Transcriptional repression of the anti-apoptotic survivin gene by wild type p53.

Survivin is a member of the inhibitor of apoptosis family. This apoptosis inhibitor also has an evolutionarily conserved role as a mitotic spindle checkpoint protein. Previous studies on p53-repressed genes have implicated several genes involved in the G(2)/M transition of the cell cycle as targets of negative regulation by p53. However, few targets of p53 repression that are anti-apoptotic have been identified. This study identifies the anti-apoptotic survivin gene as a p53-repressed gene. Notably, Survivin repression by p53 is shown to be distinct from p53-dependent growth arrest. Chromatin immunoprecipitations indicate that p53 binds the survivin promoter in vivo; immunobinding studies indicate that this site overlaps with a binding site for E2F transcription factors and is subtly distinct from a canonical p53-transactivating element. The survivin-binding site contains a 3-nucleotide spacer between the two decamer "half-sites" of the p53 consensus element; deletion of this spacer is sufficient to convert the survivin site into a transactivating element. Finally, we show that overexpression of Survivin in cells sensitive to p53-dependent cell death markedly inhibits apoptosis induced by ultraviolet light. The identification of survivin as a p53 repressed gene should aid in the elucidation of the contribution of transcriptional repression to p53-dependent apoptosis.

Survivin is a member of the inhibitor of apoptosis family. This apoptosis inhibitor also has an evolutionarily conserved role as a mitotic spindle checkpoint protein. Previous studies on p53-repressed genes have implicated several genes involved in the G 2 /M transition of the cell cycle as targets of negative regulation by p53. However, few targets of p53 repression that are antiapoptotic have been identified. This study identifies the anti-apoptotic survivin gene as a p53-repressed gene. Notably, Survivin repression by p53 is shown to be distinct from p53-dependent growth arrest. Chromatin immunoprecipitations indicate that p53 binds the survivin promoter in vivo; immunobinding studies indicate that this site overlaps with a binding site for E2F transcription factors and is subtly distinct from a canonical p53transactivating element. The survivin-binding site contains a 3-nucleotide spacer between the two decamer "half-sites" of the p53 consensus element; deletion of this spacer is sufficient to convert the survivin site into a transactivating element. Finally, we show that overexpression of Survivin in cells sensitive to p53-dependent cell death markedly inhibits apoptosis induced by ultraviolet light. The identification of survivin as a p53 repressed gene should aid in the elucidation of the contribution of transcriptional repression to p53-dependent apoptosis.
survivin was first identified as a gene with a coding region complementary to the effector cell protease receptor, EPR-1. Although these genes share homology in their coding regions, they are transcribed in a reverse orientation and appear to share no regulatory or promoter regions (1). Sequence analysis of the survivin coding region revealed a conserved motif present at the amino terminus that identified it as a member of the inhibitor of apoptosis (IAP) 1 family. This motif, or baculovirus IAP repeat domain, has been shown to mediate the interaction with, and inhibition of, the caspase family of proteolytic enzymes, which are the penultimate mediators of apoptosis (2,3). Consistent with its predicted anti-apoptotic function, expression of antisense RNA for Survivin is sufficient to induce apo-ptosis in a number of human tumor cell lines (4 -6). Elevated expression of Survivin would be predicted to promote tumorigenesis, and in fact Survivin is highly expressed in a number of tumor types, including neuroblastoma, colorectal carcinoma, and gastric carcinoma; in these tumors, Survivin overexpression is correlated with poor prognosis (7)(8)(9). Additionally, analysis of the differences in gene expression between normal and tumor cells has revealed that survivin is one of the genes most consistently overexpressed in tumor cells relative to normal tissue (10).
Survivin is expressed widely in fetal tissues, but becomes restricted during development, and appears to be negligibly expressed in the majority of adult tissues (1,11). The expression of Survivin is also cell cycle-regulated. This gene is repressed in the G 1 phase of the cell cycle and is highly expressed in G 2 /M (12). This cell cycle regulation appears to rely on the presence of two CDE elements (cell cycle-dependent elements) that are downstream of the transcriptional start site (12), although a proximal CHR element (cyclin homology region) may also play a role. During mitosis, Survivin protein binds the mitotic spindle, and there is evidence that via this interaction this protein monitors mitotic spindle integrity; it is hypothesized that Survivin controls the elimination by apoptosis of those cells with aberrantly formed mitotic spindles (5,13). This role for Survivin is evolutionarily conserved, as it is shared among homologues in yeast (14) and Caenorhabditis elegans (15). Although deregulation of survivin gene expression appears to be a common and significant event in tumorigenesis, little is known regarding the important regulators of the expression of this gene in normal and tumor cells.
Like Survivin, the p53 tumor suppressor protein is also a critical mediator of apoptosis and tumorigenesis. p53 is a nuclear transcription factor that is latent in normal cells but becomes activated by a variety of cellular stresses such as DNA damage, hypoxia (insufficient oxygen), and the presence of activated oncogenes (for review see Ref. 16). Following induction of p53 by these stresses, p53 up-regulates a set of genes that can promote cell death and growth arrest, such as p21 waf1 , bax, fas, and KILLER/DR5 (for review see Ref. 17). p53 also negatively regulates the expression of a separate set of genes; in some cases this negative regulation has been shown to be important for the induction of apoptosis (18 -20). The exact nature of the binding site for p53 in repressed promoters, and how this site differs from that in transactivated promoters, has only begun to be elucidated. The mechanism of repression by p53 is also becoming more clearly elucidated; specifically, at least one mechanism whereby p53 negatively regulates gene expression involves an association between p53 and histone deacetylases (HDACs). This p53-HDAC interaction is mediated by binding of p53 to the co-repressor protein Sin3. The p53-Sin3 interaction targets HDACs to the promoters of p53-repressed genes, where HDACs serve to deacetylate histones and create a chromatin environment that is unfavorable for transcription (21).
In several studies the transcriptional repression activity of p53 has been implicated in p53-dependent apoptosis; notable in these studies is the finding that deletion of the proline-rich domain of p53 renders this protein competent as a transactivator but unable to induce apoptosis or to repress transcription (22)(23)(24). Additionally, tumor-derived mutant forms of p53 that are impaired for apoptosis induction are likewise unable to repress transcription, yet retain the ability to activate transcription (25). In support of the positive association between p53-mediated repression and apoptosis, we recently mapped the Sin3-binding domain of p53 to the proline-rich domain, which is critical for apoptosis induction by p53 (26). These and other studies raise the possibility that p53 may transcriptionally repress genes with antiapoptotic activity. In a search for genes that are negatively regulated by p53, we and others identified several genes with roles in the control of the G 2 /M phase of the cell cycle that are repressed following induction of wild type p53. These genes include stathmin, Map4, cyclin B1, cdc2, and cdc25c (27)(28)(29)(30). Repression of these genes following DNA damage has been shown to require wild type p53 and is hypothesized to constitute a DNA damageinduced G 2 /M checkpoint (31,32). These studies prompted us to test the possibility that Survivin, which binds to the mitotic spindle and exhibits anti-apoptotic activity, might likewise be subject to negative regulation by p53. Our studies have identified survivin as a gene that is potently repressed, at both the RNA and protein levels, following p53 induction in cells with both endogenous and inducible p53. The identification of survivin as a p53-repressed gene should aid in the elucidation of the mechanism of transcriptional repression by p53 and in the estimation of the contribution of this activity to p53-dependent programmed cell death.

