One Mechanism for Cell Type-specific Regulation of thebax Promoter by the Tumor Suppressor p53 Is Dictated by the p53 Response Element*

Key to the function of the tumor suppressor p53 is its ability to activate the transcription of its target genes, including those that encode the cyclin-dependent kinase inhibitor p21 and the proapoptotic Bax protein. In contrast to Saos-2 cells in which p53 activated both the p21 andbax promoters, in MDA-MB-453 cells p53 activated thep21 promoter, but failed to activate the baxpromoter. Neither phosphorylation of p53 on serines 315 or 392 nor an intact C terminus was required for p53-dependent activation of the bax promoter, demonstrating that this differential regulation of bax could not be explained solely by modifications of these residues. Further, this effect was not due to either p73 or other identified cellular factors competing with p53 for binding to its response element in the bax promoter. p53 expressed in MDA-MB-453 cells also failed to activate transcription through the p53 response element of the bax promoter in isolation, demonstrating that the defect is at the level of the interaction between p53 and its response element. In contrast to other p53 target genes, like p21, in which p53-dependent transcriptional activation is mediated by a response element containing two consensus p53 half-sites, activation by p53 of the bax element was mediated by a cooperative interaction of three adjacent half-sites. In addition, the interaction of p53 with its response element from the bax promoter, as compared with its interaction with its element from the p21promoter, involves a conformationally distinct form of the protein. Together, these data suggest a potential mechanism for the differential regulation of p53-dependent transactivation of thebax and p21 genes.

The tumor suppressor protein p53 is an important regulator of cellular growth. The p53 gene is mutated in the majority of human cancers (1,2), suggesting that loss of p53 may play an important causative role in oncogenesis. The p53 protein has been implicated in several diverse growth-related pathways, including apoptosis, cell cycle arrest, and senescence (3)(4)(5). The ability of p53 to function as a sequence-specific DNA-binding protein appears to be central to the function of p53 as a tumor suppressor (6,7). At its N terminus, the p53 protein contains a potent transcriptional activation domain (8) that is linked to a central core domain that mediates sequence-specific DNA binding (9 -11). Both of these domains have been shown to be important for p53-mediated growth suppression (12). The importance of the DNA binding domain is further highlighted by the fact that the major mutational hot spots from human cancers are found in this domain (13), and several of these mutations have been shown to abolish the ability of p53 to function as a transcriptional activator (14 -16).
A DNA consensus sequence through which p53 binds and activates transcription has been identified. This sequence consists of two palindromic decamers of 5Ј-RRRCWWGYYY-3Ј (where R is a purine, Y is a pyrimidine, and W is an adenine or thymine) separated by 0 -13 bp, 1 forming four repeats of the pentamer 5Ј-RRRCW-3Ј alternating between the top and bottom strands of the DNA duplex (17)(18)(19). This arrangement is consistent with the notion that p53 binds DNA as a homotetramer (20 -23). Through sequences similar to this consensus, p53 has been shown to activate the transcription of many genes, including bax, p21, mdm2, gadd45, IGF-BP3, and cyclin G (24 -31). Data are consistent with a model in which DNA damage leads to the phosphorylation of p53 as well as the subsequent stabilization of p53 and activation of its DNA binding capability (32)(33)(34)(35). Consequently, p53-mediated transcription of its target genes increases. When compared with alternate p53 targets, such as the cyclin-dependent kinase inhibitor p21, evidence suggests that the bax gene is differentially regulated by p53. Several tumor-derived p53 mutants have been identified that are capable of activating transcription through the promoter of the p21 gene but not through the bax promoter (36 -39). This has been correlated with an inability of the mutants both to bind the p53 response element of the bax promoter and to trigger apoptosis (36,38,39). Such studies with these tumor-derived p53 mutants suggest that a failure in the ability of p53 to activate the bax gene may play an important role in tumor formation and progression. As such, a complete understanding of the transcriptional regulation of the bax promoter by p53 may yield important information relevant to our understanding of tumorigenesis.
Previous studies have demonstrated that the bax promoter is differentially regulated by wild-type p53 in a cell type-specific manner (40). Here the osteosarcoma Saos-2 and the breast carcinoma MDA-MB-453 cell lines were used as a model system to explore the potential mechanisms for this differential regulation. In the Saos-2 cell line, transfected wild-type p53 effectively activated transcription through both the p21 and bax promoters. In contrast, p53 expressed in the MDA-MB-453 cell line was capable of activating transcription through the p21 promoter as well as the p53 response elements of the p21, cyclin G and cdc25C promoters but failed to do so through either the bax promoter or the isolated p53 response element derived from the bax promoter. Neither p53 phosphorylation at serine 315 or serine 392 nor an intact C terminus was required for activation of the bax promoter, demonstrating that the observed defect in MDA-MB-453 cells could not be explained solely by modifications of these residues. In addition, neither the p53 homolog p73 nor other cellular factors that are capable of binding the p53 response element of the bax promoter explained the differential regulation of the bax promoter. Detailed analysis of the interaction of p53 with the bax promoter, however, demonstrated that unlike other well characterized p53 response elements, like that of the p21 gene, in which p53-dependent transcriptional activation is mediated by a response element containing two consensus p53 half-sites, the response element of the bax promoter consists of three adjacent half-sites that cooperate to bring about complete activation by p53. In addition, it appears that p53 exists in a distinct conformation when bound to its response element from the bax promoter as compared with when it is bound to the 5Ј-response element of the p21 promoter. Together, these data suggest a potential mechanism for the cell type-specific differential regulation of bax by p53.
