Two Functionally Divergent p53-responsive Elements in the Rat Bradykinin B2 Receptor Promoter*

Although p53 is known to have dual functions as a transcriptional activator and repressor, there has not been an example where both p53-activating and -repressing elements reside within one target promoter. Previous work from this laboratory defined two different p53 response elements, termed P1 and P2, located at nucleotide positions –70 and –707, respectively, in the rat bradykinin B2 receptor promoter. In this study, through manipulation of the DNA sequence and context, we demonstrate opposing roles for P1 and P2 as transcriptional activator and repressor, respectively. Deletion of P1 abrogates p53-mediated activation. P1 maintains its role as an activator upon relocation to the P2 site and activates transcription from a heterologous promoter construct. Thus, P1 is a bona fide positive p53-response element. In contrast, deletion of P2 enhances P1-induced activation. P2 represses transcription when substituted for P1 or when relocated midway between P1 and P2. P2-mediated repression is sequence-dependent, because it is reversed to activation when P2 is substituted by the P1 or p53 consensus sequences. Moreover, site-directed mutagenesis that converts P2 to a higher affinity p53-binding site results in transcriptional activation rather than repression. Surprisingly, P2 strongly activates a heterologous promoter. Thus, P2-mediated transcriptional repression is both sequence- and context-dependent. Investigations into the mechanisms of P2-mediated repression indicate that it is trichostatin-insensitive and unaffected by CBP or mutation of the minimal repression C-terminal domain of p53. However, gel shift assays suggest that p53 competes with other transcriptional activators for binding to overlapping binding sequences within the P2 element. In conclusion, this study provides a rare example of a transcription factor having two divergent functional effects that are sequence- and context-dependent. The interplay of P1 and P2 may be important in the regulation of bradykinin B2 receptor gene expression in response to inflammatory stress and during development.