EXPERIMENTAL PROCEDURES
Cell Culture, p53 Induction-MCF-7 cells were maintained in RPMI 1640 supplemented with 10% fetal bovine serum and 100 units/ml of penicillin and streptomycin. The human melanoma cell line CaCl and the derivative clone expressing the HPV E6 gene were cultured as described (27). The human osteosarcoma cell lines Saos2 and U2OS (kindly provided by Peter Adams, Fox Chase Cancer Center), the immortalized murine p21 knock-out fibroblasts (WD-50-5 cells, kindly provided by James Sherley, MIT, and established from cultures derived from Tyler Jacks, MIT), and the human lung adenocarcinoma cell line H1299 were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 100 units/ml of penicillin and streptomycin. Murine Val5 cells, containing the temperature-sensitive valine 135 allele of p53, human Val138 cells containing the valine 138 allele, and human tsp73 cells were maintained as described (33,34). Unless otherwise noted all cells were grown at 37°C in a 5% CO 2humidified atmosphere. For treatment with ultraviolet light, cells were irradiated with 0 -10 J/m2 (as noted) of UV-C with a Spectroline X series ultraviolet lamp, and output was monitored with a traceable UV light meter (Fisher). For drug treatment, cells were treated with 0.5 g/ml adriamycin (doxorubicin, Sigma) or dilution vehicle alone for 24 h. p53 induction was monitored by Western analysis of 100 g of protein as described (21), using 0.1 g/ml DO-1 (Ab-6, Calbiochem) or 421 (Ab-1, Calbiochem).
Reverse Transcription-PCR, Northern Blots-The full-length coding region for Survivin was generated by reverse transcription-polymerase chain reaction using 5 g of total RNA from the H1299 human lung adenocarcinoma cell line, using the Access Reverse Transcription-PCR kit (Promega) and the following oligonucleotides: forward 5Ј-ATGAGA-TACCATGGGTGCCCCGACG-3Ј, reverse 5Ј-TTAAGGATCCCTGCTC-GATGGCACG-3Ј. The cDNA was cloned into the pCR3.1 vector (Invitrogen) and subjected to DNA sequence analysis for sequence confirmation. Total RNA was isolated from cells using CsCl purification (19) or using Trizol, as per the manufacturer (Invitrogen). Northern analyses were performed as described (19). Probes for Northerns were radiolabeled using random primers (Prime-It-II, Stratagene) and [␣-32 P]dCTP (PerkinElmer Life Sciences). Autoradiographs were quantitated using NIH image software.
PCR of the survivin Promoter, Generation of Deletion Constructs-Approximately 1.1 kb of the survivin promoter region was generated by polymerase chain reaction and cloned into the vector pCR 2.1 (Invitrogen) using the following oligonucleotides: forward 5Ј-CTGGCCATAGA-ACCAGAGAAGTGA-3Ј, reverse 5Ј-CCACCTCTGCCAACGGGTCCCG-CG-3Ј, to generate the plasmid SpI. This sequence represents nucleotides 1821-2912 of the human survivin gene, GenBank TM accession number U75285. DNA sequence analysis confirmed the sequence of this region. The survivin promoter was then cloned into the promoterless luciferase vector pGL2-basic (Promega) to generate the plasmid SpII, for use in transfections and luciferase assays. The SpV deletion construct was generated from the SpI plasmid following digestion with SmaI and HindIII, generation of blunt ends by Klenow fill-in, and re-ligated; this promoter construct contains nucleotides 2331-2912 of the human survivin gene/promoter. For deletion of the p53-binding site in the survivin promoter, the SpV plasmid was digested with SacII, which flanks the p53-binding site. Digestion of SpV with SacII and religation of this construct eliminates the p53-binding site, as well as approximately 30 nucleotides upstream of the site, to generate the construct SpV-⌬p53BS. The Sp-min construct consists of the p53-binding site and the adjacent E2F-binding site; this construct was generated using PCR using the following overlapping oligonucleotides: 5Ј-AAAG-GTACCGGGCGTGCGCTCCCGACATG-3Ј and 5Ј-AAACTCGAGGGC-GCGCCGCGGGGCATGTCG-3Ј. This fragment was cloned into the plasmid pGL3-E1B-TATA, kindly provided by James Manfredi (Mount Sinai School of Medicine). The SpVII construct was generated in an identical manner, except that oligonucleotides were designed that deleted the 3-nucleotide spacer in the p53-binding site.
Transfections, Luciferase Assays-H1299, SaoS-2, or WD-50-5 (p21 Ϫ/Ϫ cells) cells were seeded in 10-cm plates at 1 ϫ 10 6 cells/plate and allowed to settle overnight. The next morning cells were transfected with 4 g of firefly luciferase reporter construct and 2-4 g of Renilla luciferase construct (pRL-CMV or pRL-tk, Promega) or SV40 ␤-galactosidase, along with the indicated amounts of p53 expression plasmid (in pRc/CMV) using Fugene, according to protocols derived from the manufacturer (Roche Molecular Biochemicals). After 24 h, the cells were harvested and lysed, and dual luciferase assays were performed as per the protocol derived from the manufacturer (Promega) on a Monolight 2010 luminometer (Analytical Luminescence Laboratory). Luciferase activity was normalized to total protein levels, as well as to Renilla luciferase activity or ␤-galactosidase activity; ␤-galactosidase assays were performed according to protocols derived from the manufacturer (Promega).
Kinase Assays, Flow Cytometry-Cyclin-dependent kinase assays were performed essentially as described (35), using 250 g of whole cell extract immunoprecipitated with 0.5 g of polyclonal antisera to Cdk2 (Santa Cruz Biotechnology), cyclin E, and cyclin A. Following immunoprecipitation and washing, 1 g of histone H1 (Sigma) was used as a substrate in kinase reaction buffer (20 mM Tris, pH 7.4, 7.5 mM MgCl 2 , 1 mM dithiothreitol) supplemented with 5 mM sodium fluoride, 1 mM sodium 32°C orthovanadate, and 125 ng/l cAMP-dependent protein kinase inhibitor (Sigma). 0.5 l of [␥-32 P]ATP (PerkinElmer Life Sciences, 800 Ci/mmol) was added to each reaction, and 0.5 g of normal rabbit IgG (Sigma) was used as a negative control. Following incubation at 37°C for 20 min, reactions were boiled in Laemmli sample buffer and loaded onto 10% SDS-PAGE. Cells for flow cytometry were fixed in ethanol and stained with propidium iodide as described (19) and analyzed on a Becton Dickinson FacScan. G 1 , S (synthesis), and G 2 /M populations were calculated using the program CellQuest.
