Requirement of the familial Alzheimer's disease gene PS2 for apoptosis. Opposing effect of ALG-3.

ALG-3, a truncated mouse homologue of the chromosome 1 familial Alzheimer's disease gene PS2, rescues T hybridoma 3DO cells from T-cell receptor-induced apoptosis by inhibiting Fas ligand induction and Fas signaling. Here we show that ALG-3 transfected 3DO cells express a COOH-terminal PS2 polypeptide. Overexpression of PS2 in ALG-3 transfected 3DO cells reconstitutes sensitivity to receptor-induced cell death, suggesting that the artificial PS2 polypeptide functions as a dominant negative mutant of PS2. ALG-3 and antisense PS2 protect PC12 cells from glutamate-induced apoptosis but not from death induced by hydrogen peroxide or the free radical MPP+. Thus, the PS2 gene is required for some forms of cell death in diverse cell types, and its function is opposed by ALG-3.

Programmed cell death (PCD) 1 is a normal event under genetic control that regulates the life span of different cell types in multicellular organisms. Among other physiological processes, PCD plays a pivotal role in the regulation of the immune system. T-cells bearing T-cell receptors (TCR) able to recognize peptides from self proteins are generated in the thymus, but mature T-cells do not normally react against selfconstituents and hence exhibit self-tolerance. Self-tolerance is therefore not an inherent feature of T lymphocytes but is acquired during development (negative selection or clonal deletion) (1)(2)(3), and the main process through which it is estab-lished involves elimination by apoptosis of self-reactive thymocytes and peripheral T-cells (4 -7). Foreign antigens, on the other hand, induce activation and expansion of antigenspecific peripheral T-cells followed, if the antigen stimulates a large fraction of T-cells and persists for a prolonged time, by their elimination (8 -11). This process, called activation-induced cell death, is essential to protect the organism against the deleterious effects of an uncontrolled T-cell expansion and subsequent production of toxic levels of cytokines. Disregulation of PCD in lymphoid cells can result in diseases such as cancer, AIDS, and autoimmune disorders. The pathological consequences of the malfunction of PCD are not limited to the immune system but can affect a number of cell types including neurons (12).
To better characterize the biochemical steps linking TCR stimulation to cell death, we have used the mouse T-cell hybridoma 3DO which, like other T-cell hybridomas (13,14), undergoes PCD when stimulated with an anti-TCR antibody. Fas ligand is induced following TCR triggering and the engagement of Fas by Fas ligand activates the cell death program (15)(16)(17). Therefore, 3DO is a suitable in vitro model to study the molecular mechanisms of TCR-induced cell death. By means of a functional selection strategy, we have isolated, from an expression cDNA library, genes involved in apoptosis induced by TCR cross-linking in 3DO cells. The experimental system, called "death trap," selects for cDNAs able to block death by: 1) expressing in the cells adequate levels of specific antisense RNAs or dominant negative mutants for "apoptotic" genes and 2) producing "anti-apoptotic" proteins. One of the transcripts identified, ALG-3 (18), was shown to be a truncated mouse homologue of the chromosome 1 familial Alzheimer's disease (AD) gene PS2 (19,20). Expression of this partial RNA, potentially coding for the 103 COOH-terminal PS2 amino acids, rescued 3DO cells from T-cell receptor-induced apoptosis by inhibiting the induction of Fas ligand and the Fas-mediated death signal. Whether ALG-3 induced resistance to receptormediated cell death by being translated into a polypeptide and whether the endogenous PS2 protein was involved in PCD remained to be determined. In the present study we show that the Alzheimer's disease gene PS2 is required for cell death and that ALG-3 codes for a dominant negative mutant of PS2.