Transfections-Unless otherwise indicated, 1 ϫ 10 5 cells were seeded into 35-mm plates. Cells were transfected 24 h later using the DOTAP liposomal transfection reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions. Cellular lysates were prepared 48 h post-transfection, total protein concentration was determined by protein assay (Bio-Rad), and luciferase assays were quantitated using a commercially available kit (Promega) and a TD-20e Luminometer (Turner).
Nuclear Extracts-All procedures were conducted at 4°C. For each 100-mm dish, cells were washed three times with 5 ml of phosphatebuffered saline. Cells then were scraped into 500 l of lysis buffer (20 mM HEPES, pH 7.5, 20% glycerol, 10 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.1% Triton X-100, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 50 M leupeptin, and 50 g/ml aprotinin) and centrifuged at 500 ϫ g for 5 min. Pellets were resuspended in 200 l of nuclear extraction buffer (lysis buffer containing 500 mM NaCl) and incubated end-over-end for 60 min. Samples were centrifuged at 18,000 ϫ g for 10 min. Nuclear extracts were aliquoted, quick-frozen in liquid nitrogen, and stored at Ϫ70°C.
Electrophoretic Mobility Shift Assays-Purification of human p53 protein and electrophoretic mobility shift assays using this purified p53 were conducted as described previously (46). In brief, Sf9 cells that were infected with recombinant baculovirus expressing His-tagged p53 were lysed in 20 mM HEPES, pH 7.4, containing 20% glycerol, 10 mM NaCl, 0.2 mM EDTA, 0.1% Triton X-100, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 50 M leupeptin, and 50 g/ml aprotinin (Buffer L). Nuclei were pelleted by centrifugation at 2300 rpm and then resuspended in Buffer L containing 500 mM NaCl. Extracts were diluted to 100 mM NaCl with Buffer L, applied to a 0.5-ml nickel nitriloacetic acid agarose column (Qiagen) that was equilibrated with 20 mM HEPES containing 100 mM NaCl and eluted with 200 mM imidazole containing 10 mM HEPES, pH 7.4, and 5 mM NaCl. Fractions of 0.5 ml were collected, dialyzed against 10 mM HEPES, pH 7.4, 5 mM NaCl, 0.1 mM EDTA, 20% glycerol, and 1 mM dithiothreitol, aliquoted, and stored at Ϫ70°C.
Purified p53 protein or nuclear extract was incubated with 3 ng of radiolabeled double-stranded oligonucleotide and hybridoma supernatant where appropriate in a total volume of 30 l of DNA binding buffer, containing 20 mM MgCl 2 , 2 mM spermidine, 0.7 mM dithiothreitol, 1 mg/ml bovine serum albumin, and 25 g/ml poly[d(I-C)] for 30 min at room temperature. Samples were loaded on a native 4% acrylamide gel and electrophoresed in 0.5ϫ TBE at 225 V for 2 h at 4°C. Gels were dried and exposed to Kodak XAR-5 film using an intensifying screen at Ϫ70°C. Bands were scanned and quantitated using the Molecular Analyst Imaging Densitometer (Bio-Rad).
SDS-Polyacrylamide Gel Electrophoresis and Western Blot-Cells were lysed in 150 mM NaCl, 50 mM Tris-HCl, pH 7.6, 1 mM EDTA, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 50 M leupeptin, and 10 g/ml aprotinin. The protein concentration of each sample was determined using the Bio-Rad Protein Assay. Samples containing equal amounts of protein were electrophoresed in a 10% polyacrylamide gel. Following electrophoresis, protein was transferred to nitrocellulose and probed with a 1:1 mixture of the anti-p53 mouse monoclonal antibodies 1801 and 421. The secondary antibody was a horseradish peroxidaseconjugated goat anti-mouse IgG, and the signal was detected by the enhanced chemiluminescence method (Amersham Pharmacia Biotech).