Modified McKay Assays (Immunobinding), Chromatin Immunopre-p53-dependent Repression of Survivin cipitations-Immunoselection of the p53-binding site in the survivin promoter was performed essentially as described (33,36 Chromatin immunoprecipitations were performed exactly as described (21) using primers specific for the survivin promoter and polyclonal antisera specific for p53 (Santa Cruz Biotechnology) and acetylated histone H3 (Upstate Biotechnology, Inc.). To ensure for linearity, PCRs were sampled and analyzed following 25, 30, and 35 cycles.

Site-directed Mutagenesis, Generation of Stable Cell Lines
Overexpressing Survivin-The Survivin cDNA was mutagenized to convert it to a constitutively active anti-apoptotic form by changing the threonine residue at amino acid 34 to glutamic acid (T34E), using the QuikChange site-directed mutagenesis protocol (Stratagene). This mutant, which had significantly less toxicity than the wild type cDNA, was cloned into a CMV-driven plasmid (CMV-neo-Bam3) and transfected into CaCl cells. Subclones were isolated and analyzed for apoptosis induction following treatment with ultraviolet radiation using annexin V assays (R & D Systems), as well as immunofluorescence and Western analysis using the p85 PARP antibody, which is specific for the caspase-cleaved form of PARP, according to protocols derived from the supplier (Promega).

Survivin Is Negatively Regulated in a p53-dependent Man-
ner-Previous studies using DNA micro-array technology to identify candidate p53-repressed genes revealed that a number of genes with roles in G 2 /M progression, including several genes that encode microtubule components, are subject to negative regulation by p53 (30). We sought to test the hypothesis that, like these genes, the microtubule-associated protein Survivin might likewise be negatively regulated by p53. For these studies the human lung adenocarcinoma cell line Val138 was utilized; these cells are derived from p53-null H1299 cells and are stably transfected with a temperature-sensitive p53 allele encoding valine at codon 138. The p53 protein in these cells exists in a mutant conformation at 39°C, and temperature shift to 32°C induces a wild type conformation of p53 and concomitant growth arrest (34). A cDNA clone for Survivin, generated by reverse transcription-PCR (see "Experimental Procedures") was utilized as a probe in Northern analyses of Val138 cells cultured at 39°C (mutant p53) or temperatureshifted to 32°C (wt p53) for 24 h. As shown in the Northern analysis in Fig. 1A, temperature shift of parental H1299 cells had no effect on Survivin RNA levels (lanes 1 and 2). However, temperature shift in Val138 cells resulted in a significant decrease in Survivin levels (lanes 3 and 4), whereas the levels of the housekeeping gene gapdh and the IAP gene XIAP remained unchanged (see Fig. 1).
In an effort to examine the effects of endogenous p53 induction on survivin gene expression, human tumor cell lines containing wild type p53 were treated with the DNA-damaging agent doxorubicin (adriamycin), which is a potent inducer of p53 protein and p53-dependent transcriptional activity. Treatment of the human MCF-7 breast carcinoma and the human melanoma cell line CaCl with doxorubicin led to significant decreases in the steady state level of Survivin messenger RNA as early as 12 h after treatment (see Fig. 1B). Similarly, treatment of the human osteosarcoma cell line U2OS (wt p53) with doxorubicin led to significant induction of p53 and a 5-fold decrease in Survivin protein after 24 h (Fig. 1C). In contrast, such decreased levels did not occur following identical treatment of Saos2 human osteosarcoma cells, which are null for p53 (Fig. 1C, lane 4). To extend these studies, we performed Western analysis of p53 and Survivin in matched human tumor cell lines that differ in p53 status, following treatment with doxorubicin. CaCl is a human melanoma cell line with wt p53; CaCl/E6 is a clonal derivative of this cell line that stably expresses the human papillomavirus protein E6, which targets p53 for degradation (27). As shown in Fig. 1D, the CaCl/E6 line has significantly reduced p53 levels compared with parental CaCl cells, and doxorubicin treatment failed to up-regulate the p53-induced protein p21 waf1 . Similarly, doxorubicin treatment failed to cause down-regulation of Survivin in CaCl/E6 cells, indicating a requirement for wt p53 for this process. Flow cytometric analyses indicated that both cell lines undergo G 2 /M arrest following doxorubicin treatment, indicating that Survivin down-regulation was unlikely to be an indirect effect of p53-induced G 1 arrest in these cells (see below).

FIG. 1. Survivin is down-regulated following induction of wild type (wt) p53.
A, Northern analysis of survivin gene expression in cells containing no p53 (human H1299 lung adenocarcinoma) and temperature-sensitive p53 (Val138, wt p53 at 32°C). Cells in lanes 2 and 4 were temperature-shifted to 32°C (wt p53) for 24 h. As loading controls, the levels of the housekeeping gene gapdh, and the inhibitor of apoptosis XIAP, are depicted. B, Survivin mRNA is down-regulated following p53 induction in cells containing wt p53. Induction of wt p53 in human breast carcinoma cells (MCF-7) and melanoma cells (CaCl) following doxorubicin (DOX) treatment results in marked decreases in mRNA levels for Survivin, as early as 12 h following treatment (0.5 g/ml doxorubicin). In contrast, doxorubicin treatment results in increased expression of the p53-induced gene p21 waf1 , whereas the RNA levels of GAPDH remain unchanged. C, Survivin down-regulation requires the presence of wt p53. Western analysis of the human osteosarcoma cell lines U2OS (wt p53) and Saos2 (p53-null) following doxorubicin treatment results in decreased Survivin protein levels only in cells with wt p53 (U2OS). A ␤-actin control is included to verify equal protein loading in the lanes. D, the human papillomavirus E6 protein, which targets p53 for degradation, inhibits doxorubicin-mediated down-regulation of survivin by p53. Western analysis of the human melanoma cell line CaCl and a clonal derivative of this cell line stably express the human papillomavirus E6 protein (CaCl/E6) following treatment with the DNA-damaging agent doxorubicin (DOX) for 24 h. Doxorubicin treatment results in down-regulation of Survivin protein levels only in parental CaCl cells, which contain functional p53. As a positive control for p53 induction, immunoblots of p53 and of the p53-induced gene p21 waf1 are included. A ␤-actin control is included to verify equal protein loading in the lanes.
