p53-dependent Aph-1 and Pen-2 Anti-apoptotic Phenotype Requires the Integrity of the γ-Secretase Complex but Is Independent of Its Activity*

The presenilin-dependent γ-secretase activity, which is responsible for the generation of amyloid β-peptide, is a high molecular weight complex composed of at least four components, namely, presenilin-1 (or presenilin-2), nicastrin, Aph-1, and Pen-2. Previous data indicated that presenilins, which are thought to harbor the catalytic core of the complex, also control p53-dependent cell death. Whether the other components of the γ-secretase complex could also modulate the cell death process in mammalian neurons remained to be established. Here, we examined the putative contribution of Aph-1 and Pen-2 in the control of apoptosis in TSM1 cells from a neuronal origin. We show by terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling and DNA fragmentation analyses that the overexpression of Aph-1a, Aph-1b, or Pen-2 drastically lowered staurosporine-induced cellular toxicity. In support of an apoptosis rather than necrosis process, Aph-1 and Pen-2 also lower staurosporine- and etoposide-induced caspase-3 expression and diminished caspase-3 activity and poly(ADP-ribose) polymerase inactivation. The Aph-1 and Pen-2 anti-apoptotic phenotype was associated with a drastic reduction of p53 expression and activity and lowered p53 mRNA transcription. Furthermore, the Aph-1- and Pen-2-associated reduction of staurosporine-induced caspase-3 activation was fully abolished by p53 deficiency. Conversely, Aph-1a, Aph-1b, and Pen-2 gene inactivation increases both caspase-3 activity and p53 mRNA levels. Finally, we show that Aph-1 and Pen-2 did not trigger an anti-apoptotic response in cells devoid of presenilins or nicastrin, whereas the protective response was still observed in fibroblasts devoid of β-amyloid precursor protein and amyloid precursor protein like-protein 2. Furthermore, Aph-1- and Pen-2-associated protection against staurosporine-induced caspase-3 activation was not affected by the γ-secretase inhibitors N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester and difluoromethylketone. Altogether, our study indicates that Aph-1 and Pen-2 trigger an anti-apoptotic response by lowering p53-dependent control of caspase-3. Our work also demonstrates that this phenotype is strictly dependent on the molecular integrity of the γ-secretase complex but remains independent of the γ-secretase catalytic activity.

Unlike PS, relatively few works document the function of Aph-1, Pen-2, and nicastrin. It is clear however that in mammals, the interaction of the four members is necessary for the functionality of the PS-dependent ␥-secretase activity. Aph-1 contributes largely to the stabilization of ␥-secretase complex by forming a subcomplex with nicastrin (20 -22), whereas Pen-2 is required for the maturation of both PS by endoproteolysis (18,23) leading to the generation of two C-and N-terminal fragments, namely CTF-PS and NTF-PS. The stringency of the necessary presence of each of the components of the complex is illustrated by the fact that the absence of any of these proteins affects A␤ production (18,19). Accordingly, the loss of function of Aph-1 and Pen-2 genes in Caenorhabditis elegans results in a Notch signaling deficiency, a phenotype that is directly linked to the ability of ␥-secretase to process Notch (24).
The implication of programmed cell death in AD has been largely documented in the last 10 years (25)(26)(27)(28)(29). In agreement with these studies, several works have consistently documented a pro-apoptotic phenotype of wild-type PS2, the extent of which appeared exacerbated by mutations responsible for familial forms of AD (30 -32), whereas, conversely, PS1 appeared biologically inert in relation to apoptosis (33). Recently, we demonstrated that PS2 and PS1 cross-talk to control the p53-dependent cell death pathway (34). Both the involvement of AICD (␤APP intracellular domain), the ␥secretase-derived C-terminal product of the ␤APP in the modulation of p53 and the blockade of this phenotype by ␥-secretase inhibitors suggest that the ␥-secretase activity remains essential for the PS-dependent control of cell death.