Wild-type p53 Fails to Activate Transcription through the p53
Response Element of the bax Promoter in the Breast Carcinoma MDA-MB-453 Cell Line-Wild-type p53 expressed in the breast carcinoma MDA-MB-453 cell line is unable to activate transcription through the bax promoter or through the isolated p53 response element of the bax promoter (Figs. 1A and 2A). Luciferase reporter plasmids containing either the p21 promoter or the bax promoter were transfected into the p53-negative Saos-2 or MDA-MB-453 cell line with pCMV vector, increasing amounts of a plasmid expressing wild-type p53, or a plasmid expressing the mutant p53 V143A . In the Saos-2 cell line, wildtype p53 effectively activated transcription of reporter constructs containing either the p21 or bax promoters. In contrast, wild-type p53 expressed in the MDA-MB-453 cell line, although still capable of activating transcription of a reporter containing the p21 promoter, failed to activate transcription through a construct containing the bax promoter (Fig. 1A). Western blots demonstrated that p53 was expressed to equivalent levels in the two cell lines (  1B, lanes 10 and 12), suggesting that the failure of p53 to activate transcription through the bax promoter is not due to decreased levels of p53 protein expression.
To determine whether the isolated p53 response element of the bax promoter was sufficient for this differential effect, synthetic oligonucleotides corresponding to the p53 response elements of the p21 and bax promoters were cloned into the pGL3-E1bTATA luciferase reporter vector, upstream from the minimal adenovirus E1b promoter. Each reporter construct again was transfected into either the Saos-2 or MDA-MB-453 cell line with pCMV vector, increasing amounts of the wild-type The indicated values are the average of three independent experiments each performed in duplicate. The numbers above each bar indicate the fold activation for each pTATA construct observed with pCMV-p53 wt or pCMV-p53 V143A as compared with pCMV. p53 expression plasmid, or the mutant p53 V143A expression plasmid. In the Saos-2 cell line, wild-type p53 effectively activated transcription of constructs containing either the 5Ј or the 3Ј p53 response elements from the p21 promoter ( Fig. 2A), as well as a construct containing the p53 response element of the bax promoter ( Fig. 2A, inset). As observed with the promoter constructs, wild-type p53 expressed in the MDA-MB-453 cell line failed to activate transcription via an E1b reporter plasmid containing the p53 response element of the bax promoter ( Fig.  2A, inset), whereas activating reporters containing either the 5Ј or the 3Ј element of the p21 promoter ( Fig. 2A). Expression of wild-type p53 in MDA-MB-453 cells also activated transcription of reporters containing the p53 response elements of the cyclin G and cdc25C genes (Fig. 2B). Thus, the defect in p53dependent transcriptional activation of the bax promoter appears to be at the level of the interaction of p53 with its response element.
The p53 Response Element of the bax Promoter Consists of Overlapping Binding Sites for p53-The data presented in Figs. 1 and 2 demonstrate that in MDA-MB-453 cells there is a defect in wild-type p53-dependent activation via the 37-bp p53 response element of the bax promoter, as compared with the 5Ј p53 response element of the p21 promoter. To understand the molecular mechanism mediating this differential regulation of the p53 response elements, the interaction between p53 and its response element from the bax promoter was examined in detail by electrophoretic mobility shift assays. Previous studies localized the p53 response element of the bax promoter to a 37-bp region at Ϫ113 to Ϫ77 from the start site of transcription (29). An examination of the nucleotide sequence of this 37-bp element revealed three potential p53 binding sites, termed Site A, Site B, and Site C (Fig. 3), that correspond to the consensus site for p53 binding (17)(18)(19). Site A consists of the first 21 bp of the 37-bp response element, with two potential p53 half-sites separated by a 1-bp insert. The first half-site contains three bases that vary from the consensus (two purine-to-pyrimidine changes in the first quarter-site and one in the second quartersite). The second half-site of Site A matches the consensus sequence in all 10 bases. Site B consists of 20 bp including this same "consensus" half-site and a second half-site downstream, separated by no intervening sequences. Site B diverges from the consensus at three bases (the A/T is a G in the last position of the third quarter-site, and there are two purine-to-pyrimidine changes in the fourth quarter-site). Site C consists of 26 bp and includes the same half-site noted in Sites A and B, separated from a second half-site by a 6-bp insert. Site C contains two variations from the consensus sequence (a C to A change and a purine-to-pyrimidine change both in the fourth quarter site). Of note is the spatial relationship of these three potential p53 binding sites. The three sites overlap one another with the consensus half-site (Ϫ102 to Ϫ93) common to each. Because of this shared half-site, the binding of p53 to one site excludes its simultaneous binding to either of the other sites. Therefore, if one assumes that p53 binds as a tetramer (20 -22), then only one site can be occupied at any given time.