p53-dependent Repression of Survivin p53-dependent Repression of survivin Occurs Independent of G 1 Arrest by p53-Because the survivin gene is known to be transcriptionally repressed during the G 1 phase of the cell cycle, it became important to eliminate the possibility that repression of Survivin by p53 was a side effect of the induction of G 1 growth arrest by this protein. Immunoblots of MCF-7 cells treated with doxorubicin indicated that, as expected, p53 levels were induced, and Survivin levels were markedly decreased following treatment ( Fig. 2A). Flow cytometric analysis of these cells indicated that they were growth-arrested in the G 2 /M phase of the cell cycle following doxorubicin treatment (Fig. 2B); such a failure of human tumor cell lines with wt p53 to growth arrest in G 1 following doxorubicin and gamma irradiation treatment has been noted by others (37,38). In an effort to confirm that these doxorubicin-treated cells were in fact arrested in G 2 /M, we performed cyclin-dependent kinase assays in MCF-7 cells before and after drug treatment. These assays indicated that the activity of the G 1 -specific cyclin E-associated kinase, as well as the Cdk2 kinase, were dramatically reduced in MCF-7 cells following doxorubicin treatment (Fig.  2C, lanes 3 and 9). Additionally, the kinase activity associated with cyclin A, the S phase cyclin, was also markedly reduced (Fig. 2C, lane 6), as would be expected for cells in G 2 /M. Because the cyclin B1 and cdc2 genes are subject to transcriptional repression by p53, the activity of these enzymes was not analyzed; however, the combined data support the premise that doxorubicin-treated MCF7 cells are arrested in G 2 /M.
We extended these studies in efforts to better separate the ability of p53 to down-regulate Survivin from its ability to induce G 1 arrest. We (27) and others (39) have noted that treatment of cells with low doses of UV radiation, between 2 and 4 J/m 2 , is sufficient to induce p53 protein and transcriptional activity but is insufficient to induce G 1 growth arrest. Treatment of MCF-7 breast carcinoma cells with ultraviolet irradiation led to dose-dependent decreases of Survivin protein 24 h after treatment, whereas flow cytometry indicated that these cells continue to proliferate following treatment ( Fig. 3A and data not shown). We also analyzed murine embryo fibroblasts from the genetically engineered p21 waf1 knock-out mouse; these cells are markedly impaired for growth arrest in response to p53 (40). p21-null mouse embryo fibroblasts demonstrated normal p53 induction and decreased Survivin levels, following treatment with UV radiation (Fig. 3B), but there was no evidence for G 1 arrest by flow cytometry (Fig. 3C). These data solidify the notion that the ability of p53 to down-regulate Survivin is distinct from its ability to induce G 1 arrest or to induce p21 waf1 .
It became of interest to test the ability of the p53 homologue p73 to down-regulate the survivin gene; although p73 is fully capable of inducing p21 waf1 and G 1 arrest, it has been reported to be incapable of repressing the p53-repressed gene cdc25c (41). For this analysis we took advantage of p53-null H1299 cell lines that have been stably transfected with temperature-sensitive versions of p53 and p73 protein (34). As indicated previously, these proteins exist in mutant (inactive) conformation at 39°C and become wild type conformation (and activity) at 32°C. As depicted in Fig. 4, only temperature-sensitive p53, and not p73, was capable of down-regulating Survivin at the permissive temperature (32°C, wt p53, lane 4). In contrast, both proteins were able to transactivate p21 waf1 to identical levels (Fig. 4A), and both were indistinguishable in their ability to growth-arrest cells (Fig. 4B). The combined data indicate that induction of growth arrest and transactivation of p21 waf1 are not sufficient to cause down-regulation of Survivin. In efforts to address whether p53 directly negatively regulates the survivin promoter, we cloned a 1-kb fragment of this promoter, and we analyzed it for the ability to confer negative regulation to a reporter gene by p53.
The survivin Promoter Is Sufficient to Confer Negative Regulation to a Heterologous Gene by p53, Even in Stably Transfected Cells-The survivin promoter has been characterized previously (5,12). We cloned ϳ1 kb of this promoter by PCR of human genomic DNA using primers designed from the published sequence (see "Experimental Procedures"). The fulllength 1-kb survivin promoter was cloned upstream of the firefly luciferase reporter gene to create the reporter construct SpII. To test the repressibility of this promoter by p53, this construct was transfected into p53 null cells (human H1299 cells) with increasing concentrations of p53. As a control for the ability of p53 to nonspecifically repress transcription (so-called "transcriptional squelching"), equal microgram amounts of the Renilla luciferase gene were transfected, and the ratio of firefly to Renilla luciferase per equal microgram amount of protein was determined following transfection using standard assays. As depicted in Fig. 5A, as little as 10 ng of wt p53 was sufficient to repress 5 g of the survivin promoter, indicating that this promoter is quite sensitive to p53-dependent repression. Increasing concentrations of p53 led to stepwise decreases in the activity of the luciferase gene driven by the survivin promoter (Fig. 5A). Deletion analysis of this promoter indicated that a truncated version (encoding nucleotides Ϫ105 to Ϫ15, where ϩ1 denotes the initiating ATG, as per 5) was also negatively regulated by p53. This construct, SpVI, demonstrated a lower basal level of transcription than the full-length survivin pro-  2 and 3), cyclin A (lanes 5 and 6), and cyclin E (lanes 8 and 9) using histone H1 as the substrate. As a negative control, normal rabbit IgG is used instead of polyclonal antisera for the immunoprecipitations and kinase assays (No Ab, lanes 1, 4, and 7). Notably, whereas these kinase activities are high in asynchronously growing MCF7 cells (Asyn, lanes  2, 5, and 8), all activities are low in cells treated with doxorubicin (lanes 3, 6, and 9). This finding is consistent with flow cytometry data, which indicate that doxorubicin-treated cells are arrested with a G 2 /M content of DNA.
p53-dependent Repression of Survivin moter, but was repressed nearly as well by wt p53 (Fig. 5A). Similar results were obtained in the human Saos2 osteosarcoma cell line (Fig. 5B); this cell line is null for both p53 and the retinoblastoma protein pRB and thus is impaired for p53-dependent growth arrest. These data also support the conclusion that p53 can down-regulate Survivin independent of p53-dependent growth arrest.