Except for a role of Aph-1 and Pen-2 in ␥-secretase complex formation and activity, little is known about other putative physiological functions, and more particularly, about their influence on the control of cell death in mammalian cells. However, the modulation of PS1 processing by RNA interference experiments targeting either Aph-1 and Pen-2 has been shown to affect the cellular sensitivity to caspase-3 activation (35) leading to the conclusion that, at least indirectly through the control of PS, Aph-1 and Pen-2 could be seen as regulators of cell death. However, the mechanisms by which cell death is affected remained a matter of questions, particularly in cells of neuronal origin. Furthermore, the above considerations did not rule out the possibility that Aph-1 and Pen-2 could control cell death in a ␥-secretase complex-independent manner or, alternatively, inside the ␥-secretase complex but independently of its activity.
Here, we examined the potential of Aph-1 and Pen-2 as regulators of cell death in the neuronal cell line TSM1, and we show that Aph-1aL, Aph-1b, and Pen-2 all decrease TSM1 susceptibility to staurosporine by lowering caspase-3 expression and activity in a p53-dependent manner. Furthermore, we establish that the Aph-1-and Pen-2-mediated anti-apoptotic phenotype is fully dependent of the presence of a complete and integral ␥-secretase complex but exert their function independently of the ␥-secretase activity.

Transient Transfections-Fibroblasts
Flow Cytometry Analysis of Propidium Iodide Incorporation-TSM-1 cells were grown in 6-well plates and incubated overnight at 37°C in the presence or absence of staurosporine (0.5 M), and then cells were harvested, pelleted by centrifugation at 1000 ϫ g for 10 min at 4°C, gently resuspended in 500 l of 0.1% sodium citrate buffer containing 50 g/ml propidium iodide, and incubated overnight under agitation. The propidium iodide fluorescence of individual nuclei was measured using a FACScan flow cytometer (program CellQuest, BD Biosciences) as described before (30).
TUNEL Analysis-Cells were cultured in 6-well plates and then treated for 24 h without or with staurosporine (0.5 M). Cells were then fixed for 30 min with 4% paraformaldehyde, rinsed in phosphate-buffered saline, permeabilized overnight with 70% ethanol, and then processed for the dUTP nick-end labeling TUNEL technique according to the manufacturer's recommendations (Roche Applied Science). Staining was performed with peroxidase-conjugated antibody and revealed with a diaminobenzidine as substrate (30). Fragmented DNA labeling corresponds to black spots. A second labeling with erythrosin B was carried out to visualize the totality of the cells.
Caspase-3 Activity Measurements-Cells were cultured in 6-well plates and then incubated for 2 or 16 h at 37°C in the absence or presence of staurosporine (1 M) or etoposide (100 M). In some cases, cells were either incubated overnight with Ac-DEVD-al (100 M, caspase-3 inhibitor, Sigma) or DAPT or DFK167 (50 M, ␥-secretase inhibitor) before treatment with staurosporine. Caspase-3 activity was then analyzed as extensively detailed (30). Fluorometry was recorded at 390 nm and 460 nm for excitation and emission wavelengths, respectively, by means of a microtiter plate reader (Labsystems Fluoroskan II). Caspase specific activity was calculated from the linear part of fluorometry recording and expressed in units/h/mg of proteins (established by the Bio-Rad procedure). One unit corresponds to 4 nmol of amidomethylcoumarin released.
p53 Transcriptional Activity and p53 Promoter Activity Measurements-The activity of p53 is measured with a bioassay by transient transfection of the PG13-luciferase cDNA (kindly provided by Dr. B. Vogelstein, Baltimore, MD), which is based on the genomic DNA consensus sequence recognized by p53 on its target genes (40). The transcriptional activation of the p53 promoter was measured after transfections of cDNAs harboring the murine (41) p53 promoter sequences in-frame with luciferase (provided by Dr. M. Oren, Rehovot, Israel). TSM-1 cells were cultivated in 12-well plates until 60% confluence and then cotransfected with 1.0 g of PG13-luciferase cDNA or 1.0 g of p53-promoter-luciferase cDNA and 0.5 g of a ␤-galactosidase transfection vector (to normalize transfection efficiency) by means of the SuperFect transfection reagent according the manufacturer's conditions (Qiagen). Forty-eight hours after transfection, luciferase and ␤-galactosidase activities were analyzed according to manufacturer's conditions (Promega kit).
