Cellular Prion Protein Sensitizes Neurons to Apoptotic Stimuli through Mdm2-regulated and p53-dependent Caspase 3-like Activation*

We examined the influence of cellular prion protein (PrP c ) in the control of cell death in stably transfected TSM1 cells. PrP c expression enhanced staurosporine-stimulated neuronal toxicity and DNA fragmentation, caspase 3-like activity and immunoreactivity, and p53 immunoreactivity and transcriptional activities. Caspase activation was reduced by the chemical inhibitor of p53, pifithrin- (cid:1) , as well as by PrP c - or p53-antisense approaches but remained insensitive to the Fyn kinase inhibitor PP2 (4-amino-5-(4-chloro-phenyl)-7-( t -butyl)pyrazolo[3,4- d ]pyrimidine). We establish that PrP c controls p53 at a post-transcriptional level and is reversed by Mdm2 transfection and p38 MAPK inhibitor. We propose that endogenous cellular prion protein sensitizes neurons to apoptotic stimuli through a p53-de-pendent caspase 3-mediated activation controlled by Mdm2 and p38 MAPK. 5 m M EDTA containing 0.5% Triton X-100 and 0.5% sodium deoxycholate). Equal amounts of protein (50 (cid:1) g) determined by the Bradford method (20) were separated on 12% SDS-PAGE gels and analyzed for their p53, active caspase 3, Mdm2, phospho-p38, and PrP c immunoreactivities by Western blot metabolizing pair-wise

Prion-associated pathologies are transmissible diseases that all have fatal issues (1,2). The elucidation of their etiology has revealed a fully original transmission mechanism. It is now generally admitted that the pathological infectious mechanism is due to a protein present ubiquitously in the brain (referred to as PrP c ) 1 that becomes deleterious after its biophysical conversion into a protease-resistant protein referred to as PrP res or PrP-scrapie (3). This "bio-transformation" and subsequent pathologies absolutely require the presence of PrP c because the absence of endogenous PrP c totally precludes the PrP sc -mediated infectivity and neurotoxicity (4,5).
Most of the works on prions have focused on these intriguing disease-related features, particularly their conformation properties in vitro (6 -8) and the species barrier allowing or limiting the propagation of the various prion strains (for reviews see Refs. 1 and 2). The invalidation of the PrP gene (Prnp) led to the conclusion that the Prnp Ϫ/Ϫ mice exhibit normal development and unaltered behavioral phenotype (9,10).
Several clues of a possible participation of PrP c in programmed cell death came from the identification of a significant but restricted sequence homology of PrP c octapeptide repeats with the anti-apoptotic oncogene Bcl-2 protein and its ability to bind to Bcl-2 in double hybrid approach (11)(12). A recent study (13) also indicated that prion protein could rescue human neurons from Bax-induced apoptosis. Alternatively, the possibility that a restricted sequence corresponding to the 106 -126 part of the molecule could elicit a cellular toxic response has also been extensively documented (14 -16).
It is noteworthy that the 106 -126 domain of the prion protein has no genuine physiological or pathological existence as it has never been described as a catabolite of PrP c that would have been proteolytically generated upon normal or pathological conditions. Therefore, the "toxic" potential of the 106 -126 peptide could either be not relevant to any normal or altered situation or, alternatively, could reflect a phenotype that would be also triggered by the whole parent prion protein.
The latter hypothesis would be in contradiction with the above studies, suggesting an anti-apoptotic PrP c -related phenotype, but would be in line with the work of Westaway et al. (17), showing that transgenic mice over-expressing "normal" wild type PrP c exhibit degeneration in central and peripheral nervous system as well as in skeletal muscle. Interestingly, the 106 -126 domain appears to be targeted by a set of proteolytic activities called disintegrins, ADAM10 and ADAM17, that cleave PrP c at the 110/111 peptide bond and thereby "inactivate" the putative toxic core of the PrP c protein (18). To establish a pro-or anti-apoptotic PrP c -mediated phenotype, we have examined the possible toxicity of PrP c and its putative involvement in the control of cell death. We show that in TSM1 cells, over-expressed and endogenous PrP c all trigger p53-dependent caspase 3 activation leading to a pro-apoptotic phenotype.