Repression of Survivin by p53 was recapitulated in stably transfected cells, where the full-length survivin promoter-luciferase construct was stably introduced into cells containing temperature-sensitive p53 (Val5-SpII cells). In two independently derived cell lines containing temperature-sensitive p53, the stably integrated luciferase gene driven by the survivin promoter was markedly down-regulated by p53 induction (Fig.  5C). Similarly, we found that the luciferase construct containing a truncated version of the survivin promoter (SpVI) was also sufficient to confer negative regulation by p53 in stably transfected cells (data not shown). In contrast, temperature shift alone had no effect on luciferase levels in parental 10.1 mouse embryo fibroblasts (data not shown). To our knowledge, the survivin promoter is the first promoter demonstrated to confer p53-dependent repression to a heterologous gene in both transiently and stably transfected cells.
p53 Binds in a Sequence-specific Manner to the survivin Promoter; the p53-binding Site Is Necessary for Transcriptional Repression of Survivin in Vivo-To test whether p53 could physically associate with the survivin promoter, the assay of McKay was utilized; this assay allows for the analysis of DNAprotein interactions using large fragments of DNA, up to several kilobases in length (36). In this assay, the survivin promoter was digested with restriction endonucleases, radiolabeled with [ 32 P]dCTP using Klenow polymerase, and incubated with whole cell extract from cells that are null for p53 (10.1 murine embryo fibroblasts) or that contain wild type p53 (Val5 cells shifted to 32°C). Following incubation of the radiolabeled survivin promoter with cell extract, samples were immunoprecipitated with p53 antisera and protein A-Sepharose. These immunoprecipitates were washed and phenol/chloroform extracted, and bound DNA fragments were resolved on nondenaturing polyacrylamide gels. McKay assays on the fulllength survivin promoter (SpII, not shown), and on the smaller SpV fragment of this promoter (Fig. 6A), indicated that this FIG. 3. Down-regulation of survivin by p53 does not require induction of G 1 arrest or p21 waf1 . A, dose-dependent decreases in survivin gene expression in MCF7 breast carcinoma cells following ultraviolet irradiation. Cells were treated with the indicated dose of radiation, harvested 24 h later, and subjected to Western analysis for p53 and Survivin. That there is less p53 in the 10 J/m 2 sample (lane 4) reflects the fact p53 has already peaked in level in these cells and that levels of p53 are returning to normal. A ␤-actin control is included to verify equal protein loading in the lanes. B, Western analysis of murine embryo fibroblasts from the p21 waf1 knock-out mouse reveals dose-dependent decreases in Survivin levels in these cells following ultraviolet irradiation. A ␤-actin control is included to verify equal protein loading in the lanes. C, flow cytometric analyses of p21 waf1 -null mouse embryo fibroblasts that are treated with ultraviolet radiation (4 J/m 2 ) or untreated. The calculated numbers indicate there is no evidence for G 1 arrest following p53 induction in this cell line; the data depicted are representative from two independent experiments read in duplicate.  1 and 2), and in H1299 cells containing stably transfected alleles for temperature-sensitive p53 (ts p53 Val138 cells, lanes 3 and 4), and in p73␤ (lanes 5 and 6). Cells were grown at 39°C (mutant conformation, lanes 3 and 5) or temperature-shifted to 32°C for 24 h (wild type p53/p73 induction) as indicated. Whereas both proteins are capable of inducing the p53-response gene p21 waf1 to comparable levels at 32°C (lanes 4 and 6), only wild type p53, and not p73␤, was capable of down-regulating survivin expression (lane 4). Like p73␤, p73␣ was likewise incapable of repressing survivin expression, despite being capable of inducing growth arrest in temperature-shifted cells (not shown). A GAPDH control is included to verify equal loading and integrity of RNA. The data depicted are representative of three independent experiments. B, flow cytometric analysis of temperature-shifted, propidium iodide-stained cells containing temperature-sensitive p53 (ts p53 Val138 cells) or p73␤. Both proteins are capable of inducing a G 1 and G 2 /M arrest following 24-h temperature shift to 32°C, whereas only ts p53 can repress survivin (A). The data presented are the average from two independent experiments, and the same plates of cells were used for the Northern analyses in A.

p53-dependent Repression of Survivin
DNA fragment could be specifically immunoprecipitated with p53 antisera, only in cells containing wild type p53 (Fig. 6A,  lane 3) and not in parental p53-null cells (lane 2). As an internal negative control, vector sequences were negligibly bound to p53 in this assay (Fig. 6A). The survivin promoter was capable of binding to p53 to a level roughly comparable with the Mdm2 promoter, which was labeled to identical specific activity (Fig.  6C, lane 8).
The sequence of the minimal p53-binding region of the survivin promoter defined from these assays is depicted in Fig. 6B. Inspection of this region revealed a consensus p53-binding site that differs from a transactivated element in that it contains a 3-nucleotide spacer element between the two pairs of palindromic pentamers, or "half-sites" (Fig. 6B). To date, the majority of p53-transactivated genes contain a spacer of 0 or 1 nucleotide; in fact, lengthening this spacer to 4 nucleotides has been shown to abolish the ability of p53 to transactivate promoters containing this binding site (42). To test the possibility that this candidate-binding site was responsible for the in vitro binding we detected, this site was deleted by restriction endonuclease digestion from the survivin promoter and tested in a McKay assay. Deletion of this candidate p53-binding site from the survivin promoter was sufficient to completely abolish p53 binding in a McKay assay (Fig. 6C, lane 4). That binding of the survivin promoter by p53 was sequence-specific was supported by the finding that it could be eliminated by incubation with a 20 5-fold excess of unlabeled DNA (Fig. 6C, lane 3).
To assess the significance of the p53-binding site on transcriptional repression of survivin by p53, these same survivin promoter/luciferase constructs (SpV and SpV⌬p53BS) were stably transfected into Val5 cells that contain a temperaturesensitive (inducible) p53. Significantly, whereas p53 negatively regulated the luciferase gene driven by the wild type survivin promoter (SpV, Fig. 6D), deletion of the p53-binding site effectively inhibited this repression (SpV-⌬p53BS). Therefore, the p53-binding site of the survivin promoter is necessary for both binding and for transcriptional repression by p53.