Real-time Quantitative PCR-Total RNA from HEK293 or TSM1 cells was extracted by the mean of the RNeasy kit (Qiagen) according to the manufacturer's recommendations, then treated with DNase I, and 4 g of RNA obtained was reverse transcribed as previously described then Real-time PCR was performed as extensively described (34) in an ABI PRISM 5700 sequence detector system (Applied Biosystems, Foster City, CA) with gene-specific primers for human Aph-1a, Aph-1b, Pen-2, p53, and human glyceraldehyde-3-phosphate dehydrogenase to control mRNA concentrations.
siRNA Approach-siRNA duplexes (Qiagen) for Aph-1a (42), Aph-1b (42), and Pen-2 (43) were previously shown to drastically reduce endogenous levels of these proteins. A control non-silencing siRNA duplex (siCT) was used as a control. HEK293 cells were cultured until they reach 70% of confluency, and then transfected with the appropriate siRNA duplex for 24 h using HiPerfect transfection reagent (Qiagen). Cells were then harvested, and the levels of mRNA for Aph-1a, Aph-1b, and Pen-2 were analyzed by realtime quantitative PCR as described above. Then, caspase-3 activity and p53 mRNA levels were analyzed by fluorometry and real-time quantitative PCR as described above.
Statistical Analysis-Statistical analysis was performed with PRISM (GraphPad, San Diego, CA) using the Newman-Keuls multiple comparison test for one-way analysis of variance.
Aph-1a, Aph-1b, and Pen-2 Overexpression Interferes with the p53-dependent Cell Death Pathway-Because p53 is directly involved in the transcriptional control of caspases (48,49), we examined whether Aph-1a, Aph-1b, and Pen-2 could modulate this tumor suppressor, thereby leading to decreased caspase-3 activation. First, we show that both proteins drastically reduce p53-like immunoreactivity (Fig. 4A). This mimics the PS1-associated decrease observed in PS1-transfected cells (Fig. 4E). By means of a PG13-luciferase construct harboring a consensus sequence targeted by p53 (40), we were able to show that p53 activity was drastically reduced by Aph-1a, Aph-1b, and Pen-2 expression (Fig. 4B). We then examined whether Aph-1 and Pen-2 could regulate p53 at a transcriptional or post-transcriptional level. By means of a construct corresponding to the promoter of murine p53 in-frame with luciferase, we observed that Aph-1a, Aph-1b, and Pen-2 all reduce p53 promoter transactivation (Fig. 4C) as does PS1 (Fig. 4E), in agreement with our observations on p53 protein levels. Interestingly, Mdm2, an E3 ubiquitin ligase that triggers p53 proteasomal degradation by increasing p53 ubiquitination, is drastically enhanced by Aph-1a,b and Pen-2 (Fig. 4D). Altogether, this indicates that Aph-1 and Pen-2 likely modulate p53 at both transcriptional and post-transcriptional levels.

Depletion of Endogenous Aph-1a and Pen-2 Increases Caspase-3 Activity and Potentiates the p53-dependent Cell
Death Pathway-To examine whether the phenotype observed after overexpression of Aph-1 and Pen-2 could be physiologically relevant, we inactivated Aph-1a, Aph-1b, and Pen-2 genes by an siRNA approach. As expected, siRNA significantly low-ered Aph-1a, Aph-1b, and Pen-2 mRNA levels, whereas control siRNA remain ineffective (Fig. 6A). Interestingly, proteins depletion increased the susceptibility of HEK293 cells to staurosporine, leading to enhanced caspase-3 activation (Fig. 6B). In agreement with these data, silencing of Aph-1a, Aph-1b, and Pen-2 increased p53 mRNA levels (Fig. 6C). Therefore, depletion of endogenous proteins trigger a fully opposite phenotype than the one observed when the corresponding proteins are overexpressed. The same observations stood for fibroblasts devoid of PS1 (Fig. 6, D and E), depletion of TMP21 by an siRNA

p53-dependent Aph-1 and Pen-2 Anti-apoptotic Phenotype
approach did not modify staurosporine-induced caspase-3 activation in TSM1 cells (data not shown). Overall, these data confirm the physiological function of Aph-1a, Aph-1b, and Pen-2 in the control of p53-dependent cell death.
