Amyloid-B Precursor-Like Protein APLP1 Is a Novel P53transcriptional Target Gene That Augments Neuroblastoma Cell Death Upon Genotoxic Stress

Home , APLP1, Biochemical cascade

Oncogene (2007) 26, 7302–7312 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Amyloid-b precursor-like protein APLP1 is a novel p53transcriptional target gene that augments neuroblastoma cell death upon genotoxic stress

X Tang, M Milyavsky, N Goldﬁnger and V Rotter

Department of Molecular Cell Biology, The Weizmann Institute of Science, Rehovot, Israel

The tumor suppressor p53is a key modulator of the predisposes cells to neoplastic transformation (Ryan cellular stress response, inducing cell-cycle arrest, apop- et al., 2001). tosis, senescence and cell differentiation. To evaluate p53 mediates many of its key functions largely by the further the molecular mechanism underlying p53function, transactivation or transrepression of its target genes the transcriptional profiles of proliferating and senescent (Bargonetti and Manfredi, 2002;Vousden and Lu, WI-38 cells, both wild-type p53 expressers and counter- 2002). Most tumor-derived mutants of p53 are defective parts with an inactivated p53, were compared by DNA in DNA binding and transactivation, supporting a microarray analysis. In particular, the amyloid-b pre- critical role for transactivation in p53’s ability to cursor-like protein 1 (APLP1) is induced in senescent cells suppress neoplasia. A number of p53 responsive genes in a p53-dependent manner. APLP1 was confirmed to be a have been identified and characterized to participate novel transcriptional target of p53by in vivo and in vitro downstream in different p53 cellular pathways, such as characterization of a p53responsive element found in the growth arrest, cellular differentiation or apoptosis. first intron of the APLP1 gene locus. APLP1 knockdown p53-dependent G1 arrest is primarily mediated by experiments demonstrate that APLP1 is required for the p21WAF1, a cyclin-dependent kinase inhibitor (el-Deiry proliferation of fibroblastic and epithelial cells. Moreover, et al., 1992). Many p53 target genes have been proposed depletion of APLP1 expression diminishes stress-induced to play roles in apoptosis, such as Bax, Bid and PUMA apoptosis of neural cells, whereas ectopic APLP1 expres- (p53-upregulated modulator of apoptosis), transmem- sion augments apoptosis. Based on these data, a mechanism brane proteins Perp, Fas and Killer/Dr5, and PIGs is proposed whereby p53-dependent induction of APLP1 is proteins. Bax, Bid and PUMA proteins induced by p53 involved in neural cell death, and which may exacerbate activate an intrinsic pathway to induce cell death. Fas neuronal cell loss in some acute or chronic neurodegenera- and Killer/Dr5 induce cell apoptosis through an tive disorders. extrinsic pathway. The PIGs, encode proteins that are Oncogene (2007) 26, 7302–7312;doi:10.1038/sj.onc.1210542; generally connected to the redox state of the cell, which published online 28 May 2007 may suggest that p53 activation affects the production of reactive oxygen species and causes cell death (Polyak Keywords: p53;APLP1;transcription;apoptosis;senes- et al., 1997). The diverse proapoptotic genes transacti- cence;neuroblastoma and Alzheimer’s disease vated by p53 may suggest that these genes may play a role in different contexts of p53-dependent apoptosis. Many acute or chronic neurodegenerative diseases such as ischemic stroke, Alzheimer’s disease and Parkinson’s disease result from degeneration and death Introduction of specific populations of neurons. Although the genetic and nongenetic factors that initiate these disorders differ The tumor suppressor p53 is a key modulator of the between the various diseases, a biochemical cascade of cellular stress response. In response to DNA damage, events executing the cell death process appears to be hypoxia, viral infection or oncogene activation, the shared (Culmsee and Mattson, 2005). Interestingly, protein is stabilized and activated. This leads to diverse activation of p53 has been observed frequently in acute biological effects, such as cell-cycle arrest, apoptosis, neuronal injury as well as in chronic neurodegenerative senescence, differentiation and anti-angiogenesis (Vousden disorders. Elevated levels of p53 messenger RNA and Lu, 2002). p53 is the most frequently altered (mRNA) and protein have been detected in brain tissue gene in human cancer (Hainaut et al., 1997). Absence derived from patients diagnosed with neurodegenerative of functional p53 allows cellular immortalization and diseases and from animal models with such disorders. This suggests a role for p53 in neuron death during neurodegeneration (Morrison and Kinoshita, 2000). Correspondence: Dr V Rotter, Department of Molecular Cell Biology, Alzheimer’s is a neurodegenerative disease character- The Weizmann Institute of Science, Rehovot 76100, Israel. E-mail: [email protected] ized by senile plaques containing amyloid-b peptide 142 Received 19 October 2006;revised 16 March 2007;accepted 18 April (Ab ) deposits, and these deposits are considered to be 2007;published online 28 May 2007 a major cause of the progressive degeneration and death APLP1 is a direct transcriptional target of p53 X Tang et al 7303 of neurons (Mattson, 2004). Neurotoxic Ab142 is a the bona fide p53 target gene p21waf1, as judged by our proteolytic product from amyloid precursor protein cluster analysis. (APP), which has two paralogues, namely amyloid-b To confirm that the induction of APLP1 gene precursor-like protein 1 (APLP1) and protein 2 (APLP2). expression in senescent cells is dependent on p53, the APP and APLP2 are expressed in tissues through- levels of APLP1 mRNA in proliferating and senescent out the body, while APLP1 expression appears to be WI-38 cells, as well as in their GSE56-infected counter- limited to the brain (Wasco et al., 1992). Knockout mice parts, were compared by quantitative reverse transcrip- studies have revealed that the functions of APP and tion (qRT)–PCR. As shown in Figure 1b, the level of its two paralogues are redundant (Heber et al., 2000). APLP1 mRNA is significantly induced in wild-type Indeed, both APLP1 and APLP2 have been found to senescent cells (p30V), but not in senescent cells where accumulate in the senile plaques of Alzheimer’s patients’ p53 has been inactivated (p30G). This is in contrast to the brains (Bayer et al., 1997;McNamara et al., 1998). The situation for APP and APLP2, two other members of the specific contribution of APLP1 and APLP2 to the APP family, which exhibit slightly increased mRNA pathogenesis of Alzheimer’s remains unclear. Notably, expression during replicative senescence, but in a p53- in mice overexpressing the Ab142, elevated p53 levels independent manner. Immunoblot analysis revealed were observed in those neurons associated with DNA that the APLP1 protein level is induced in senescent strand breaks (LaFerla et al., 1996;de la Monte et al., cells, and that this induction is absent in GSE56- 1997). These results raise the possibility that there are infected cells (Figure 1c). The basal level of APLP1, functional connections between p53 and APP or its different from p21Waf1, is not strongly affected by paralogues and neural cell death. p53. It is noteworthy that APLP1 protein undergoes In this work, the transcriptional profiles were post-translational glycosylation (Eggert et al., 2004), compared of senescent and proliferating WI-38 human giving rise to several bands, as was detected in this primary fibroblasts with their counterparts where p53 study. Western blot and RT–PCR analyses were used had been inactivated by DNA microarray analysis. to confirm that GSE56 infection abrogates p53 activity; Specific genes regulated by p53 during senescence were indeed massive p53 protein stabilization was observed, identified, in particular the APLP1. Various assays but p53 transactivation activity was eliminated com- demonstrated that APLP1 is transactivated directly by pletely, as indicated by reduced RNA and protein p53. In light of the putative role of APLP1 in expression of p21Waf1 (Figures 1b and c). Alzheimer’s disease, it was tested whether APLP1 Together, these experiments indicate that APLP1 influences neural cell death. Ectopic APLP1 expression expression is induced in a p53-dependent manner during was found to enhance neuroblastoma cell death, replicative senescence. whereas its depletion diminished cell death upon genotoxic stress. These results identify a novel p53 target gene that potentially influences neural cell death. APLP1 is induced by genotoxic stress in a p53-dependent manner It is well accepted that activated p53 induces the expression of specific target genes (el-Deiry, 1998). p53 Results can be activated by various stimuli, therefore it was interesting to examine whether genotoxic stress, the Induction of the APLP1 in senescent cells is p53 archetypal p53 inducer, results in elevated APLP1 dependent expression. To that end, APLP1 expression was Normal human primary fibroblasts undergo replicative examined in wild type (infected with control virus) and senescence after a limited number of cell divisions. p53 inactive (GSE56-infected) proliferating WI-38 cells Endogenous p53 can be inactivated using the GSE56 following exposure to doxorubicin (Dox). An increase in polypeptide that works by a negative-dominant mechan- APLP1 mRNA levels is observed only in control cells ism (Ossovskaya et al., 1996). This approach was taken following Dox exposure, and not in the GSE56-infected to inactivate p53 and was found to extend effectively the cells. A similar mRNA pattern is observed for p21Waf1 replicative life span of WI-38 human primary cells (Figure 2a). In contrast, APP and APLP2 mRNA levels (Figure 1a), suggesting that p53 is activated during do not change detectably in response to Dox. Immuno- cellular senescence and inhibits cell growth. To delineate blot analysis revealed that the amount of APLP1 protein further the role of p53 in cellular senescence, DNA accumulates in a relatively slow manner following Dox microarray analysis was used to compare the gene- treatment, and that this induction requires p53 activity, expression profiles from four types of WI-38 cells: since it is absent in GSE56-infected cells (Figure 2b). As proliferating ‘wild-type’ cells (infected with control expected, p21Waf1 protein is induced only in treated virus), senescent wild-type cells and their ‘p53 inactive’ control cells (Figure 2b). In addition, p53-dependent (GSE56-infected) counterparts. Clusters of genes were induction of APLP1 upon stress was also abolished observed that are either upregulated or downregulated by p53-specific RNAi in WI-38 proliferating cells during senescence only when p53 is wild type (Tables 1 under stress condition (Supplementary Figure S1). and 2 in Supplementary data). In particular, it was These results indicate that APLP1, but not APP and notable that APLP1, a paralogue of Alzheimer’s protein APLP2, is regulated specifically by p53 during genotoxic APP, is induced in senescent cells in a manner similar to stress.