Western Blot Analyses-Cells were scraped and homogenized in lysis buffer (50 mM Tris-HCl, pH7.5, 150 mM NaCl, 5 mM EDTA containing 0.5% Triton X-100 and 0.5% sodium deoxycholate). Equal amounts of protein (50 g) determined by the Bradford method (20) were separated on 12% SDS-PAGE gels and analyzed for their p53, active caspase 3, Mdm2, phospho-p38, and PrP c immunoreactivities by Western blot and * This work was supported in part by the Groupe d'Intérél-Scientifique "Infections à prions" and by the CNRS and by INSERM. 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.
Flow Cytometry Analysis-Cells were grown in 6-well plates and incubated for 15 h at 37°C in the presence or absence of staurosporine (0.5 M). Cells were rinsed with phosphate-buffered saline, resuspended with 750 l of buffer (20 mM Tris, 0.1% tri-natrium citrate, 0.05% Triton X-100) containing 50 g/ml propidium iodide (PI) and incubated overnight at 4°C. The PI fluorescence of individual nuclei was measured using a FACSCalibur flow cytometer (CellQuest software; BD Biosciences). Nuclei were characterized for their side angle scatter and forward-angle scatter parameters. Red fluorescence due to PI staining of DNA was then counted on a logarithmic scale. All measurements were performed under identical conditions. This technique allows discrimination of populations of fragmented nuclei from debris and nonviable cells and also from intact diploid nuclei those show higher fluorescence staining. The number of apoptotic nuclei is expressed as a percentage of the hundred thousand events gated.
XTT Assay of Cell Viability-Cells were grown in a 5% CO 2 atmosphere in 96-well plates in a final volume of a 100 l per well and incubated for 24 h at 37°C in the presence or absence of 2 M of staurosporine. Briefly, XTT is metabolized by mitochondrial dehydrogenase to a water soluble formazan salt only by metabolically active viable cells. XTT metabolizing activity was determined as previously described (22).
Caspase 3-like Activity Assay-Cells were cultured in 6-well plates for 15 h at 37°C in the absence or presence of etoposide, ceramide C2, or staurosporine (0.5 M) or 15 M of the Fyn kinase inhibitor PP2. In some cases, cells were pre-incubated for 24 h with Ac-DEVD-al (100 M) or for 2 h with the p38 MAPK inhibitor SB203580 (1 M, Calbiochem). Cells were then rinsed, gently scraped, pelleted by centrifugation, then resuspended in 40 l of lysis buffer and analyzed as previously detailed (22).
p53 Transcriptional Activity-The PG13-luciferase p53 gene reporter construct (provided by Dr. B. Vogelstein) has been previously described (23). One g of PG13-luciferase was co-transfected with 1 g of a ␤-galactosidase transfection vector (to normalize transfections efficiencies) in TSM1 neurons and in PrP ϩ/ϩ and PrP Ϫ/Ϫ primary cultured neurons. Forty-eight hours after transfection, luciferase and ␤-galactosidase activities were measured according to previously described procedures (23).
Statistical Analysis-Statistical analyses were performed with Prism software (Graphpad Software, San Diego, CA) using Newman-Keuls multiple comparison test for one-way analysis of variance and the unpaired Student's t test for pair-wise comparisons.

PrP c Over-expression Increases Basal and Staurosporine-induced Toxicity and DNA Fragmentation in TSM1 Neurons-
We have obtained stable TSM1 transfectants over-expressing 3F4MoPrP c (Fig. 1A, upper panel). As expected, these cells secrete higher amounts of a 11-12-kDa fragment (Fig. 1A, lower panel), the immunological characterization of which previously led to its identification as the N-terminal product (called N1) derived from proteolysis of PrP c at the 110/111-112 peptide bond (18). The over-expression of PrP c in TSM1 neurons increases both basal and staurosporine-induced toxicity (Fig. 1B) and DNA fragmentation (Fig. 1C) as measured by XTT and PI incorporation, respectively. This phenotype has been observed for all clones examined (not shown).

Staurosporine-induced Caspase 3 Activation is Increased by PrP c Over-expression in TSM1 Neurons-Over-expression
of PrP c increases staurosporine-stimulated but not basal Ac-DEVD-al-sensitive Ac-DEVD-7AMC-hydrolyzing caspase 3-like activity in TSM1 neurons (Fig. 1D). This was accompanied by an augmentation of staurosporine-induced active caspase 3-like immunoreactivity (Fig. 1D). It should be empha-sized that PrP c expression potentiated the increase of caspase 3 activity triggered by several other apoptotic stimuli such as ceramide C2 and etoposide in TSM1 (Fig. 1E). PP2, a selective inhibitor of kinases belonging to the Src family, such as Lck and Fyn, (24) did not affect the extent of PrP c -induced staurosporine-stimulated caspase 3 activation in TSM1 neurons (Fig.  1F).