Of interest in the survivin promoter is the observation that the p53-binding site overlaps with a site homologous to those utilized by E2F family members (2 mismatches). This prompted us to test the possibility that E2F family members could transactivate the survivin promoter and furthermore that p53 might repress survivin in part by interfering with E2F-mediated transactivation of this gene. To test these possibilities, a minimal survivin promoter construct, containing only the p53-binding site and the overlapping E2F site, was constructed in the plasmid pGL3-E1B-TATA, which contains the luciferase gene and the TATA box of the adenovirus E1B gene (see "Experimental Procedures"). This construct, designated Sp-min, was transfected into H1299 cells with increasing concentrations of p53 plasmid. As indicated in Fig. 7, when normalized to control and protein concentrations, the Sp-min construct was repressed by p53 in a dose-dependent manner (Fig. 7A). Additionally, transfection with E2F-1 was sufficient to activate transcription from this promoter; typically this induction ranged from 3-to 10-fold (Fig. 7A). Additionally, the SpII construct, containing the full-length survivin promoter, was also potently up-regulated by E2F-1 (data not shown). Notably, E2F-mediated activation of Sp-min was inhibited by increased concentrations of p53 (Fig. 7A). These data support the hypothesis that p53 may repress the survivin promoter in part by inhibiting the ability of E2F proteins to transactivate this gene. Interestingly, the ability of p53 to repress E2Fmediated transactivation of this construct did not rely on FIG. 5. A, the survivin promoter is sufficient to confer negative regulation of the firefly luciferase gene by p53. The survivin promoter, ligated to the promoterless firefly luciferase gene (pGL2-basic, Promega) to create the construct SpII, was transiently transfected with increasing concentrations of wild type p53 (0, 0.01, 0.1, and 1 g of p53). Deletion constructs of the survivin promoter (SpV and SpVI, depicted to the right and numbered as per Ref. 5, where ϩ1 represents the ATG), which contain the nucleotides depicted, were separately transfected with 10 ng of p53. As a control for transfection efficiency and nonspecific repression, equal micrograms of Sp construct and CMV-Renilla luciferase were transfected with the p53 amounts indicated, and the numbers shown are normalized to Renilla values using the Dual Luciferase assay (Promega) and to protein levels using the Bio-Rad Dc assay (Bio-Rad); the level of SpII was set at 100%. The results depicted are the average of three independent experiments; error bars mark S.D. B, the survivin promoter is negatively regulated by p53 in cells that are incapable of G 1 arrest. Human Saos-2 cells (p53 Ϫ/Ϫ, pRB Ϫ/Ϫ) were transiently transfected with 4 g of SpII (full-length survivin promoter driving firefly luciferase expression) and increasing doses of p53. The numbers shown are normalized to equal protein levels (Bio-Rad Dc assay) and to the activity of equal microgram amounts of co-transfected Renilla luciferase (CMV-Renilla, Promega) or SV40-␤-galactosidase (Promega). C, the full-length survivin promoter (SpII) is sufficient to confer negative regulation by p53 in stably transfected cells. The survivin construct SpII was co-transfected into cells containing temperature-sensitive p53 (Val5 cells) in 5-fold excess with the drug selectable plasmid pGK-hygro, which confers hygromycin resistance. Hygromycin-resistant colonies were pooled and analyzed. Temperature shift to 32°C induces wt p53 protein and down-regulation of luciferase mRNA, as depicted in this Northern analysis. Expression of GAPDH is shown as a control for RNA loading and integrity. The results shown were recapitulated in three independent experiments, in two independent sets of pooled, stably transfected cells.
p53-dependent Repression of Survivin p21 waf1 , as p53-mediated repression of the Sp-min construct was also evident in p21 Ϫ/Ϫ cells (Fig. 7B), albeit to a slightly reduced magnitude. The combined data raise the possibility that p53 may repress survivin transcription in two ways as follows: one may involve inhibition of E2F by induction of p21 waf1 , and subsequent binding of hypophosphorylated pRB to E2F to create a transcriptional repressor. Additionally, however, p53 appears to interact directly with the survivin promoter and may interfere with the ability of E2F proteins from activating transcription from an overlapping E2F site.
p53 Interacts with the survivin Promoter in Vivo; Deletion of the Sin3 Binding Domain of p53 Impairs Its Ability to Repress Survivin-We have shown previously that one mechanism whereby p53 can repress gene expression is via binding to the co-repressor Sin3, which recruits HDACs to the promoters of p53-repressed genes like Map4 (21). That p53 binds to the Map4 promoter in vivo, and recruits HDACs, was demonstrated using the technique of chromatin immunoprecipitations. Chromatin immunoprecipitations were performed on the survivin promoter in cells containing wild type (Val5-SpII cells at 32°C) and mutant (cells at 39°C) p53. These data indicated that wt but not mutant p53 could be found complexed to the survivin promoter in vivo (Fig. 8A); this binding was comparable with that for the Map4 promoter in these cells (Ref. 21 and data not shown). Additionally, in cells with wt p53 (cells at 32°C), but not mutant p53 (39°C), the survivin promoter was preferentially associated with deacetylated histone H3, consistent with the action of HDACs (Fig. 8A, lanes 5 and 6). Identical reactions performed in the absence of antibody, with irrelevant antibody, or in wash buffers alone, failed to reveal a PCR product for the survivin promoter (Fig. 8A, lanes 1-3). These data were consistent in three independent experiments, and they support our data from the McKay assay, which indicate that wild type p53 can interact with a consensus p53-binding site within the survivin promoter.
We recently mapped the interaction domain between Sin3 and p53, and we have found that amino acids 61-75 of wild type p53 are necessary for interaction with Sin3 (26). Therefore, it became of interest to determine whether a deletion mutant of p53 lacking the Sin3-binding domain (⌬61-75) was impaired for repression of the survivin promoter. p53-null H1299 cells were transfected with the luciferase construct SpVI, in the presence of 10 ng of either wt p53 or the ⌬61-75 mutant of p53. Significantly, whereas wt p53 was capable of FIG. 6. p53 binds in a sequence-specific manner to a fragment of the survivin promoter. A, McKay immunobinding assays indicate that the radiolabeled survivin promoter (SpV), but not the vector internal control, is immunoprecipitated by p53 antisera in cells with wt p53 (Val5 32°C, lane 3) but not in cells that are null for p53 (parental 10.1 cells, p53 Ϫ/Ϫ, lane 2). 20% of the radiolabeled vector/insert is loaded in the input lane. B, sequence of the survivin promoter in the minimal p53-binding region. The arrow denotes the start site of survivin transcription; just upstream of the transcriptional start site is depicted the p53-binding site (boxed) and an overlapping consensus for E2F transcription factors (underlined). Downstream elements responsible for preferential G 2 /M transcription, the CDE/CHR, are underlined. C, immunobinding (McKay) assays using the radiolabeled promoter constructs listed, incubated with 100 g of whole cell extract from cells with wt p53 and monoclonal antisera specific for p53. Following binding reactions, bound protein-DNA complexes are immunoprecipitated, washed, phenol/chloroform extracted, and resolved on a non-denaturing polyacrylamide gel. The radiolabeled vector fragment serves as an internal negative control and the Mdm2 promoter serves as a positive control (lanes 7-9). The SpV construct encodes an ϳ500-bp fragment of the survivin promoter; the SpV-⌬p53BS encodes the same fragment, with the p53-binding site deleted using the restriction endonuclease SacII. As a control for sequence-specific binding, a 25-fold molar excess of each construct was preincubated with the extract for 10 min prior to the binding reaction (ϩ competitor). 10% of the radiolabeled vector/insert is loaded in the input lane. The results depicted are representative of several independent experiments. D, Northern analysis of cells with temperature-sensitive p53 (Val5 cells, murine p53-null cells transfected with the valine 135 temperature-sensitive p53 gene) that are stably transfected with the SpV-survivin promoter/luciferase construct or the SpV construct in which the p53-binding site is deleted (SpV⌬p53BS). Northern hybridization to a probe specific for the luciferase gene indicates that p53 is capable of down-regulating the SpV construct but not the construct in which the p53-binding site is deleted (SpV⌬p53BS). The data depicted are from pooled, stably transfected clones and are representative of three independent experiments on two independently derived sets of cell lines for each construct.