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7304

Figure 1 Induced amyloid-b precursor-like protein 1 (APLP1) expression in senescent WI-38 cells is p53 dependent. (a) Life-span curves of empty vector (puro) and GSE56-infected WI-38 cells. Mean population doublings (PDLs) were calculated as detailed in the Material and methods section. (b) Quantitative PCR analysis of gene expression in proliferating (p22V), senescent (p30V), proliferating GSE56 expressing (p22G) and senescent GSE56-expressing (p30G) WI-38 cells. Values shown are means7s.d. of duplicate readings and represent expression patterns normalized using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression. (c) Immunoblot analysis of protein expression in whole cell lysates from proliferating (p22), senescent (p30) cells, proliferating GSE56 expressing (p22G) and senescent GSE56-expressing (p30G) WI-38 cells. b-Tubulin expression was used as a control for protein loading.

To corroborate these data, similar experiments were from a bone metastasis and possess a wild-type p53. performed using a different cellular system. MCF7 is a Viral oncoprotein E6 expression was used to create a breast cancer cell line characterized to possess wild-type counterpart cell line where p53 is inactive, as this viral p53. A counterpart cell line (p53RNAi) was created by protein triggers p53 degradation (Scheffner et al., 1990). transfection with specific p53 RNAi (Brummelkamp As shown in Figure 2e, the level of APLP1 mRNA is et al., 2002). APLP1 expression was assessed in these induced in SK-N-SH-puro cells (infected with empty cells before and after treatment with Dox. APLP1 virus) following treatment with either Dox or etoposide mRNA increases in MCF7-puro cells following Dox (Etp). However, no alteration in APLP1 levels is exposure. In contrast, there is no significant alteration in observed in E6-infected cells. Consistently, APLP1 pro- the level of APLP1 mRNA in MCF7-p53RNAi cells tein level is induced after treatment only in SK-N-SH (Figure 2c). Similarly, at the protein level, APLP1 control cells (Figure 2f). As described above, the p53 induction is only evident in Dox-treated MCF7-puro status of each cell line was checked by examining p53 cells (Figure 2d). The functionality of p53 RNAi was and p21Waf1 induction upon stress, which as expected is validated by the significantly reduced levels of p53 and absent in E6-infected cells (Figures 2e and f). p21Waf1 protein in MCF7-p53RNAi cells following Dox Taken together, these data demonstrate that APLP1 treatment (Figures 2c and d). These results using the expression is induced upon genotoxic stress in a p53- MCF7 system confirm that APLP1 induction after dependent manner in various cell types. genotoxic stress is p53 dependent. Previously, it was reported that APLP1 is a brain- specific protein (Kim et al., 1995). Therefore, it was APLP1 gene is a direct target of p53 important to check APLP1 expression in a neural The p53-dependent induction of APLP1 mRNA in system, and so similar experiments were carried out various cell types pointed to the possibility that p53 using SK-N-SH neuroblastoma cells, which established transactivates APLP1 gene expression directly. Almost