PrP c Antisense Approach Lowers Staurosporine-induced Caspase 3 Activation and DNA Fragmentation in TSM1 Neurons-To examine the endogenous contribution of PrP c to TSM1 susceptibility to staurosporine, we set up stable antisense-PrP c (ASPrP c ) TSM1 transfectants. This antisense approach led to a drastic reduction of endogenous PrP c expression and to a barely detectable N1 secretion ( Fig. 2A). Interestingly, the staurosporine-stimulated, but not the ba-

p53-dependent PrP c -induced Apoptosis in Neurons
sal caspase 3 activity observed in mock-transfected TSM1 neurons, was drastically reduced in ASPrP c -TSM1 cells (Fig.  2B). Concomitantly, PrP c -antisense abolished the staurosporine-induced augmentation of active caspase 3-like immunoreactivity (Fig. 2B). In line with these observations, the staurosporine-induced DNA fragmentation was reduced by about 50% in ASPrP c -expressing TSM1 neurons (Fig. 2C), a percentage that matched well the extent of the reduction of caspase 3 activity (see Fig. 2B).
PrP c -induced Caspase 3 Activation is p53-dependent in TSM1 Neurons-We examined the putative contribution of the tumor suppressor oncogene p53 to the PrP c -induced caspase 3 activation and toxicity by antisense and pharmacological approaches. We set up TSM1 stable transfectants expressing p53 antisense cDNA either alone (ASp53) or in combination with 3F4MoPrP c (3F4MoPrP c /ASp53)(see Fig. 4B). Our data show that all 3F4MoPrP c -related phenotypes can be prevented by antisense down-regulation of p53. Thus, Fig. 3A clearly shows that ASp53 drastically protects TSM1 cells from both basal and staurosporine-stimulated PrP c -mediated toxicity measured by the XTT assay (compare 3F4MoPrP c with 3F4MoPrP c /ASp53). Furthermore, ASp53 fully prevents the 3F4MoPrP c -induced DNA fragmentation (Fig. 3D, compare 3F4MoPrP c -sts with 3F4MoPrP c /ASp53-sts). Finally, ASp53 totally reverses the staurosporine-induced caspase activation triggered by 3F4MoPrP c (Fig. 3B, compare 3F4MoPrP c and 3F4MoPrP c / ASp53 transfectants).
We took advantage of the recent design of a specific p53 inhibitor, pifithrin-␣ (25), to further confirm the role of endogenous p53 in PrP c -mediated caspase activation. This pharmacological approach allowed us to show that pifithrin-␣ abolishes the 3F4MoPrP c -mediated staurosporine-stimulated increase of caspase 3 activity (Fig. 3C, compare 3F4MoPrP c and Mock Ϫ/ϩpft). Interestingly, pifithrin-␣ appeared inactive in ASp53-TSM1 cells (Fig. 3C, compare ASp53 Ϫ/ϩpft), indicating that the antisense and pharmacological approaches indeed inhibited the same expected molecular target, i.e. p53. This conclusion was reinforced by the fact that pifithrin-␣ did not significantly modify caspase 3 activity displayed by doubly transfected 3F4MoPrP c /ASp53-TSM1 neurons that remain at the level of pifithrin-␣-treated mock-and 3F4MoPrP-transfected cells (Fig. 3C).