To identify which of these putative binding sites are responsible for the interaction between p53 and the bax promoter, synthetic double-stranded oligonucleotides were constructed to model each site (Table I). The Bax oligonucleotide contained the complete 37-bp p53 response element from the bax promoter. Oligo A contained the 21 bp corresponding to Site A, whereas Oligo B contained the 20 bp corresponding to Site B. Oligo C consisted of the 26 bp corresponding to Site C; however, because of the sequence overlap between Sites B and C the 6 bp separating the two half-sites in Site C were scrambled to abolish any potential contribution from Site B. Each oligonucleotide contained identical flanking sequences that allowed for its subsequent cloning into a luciferase reporter plasmid. The relative affinities of these oligonucleotides for p53 were assessed by electrophoretic mobility shift assay. Purified p53 bound the labeled Bax oligonucleotide containing the entire 37-bp p53 response element (Fig. 4A, lane 1), and this binding was effectively competed by an excess of the same, unlabeled oligonucleotide (Fig. 4A, lanes 2-4). Unlabeled Oligo A, Oligo B, and Oligo C also successfully competed for p53 binding (Fig. 4A, lanes 5-7, 8 -10, and 11-13). For comparison, an unrelated control oligonucleotide, Sens-1, was unable to compete for p53 binding (Fig. 4,A, lanes 14 -16, and B), demonstrating that the binding of p53 to Oligos A, B, and C is specific. In each case, however, the binding of p53 to the isolated sites was weaker than that observed with the entire 37-bp response element (Fig. 4B). These data suggest the possibility that in the context of the entire p53 response element of the bax promoter there is a cooperative interaction between the overlapping p53 binding sites that allows for enhanced p53 binding.
The ability of purified p53 to directly bind to these oligonucleotides in electrophoretic mobility shift assays was then examined. A labeled oligonucleotide corresponding to the 5Ј p53 response element of the p21 promoter was used as a positive control for p53 binding (Fig. 5, lanes 1-3). The p21-5Ј oligonucleotide was bound by p53 and was effectively supershifted by mAb 1801, a p53 N-terminal-specific monoclonal antibody (Fig.  5, lane 2). In addition, the labeled Bax oligonucleotide, corresponding to the entire p53 response element of bax, as well as those corresponding to Site A, Site B, and Site C were also bound by purified p53 (Fig. 5, lanes 4, 7, 10, and 13) and were supershifted by mAb 1801 (Fig. 5, lanes 5, 8, 11, and 14). This binding, however, was weaker than that observed with the FIG. 3. Schematic of the p53 response element of the human bax promoter. The previously identified p53 response element of the bax promoter is located at Ϫ113 to Ϫ77 from the transcriptional start site. Based on the p53 consensus binding site, there exists, within this 37-bp sequence, three potential, overlapping p53 binding sites. These putative binding sites are labeled Site A (Ϫ113 to Ϫ93), Site B (Ϫ102 to Ϫ83), and Site C (Ϫ102 to Ϫ77). The arrows indicate the four quarter sites that constitute each proposed p53 binding site. The p53 consensus sequence is indicated above each arrow with r representing purine and w representing either an adenine or thymine base. Bases in the bax sequence that vary from this consensus are indicated by asterisks. The perfect half-site shared by each potential binding site is highlighted by the gray box. The position of the TATA box for the bax promoter (Ϫ22 to Ϫ26) also is indicated. p21-5Ј site, requiring approximately 10-fold more p53 to generate a detectable band shift.
Previously our laboratory reported two distinct classes of p53 binding sites based on their responses to the C-terminal-specific mAb 421 (46). p53 binding to one class of sites, which includes the p21-5Ј site, is enhanced in the presence of mAb 421, whereas binding to the second class of sites is inhibited by mAb 421. Confirming our original observation, p53 binding to the p21-5Ј site was enhanced in the presence of mAb 421 (Fig.  5, lane 3). Binding of p53 to the Bax oligonucleotide as well as to Oligo C, however, was inhibited in the presence of mAb 421 (Fig. 5, lanes 6 and 15). The binding of p53 to Oligos A and B displayed an intermediate phenotype, in which mAb 421 failed to effectively supershift the p53-oligonucleotide complexes and failed to enhance p53 binding to the oligonucleotides (Fig. 5,  lanes 9 and 12). In either case, the data are consistent with the notion that binding to each of the bax sites as compared with the p21-5Ј site may require a conformationally distinct form of p53.
Overlapping, Low Affinity p53 Binding Sites Synergize for Complete p53-dependent Transactivation through the p53 Response Element of the bax Promoter-The Bax oligonucleotide as well as Oligo A, Oligo B, and Oligo C were cloned into the pGL3-E1bTATA luciferase reporter vector upstream from the adenovirus minimal E1b promoter. Each reporter construct was transfected with the pCMV empty vector, a plasmid expressing wild-type p53, or a plasmid expressing the temperature-sensitive p53 V143A mutant into the p53-negative Saos-2 cell line (Fig. 6). At 37°C the p53 V143A mutant fails to activate transcription through p53-responsive promoters. At 32°C, however, this mutant adopts a wild-type conformation and has been shown to activate some p53-responsive promoters (such as p21) but not others (such as bax) (36,38). At 37°C, wild-type p53 activated transcription through the complete 37-bp response element of the bax promoter (Fig. 6A). In addition, wild-type p53 activated transcription through Oligo B; however, this activation was significantly lower than that observed with the complete response element (21-fold compared with 67-fold). Although Oligos A and C both showed sequence-specific binding to p53 in an electrophoretic mobility shift assay (Fig. 4), p53 failed to activate transcription, to any significant

5Ј-aattcggtaccGAACATGTCCCAACATGTTGgctagcgaatt-3Ј
a The bold capital letters represent the sequences taken from the bax and p21 promoters. Bases that participate in the formation of potential p53 binding sites are indicated by underlining. The lowercase letters indicate sequences not derived from either the bax or p21 promoters. p53-dependent Transactivation of bax degree, through either sequence (Fig. 6A, 2-and 1-fold, respectively). The same pattern of activation was observed with wildtype p53 at 32°C (Fig. 6B). Similar to observations made with the bax promoter (36,38), the temperature-sensitive p53 V143A mutant at 32°C failed to activate transcription through any of the isolated p53 binding sites of the bax promoter (Fig. 6B, gray  bars). The p53 V143A mutant, however, did successfully activate transcription through the p21-5Ј response element inserted into the same pGL3-E1bTATA reporter vector (Fig. 6B, inset).