Aph-1a, Aph-1b, and Pen-2 Antiapoptotic Phenotype Is Independent of the ␥-Secretase Activity-We previously established that the PS-dependent control of cell death was tightly associated with the ␥-secretase complex activity (34). Thus, we demonstrated that PS-associated   APRIL 6, 2007 • VOLUME 282 • NUMBER 14 control of p53 was linked to the production of the ␥-secretasederived ␤APP and APLP2 fragments, AICD and APP-like intracellular domain 2 (ALID2), respectively, and fully prevented by ␥-secretase inhibitors. We therefore examined whether Aph-1a, Aph-1b, and Pen-2 remained protective in ␤APP/APLP2deficient fibroblasts, which do not harbor AICD/ALID fragments. Fig. 9 (B-E) indicates that Aph-1-and Pen-2-induced decrease in caspase-3 activity remained similar in wild-type and APP/APLP2-deficient fibroblasts. Therefore, Aph-1 and Pen-2, unlike PS, appear to control cell death in a ␥-secretase activity-independent manner. To further confirm this conclusion, we examined the direct effect of ␥-secretase inhibitors on Aph-1-and Pen-2-associated phenotypes. Fig. 10 shows that DAPT (Fig. 10A) and DFK167 (Fig. 10B) did not affect the inhibition of staurosporine-stimulated caspase-3 activation triggered by Aph-1a, Aph-1b, and Pen-2 in TSM1 neurons in conditions where A␤ production is fully prevented (not shown).

DISCUSSION
Two of the key histological stigmata in AD-affected brains are the senile plaques and neuronal loss (50). Senile plaques are mainly due to the exacerbated production of a set of hydrophobic A␤-like peptides, whereas several histochemical clues indicate that cellular loss could be associated with increased apoptotic cell death (25)(26)(27). Thus, AD pathology selectively affects a set of proteins involved in cell death (51)(52)(53), including caspase-8 (54).
Several previous studies also indicate that the tumor suppressor p53 could participate in neuronal cell death in AD. First p53-like immunoreactivity was enhanced in a subpopulation of degenerative cortical neurons in sporadic AD brains (55)(56)(57)(58). Second, p21, which is transcriptionally activated by p53, was increased in AD brains (52). Whether A␤ overproduction and exacerbated cell death are linked is still a matter of discussion, but it should be noted that several studies indicated that A␤ could trigger apoptosis in neuronal cell cultures (59) and that this A␤-induced neuronal cell death appeared caspase-8-mediated (60) and associated with a transcriptional activation of p53 (58).
It was interesting to note that both phenotype appeared kinked to the catalytic activity of the ␥-secretase complex. Thus, we demonstrated that both PS2 and PS1 contributed to control the intracellular level of AICD and ALIDs, the C-terminal fragments of ␤APP and APLPs from which production is derived by the ␥and ⑀-secretase cleavages (34). In agreement, ␥-secretase inhibitors fully prevent AICD-mediated control of p53 in wild-type but not in PS-deficient fibroblasts. Overall, it was concluded that PS1-and PS2-mediated control of p53-dependent cell death was directly linked to their associated ␥-secretase activity (34).
Very little is known concerning the contribution of the other components of the PS-dependent ␥-secretase complex in the control of cell death in mammalian cells and more particularly in neurons, but two studies suggested that they could have a role. Thus, selective apoptosis in the neuronal tube was observed in Aph-1a knock-out mice (66). Furthermore, while the present work was being completed, an elegant study demonstrated that the depletion of Pen-2 led to neuronal loss and apoptosis in zebrafish (67). We therefore examined the influence of overexpressed and endogenous Aph-1a and Pen-2 expression in the TSM1 neuronal cell line.