PrP c -expression Increases p53-like Immunoreactivity and Activity-Because PrP c -proapoptoptic phenotype appeared p53dependent, we examined whether PrP c could also affect p53 expression and transcriptional activity. First, our data show that TSM1 over-expressing 3F4MoPrP c displays higher endogenous p53-like immunoreactivity than mock-transfected cells (Fig. 4A). This was also observed in transfected HEK293 cells (Fig. 4E). Conversely, ASPrP c -TSM1 cells exhibit lower p53like immunoreactivity (Fig. 4A). To correlate the modulation of p53 immunoreactivity with its biological activity, we examined the p53 transcriptional activity by means of the PG13luciferase construct classically used as the p53 gene reporter (23). Interestingly, p53 transcriptional activity was drastically enhanced in 3F4MoPrP c -TSM1 cells (Fig. 4C) and lowered in ASPrP c -expressing neurons (Fig. 4C). As expected, ASp53-expressing cells exhibit lower p53 immunoreactivity (Fig. 4B, compare ASp53 and Mock) and totally prevents the p53 increase observed with singly transfected 3F4MoPrP c cells (Fig. 4B, compare 3F4MoPrP and 3F4MoPrP/ASp53 with ASp53). PrP c Controls p53 at a Post-transcriptional Level-We took advantage of a cell system in which p53 is functionally deficient (26) to examine the ability of PrP c to increase p53 expression in co-transfection experiments. Any PrP c -dependent increase in p53 could not be explained by a transcriptional activation as the p53 cDNA is driven by a cytomegalovirus promoter and therefore would imply post-transcriptional PrP c -dependent p53 modulation. As expected, mock-transfected cells do not display any p53 transcriptional activity (Fig. 5A). PrP c cDNA transfection does not modify this null phenotype. However, we establish that transient co-transfection of PrP c and p53 cDNAs drastically potentiates the p53 transcriptional activity associated with p53 cDNA transfection alone (compare p53 and 3F4/p53 in Fig. 5A). These data led us to search for post-transcriptional modifications triggered by PrP c that would have altered p53 transcriptional activity. Mdm2 was recently shown to decrease p53 metabolic stability thereby lowering its activity (27). We demonstrate here that Mdm2 fully reverses the PrP c -induced increase of p53 transcriptional activity (Fig. 5E).
As p38 MAPK was reported to up-regulate p53 activity via upstream decrease of Mdm2 expression (28), we examined the effect of a selective p38 MAPK inhibitor (29) and the endogenous levels of the active phosphorylated counterpart of p38 MAPK. SB203580 significantly reduced but did not abolish the staurosporine-induced caspase activation in mock-and 3F4MoPrP c -expressing neurons (Fig. 5D). Accordingly, overexpression of PrP c increases phospho-p38 MAPK immunoreactivity in both TSM1 neurons (Fig. 5B) and p53-inactive fibroblasts (Fig. 5C). Altogether, our data suggest that PrP c increases p53 activity by p38 MAPK activation and concomitant reduction of Mdm2.

DISCUSSION
Although a series of studies (1-9) attempted to delineate the prions-related pathogenic mechanisms, very few data concern the putative physiological function of PrP c . This lack of attention perhaps came in part from the fact that the absence of the protein does not seem absolutely crucial because mice in which the prnp gene had been invalidated are safe and healthy, with apparently poorly detectable alterations in their development and behavior (9). However, Kuwahara et al. (30) reported that hippocampal neurons prepared from Prnp Ϫ/Ϫ mice undergo exacerbated cell death triggered by serum deprivation, thereby suggesting that PrP c could display an antiapoptotic tonus. More recently, an interesting study docu-mented the fact that PrP c protected human neurons from Bax-induced cell death (13). These data are apparently in opposition with a previous work demonstrating that transgenic mice over-expressing wild type PrP c could exhibit severe neurodegeneration (17). On the other hand, a recent article indicated that the post-natal deletion of the PrP c alters CA1 hippocampal neuron excitability but does not induce neurodegeneration (10) as would have been expected if the protein displayed an important anti-apoptotic function at adulthood. Furthermore, an infected hypothalamic neuronal cell line exhibits increased apoptosis (31). Whether this results from a dysfunction in the normal control of cell death by PrP c remains a possibility.
The 106 -126 domain of PrP c was shown to be toxic (14,15) and appears to activate several pro-apoptotic markers (16,32). This fragment is never proteolytically released as such from PrP c . In this context, either the bulk of studies examining the effect of 106 -126 just describe not relevant and artifactual data, or, more likely, the studies carried out with this fragment can be seen as a revelator of a PrP c function that would be pro-apoptotic. Our study clearly establishes by means of neuronal cell systems, stable transfections, antisense, and pharmacological approaches that PrP c drastically exacerbates cell responsiveness to pro-apoptotic stimuli. Thus, over-expression or specific induction of PrP c enhances staurosporine-induced cell toxicity, DNA fragmentation, and increases both caspase 3-like activity and immuno-reactivity that are all reversed by PrP c antisense approach. The fact that PrP c -mediated caspase activation also occurs in HEK293 cells (33) importantly indicates that the pro-apoptotic PrP c -related phenotype is not cell-specific.