The transfection data demonstrate that Site B can mediate p53-dependent activation but that the level of activation conferred by this sequences is one-third of that observed with the complete 37-bp response element. To analyze which additional sequences in the 37-bp element are necessary for full activation, another set of synthetic double-stranded oligonucleotides was constructed (Table I). Oligo AB contained the 31 bp that correspond to the overlapping Sites A and B. Oligo AC consisted of the 37-bp response element; however, the 6 bp separating the two half-sites in Site C were scrambled to abolish any potential contribution from Site B. Oligo BC contained the 30 bp corresponding to the overlapping Sites B and C. Again, each oligonucleotide contained identical flanking sequences that allowed for its subsequent cloning into a luciferase reporter plasmid. These oligonucleotides were analyzed by electrophoretic mobility shift assay. Purified p53 bound the labeled Bax oligonucleotide containing the entire 37-bp p53 response element of the bax promoter (Fig. 7A, lane 1), and this binding was effectively competed by an excess of the same, unlabeled oligonucleotide (Fig. 7A, lanes 2-4). Oligo BC, as well as Oligo AC failed to compete for p53 binding to any greater degree than Oligo B (Fig. 7A, compare lanes 11-13 and 14 -16 with lanes  5-7). Oligo AB, however, effectively competed for p53 binding (Fig. 7A, lanes 8 -10). This competition was in the same range as that observed with the complete Bax oligonucleotide (Fig.  7B), suggesting that the two oligonucleotides share a similar affinity for the purified p53.
Each double-stranded oligonucleotide was inserted into the pGL3-E1bTATA reporter vector upstream of the adenovirus minimal E1b promoter and transfected into Saos-2 cells with either empty vector or the wild-type p53 expression vector (Fig.  8). Wild-type p53 effectively activated transcription through the 37-bp p53 response element of the bax promoter (60-fold) and to a lesser extent through Oligo B (13-fold). In contrast, p53 failed to significantly activate transcription through either Oligo A (2-fold) or Oligo C (1-fold). Consistent with the results of the electrophoretic mobility shift assays, p53 activated transcription through Oligo AB to a greater extent than through Oligo B (61-fold compared with 13-fold). This activation was in the same range as that observed with the complete p53 response element (61-fold compared with 60-fold). Both Oligos BC and AC failed to mediate any significant p53-dependent transactivation (4-fold and 1-fold respectively). These data confirm that in contrast to other p53 response elements, like the p21-5Ј site, in which two adjacent p53 half-sites mediate transcriptional activation, the p53 response element of the bax promoter consists of three half-sites that cooperate to bring about full activation.