We establish that the expression of either Aph-1a, Aph-1b, or Pen-2 led to reduced susceptibility to staurosporine-induced toxicity. Thus, Aph-1 and Pen-2 lower the number of TUNEL-positive cells and DNA fragmentation in TSM1 neurons. This was accompanied by a reduction of caspase-3 activity and expression and concomitant lowered PARP cleavage. Here again, the influence of Aph-1 and Pen-2 on cell death appears to involve the control of p53. First, Aph-1 and Pen-2 reduce p53 expression and activity and diminish p53 promoter transactivation. Second, the Aph-1and Pen-2-associated anti-apoptotic phenotypes appear fully p53-dependent, because they were both abolished by p53 depletion. Most importantly, the depletion of both Aph-1 and Pen-2 by the siRNA approach led to a drastic increase in both STS-induced caspase-3 activation and p53 mRNA levels. Therefore, Aph-1and Pen-2-associated phenotypes were not the result of an artifact due to the overexpression of these proteins.
Several types of ␥-secretase complex could occur, because there exists two different PSs, three Aph-1 homologs, and even two splice isoforms of Aph-1a. Whether selective functions could be specifically associated to a subset of complex remains to be established. A recent study indicates that this could indeed be the case, because Serneels and colleagues (66) reported differential participation of the Aph-1 proteins to PS- FIGURE 9. Aph-1a, Aph-1b, and Pen-2 still lower STS-stimulated caspase-3 activity in fibroblasts devoid of APP, APLP1, and APLP2. A, wild-type (WT) or APLP1 Ϫ/Ϫ APLP2 Ϫ/Ϫ APP Ϫ/Ϫ triple knock out fibroblasts (TKO) were treated for 4 h without (CT) or with staurosporine (STS,1 M), and then caspase-3 activity was then measured as described under "Materials and Methods." WT (B) and TKO (D) fibroblasts were transiently transfected with empty vector, Aph-1a, Aph-1b, or Pen-2cDNA. 24 h after transfection, cells were treated with STS, harvested, lysed, and analyzed for caspase-3 activity. Bars are the mean Ϯ S.E. of eight to ten independent experiments. Expressions of indicated proteins in WT (C) and TKO fibroblasts (E) were analyzed by Western blot as described under "Materials and Methods" using 9E10 antibody. Actin was used as loading control. p values compare with mock transfected cells. dependent ␥-secretase activity. Our data indicate that PS, Aph-1, and Pen-2 all control p53-dependent cell death. However, the question remained as to whether Aph-1 and Pen-2 modulate p53 within the ␥-secretase complex and, if so, whether Aph-1-or Pen-2-associated phenotypes could be strictly dependent of the catalytic activity linked to the ␥-secretase complex. Our study clearly shows that the integrity of the PS-dependent ␥-secretase complex is a strict requirement for Aph-1-and Pen-2-associated apoptosis-related phenotypes. Thus, Aph-1 and Pen-2 expression in fibroblasts devoid of both PS or nicastrin did not modulate staurosporine-induced caspase-3 activation. However, ␤APP and APLP2 depletion did not abolish Aph-1-and Pen-2-associated phenotypes. Therefore, unlike for PS, the control of p53 by Aph-1 and Pen-2 was not mediated by the ␥-secretase-derived fragments, AICD and ALIDs. Another possibility would have been that the ␥-secretase cleavage of another protein unrelated to ␤APP and APLPs but also able to affect p53 could be controlled by Aph-1 and Pen-2. However, two well characterized ␥-secretase inhibitors DAPT and DFK167 did not modify Aph-1 and Pen-2 anti-apoptotic phenotype. Therefore, one can conclude that Aph-1-and Pen-2-protective function was dependent on the structural integrity of the ␥-secretase complex but remained unrelated to ␥-secretase activity.
The physiology of the proteins contributing to the PS-dependent ␥-secretase complex is likely much more complicated than previously anticipated. These proteins could display a large spectrum of functions either inside or outside of the complex. When complex-dependent, some of the functions could appear fully independent of the ␥-secretase activity as is the case for the control of cell death by Aph-1 and Pen-2.