p53-dependent Repression of Survivin
inducing 4-fold repression of the SpVI construct, the ⌬61-75 mutant of p53 was unable to repress this promoter construct (Fig. 8B). At higher concentrations of p53 (0.5 g), however, this Sin3-binding mutant was able to repress the survivin promoter, albeit at reduced levels compared with wt p53 (data not shown).
In similar studies we sought to test the significance of the 3-nucleotide spacer in the p53-binding site for repression of Survivin. Specifically, we sought to test the hypothesis that deletion of the 3-nucleotide spacer could convert this element to a p53-transactivated site. Deletion of this spacer, in the background of the p53-repressible Sp-min construct (SpVII), rendered this construct potently transactivated by p53 (Fig. 8C). Notably, the ⌬61-75 construct of p53, which fails to interact with Sin3, is equally capable of transactivating this construct. The ⌬61-75 mutant also transactivated the p21 waf1 promoter to levels indistinguishable from wt p53 (26 and data not shown). Therefore, deletion of the 3-nucleotide spacer region in the p53-binding site of the survivin promoter is sufficient to render this site equivalent to a transactivated element.
Overexpression of Survivin Can Inhibit Apoptosis in Cells with wt p53-To test the contribution of p53-mediated repression of Survivin to apoptosis, it became logical to test the ability of overexpressed Survivin to inhibit apoptosis induced by p53. Toward this end, we generated stable transfectants of CaCl melanoma cells, which are sensitive to p53-dependent apoptosis, with Survivin constructs driven by the cytomegalovirus immediate-early promoter. Because we and others (43) noted that overexpression of wild type Survivin was associated with some toxicity in cells, we used for these studies a Survivin construct that was mutated to mimic constitutive phosphorylation at the threonine residue at amino acid 34. The available data indicate that the threonine 34-phosphorylated form of Survivin is the anti-apoptotic form of this protein (43). Stable transfection of the T34E variant of Survivin yielded several subclones of CaCl cells with greatly overexpressed Survivin protein (20 -30-fold overexpression, see Fig. 9). CaCl-Survivin clones, as well as vector-transfected control, were treated with increasing doses of ultraviolet light to induce apoptosis, and the extent of apoptosis after 24 h was monitored by immunopositivity for the caspase-cleaved p85 fragment of poly(ADP) ribose polymerase (PARP). As depicted in Fig. 9, Western analysis of CMV-and Survivin-transfected cells treated with increasing doses of ultraviolet radiation led to significant p53 induction in both cell lines (Fig. 9). However, only in CMV-transfected cells did this p53 induction lead to dose-dependent increases of p85 PARP, a marker for caspase activation and apoptosis. Similar results were obtained in another independently generated Survivin-transfected clone (CaCl-T34E-Survivin clone 1, data not shown). We also noted reduced annexin V staining in UVtreated cells that overexpress Survivin, relative to vectortransfected control (data not show). These data support the hypothesis that overexpression of Survivin can inhibit UVinduced apoptosis and that repression of Survivin by p53 would be predicted to either directly induce, or sensitize cells to, apoptosis.

DISCUSSION
Although originally identified as an IAP (1) and shown to inhibit caspase activity in vitro (2), that Survivin is a bona fide caspase inhibitor has been the subject of considerable debate. Nonetheless, several groups (4 -6) have shown that antisense down-regulation of Survivin is sufficient to induce apoptosis in human tumor cell lines. Interestingly, we found that expression of wt Survivin in human tumor cells was associated with considerable toxicity and that only by mutating this protein at amino acid 34, to mimic phosphorylation at this site (as indicated in Ref. 43), were we able to generate stably transfected cell lines that overexpress this protein. Along these lines it has been proposed that the non-phosphorylated form of Survivin protein may in fact function as a dominant negative protein in vivo (43), and this may explain the toxicity evident in our studies. We were able to generate two independent subclones of tumor cells that express 20 -30-fold increased levels of the T34E Survivin protein. Notably, both of these cell lines demonstrated resistance to apoptosis induced by ultraviolet light, as determined by reduced cleavage of the protein PARP, a marker for apoptosis. Therefore, although the mechanism

p53-dependent Repression of Survivin
whereby Survivin inhibits apoptosis is subject to some debate, this protein clearly functions to inhibit apoptosis induction in this cell line system. Further analysis of these Survivin-overexpressing cells should aid in the clarification of the mechanism whereby this protein inhibits apoptosis.
In this study we have identified the survivin gene as a member of a growing class of genes with a role in the G 2 /M transition of the cell cycle that are also negatively regulated by p53; these genes include cdc2, cdc25c, and cyclin B1 (28,29,41). The promoters of these genes, like survivin, each contain a bipartite element near the start site of transcription, termed a CDE/CHR (cell cycle-dependent element/cyclin homology region). This element interacts with an as yet uncloned binding protein termed CDF-1 in the G 1 phase of the cell cycle; this binding is believed to lead to transcriptional repression of these genes in G 1 and enhanced expression in G 2 /M (44,45). Whereas this element is clearly important for the cell cycle-regulated expression of these genes, our data indicate that the CDE/CHR element is not required for the repression of survivin promoter constructs by p53 in transient assays. Additionally, we have found that G 1 arrest alone, induced by the p53-homologue p73, is not sufficient to cause repression of survivin.
Two mechanisms for the p53-mediated repression of genes like cdc2, cdc25c, and cyclin B1 have been proposed. Repression of these genes by p53 has been proposed to involve inhibition by p53 of the NF-Y transcription factor, which binds to CCAAT boxes in these promoters and normally transactivates these genes (39,46,47). As the survivin promoter does not contain an obvious CCAAT box-binding site for NF-Y, it is unlikely that this mechanism plays a role in p53-dependent repression of survivin. It has also been proposed that p53-dependent repression of cdc2 by p53 relies on up-regulation of p21 waf1 . This leads to hypophosphorylation of pRB family proteins and transcriptional repression via E2F family members (29). Our data indicate that p53 can repress survivin in p21-null cells, which would argue against such a mechanism; however, we have consistently noted that the repression of survivin by p53 shows decreased magnitude in cells lacking p21 or pRB. Therefore, it is possible that p53 represses survivin in part by transactivation of p21 waf1 and subsequent conversion of E2F complexes to E2F-RB repressor complexes. Our data indicate, however, that this may be one of two overlapping, redundant mechanisms for the repression of survivin.