Two Nuclear Factors Selectively Interact with the p53 Response Element of the bax Promoter but Are Not Responsible for Its Differential Regulation in MDA-MB-453 Cells-Given that the defect in the ability of p53 to activate transcription of bax is at the level of the interaction between p53 and its response element in the bax promoter, one potential mechanism to explain the failure of p53 to activate transcription of bax in MDA-MB-453 cells might be that cellular factors exist in this cell line that can selectively compete p53 for binding to the bax promoter. To investigate this possibility, the labeled Bax oligonucleotide was used as a probe with MDA-MB-453 cell nuclear extract in an electrophoretic mobility shift assay. Four distinct nuclear factors bound this oligonucleotide (Fig. 8A, lane 1). Three of these factors, labeled BoB1 and BoB2 (binder of bax 1 and 2), and n.s., were effectively competed by an excess of this same unlabeled oligonucleotide (Fig. 8A, lanes 2-4). The band labeled n.s. also was competed effectively by Oligos A, B, and C, as well as by the p21-5Ј oligonucleotide (Fig. 8A, lanes 2-16), suggesting that this factor is a nonspecific (n.s.) DNA-binding protein. In contrast, the bands labeled BoB1 and BoB2 were effectively competed by an excess of unlabeled Oligo B but were not competed by Oligo A, Oligo C, or the p21-5Ј oligonucleotide, demonstrating sequence specificity for Oligo B (Fig. 8A, compare lanes 8 -10 with lanes 5-7 and 11-16). The band shifts produced with nuclear extract of MDA-MB-453 cells were unaffected by the presence of anti-p53 antibodies (data not shown). In addition, BoB1 and BoB2 failed to bind the p21-5Ј oligonucleotide, as well as oligonucleotides corresponding to the p53 response element of the gadd45 gene and the 3Ј element of the mdm-2 gene (Fig. 8A, lanes 14 -16, and data not  shown). These results demonstrate the identification of two novel nuclear factors that display sequence specificity for the same region of the bax promoter that we have shown to be FIG. 6. Site B is sufficient to confer p53-dependent transactivation, but the level of transactivation is lower than that observed with the complete 37-bp response element. Saos-2 cells were transfected as described under "Materials and Methods" with 2 g of the indicated reporter constructs and 50 ng of either empty pCMV (white bars), the wild-type p53 expression vector pCMV-p53 wt (black bars), or the temperature-sensitive p53 expression vector pCMV-p53 V143A (gray bars). Cells were maintained either at 37°C (A) or shifted to 32°C 24 h prior to lysis (B). Luciferase activity and total protein levels were assayed as described under "Materials and Methods." The pTATA-p21-5Ј reporter construct (B inset) was used as a positive control for the pCMV-p53 V143A expression vector. The indicated values are the averages of three independent experiments each performed in duplicate. The numbers above each bar indicate the fold activation for each pTATA construct observed with pCMV-p53 wt or pCMV-p53 V143A as compared with pCMV. essential for p53-dependent transcriptional activation.
The identification of nuclear factors that showed sequence specificity for the p53 response element of the bax promoter suggests a potential mechanism for the differential activation of a reporter construct containing the bax promoter in MDA-MB-453 cells. To explore this possibility, the levels of BoB1 and BoB2 in Saos-2 (Fig. 8B, lanes 1-5) and MDA-MB-453 (Fig. 8B,  lanes 7-11) nuclear extracts were compared by electrophoretic mobility shift assay, using the Bax oligonucleotide as radiolabeled probe. No significant difference in BoB1 or BoB2 levels was observed between nuclear extracts from these two cell lines (Fig. 8B, compare lanes 1-5 with lanes 7-11) that had been normalized by total protein. These results suggest that BoB1 and BoB2 levels, as assessed by electrophoretic mobility shift assay, cannot explain the differential effects observed with wild-type p53 on its response element from the bax promoter in MDA-MB-453 cells as compared with Saos-2 cells.
The p53 Homolog p73 Does Not Selectively Inhibit the Ability of p53 to Activate Transcription through the bax Promoter-In  3, 6, 9, 12, and 15), or 1500-fold (lanes 4, 7, 10, 13, and 16) molar excess of the indicated unlabeled competitors. The arrow indicates the position of the p53-DNA complexes. B, bands were quantitated by densitometry and expressed as a percentage of the no competition signal (lane 1). C, Saos-2 cells were transfected as described under "Materials and Methods" with 2 g of the indicated reporter constructs and 50 ng of either pCMV (white bars) or the wild-type p53 expression vector pCMV-p53 wt (black bars). 48 h post transfection luciferase activity and total protein levels were assayed as described under "Materials and Methods." The indicated values are the averages of three independent experiments each performed in duplicate. The numbers above each black bar indicate the fold activation for each pTATA construct observed with pCMV-p53 wt as compared with pCMV.  1 and 7), 8 (lanes 2 and 8), 12 (lanes 3 and 9),  16 (lanes 4 and 10), and 20 g (lanes 5 and 11) of either Saos-2 (lanes 1-5) or MDA-MB-453 (lanes 7-11) nuclear extract was incubated with 3 ng of the probe. BoB1 and BoB2 indicate the positions of the two sequence-specific binding factors.
p53-dependent Transactivation of bax addition to BoB1 and BoB2, the p53 homolog p73 was examined as a potential explanation for the inability of wild-type p53 to activate transcription through the bax promoter in MDA-MB-453 cells. Saos-2 cells were transfected with a wild-type p53 expression vector, increasing amounts of an expression vector for p73␣ and either the p21P or pBax luciferase reporter constructs (Fig. 9). In the absence of p73, p53 activated transcription through both the p21 (12-fold) and bax (48-fold) promoters. The addition of increasing amounts of p73 failed to inhibit the ability of p53 to activate transcription through either the p21 or bax promoters, suggesting that p73 is not responsible for the differential activation observed with these two promoters in the MDA-MB-453 cell line.