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7305

Figure 2 Amyloid-b precursor-like protein 1 (APLP1) expression is induced by genotoxic stress in a p53-dependent manner in several cell types. (a) Quantitative real-time (qRT)–PCR analysis of gene expression in vector or GSE56-infected proliferating WI-38 ﬁbroblasts following exposure to doxorubicin (Dox;0.2 mg/ml) for 24 h. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expression was used for normalization. (b) Immunoblot analysis of protein expression in vector or GSE56 infected proliferating WI-38 ﬁbroblasts following exposure to Dox (0.2 mg/ml) at the indicated times. b-Tubulin expression was used as a control for protein loading. (c) qRT–PCR analysis of gene expression in MCF7-puro and MCF7-p53RNAi cells treated with Dox (0.2 mg/ml) for 24 h. (d) Immunoblot analysis of protein expression in MCF7 cells treated with Dox (0.2 mg/ml) at the indicated times. (e) qRT–PCR analysis of APLP1 expression in SK-N-SH puro and SK-N-SH E6 cells treated with Dox (0.1 mg/ml) or etoposide (Etp;10 mM) for 14 h. (f) Immunoblot analysis of protein expression in SK-N-SH cells treated with Dox (0.1 mg/ml) or Etp (10 mM) at the indicated times.

all genes transactivated by p53 contain one or more p53 performed. MCF7 cells, control and Dox treated, were responsive elements (p53REs), with a sequence that subjected to cross-linking procedures. Cross-linked conforms generally to the consensus defined by el-Deiry protein–DNA complexes were immunoprecipitated et al. (1992). Therefore, the APLP1 gene locus was using either an anti-p53 antibody, or an anti-APLP1 screened at the sequence level for potential p53REs antibody, which served as a negative (nonspecific (NCBI human genomic database). Several were found in binding) control. The precipitated DNA was amplified the APLP1 gene locus;the three best candidates were by PCR using primers specific to each of the three designated as p53RE-1, p53RE-2 and p53RE-3 potential p53REs. As seen in Figure 3b, only APLP1 (Figure 3a), and were analysed further. To verify p53RE-3 is amplified specifically from anti-p53 immu- whether p53 binds to these APLP1 loci in vivo, noprecipitated complexes. The two other genomic chromatin immunoprecipitation (ChIP) assays were regions (p53RE-1 and p53RE-2) are each amplified

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7306

Figure 3 Amyloid-b precursor-like protein 1 (APLP1) is a direct transcriptional target of p53. (a) Schematic representation of APLP1 promoter constructs. Genomic fragments containing the APLP1 promoter, the first exon, the partial first intron and combinations were inserted into the basic luciferase reporter plasmid PGL3. The sequences of three potential p53 responsive elements (p53Res;spacer sequences are omitted) in the APLP1 gene locus and a mutation of p53RE-3 are detailed in the insert panel. The arrows indicate the approximate position of three potential p53REs in the APLP1 gene locus. The boxes indicate the first exon of APLP1 gene. (White box represents untranslated region and black box represents coding region.) (b) Chromatin immunoprecipitation (ChIP) assay in MCF7 cells which treated with ( þ ) or without () 0.2 mg/ml doxorubicin (Dox) for 24 h. (c) Various APLP1 reporters were co-transfected with a wild-type p53 expression plasmid or empty vector control plasmid into H1299 cells. The reporter activity was measured 48 h after transfection and is normalized to p53-independent b-galactosidase activity to correct for transfection efficiency. (d) The wild-type APLP1-Junc or the mutated APLP1-Junc-Mut reporter were co-transfected with increasing amounts (2.5 and 10 ng, respectively) of wild-type p53 expression plasmid into H1299 cells. (e) The APLP1-Junc reporter was co-transfected with wild-type p53 or p53-22,23, R175H p53 or R273H p53 mutant expression plasmids into H1299 cells.

equivalently from anti-p53 and control antibody immu- transcriptional induction for p53 mutants, such as the noprecipitated complexes. The integrity and specificity transcriptionally inactive mutant p53-22, 23 and several of this ChIP assay is validated by amplification of p53 core mutants (Figure 3e). This corroborates further the p21Waf1 promoter only from complexes immunopre- that p53RE-3 is a bona fide locus capable of mediating cipitated by anti-p53 (Figure 3b). These results suggest specifically wild-type p53 transactivation. that the p53RE-3, which is located in the first intron of In all, these results indicate that APLP1 is a direct the APLP1 gene, is a functional p53-binding site. It is transcriptional target of p53, where p53 binding is noteworthy that the left part of p53RE-3 shares mediated by a specific p53RE-3 in the first intron of the homology with the p53RE consensus, while the right APLP1 gene locus. part is less homologous due to three nucleotides inserted within the core region. Having established that p53 binds directly to the APLP1 expression is required for fibroblastic and APLP1 gene in vivo via the p53RE-3, it was necessary to epithelial cell proliferation verify the functionality of this site. To this end, various Although APLP1 is considered a neural-specific protein, APLP1 genomic fragments (Figure 3a) were inserted this study has demonstrated that it is induced in a p53- into a PGL3 basic luciferase reporter plasmid, and their dependent manner in cultured fibroblastic and epithelial ability to confer p53-dependent luciferase activity was cell lines, during senescence or in response to genotoxic measured. As seen in Figure 3c, each of the APLP1 stress. To illuminate the biological role of APLP1 in reporters containing p53RE-3 (1.2Kbpro-Junc or Junc these cell types, its expression was modulated ectopically constructs) is induced by coexpression of wild-type p53. and the phenotypic consequences were evaluated. APLP1 reporters without the p53RE-3 region are not Unexpectedly, ectopic expression of APLP1 in WI-38 induced and interestingly are inhibited slightly by p53 cells has no detectable effect on cell proliferation or life coexpression. The capacity of p53RE-3 to confer p53- span (Figure 4a). Moreover, genotoxic stress induces dependent luciferase activity is abrogated if three apoptosis in APLP1 overexpressing cells to the same nucleotides are changed in its left part (GGA- degree as in control cells overexpressing green fluore- CATGTCC to GGAtccGTCC;Figure 3a), indicating scent protein (GFP, data not shown). that this locus within the APLP1-Junc reporter indeed As no phenotypic changes were detected upon APLP1 represents the p53-binding site (Figure 3d). The p53RE- overexpression, another approach was taken whereby 3 in the APLP1-Junc reporter does not mediate APLP1 expression was reduced using RNAi method,

