD the ␣-secretase levels are slightly affected, then this means that the potential of formation of N1 remains and that the formation of this PrP c catabolite could, at least at the early stages of the disease, contribute to the cellular mechanisms aimed at compensating for sAPP␣ deficiency. Thus, our FIGURE 7. N1 protects primary cultured mouse cortical neurons from A␤ oligomer-enriched extracts from AD patient brain tissue. A, extracts of frontal cortex samples from an AD patient with Braak stage VI or a control patient were prepared as described under "Materials and Methods" and analyzed for A␤ contents by immunoprecipitation and Western blots as described under "Materials and Methods" by means of FCA3542 (A␤42), FCA3340 (A␤40), or 6E10 (total A␤) antibody. B, mouse neurons seeded on glass coverslips were treated with recombinant N1, N2 (1 M), or an equivalent volume of control buffer (Ct) in media containing 1 ⁄1000 of a brain extract preparation. Twenty-four hours after incubation, cells were treated as above for an additional 24-h period. Representative pictures of TUNEL analyses of neurons are presented in B, and TUNEL-positive nuclei were counted in eight independent optical fields (C). Bars correspond to the percentage of labeled nuclei of total DAPI-stained nuclei. *, p Ͻ 0.05; **, p Ͻ 0.01. data reinforce the interest in a strategy aimed at stimulating the ␣-secretase pathway that has been envisioned as a therapeutic track for Alzheimer disease (67)(68)(69). Furthermore, this could open new therapeutic avenues for the design of drug candidates that could potentially prevent A␤-associated toxicity to delay the early symptoms of Alzheimer disease.