p53-dependent PrP c -induced Apoptosis in Neurons
A recent study (34) suggested that PrP c could behave as a signaling molecule, transducing the signal after its coupling to the Src kinase Fyn. In our experiments, the selective Src inhibitor PP2 does not affect the PrP c -induced caspase activation, indicating that Fyn kinase phosphorylating activity was clearly not essential for PrP c -induced apoptosis in our experimental conditions. It should be noted that distinct cell types were used in the two studies, and more particularly the Fyn-dependent PrP c -mediated signaling appeared drastically dependent on the differentiation state of the 1C11 cells used in Mouillet-Richard et al. (35). Clearly the lack of Fyn dependence observed for PrP c -mediated phenotype in TSM1 cells indicates that Fyn requirement is likely restricted to few specialized PrP c phenotypes much more than for more general PrP function in cell death control.
We demonstrate that the PrP c -induced caspase 3 activation is mediated by p53. Indeed, PrP c expression increases p53 immuno-reactivity and transcriptional activity, whereas the PrP c antisense approach led to the opposite phenotype. Furthermore, p53 antisense blocked the PrP c -induced toxicity, DNA fragmentation, and caspase 3 activation. The opposite phenotype triggered by down-regulation of endogenous PrP c importantly indicates that the pro-apoptotic phenotype observed in PrP c -transfected cells was not artifactually related to the procedure involving over-expression of the protein. This is in agreement with our other data showing that Rov9 cells display a pro-apoptotic phenotype only after selective PrP c induction (33).
PrP c controls p53 expression and activity at a posttranscriptional level. Thus we were able to demonstrate that in a cell line deficient in active p53, PrP c drastically increases p53 activity after co-transfection of p53 and PrP c cDNAs. As p53 and PrP c cDNA are driven by a constitutively active cytomegalovirus promoter, p53 enhanced activity cannot be derived from increased transcription of p53, the genuine promoter of which leads to an inactive protein in our cell system. Interestingly, Mdm2 fully prevents the PrP c -induced p53 increase observed in the above cell system. Mdm2 was recently shown to control p53 expression at post-transcriptional levels (27) and particularly modulates p53 ubiquitination and degradation. Therefore, one could conclude that PrP c increases p53 expression and transcriptional activity via a decrease of the p53 down-regulator Mdm2. To support this hypothesis we examined the putative contribution of the p38 mitogenactivated protein kinase that was recently shown to downregulate Mdm2 and concomitantly increase p53 expression in neurons in a model of hypoxia. Indeed, we clearly showed that phospho-38 MAPK immuno-reactivity, the activated form of this kinase, was higher in PrP c -expressing TSM1 cells, in PrP ϩ/ϩ than in PrP Ϫ/Ϫ neurons, and could be increased by PrP c in fibroblasts (see Fig. 5). Furthermore, p38 inhibitor could partly block the PrP c -associated caspase 3 activation in transfected TSM1 neurons. Altogether, our data show that PrP c exerts a pro-apoptotic phenotype through the modulation of caspase 3 activity via a p53-dependent pathway that is positively and negatively controlled by p38 MAPK and Mdm2, respectively.
It is interesting to note that a recent study indicated that misfolded PrP c could be neurotoxic and could induce neurodegenerescence when accumulating in the cytosol, even at small amounts (35). Whether such a small amount of PrP c could account for the phenotype observed in our over-expression system is difficult to examine. However, the fact that our antisense approach led to an anti-apoptotic phenotype would suggest that even in normal conditions, a small fraction of endogenous PrP c could be responsible for our observed p53-dependent caspase 3 activation.
We recently documented the fact that PrP c undergoes basal and protein kinase C-regulated cleavage in the middle of its 106 -126 domain by proteases of the disintegrins family, thereby releasing a fragment that we called N1 (for review see Ref. 36). This PrP c cleavage also observed by Chen et al. (37) is reminiscent of the one taking place in the middle of the A␤ domain borne by the ␤-amyloid precursor protein (38). Interestingly, this intra-A␤ cleavage not only lowers A␤ production but also gives rise to a cytotrophic and neuroprotective secreted fragment called sAPP␣. Whether the cleavage generating N1 not only abolishes a pro-apoptotic control of PrP but also produces a fragment with opposite neuroprotective influence is currently being examined in our laboratory.