An Intact C Terminus Is Not Required for p53-dependent Transcriptional Activation of the bax Promoter-Previous studies have demonstrated that C-terminal phosphorylation on serines 315 (47)(48)(49) and 392 (50) as well as acetylation of the C terminus (51) functionally alter the DNA binding characteristics of p53. Further, the ability of the C-terminal-specific mAb 421 to enhance the DNA binding activity of p53 has been proposed to be functionally similar to deletion of the last 30 amino acids of p53. In both cases, the binding of p53 to certain response elements is enhanced (50). As mAb 421 inhibits binding of p53 to the bax element, the effect of deletion of the terminal 30 amino acids was also examined. Saos-2 cells were transfected with either the p21P or pBax luciferase reporter plasmid and increasing amounts of pCMV-p53 wt , pB-p53 S315A , pB-p53 S315D , pCMV-p53 S392A , or pCMV-p53 ⌬370 -393 expression vector (Fig. 10). In each case p53 effectively activated transcription through both the p21 and the bax promoters, suggesting that neither phosphorylation of serine 315 or serine 392 nor an intact C terminus is required for the p53-dependent transactivation of the bax promoter. As compared with wildtype p53, each phosphorylation mutant activated transcription through the p21 promoter to an equal or greater extent. Although these mutants, S315A, S315D, and S392A, also clearly activated transcription through the bax promoter (up to 18-, 16-, and 24-fold, respectively), this level of activation was consistently lower than that observed with the wild-type p53 (up to 72-fold), suggesting that although loss of phosphorylation on either of these residues alone does not completely inhibit the ability of p53 to activate transcription through the bax promoter they may contribute in a partial manner. DISCUSSION The data presented in this report demonstrate that wild-type p53 expressed in the osteosarcoma Saos-2 cell line successfully activated transcription through the promoters of both the cyclindependent kinase inhibitor p21 and the proapoptotic bax. In contrast, p53 expressed in the breast carcinoma MDA-MB-453 cell line was capable of activating transcription through the p21 promoter but failed to do so through the bax promoter (Fig.  1A). A luciferase reporter construct containing the 37-bp p53 response element from the bax promoter displayed the same differential response to p53 as the reporter containing the complete promoter (Fig. 2). This suggests that the 37-bp p53 response element alone is sufficient to mediate this differential regulation and argues in favor of the notion that the differential effect depends on an inherent difference in the interaction of p53 with its response elements in the bax and p21 promoters. In this regard, the data demonstrate three distinct differences FIG. 9. The p53 homolog p73 does not selectively inhibit the ability of p53 to activate transcription through the bax promoter. Saos-2 cells were transfected as described under "Materials and Methods" with 2 g of either the p21P or pBax luciferase reporter plasmids, 0 ng (Ϫ) or 50 ng (ϩ) of pCMV-p53 wt , and 0 (Ϫ), 50, or 100 ng of pCMV-p73␣. 48 h post transfection cells were lysed and assayed for total protein and luciferase activity as described under "Materials and Methods." Appropriate amounts of the vector pCMV were added to each transfection mixture to maintain a constant level of plasmid DNA of 2.1 g/sample. The indicated values are the average of three independent experiments each performed in duplicate. The numbers above each bar indicate the fold activation for each reporter construct observed with pCMV-p53 wt and/or pCMV-p73␣ as compared with pCMV. between the p53 response elements from these two promoters. First, unlike the p21-5Ј element, which consists of two consensus p53 half-sites that form a high-affinity p53 response element, the response element of the bax promoter consists of three half-sites that cooperate in mediating p53-dependent transactivation (Fig. 7). Second, the studies with the C-terminal-specific mAb 421 suggest that the binding of p53 to its response element in the bax promoter, as compared with its binding to other response elements, involves a conformationally distinct form of p53 (Fig. 5). Finally, two novel nuclear factors, termed BoB1 and BoB2, were identified that demonstrated sequence-specific binding to the same region of the bax promoter that was essential for p53-dependent transactivation and failed to bind to the 5Ј element of the p21 promoter (Fig. 8).
The fact that the binding of p53 to the bax element, unlike that to the p21-5Ј element, failed to be enhanced by the addition of mAb 421 (Fig. 5) indicates that the binding of p53 to these two sequences may require conformationally distinct forms of p53. Thus, the inability of p53 to activate transcription through the bax promoter in certain cell lines, like MDA-MB-453, may be due to an altered post-translational modification that prevents p53 from acquiring the correct conformation for binding. Alternatively, binding to the bax element may induce a distinct conformational change in p53, as compared with when it is bound to the p21-5Ј element, that subsequently allows it to interact with a distinct set of additional regulatory factors, and the cell type-specific regulation is at the level of these additional regulators. This latter scenario has been observed with the transcription coactivator OCA-B. OCA-B is a B-cell-specific coactivator that markedly enhances transcription mediated by Oct-1 or Oct-2 through the octamer sequence of immunoglobulin promoters but fails to activate transcription mediated by the same Oct-1 or Oct-2 activators through octamer sequences in the histone H2B gene (52). Consistent with the notion that mAb 421 is revealing a conformational distinction significant to the observed differential regulation of bax, the ability of wild-type p53 to activate transcription through the p21-3Ј response element, to which the binding of p53 also is inhibited by mAb 421 (46), was significantly decreased in MDA-MB-453 cells as compared with Saos-2 (Fig. 2).