EXPERIMENTAL PROCEDURES
Cloning of Mouse PS2-A cDNA library from mouse liver (Clontech) was screened with the ALG-3 fragment, and several positive phages were double strand sequenced. Several shorter cDNAs corresponding to the ϳ1.1 kilobases most 3Ј end of the mouse PS2 were also isolated and named PS2 short (PS2s). A detailed description of PS2s will be published elsewhere.
In Vitro Transcription and Translation, Immunoprecipitation, and Immunoblot Analysis-PS2, PS1, and ALG-3 cDNAs were cloned into pcDNA3. Briefly, the cDNAs fragments coding for the mouse PS2, PS2s, and PS1 proteins were cloned in the EcoRI site of pcDNA3 (Invitrogen) in both orientation. ALG-3 cDNA was cloned in the EcoRV/NotI sites of pcDNA3. 1 g of DNA was used as a template for in vitro RNA transcription using T7 RNA polymerase (Promega). The RNAs were translated in vitro using rabbit reticulocyte lysate (Promega) and [ 3 H]leucine. For immunoprecipitation, immunocomplexes were bound * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U49111 (ALG-3) and U57324 (mouse PS2).
to protein A-Sepharose beads (Pharmacia), separated, and washed. Immunoprecipitates were separated on SDS-PAGE and detected by x-ray exposure. Total cell lysates, immunoprecipitated with the indicated antiserum, were separated by SDS-PAGE, transferred to nitrocellulose membranes, and probed with ␣ALG-2, ␣PS2n, or ␣PS2c antiserum. Western blot analysis of transfected PC12 was accomplished by harvesting cells 3 days after transfection. Cells were lysed in RIPA buffer plus the proteases inhibitors aprotinin, pepstatin, and leupeptin (100 g/ml, Sigma). Modified 5 ϫ loading buffer (320 mM Tris, pH 6.8, 50% glycerol, 0.5% bromphenol blue, 10% SDS, 100 mM dithiothreitol, and 8 M urea), used to reduce aggregation that typically occurs with seven transmembrane proteins, was added to the lysates (10 g). Samples were heated at 37°C for 45 min and at 75°C for 5 min and then separated on a 4 -20% polyacrylamide-SDS gradient gel, blotted onto nitrocellulose membranes (Gelman Science), and probed with either affinity purified ␣PS2n or anti-␤-tubulin (Boehringer Mannheim) antibodies. Immunoblots were developed using the ECL System (Amersham Corp.).
Transient Transfection in 3DO cells and ␤-Gal Activity-ALG-3.2 cells were transiently transfected with pcDNA3 expressing the indicated cDNAs (1 g) together with equal amounts of CMV ␤-Gal (Clontech) using DEAE-dextran (450 g/ml). 8 h after transfection, cells (10 5 /well) were cultured for 10 h in 96-well plates coated with 1 g/ml of 2C11 to induce apoptosis. The ␤-Gal activity of living cells was visualized by fixing the cells for 10 min followed by staining in PBS containing 20 mM each K 3 Fe(CN) 6 and K 4 Fe(CN) 6 ⅐H 2 O and 1 mg/ml 5-bromo-4-chloro-3-indoyl ␤-D-galactopyranosidase for 1-3 h at 37°C. The number of blue cells present in at least four wells were counted. The percentage of reduction corresponds to the ratio of 2C11 treated versus untreated ␤-Gal ϩ cells for each individual transfection. The transfection efficiency was comparable in all experimental samples.
Transient Transfection in PC12 Cells and Cell Death Assay-Freshly plated PC12 cells were transfected with 2 g of DNA ϩ 12 l of lipofectamine/ml Dulbecco's modified Eagle's medium containing 100 ng/ml nerve growth factor for 5 h. The cells were then washed and transferred to growth medium (10% fetal bovine serum, 5% horse serum in Dulbecco's modified Eagle's medium, plus 100 ng/ml nerve growth factor). The day after transfection, the cells were plated on 96-well dishes or Nunc chamber slides at a density of 5000 cells/well. On the second day after transfection, the cells were treated with toxic agents. Glutamate toxicity experiments were performed using undifferentiated PC12 cells grown in growth medium containing dialyzed sera (Life Technologies, Inc.); all glutamate treatments were for 24 h. MPP ϩ was used at 1 and 10 M. Hydrogen peroxide was used at 2, 20, and 200 M. The MTT assay (see Fig. 5A) was performed by incubating for 3 h in 0.5 mg/ml MTT and then adding lysis buffer (20% SDS/50% N,N-dimethylformamide/3.3 mM HCl), rocking, and measuring the A 595 the following day. For in situ detection of DNA fragmentation (see Fig. 5B), the cells were fixed in 4% paraformaldehyde/PBS for 10 min, washed twice, incubated 5 min in 0.1% Triton X-100/PBS, washed three times with PBS, and then incubated with 0.1 unit/l unmodified T7 polymerase and 0.4 mM biotin-14-dATP. The color was developed using streptavidin-peroxidase and 0.125 mg/ml 3,3Ј-diaminobenzidine as described previously (21). The data shown represent the means Ϯ standard deviations on at least triplicate counts of data reproduced in six independent experiments. The efficiency of transfection was determined by transfecting CMV ␤-Gal (Clontech). The ␤-galactosidase activity was visualized as described above. The number of cells present in at least four wells were counted with a phase contrast microscope.