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7307

Figure 4 Amyloid-b precursor-like protein 1 (APLP1) is required for fibroblastic and epithelial cell proliferation. (a) Life-span curves of green fluorescent protein (GFP) and APLP1-infected WI-38 cells. Immunoblot analysis demonstrating ectopic APLP1 expression is shown in the insert panel. (b) Immunoblot analysis of protein expression in WI-38 cells infected with LacZi, APLP1 RNAi-1 or RNAi- 2 producing plasmids was selected. (c) Growth curves of LacZi, APLP1–RNAi-1 and APLP1–RNAi-2 infected proliferating WI-38 cells. (d) Life-span curves of LacZi, APLP1–RNAi-1 and APLP1–RNAi-2-infected WI-38 cells. (e) Growth curves of various MCF7- infected cells. MCF7 puro and p53RNAi cells were infected with APLP1 antisense (APLP1-AS) or GFP-producing plasmids. and any consequent phenotypic changes were assessed. In summary, the studies using APLP1 RNAi or Two different RNAi constructs (APLP1–RNAi-1 and antisense method suggest that APLP1 is required for APLP1–RNAi-2) were created using pRetroSuper vec- fibroblastic and epithelial cell growth under normal tor, each designed to knockdown APLP1 expression culture conditions that is independent on the function when infected into WI-38 cells. Western blot analysis of p53. indicates that the endogenous APLP1 protein level is reduced successfully by APLP1–RNAi-1, but is not affected by APLP1–RNAi-2. Indeed, APLP1–RNAi-2- Knockdown of APLP1 expression diminishes stress- infected cells exhibit similar levels of APLP1 protein as induced cell death in neuroblastoma cells LacZi-infected control cells (Figure 4b). Strikingly, APLP1 is a neural-specific protein and its function in depletion of APLP1 expression by APLP1–RNAi-1 neural cells remains unclear (Wasco et al., 1992;Bayer reduces dramatically the proliferation of proliferating et al., 1997). Based on our observations that p53- WI-38 cells (Figure 4c). This inhibitory effect on growth dependent regulation of APLP1 occurs in neural cells, it is is not attributable to induction of classical growth interesting to delineate the neural function of this protein. inhibitory genes, since there are no detectable differ- To that end, RNAi methods were employed to knock- ences in the protein levels of p21Waf1, p16 and p27 in down APLP1 expression in neuroblastoma cells and APLP1–RNAi-1 cells (Figure 4b). Furthermore, when consequent phenotypic changes were evaluated. APLP1– approaching senescence, APLP1–RNAi-1 cells perform RNAi-1 expression reduces significantly APLP1 protein fewer population doublings (PDLs) than control cells expression in two human neuroblastoma cell lines, SK-N- (Figure 4d). Similar results were obtained with MCF7 SH and CHP134. No effect was found with APLP1– cells. Notably, depletion of APLP1 expression by RNAi-2 or with the LacZ RNAi (Figures 5b antisense (APLP1-AS) causes a reduction in cell and e). Unexpectedly, depletion of APLP1 in SK-N-SH proliferation both in MCF7-puro cells and in MCF7- and CHP134 cells has no significant effects on cell p53RNAi cells, where p53 itself has been depleted proliferation (data not shown). This is in contrast to our by RNAi (Figure 4e). This implies that the findings described above, where depletion of APLP1 in inhibitory growth effect due to APLP1 depletion is both WI-38 and MCF7 cells causes a significant reduction p53-independent. in cell proliferation. However, when APLP1-depleted

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7308

Figure 5 Knockdown of amyloid-b precursor-like protein 1 (APLP1) expression by RNAi reduces stress-induced neuroblastoma cell death. (a) WST1 cell survival assay. SK-N-SH neuroblastoma cells infected with LacZi, APLP1–RNAi-1 or APLP1–RNAi-2 producing plasmids were treated either with or without 50 ng/ml doxorubicin (Dox) or 5 mM etoposide (Etp) for 24 h, then subjected to the WST1 assay. The readouts of treated cells are normalized to those of the nontreated cells, which were set to 100%. These data represent the average and standard deviation of independent experiments (a P-value o0.05 is indicated by *n ¼ 3–5). (b) Immunoblot analysis of protein expression in SK-N-SH cells treated with 0.1 mg/ml Dox for 24 h. b-Tubulin expression was used as a control for protein loading. (c) Quantitative analysis of the sub-G1 population by ﬂow cytometric analyses (FACS). Various SK-N-SH cells were treated with or without Dox (0.1 mg/ml) for 24 or 36 h. Cells were ﬁxed, propidium iodide (PI) stained and subjected to FACS. (d) CHP134 neuroblastoma cells infected with LacZi, APLP1–RNAi-1 or APLP1–RNAi-2 producing plasmids were treated either with or without Dox or Etp at the indicated concentration for 24 h, then subjected to the WST1 cell survival assay (a P-value o0.05 is indicated by #n ¼ 3). (e) Immunoblot analysis of protein expression in CHP134 cells were treated either with or without 40 ng/ml Dox or 2 mM Etp for 24 h.