In this study we show that p53 binds in vivo to the survivin promoter to a consensus p53-binding site, raising the possibility that p53 represses survivin by interfering with E2F-mediated transactivation. We have found that binding of p53 to this region is accompanied by a decreased association of this pro- Antisera specific for acetylated histone H3 (Upstate Biotechnology, Inc.) and p53 polyclonal antisera (Santa Cruz Biotechnology) were utilized for these experiments. The data depicted to the right are from a single experiment following Southern transfer and hybridization to a probe specific for the survivin promoter. The quantitated data are from three independent experiments; mean Ϯ S.E. error bars are shown. B, transient transfection of the survivin promoter construct SpVI and 10 ng of either wt p53 or the p53 mutant that lacks the Sin3-binding domain, amino acids 61-75 (⌬61-75 p53). Whereas wt p53 is capable of repressing the survivin promoter, the Sin3-binding domain mutant is unable, indicating that transcriptional repression of survivin requires Sin3 and supporting a direct repression of survivin by p53. As a control for transfection efficiency and nonspecific repression, equal microgram amounts of SpVI and the SV40-driven ␤-galactosidase gene were transfected and assayed. The results depicted are the levels of luciferase activity normalized to ␤-galactosidase and are the averages of three independent experiments where the activity of SpVI in the absence of p53 was set to 100%. Error bars mark S.D. in the three experiments. C, the survivin p53-binding site functions as a transactivated element when the 3-nucleotide spacer element is deleted. The survivin construct SpVII was generated by polymerase chain reaction as described under "Experimental Procedures" to recreate the p53-binding element of the survivin promoter, minus the TCC spacer. This construct was transfected into p53-null H1299 cells in the absence or presence of 10 ng of wt p53 or the ⌬61-75 mutant of p53, which fails to repress transcription from the normal survivin promoter. This artificial element is transactivated over 100-fold by either wild type p53 or the ⌬61-75 mutant. The averaged data from three independent experiments are depicted, and error bars mark S.D. from the three experiments. moter with acetylated histones; this would be consistent with the action of the p53-Sin3-HDAC complex. We propose a model whereby direct binding of p53 to the survivin promoter confers repression by p53, via the p53-Sin3-HDAC complex. Interestingly, we have found that, when placed alone upstream of a minimal promoter, the p53-binding site of the survivin promoter is not sufficient to confer repression to a heterologous gene. Rather, the overlapping E2F-site is required along with the p53-binding site. 2 Therefore, we favor the hypothesis that the p53-Sin3-HDAC complex binds to the survivin promoter and modifies the chromatin conformation such that E2F-binding and/or transactivation is impaired or abrogated. Such a mechanism of repression by p53 has been seen for other promoters, such as the ␣-fetoprotein promoter, on which p53 binds and competes for binding with the HNF-3 transcription factor (48).
It is formally possible that p53 and E2F proteins occupy the survivin promoter at the same time and together create a transcriptional repressor. Indeed, p53 and E2F-1 have been reported to interact, and association with p53 has been reported to inhibit transactivation by E2F-1 (49). Overall, the combined data favor both direct (p53 binding, histone deacetylation) and indirect (induction of p21 waf1 ) mechanisms for the repression of survivin by p53. That p53 uses both direct and indirect mechanisms of repression of the survivin promoter may simply reflect the existence of back-up mechanisms to ensure efficient and timely repression. Alternatively it is possible that each mechanism may be specific to particular stresses, phases of the cell cycle, or cell types.
We have identified a p53-binding site in the survivin promoter that conforms to the consensus found for the first p53binding sites identified (50). This consensus site was defined as two copies of the sequence 5Ј-RRR-C(A/T)(T/A)G-YYY-3Ј (where Pu is ? and Pyr is ?), separated by 0 -13 base pairs; functional p53-binding elements can contain up to four mis-matches in this consensus, but the C and G residues are invariant (50). The human survivin promoter contains a spacer of 3 nucleotides; this site is conserved in the murine promoter, which has two overlapping p53-binding sites, one with a 4-nucleotide spacer and one with a 2-nucleotide spacer. Whereas the consensus p53-binding site identified by El Deiry and coworkers (50) contained a spacer of 0 -13 nucleotides, the functionally defined p53 sites contained within p53-induced genes typically contain spacer elements of less than 2 nucleotides. In fact, increasing the spacer from 1 to 4 nucleotides has been shown to eliminate the ability of p53 to transactivate from this site; interestingly, increasing the spacer to 10 nucleotides restores transactivation (40). We propose that increasing the spacer changes the orientation of the p53 dimers to each other and that transactivation can occur only if the dimers are on the same face of the DNA helix. Otherwise, this site functions passively as a p53-binding site and is inactive for transactivation; instead p53 bound to this site may interfere with the binding or activity of other transcription factors, such as E2F-1. That the DNA-binding domain of p53 is able to accommodate such spacing changes in the DNA-binding site is supported by the fact that a flexible linker region connects the DNA binding domain to the oligomerization domain (51). Similar subtle changes in the p53 consensus element have been shown to influence the p53 transcriptional response (52,53). Interestingly, two p53-binding sites, both with 3-nucleotide spacers, are present in the promoter of the p202 gene, which has recently been identified as a p53-repressed gene (54). The contribution of this type of binding site to the repression of the p202, and other p53-repressed, promoters remains the subject of future study. FIG. 9. Overexpression of Survivin (T34E) inhibits PARP cleavage following apoptosis induced by ultraviolet radiation. Western analysis of clones of human CaCl melanoma cells stably transfected with vector alone (CaCl-CMV) or a CMV-driven Survivin T34E construct (CaCl-Survivin clone 4; amino acid 34 of Survivin has been mutated to glutamic acid to mimic phosphorylation of this residue, see "Experimental Procedures"). Clones were subjected to ultraviolet radiation at the does indicated and harvested after 24 h, at which time they were subject to Western analysis for antibodies specific to the caspasecleaved form of PARP (p85 PARP, Promega), p53, Survivin, and ␤-actin. Although p53 induction is comparable in both cell lines, only CaCl-CMV cells, and not cells overexpressing Survivin, show an increase in the presence of p85 PARP, which is a marker for apoptosis. Representative data from a single experiment are shown and are consistent with results from two independently derived subclones of CaCl-Survivin cells (clone 1 and clone 4).