Within the C terminus, phosphorylation of serines 315 (47)(48)(49) and 392 (50,(53)(54)(55) as well as acetylation of lysines 370, 372, and 373 (51) have been shown to enhance the DNA binding (47)(48)(49)(50)(51), transcriptional activation (53,54), and growth suppressor (55) functions of p53. In fact, Scheidtmann and coworkers (49,54) have suggested that phosphorylation of serines 315 and 392 alters the ability of p53 to both bind to and activate transcription through the p53 response element of the bax promoter, in particular. Given these results and the observation that the C-terminal-specific mAb 421 inhibits the binding of p53 to the bax element (Fig. 5), we investigated whether or not these particular post-translational modifications could explain the observed defect in the ability of wild-type p53 to activate transcription through the bax promoter in the MDA-MB-453 cell line. The results in Fig. 10 demonstrate that although mutation of either serine 315 or serine 392 to alanine slightly decreases the ability of p53 to activate transcription through the bax promoter, as compared with the p21 promoter neither phosphorylation of 315 or 392 nor an intact C terminus is required for p53 to effectively activate transcription through either the bax or p21 promoters. Because the data presented here address each modification independently of the others, the possibility still exists that some combination of these modifications, or other C-terminal modifications not addressed here, may have a more significant impact on the ability of p53 to activate transcription through the bax promoter.
The identification of two novel nuclear factors, BoB1 and BoB2, that showed sequence specificity for the same region of the bax promoter that was essential for p53-dependent transactivation (Figs. 6 and 8) suggested an alternate explanation for the observed defect in MDA-MB-453 cells. Preliminary results indicated that the binding of p53 and BoB1 or BoB2 to the p53 response element of the bax promoter were mutually exclusive, suggesting that these factors may compete with p53 for binding (data not shown). These factors demonstrated a strong affinity for the bax element and poor affinity for the p21-5Ј element. In addition, BoB1 and BoB2 were found to display a moderate affinity for the p21-3Ј element (data not shown). Correspondingly, the level of p53-dependent activation of the reporter construct containing this 3Ј element was reduced in MDA-MB-453 cells when compared with its level of activation in Saos-2 cells (Fig. 2). These results suggested an inverse relationship between the affinity of these binding factors for a particular sequence and the ability of that sequence to mediate p53-dependent transcriptional activation in MDA-MB-453. When the levels of these factors in MDA-MB-453 and Saos-2 cells were compared, however, there was no discernable difference observed (Fig. 8B), suggesting that although these factors still may have some significance to the p53-dependent transactivation of bax, they do not explain the observed defect in the MDA-MB-453 cell line. One could hypothesize that the p53 homolog p73 might function in a manner analogous to that originally proposed for the BoB1 and BoB2 binding factors. Given the sequence homology between the DNA-binding domains of p53 and p73, it is reasonable to speculate that p73 can bind DNA at p53 response elements and, therefore, may compete with p53 for binding. The results presented here, however, do not support such a hypothesis. Expression of p73␣ was unable to inhibit the ability of p53 to activate transcription through either the bax or p21 promoters (Fig. 9). In fact, p73 was found to be a potent activator of transcription through the bax promoter (Fig. 9, up to 30-fold).
The identification of tumor-derived p53 mutants that selectively fail to activate transcription through the bax promoter and subsequently fail to undergo apoptosis (36 -39) suggests that the ability of p53 to activate transcription through the bax promoter is important to the tumor suppressor function of p53. The Bax protein, in fact, has been shown to play an important role both in inhibiting tumor progression and in promoting the apoptosis of tumor cells in response to DNA-damaging agents like those used in the treatment of cancer (56 -62). Studies have shown that decreased Bax levels are significantly associated with tumor cell resistance to chemotherapy (56,58) and that increased expression of Bax is sufficient to sensitize at least certain tumor cell types to apoptotic stimuli (57,60,61,63). In addition, the p53-dependent transcriptional activation of the bax gene has been shown to be important both in inhibiting tumor formation and progression (59,62,64) and in promoting apoptosis in response to radio and chemotherapy (59,63). As such, understanding the mechanism of p53-dependent regulation of the bax gene will provide new insights into the processes of tumor formation and progression, as well as the development of tumor resistance to treatment. The data presented here identify several characteristics that differentiate the p53 response element of the bax promoter from other p53 response elements, such as the p21-5Ј element. These characteristics suggest a potential mechanism for the cell type-specific regulation of the bax promoter by p53, as seen with the MDA-MB-453 and Saos-2 cell lines. The data demonstrate that in this model system the defect in the ability of wild-type p53 to activate transcription through the bax promoter is at the level of the interaction between p53 and its response element and that this interaction appears to involve a conformationally distinct form of p53 interacting with a unique arrangement of three half-sites. It is reasonable to speculate that the mechanism responsible for the failure of wild-type p53 to activate transcription through the bax promoter in MDA-MB-453 cells may also be relevant to the inhibition of bax induction observed both in tumor formation and progression and in tumors that are resistant to apoptosis-inducing treatments.