SK-N-SH cells are exposed to Dox or Etp, these assay (Figure 5d). This resistance to stress-induced cell cells exhibit less apoptosis (15–20%) than control cells death is manifested by the partial cleavage of PARP1 (LacZ RNAi or APLP1–RNAi-2 cells) as assessed by protein observed in APLP1-depleted cells, in contrast to WST1 survival assays, which measures the metabolic the complete cleavage observed in the two control cell activity of viable cells (Figure 5a). An assay for lines under the same stress conditions (Figure 5e). There apoptosis, where caspase cleavage of PARP1 protein is is a poor p21Waf1 and APLP1 induction in response to evaluated by immunoblot analysis corroborates that stress (Figure 5e), which may be due to the documented stress-induced apoptosis is reduced in APLP1-depleted cytoplasmic sequestration of wild-type p53 in this cell SK-N-SH cells as compared to control cells (Figure 5b). line (Moll et al., 1996). Additionally, flow cytometric analysis of propidium Overall, these data suggest that depletion of APLP1 iodide (PI)-stained cells, which reveals apoptotic cells in neuroblastoma cells reduces apoptosis following as a result of their sub-G1 DNA content, confirms genotoxic stress, but does not influence detectably cell that APLP1–RNAi-1 SK-N-SH cells undergo less proliferation. stress-induced apoptosis, since their sub-G1 population is significantly smaller than the sub-G1 population Ectopic expression of APLP1 enhances stress-induced of control cells (Figure 5c). Depletion of APLP1 does neuroblastoma cell death not alter stress induced p53 accumulation or transcrip- To explore further the role of APLP1 in neuroblastoma tional activation of the p53 target p21waf1, suggesting cell death upon genotoxic stress, APLP1 was over- that APLP1 might act downstream of p53 (Figure 5b expressed ectopically and any effects on apoptosis and data not shown). monitored. SK-N-SH cells were infected with a virus In CHP134 neuroblastoma cells, depletion of APLP1 encoding APLP1 and a stable producer cell line selected results in a more pronounced resistance to stress- for further analysis. After exposure to Dox, the APLP1 induced cell death. There is a B30% increase in Dox overexpressing cells exhibit enhanced cell death resistance and a B20% increase in Etp resistance in (15B20%), when compared to GFP-infected control comparison to control cells according to the WST1 cells, according to WST1 assays (Figure 6a).

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7309 Similar to the APP paralogue, APLP1 also appears to of APP generated by caspase cleavage has been shown undergo proteolytic processing by various enzymes such to be toxic to cells (Galvan et al., 2002). However, as a-, b-, g-secretase and caspases (Eggert et al., 2004;Li comparable caspase-cleaved products of APLP1 were and Sudhof, 2004). The C-terminal proteolytic product not detected in the present study, making it unlikely that this is a feature of APLP1 activity (data not shown). It has been reported that the acidic domain of APP is essential for its ability to inhibit heme oxygenase activity and for targeting APP protein to mitochondrial compartments (Takahashi et al., 2000;Anandatheertha- varada et al., 2003). It is likely that the analogous domain is functionally significant. To test this premise, an APLP1 mutant where the zinc acidic domain had been deleted (APLP1-ZA) was introduced into cells and apoptosis monitored. It was observed that ectopic expression of APLP1-ZA does not affect stress-induced cell death (Figure 6a). This result suggests that the zinc and acidic domains are indeed required for APLP1’s role in apoptosis. This finding was further corroborated using PARP1 and flow cytometric analyses (FACS) (Figures 6b and c). PARP1 is cleaved more effectively and the sub-G1 population is larger in APLP1 overexpressing cells as compared with APLP1-ZA overexpressing cells following stress treatment (Figure 6b). Specifically, the APLP1-ZA overexpressing cells be- haved like the control cells. Taken together, these results suggest that the enhance- ment of apoptosis induced by ectopic APLP1, is dependent on the integrity of its zinc acid domain.

Discussion

This study identifies the APLP1 gene as a novel target gene for the tumor suppressor protein p53. Modulation of APLP1 protein expression affects the neuroblastoma cell death upon stress. Importantly, APLP1 expression is associated with Alzheimer’s disorders (McNamara et al., 1998;Bayer et al., 1999). Such a functional relationship between p53 and APLP1 might shed light on the pathogenesis of these disorders. The conclusion that APLP1 is a novel direct transcriptional target of p53 is based on the following three criteria. First, exposure of primary fibroblasts, breast carcinoma or neuroblastoma cells to genotoxic stress induces an increase in APLP1 mRNA and protein that is paralleled by increases in p21WAF1 mRNA and protein. This induction of APLP1 is wild-type p53 Figure 6 Ectopic expression of amyloid-b precursor-like protein 1 dependent, since cells where p53 has been inactivated (APLP1) enhances stress-induced cell death of neuroblastoma cells. (a) WST1 cell survival assay. SK-N-SH cells infected with green by various methods, such as GSE56 polypeptide, p53- fluorescent protein (GFP), APLP1 or zinc-acid domain deleted specific RNAi or the oncoprotein E6, show consistently APLP1 (APLP1-ZA) expression plasmids were treated either with a striking reduction in the stress-induced expression of or without 50 ng/ml doxorubicin (Dox) or 5 mM etoposide (Etp) for APLP1. Second, in silico analysis of the APLP1 gene 24 h, and then subjected to the WST1 assay. The readouts of locus identifies three potential p53-binding sites that treated cells are normalized to those of the nontreated cells. This data represents the average and standard deviation of independent conform to the consensus defined by el-Deiry et al. experiments (a P-value o0.01 is indicated by *n ¼ 4). (b) (1992). It is well established that p53 activates its targets Immunoblot analysis of protein expression in SK-N-SH cells were by binding to one or more p53REs, commonly located treated either with or without Dox (0.1 mg/ml) for 20 h. (c) in promoter or intron regions. The p53RE consensus is Quantitative analysis of the sub-G1 population by flow cytometric analyses (FACS). Various SK-N-SH cells were treated either with two copies of the 10 base-pair motif RRRCWWGYYY or without Dox (0.1 mg/ml) for 24 h. A P-value o0.05 is indicated separated by 0–13 bp. ChIP assays indicate that the by #(n ¼ 3). genomic region containing p53RE-3 is bound by p53