ALG-3 Is Translated into a COOH-terminal PS2 Polypeptide
Both in Vitro and in Vivo-ALG-3 was found to be a truncated mouse homologue of the familial Alzheimer's disease gene PS2. Cloning of the full-length mouse PS2 transcript revealed that the mouse protein is 95% identical to the human PS2 counterpart (Fig. 1). To verify whether ALG-3 can be translated into a truncated PS2 polypeptide, we generated two polyclonal anti-

FIG. 2. ALG-3 is translated into a truncated PS2 polypeptide.
In vitro metabolically labeled PS2 (A), PS1 (B), and ALG-3 (C) proteins were immunoprecipitated using either preimmune antiserum (P.I.) or the indicated antiserum, separated on SDS-PAGE, and detected by x-ray exposure. none indicates the reaction products before immunoprecipitation.

FIG. 3. A COOH-terminal PS2 polypeptide is expressed in ALG-3 transfected 3DO cells.
Total cell lysates prepared from 3DO or 3DO expressing the ALG-3 transgene or transfected with the empty vector were immunoprecipitated with either the ␣PS2c or the ␣ALG-2 antiserum on a Western blot (W. B.). After SDS-PAGE separation, proteins were transferred to nitrocellulose membranes and probed with ␣PS2n, ␣PS2c, or ␣ALG-2. The ␣ALG-2 antiserum was used to normalize the amount of proteins loaded in each well. The gel was cut above 35 kDa to remove the signal derived from the heavy chain of the antiserum used for immunoprecipitation (IP). The endogeneous PS2 protein (ϳ50 kDa) and aggregated forms comigrate with the heavy chain and are therefore not detectable. sera specific for amino acids 341-377 (␣PS2n) and 438 -448 (␣PS2c) of PS2. Both sequences are potentially coded by the ALG-3 cDNA. As determined by immunoprecipitation of in vitro translated proteins, the two antisera are specific for PS2 and do not cross-react with mouse PS1, the chromosome 14 Alzheimer's disease gene that is 65% identical to PS2 (22) (Fig.  2, A and B). The ALG-3 RNA is translated into a polypeptide of ϳ10 kDa, recognized only by the polyclonal antiserum specific for the COOH-terminal 11 amino acids of PS2 (Fig. 2C). Be-cause the first AUG codon present in the translated frame codes for Met 438 and would give rise to only an 11-amino acid peptide (Fig. 1A) and there is not in-frame AUG in the transcribed 5Ј plasmid sequence, the observed ϳ10-kDa polypeptide must be translated from a noncanonical CUG (Lys 352 , Lys 364 , Lys 373 , Lys 389 , Lys 404 , Lys 413 , Lys 433 , and Lys 441 ) or ACG (Thr 388 ) start codon (23)(24)(25).
Next, we determined whether ALG-3 could give raise to a truncated PS2 protein in vivo. To this end, we analyzed cell lysates from a 3DO T-cell hybridoma clone, named ALG-3.2, expressing the transgenic ALG-3 transcript that was protected from TCR-induced cell death (18). Immunoprecipitation followed by Western blot revealed that a truncated PS2 protein of ϳ10 kDa, detected only by the ␣PS2c antiserum, is specifically detected in the ALG-3.2 cell clone but not in 3DO cells and the mock transfected 3DO clone pc.2 (Fig. 3).
PS2 Is Required for Apoptosis and Its Function Is Opposed by ALG-3-ALG-3 could represent a gain of function mutant of an endogenous anti-apoptotic protein, whose activity is normally downregulated during receptor-induced PCD. Alternatively, this molecule could function as a dominant negative mutant of a gene necessary to carry out the apoptotic program. If the latter hypothesis is correct, overexpression of PS2 should reconstitute sensitivity to TCR-induced PCD. To test for this, we cotransfected a plasmid carrying the ␤-Gal gene together with various pcDNA3 vector constructs into ALG-3.2 cells, which are resistant to TCR-induced apoptosis (18). As shown in Fig. 4, ϳ40% of the blue cells, transfected with both PS-2 and ␤-Gal, undergo PCD upon TCR triggering. This phenotype was specific to the PS2 transfected cells and was not observed with pcDNA3 expressing other cDNAs (Fig. 4).
The importance of PS2 for AD lead us to examine the role of PS2 levels in apoptosis in a different cell type. Pheochromocytoma PC12 cells, which express PS2 (Fig. 5A), were induced into apoptosis by treatment with glutamate, hydrogen peroxide, or the free radical MPP ϩ . Cellular viability was measured using the MTT assay. Transfection with antisense PS2, which reduced the expression level of endogeneous PS2 (Fig. 5A), and ALG-3 completely blocked the glutamate toxicity ( Fig. 5B) but did not affect sensitivity to hydrogen peroxide or MPP ϩ . Independent transfections performed with a ␤-galactosidase vector indicated a transfection efficiency of 32 Ϯ 4.6%. Cellular apo- ptosis was then measured by in situ DNA fragmentation assay. Transfection with antisense PS2 and ALG-3 reduced the number of apoptotic PC12 cells upon exposure for 24 h to 10 mM glutamate (Fig. 5C) but did not affect sensitivity to the other two insults (data not shown). On the contrary, overexpression of PS2 sense augmented the number of apoptotic PC12 cells (Fig. 5C). Thus, PS2 is involved in glutamate-induced cell death, and the death pathway requiring PS2 is blocked by ALG-3.