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7310 in vivo, suggesting that this indeed represents a functional 1998). Based on our results, we suggest that induction of p53-binding site. Finally, in vitro reporter assays demon- APLP1 expression by p53, which responds to some strate that p53RE-3 does confer specifically p53-dependent intrinsic stress in brain, may further enhance neuronal expression. apoptosis and augments neurodegeneration. Having discovered that transcription of APLP1 is In the case of fibroblastic and epithelial cells, induced by wild-type p53 during senescence, it was proliferation was reduced when APLP1 expression was interesting to find out if APLP1 expression is induced by reduced by RNAi or antisense. The results argue that p53 in other circumstances and to investigate the APLP1 is required for the proliferation in epithelial and functional role of APLP1. APLP1 is indeed induced in fibroblastic cell types. Such a conclusion may be a p53-dependent manner in response to genotoxic stress supported by a report that the APLP1 gene knockout in fibroblastic (WI38), epithelial (MCF7) and neural mouse survives but with a slight growth deficit (Heber (SK-N-SH and CHP134) cell types. Unexpectedly, et al., 2000). Moreover, it has been shown that APP ectopic APLP1 expression has no significant effects on protein is required for the proliferation of epithelial cells cell proliferation or apoptosis in fibroblastic and and some human carcinomas (Pietrzik et al., 1998;Meng epithelial cell types. However, in neural cell types, where et al., 2001;Ko et al., 2004). However, these data raise APLP1 is expressed exclusively (Kim et al., 1995;Lorent the question, why is the proliferation of neural cell types et al., 1995), the function of APLP1 is in stress-induced unaffected when APLP1 expression is disturbed? Since apoptosis. Ectopic APLP1 expression enhances stress- neural cells rarely divide in vivo, it may be that the induced cell death. In a complimentary approach, where function of APLP1 is cell type specific. RNAi was used to reduce APLP1 expression, neuro- In conclusion, APLP1 is a novel direct transcriptional blastoma cells exhibited the reduction of stress-induced target of the p53 tumor suppressor. APLP1 is a tissue- apoptosis. specific target;indeed it is required for cell proliferation Many proapoptotic genes regulated by p53 have been in fibroblastic and epithelial cells, but not for neural cell found to play an important role in neural apoptosis. For proliferation. However, in neural cells, APLP1 expres- example, the Bcl-2 family protein Bax is a major sion influences stress-induced cell death. mediator of p53-dependent neuronal apoptosis (Cregan et al., 1999). Other p53-regulated proteins, such as PUMA and Noxa that are characterized to be involved Materials and methods in endoplasmic reticulum stress, and Bid that is considered to influence glutamate toxicity and ischemia, Cell culture may play also key roles in the specific case of neuronal Primary human embryonic lung fibroblasts (WI-38, ATCC) apoptosis (Ward et al., 2004). Notably, in the present were grown in minimum essential medium supplemented with study, ectopic expression of APLP1 alone does not 10% fetal calf serum (FCS), 1 mM Sodium Pyruvate, 2 mM induce neuroblastoma apoptosis, but only enhances cell L-glutamine. Cells were trypsinized and seeded at a cell density of 1 Â 105 cells/plate for proliferating cells and at a cell density death upon stress stimuli, which implies that APLP1 of 2 Â 105 cells/plate for near senescent cells and were grown is not a neural cell death executor, but more likely a for 1 week. The cumulative mean PDLs were calculated using co-effector. Interestingly, APLP1 is a transmembrane the formula: PDLs ¼ log (cell output/cell input)/log 2. MCF7 protein at the plasma membrane (Wasco et al., 1992), breast carcinoma, SK-N-SH and CHP134 human neuroblas- which relates it to several p53 targets such as KILLER/ toma cells (provided by Dr Ute M Moll, Stony Brook DR5, Fas and PERP (el-Deiry, 1998;Attardi et al., University, New York, USA), and Phoenix cells (Clontech, 2000). This raises the possibility that APLP1 may be Mountain View, CA, USA) were grown in Dulbecco’s functionally analogous to KILLER/DR5 and Fas, and modified Eagle’s medium supplemented with 10% FCS and serves as a membrane receptor to receive either 2mML-glutamine. H1299 cells (ATCC) were grown in RMPI autocrine or paracrine signals. Depletion of APLP1 supplemented with 10% FCS. All the cells were maintained in a humidified incubator at 37 1C and 5% CO . causes neuroblastoma cells to acquire resistance for 2 genotoxic stress. Thus, loss or reduction of APLP1 Microarray hybridization, RNA isolation and quantitative PCR expression may contribute to chemotherapy resistance DNA microarray hybridization and processing was carried out of neuroblastoma. as described previously (Milyavsky et al., 2005). Total RNA More research is required to understand the mechan- was isolated using the RNeasy Mini Kit (Qiagen, Valencia, ism by which APLP1 enhances stress-induced cell death. CA, USA), and transcribed using the cDNA Synthesis Kit It has been reported that the acidic domain of APP can (Promega, Madison, WI, USA) according to the manufac- interact with heme oxygenase and inhibit its enzymatic turer’s protocol. Quantitative PCR was performed using the activity;thus neurotoxicity is augmented by heme or ABI Prism 7900 sequence detection system (Perkin-Elmer, Foster City, CA, USA). The qRT–PCRs were performed in H2O2 treatments (Takahashi et al., 2000). Deletion of the zinc and acidic domain from APLP1 does diminish duplicate, and the results were averaged. The following primer its potency as a cell death promoter, suggesting that pairs, forward and reverse respectively, were used to detect the indicated gene: APLP1 (AAGGGTCCACAGAACAAGATG APLP1 may promote cell death by certain mechanisms and GCATTCACCTTTCGCTCATA), APP (CTTCCCGT similarly to APP. It is interesting to note that APLP1 GAATGGAGAGTT and TCAACAGGCTCAACTTCGTT), may also be involved in Alzheimer’s disease. Accumula- APLP2 (CTCGCTTCCAGAAGGCTAAG and ATGGCTT tion of APLP1 has been found in the senile plaques and GGAAGTGCTGAAT), p21WAF1 (GGCAGACCAGCATGA in aged brains (Bayer et al., 1997;McNamara et al., CAGATT and GCGGATTAGGGCTTCCTCTT) and

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7311 glyceraldehyde-3-phosphate dehydrogenase (ACCCACTCCT fragment into the pWZL vector. The RNAi-producing plasmid CCACCTTTGA and CTGTTGCTGTAGCCAAATTCGT). was constructed as described previously (Tang et al., 2004). Each primer is described 50 to 30. APLP1 RNAi target sequences are GTAGGATGCGCCA GATTAA (APLP1–RNAi-1) and CACTCATCGGAGATT Immunoblot analysis CAGA (APLP1–RNAi-2), respectively. Immunoblot analysis was performed using standard proce- Retroviral infection was performed as described previously dures. Cells were lysed in NP-40 buffer (50 mM Tris, pH 8.0, (Tang et al., 2004). Infected cells were selected as follows: 1 mg/ 120 mM NaCl, 0.5% NP-40) supplemented with protease ml puromycin (for 1 week), 400 mg/ml G418 (for 2 weeks) or Inhibitor cocktail (Roche, Mannhein, Germany) for 20 min 5 mg/ml blasticidin (for 1 week). on ice to extract total protein. Protein concentration was determined using the BCA protein assay kit (Pierce, Rockford, ChIP IL, USA). Polyclonal anti-p53 (produced in our laboratory), ChIP was performed as described previously (Stambolsky antibodies for the CDK inhibitors p21Waf1, p16 and p27 from et al., 2006). DNA–protein complexes were immunoprecipi- Santa Cruz, polyclonal anti-APLP1 (Calbiochem, Beeston, tated using anti-p53 or anti-APLP1 antibodies, the latter Nottingham, UK), monoclonal anti-PARP (Biomol, Poly- serving as a control for nonspecific binding. The precipitated mouth Meeting, PA, USA) and monoclonal anti-b-tubulin DNA was subjected to PCR amplification using primers (Sigma, St Louis, MI, USA) were used for immunoblotting. specific to each of the three potential p53REs of the APLP1 Horseradish peroxidase anti-mouse and anti-rabbit (Sigma) gene and the p53RE in the promoter of p21Waf1. The following were used as secondary antibodies. The signal was detected by primer pairs, forward and reverse respectively, were used to the super-signal-enhanced chemiluminescence system (Pierce). detect APLP1 promoter sequences relative to the exon-1 starting site: B þ 250 bp region (CGCTGCTGCTGCCAC APLP1 promoter reporters and luciferase assay TATTG and GTTCCTATGCTCGAGATGGG); B1000 bp To generate the APLP1 promoter reporter constructs, region (CCTTCCACCTCAGCCTCCCAAGTA and ATGG individual APLP1 genomic fragments obtained either by CTGGGCACAGTGGCTCAT);and B3000 bp region PCR from WI-38 genomic DNA or by subrestriction, were (CCATTCTCTTGGCCTGAAACAC and GATGGTAGA cloned into the pGL3-basic luciferase reporter vector (Prome- GCCCTCAGCATA). The primers used for the detection of ga). The APLP1-3.2pro fragment was obtained by Sac1 p21 promoter sequences, which amplify the region near the restriction cleavage of the PCR products generated using the p53-binding site, were: forward-GCACTCTT-GTCCCCCAG following primers (forward-CCATTCTCTTGGCCTGAAA and reverse-TCTATGCCAGAGCTCAACAT. CAC and reverse-CGCGCAGAAGCAGCAGCAATAG). The APLP1-1.2pro promoter comprises the Sph1/Sac1 restriction fragment of APLP1-3.2pro. The APLP1-Junc fragment WST1 proliferation assay and FACS analysis was produced by Sac1/Xho1 restriction cleavage of PCR Cell survival in response to genotoxic stress treatments was products generated using the following primers (forward- assessed using a WST1 assay according to the manufacturer’s GGGTCTAAAGAGGGTGAGAGTC and reverse-GTTC instructions (Roche). The WST1 reagent was added and CTATGCTCGAGATGGG). APLP1-Exo1 contains the incubated with the cells for 1 h before reading the plate. Sac1/Bssh2 restriction fragment from APLP1-Junc. Wild-type Each treatment was conducted in triplicate. For FACS p53 (pC53-SN3), p53-175R/H and p53-273R/H plasmids were analysis, cells were collected and fixed in 70% ethanol in provided by Dr B Vogelstein (The Johns Hopkins University Hank’s balanced salt solution (HBSS). The following day, cells School of Medicine, Baltimore, MD, USA). The p53-22,23 were washed and resuspended in phosphate-buffered saline mutant plasmid was provided by Dr A Levine (The Rocke- buffer containing 50 mg/ml PI, 50 mg/ml RNAse A and 0.1% feller University, New York, NY, USA). Triton. Flow cytometry was performed using a Becton Luciferase assays were performed in p53 null H1299 cells as Dickinson flow cytometer. Statistical analyses were carried described previously (Tang et al., 2004). out by Student’s t-test.

APLP1 expression and RNAi constructs and retroviral infection Acknowledgements The APLP1 cDNA was ampliﬁed by PCR using the following primers (forward-GAGGGCGCAAGGGCCGGGACA and This research was supported by a Center of Excellence grant reverse-GGGCCGGGTCAGGGTCGTTCC) and cloned into from the Flight Attendant Medical Research Institute (FAM- the pWZL-blasticidin retroviral vector (kindly provided by Dr RI), EC FP6 grant LSHC-CT-2004-503576 and Yad Abraham W Hahn, Dana-Farber Cancer Institute, Harvard Medical Center for Cancer Diagnosis and Therapy. This publication School, Boston, MA, USA). The APLP1 zinc acidic domain- reﬂects our views and not necessarily those of the European deleted mutant (APLP1-ZA, 194-300 aa deletion) was created Community. The EC is not liable for any use that may be by Sph1 restriction of the pWZL-APLP1 plasmid, followed by made of the information contained herein. VR is the self-ligation. The APLP1 antisense (APLP1-AS) construct was incumbent of the Norman and Helen Asher Professorial Chair created by cloning the BamH1/EcoR1 restricted APLP1 cDNA Cancer Research at the Weizmann Institute.

References

Anandatheerthavarada HK, Biswas G, Robin MA, Avadhani target of p53, is a novel member of the PMP-22/gas3 family. NG. (2003). Mitochondrial targeting and a novel transmem- Genes Dev 14: 704–718. brane arrest of Alzheimer’s amyloid precursor protein Bargonetti J, Manfredi JJ. (2002). Multiple roles of the tumor impairs mitochondrial function in neuronal cells. J Cell Biol suppressor p53. Curr Opin Oncol 14: 86–91. 161: 41–54. Bayer TA, Cappai R, Masters CL, Beyreuther K, Multhaup Attardi LD, Reczek EE, Cosmas C, Demicco EG, McCurrach G. (1999). It all sticks together – the APP-related family of ME, Lowe SW et al. (2000). PERP, an apoptosis-associated proteins and Alzheimer’s disease. Mol Psychiatry 4: 524–528.