DISCUSSION
In this study, we have demonstrated a requirement for PS2 in some forms of PCD in both lymphoid and pheochromocytoma cells. PS2 function is opposed by ALG-3, an artificial truncated form of PS2 that encodes a COOH-terminal PS2 polypeptide. The involvement of PS2 in TCR-induced cell death suggests that this gene may play a role in negative selection and/or activation-induced cell death. In agreement with this hypothesis is also the fact that PS2 is expressed in both thymus and peripheral lymphoid tissues. The generation of transgenic mice expressing ALG-3 in T-cells, in which PS2 should be functionally inactivated, and of PS2 null mice should address this question.
The involvement of PS2 in the PCD pathway emphasizes the potential importance of PCD in the pathophysiology of AD. It has been recently shown that ␤-amyloid, a 42-amino acid peptide that derives from the proteolytic cleavage of the ␤-amyloid precursor protein and accumulates in senile plaques in Alzheimer's disease, can cause apoptosis in a cell type-dependent fashion (26,27). Moreover, mice expressing a transgenic ␤-amyloid peptide in neurons present extensive neuronal degeneration followed by apoptotic death (28). In addition, in situ apoptosis has been described in Alzheimer's disease (29,30). Based on this evidence, a potential role for PCD in Alzheimer's disease has been proposed (31). This hypothesis is reinforced by the evidences that PS2 also participates in cell death of neuronally differentiated PC12 cells and that a PS2 mutation associated with Alzheimer's disease generates a molecule with constitutive apoptotic activity (32).
The PS2 protein has substantial structural and amino acid similarity with the chromosome 14 Alzheimer's disease gene product PS1. These two proteins share a 63% overall identity and are strikingly homologous in the COOH-terminal 103 amino acids (80% sequence identity) and are both predicted to be seven transmembrane domain proteins. Therefore, it has been postulated that PS1 and PS2 are functionally related and subserve the same biochemical pathway. However, overexpression of PS1 in the ALG-3.2 cell clone does not reconstitute sensitivity to TCR-induced cell death. It is possible that the two proteins are both necessary but not sufficient to carry out the apoptotic program. Alternatively, like members of the BCL-2/ ced-9 gene family, PS1 and PS2 could regulate PCD in a coordinate but opposite fashion.