Oncogene APLP1 is a direct transcriptional target of p53 X Tang et al 7312 Bayer TA, Paliga K, Weggen S, Wiestler OD, Beyreuther K, Mattson MP. (2004). Pathways towards and away from Multhaup G. (1997). Amyloid precursor-like protein 1 Alzheimer’s disease. Nature 430: 631–639. accumulates in neuritic plaques in Alzheimer’s disease. Acta McNamara MJ, Ruff CT, Wasco W, Tanzi RE, Thinakaran Neuropathol (Berl) 94: 519–524. G, Hyman BT. (1998). Immunohistochemical and in situ Brummelkamp TR, Bernards R, Agami R. (2002). Stable analysis of amyloid precursor-like protein-1 and amyloid suppression of tumorigenicity by virus-mediated RNA precursor-like protein-2 expression in Alzheimer disease and interference. Cancer Cell 2: 243–247. aged control brains. Brain Res 804: 45–51. Cregan SP, MacLaurin JG, Craig CG, Robertson GS, Meng JY, Kataoka H, Itoh H, Koono M. (2001). Amyloid Nicholson DW, Park DS et al. (1999). Bax-dependent beta protein precursor is involved in the growth of caspase-3 activation is a key determinant in p53-induced human colon carcinoma cell in vitro and in vivo. Int J apoptosis in neurons. J Neurosci 19: 7860–7869. Cancer 92: 31–39. Culmsee C, Mattson MP. (2005). p53 in neuronal apoptosis. Milyavsky M, Tabach Y, Shats I, Erez N, Cohen Y, Tang X Biochem Biophys Res Commun 331: 761–777. et al. (2005). Transcriptional programs following genetic de la Monte SM, Sohn YK, Wands JR. (1997). Correlates of alterations in p53, INK4A, and H-Ras genes along p53- and Fas (CD95)-mediated apoptosis in Alzheimer’s defined stages of malignant transformation. Cancer Res 65: disease. J Neurol Sci 152: 73–83. 4530–4543. Eggert S, Paliga K, Soba P, Evin G, Masters CL, Weidemann Moll UM, Ostermeyer AG, Haladay R, Winkfield B, Frazier A et al. (2004). The proteolytic processing of the amyloid M, Zambetti G. (1996). Cytoplasmic sequestration of wild- precursor protein gene family members APLP-1 and APLP- type p53 protein impairs the G1 checkpoint after DNA 2 involves alpha-, beta-, gamma-, and epsilon-like cleavages: damage. Mol Cell Biol 16: 1126–1137. modulation of APLP-1 processing by n-glycosylation. J Biol Morrison RS, Kinoshita Y. (2000). The role of p53 in neuronal Chem 279: 18146–18156. cell death. Cell Death Differ 7: 868–879. el-Deiry WS. (1998). Regulation of p53 downstream genes. Ossovskaya VS, Mazo IA, Chernov MV, Chernova OB, Semin Cancer Biol 8: 345–357. Strezoska Z, Kondratov R et al. (1996). Use of genetic el-Deiry WS, Kern SE, Pietenpol JA, Kinzler KW, Vogelstein suppressor elements to dissect distinct biological effects B. (1992). Definition of a consensus binding site for p53. Nat of separate p53 domains. Proc Natl Acad Sci USA 93: Genet 1: 45–49. 10309–10314. Galvan V, Chen S, Lu D, Logvinova A, Goldsmith P, Koo EH Pietrzik CU, Hoffmann J, Stober K, Chen CY, Bauer C, Otero et al. (2002). Caspase cleavage of members of the amyloid DA et al. (1998). From differentiation to proliferation: the precursor family of proteins. J Neurochem 82: 283–294. secretory amyloid precursor protein as a local mediator of Hainaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M, growth in thyroid epithelial cells. Proc Natl Acad Sci USA Hovig E et al. (1997). Database of p53 gene somatic 95: 1770–1775. mutations in human tumors and cell lines: updated Polyak K, Xia Y, Zweier JL, Kinzler KW, Vogelstein B. compilation and future prospects. Nucleic Acids Res 25: (1997). A model for p53-induced apoptosis. Nature 389: 151–157. 300–305. Heber S, Herms J, Gajic V, Hainfellner J, Aguzzi A, Rulicke T Ryan KM, Phillips AC, Vousden KH. (2001). Regulation and et al. (2000). Mice with combined gene knock-outs function of the p53 tumor suppressor protein. Curr Opin Cell reveal essential and partially redundant functions of Biol 13: 332–337. amyloid precursor protein family members. J Neurosci 20: Scheffner M, Werness BA, Huibregtse JM, Levine AJ, Howley 7951–7963. PM. (1990). The E6 oncoprotein encoded by human Kim TW, Wu K, Xu JL, McAuliffe G, Tanzi RE, Wasco W papillomavirus types 16 and 18 promotes the degradation et al. (1995). Selective localization of amyloid precursor-like of p53. Cell 63: 1129–1136. protein 1 in the cerebral cortex postsynaptic density. Brain Stambolsky P, Weisz L, Shats I, Klein Y, Goldfinger N, Res Mol Brain Res 32: 36–44. Oren M et al. (2006). Regulation of AIF expression by p53. Ko SY, Lin SC, Chang KW, Wong YK, Liu CJ, Chi CW et al. Cell Death Differ 13: 2140–2149. (2004). Increased expression of amyloid precursor protein in Takahashi M, Dore S, Ferris CD, Tomita T, Sawa A, oral squamous cell carcinoma. Int J Cancer 111: 727–732. Wolosker H et al. (2000). Amyloid precursor proteins inhibit LaFerla FM, Hall CK, Ngo L, Jay G. (1996). Extracellular heme oxygenase activity and augment neurotoxicity in deposition of beta-amyloid upon p53-dependent neuronal Alzheimer’s disease. Neuron 28: 461–473. cell death in transgenic mice. J Clin Invest 98: 1626–1632. Tang X, Milyavsky M, Shats I, Erez N, Goldfinger N, Rotter Li Q, Sudhof TC. (2004). Cleavage of amyloid-beta precursor V. (2004). Activated p53 suppresses the histone methyl- protein and amyloid-beta precursor-like protein by BACE 1. transferase EZH2 gene. Oncogene 23: 5759–5769. J Biol Chem 279: 10542–10550. Vousden KH, Lu X. (2002). Live or let die: the cell’s response Lorent K, Overbergh L, Moechars D, De Strooper B, Van to p53. Nat Rev Cancer 2: 594–604. Leuven F, Van den Berghe H. (1995). Expression in mouse Ward MW, Kogel D, Prehn JH. (2004). Neuronal apoptosis: embryos and in adult mouse brain of three members of the BH3-only proteins the real killers? J Bioenerg Biomembr 36: amyloid precursor protein family, of the alpha-2-macro- 295–298. globulin receptor/low density lipoprotein receptor-related Wasco W, Bupp K, Magendantz M, Gusella JF, Tanzi RE, protein and of its ligands apolipoprotein E, lipoprotein Solomon F. (1992). Identification of a mouse brain cDNA lipase, alpha-2-macroglobulin and the 40 000 molecular that encodes a protein related to the Alzheimer disease- weight receptor-associated protein. Neuroscience 65: associated amyloid beta protein precursor. Proc Natl Acad 1009–1025. Sci USA 89: 10758–10762.

Supplementary Information accompanies the paper on the Oncogene Web site (http://www.nature.com/onc).

Oncogene