p53-mediated activation of transcription is largely dependent on sequence-specific DNA binding. The consensus p53binding motif consists of two half-sites separated by 0 -13 nucleotides (18,23). Each half-site consists of the sequence RRRC(A/T)(T/A)GYYY. p53 binds the response element as a tetramer, each half-site bound to a p53 dimer (24). p53 also represses transcription of a number of genes. The mechanisms of p53-mediated repression are not well understood, because the majority of p53-repressed genes do not contain a consensus response element. Although initial studies suggested that p53 selectively represses TATA box-containing promoters, more recent studies indicated that p53 is capable of repressing TATAless promoters as well (25). Protein-protein interactions with the basal transcription machinery have been proposed as the underlying mechanism for p53-mediated repression of these promoters. Binding to and interference with the binding or function of upstream transcriptional activators such as SP1 and AP-1 has also been shown to mediate p53-induced repression (26 -29). In addition, recruitment of chromatin-modifying enzymes or modification of the acetylation status of p53 by interaction with and recruitment of histone deacetylase plays an important role in p53-mediated repression, for some (e.g. MAP-4 and ␣-fetoprotein (AFP) 1 ) but not all genes (e.g. p202) (30 -33). In several instances, p53 mediates promoter repression via direct DNA binding to a canonical p53-binding motif, as has been demonstrated in the AFP, p202, and survivin genes (31, 34 -36). At least in one case, the AFP gene, p53-mediated repression was dependent on displacement of an adjacent over-lapping transcriptional activator, HNF-3. In cells lacking HNF-3, p53 mediates activation rather than repression of the AFP promoter (35). In another case, the MDR1 gene, p53mediated repression was dependent on binding to a novel p53binding motif in which the two copies of the p53-binding site are arranged in head-to-tail orientation. Replacement of the head-to-tail site with a head-to-head site converted the activity of p53 from repression to activation (37). This finding provided the initial demonstration that inherent features in the p53response element can confer differential functional responses. More recently, a novel 21-bp DNA element, called the p53 transcriptional repressor element (p53TRE), was identified in the promoter of the placental transforming growth factor ␤ gene (38). This sequence (GYYDCCCAGGCTGGARTGSWR, where R is purine, Y is pyrimidine, W is adenine or thymine, S is guanine or cytosine, and D is guanine, adenine, or thymine) diverges from the consensus p53 sequence and represses p53 transactivation in a position-and promoter-independent fashion.
We recently identified the BdkrB2 gene, which encodes the bradykinin B 2 receptor (B 2 R), as a direct transcriptional target for p53 (39). B 2 R is a G protein-coupled receptor that is involved in the regulation of cardiovascular and renal functions as well as inflammation. The rat BdkrB2 promoter contains two functional p53-binding sites: a P1 site located at nucleotide positions Ϫ70; and P2, located at Ϫ707 bp, relative to the transcription start site. The P1 and P2 sequences bear 80% or greater homology to the p53 consensus sequence and bind to recombinant p53 with high affinity (22,39). Initial promoter function analysis suggested that p53 activates the BdkrB2 promoter via the P1 site and that P2 down-modulates P1 function (39). However, the potential mechanisms of such effects were not addressed. By manipulating the sequence and context (location and spacing) of the P1 and P2 sites, we investigated in this study whether p53-mediated promoter activation and repression are dictated by the sequence of the p53-response element and/or its context. In addition, we have examined some of the potential mechanisms of P2-mediated repression within the context of the BdkrB2 promoter. The results indicate that the functions of P1 and P2 elements as activator and repressor, respectively, are dependent on their intrinsic DNA sequence. Moreover, the function of P2 as a repressor is dependent on its presence within its native promoter context. p53-mediated repression of BdkrB2 via P2 is unaffected by histone acetyltransferases, histone deacetylases, and mutation of the minimal repression C-terminal domain of p53 but may involve competitive displacement of transcriptional activators.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-HeLa cells (American Type Culture Collection) were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, penicillin (100 units/ml), and streptomycin (100 g/ml) (Invitrogen) at 37°C in a humidified incubator with 5% CO 2 . Cells were plated in duplicates in 6-well plates at 4 ϫ 10 5 cells/well in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum 1 day prior to transfection. Cells were co-transfected with 1.5 g of reporter DNA with or without the p53-expression vector, pCMV-p53. A control ␤-galactosidase vector pSVZ (Promega, 0.5 g of DNA/well) was cotransfected to correct for transfection efficiency. Transfections were performed using LipofectAMINE Plus Reagent (Invitrogen) according to the manufacturer's recommendations. Four hours after transfection, fresh medium was replaced and cell extracts were prepared 48 h later using a reporter lysis reagent (Promega). CAT activity was measured by thin layer chromatography as described after normalization for protein concentrations or ␤-galactosidase activity (39).
Promoter Constructs-The original promoter-reporter construct pB-dkrB2 Ϫ1184/ϩ55-CAT and its derivative lacking the P2 site, Ϫ1184/ CAT-⌬ P2, were described (39). Utilizing the above two plasmids as templates, a series of point mutations, deletion, substitution, swap, and insertion constructs were engineered using the QuikChange site-directed mutagenesis kit (Stratagene). The promoter constructs and mutagenesis primers are depicted in Table I. The mutant clone p53(M340P,F341D) was obtained by the same method. The Ϫ358P2/P1 construct was generated by inserting P2 at nt position Ϫ358 in the ⌬P2/P1 construct. All clones were subjected to DNA sequencing. Transfections were performed in duplicate and performed at least three times. In experiments utilizing Trichostatin A (TSA, Sigma Chemicals), TSA was added concentrations of 100 M for 18 h or 100 ng/ml for 2 h.

RESULTS
The BdkrB2-P1 Site Is a Positive p53 Response Element-In a previous study, we established that the rat BdkrB2 promoter has two p53-binding sequences located at nucleotide positions Ϫ70 (P1) and Ϫ707 (P2), respectively, relative to the transcription start site (39) (Fig. 1A). Transient transfection assays performed in HeLa cells revealed that expression of wild-type p53 dose dependently stimulates BdkrB2 promoter-driven transcription of the CAT reporter (Fig. 1B). Deletion analysis has previously localized a positive p53-response element in the DNA fragment extending from Ϫ38 to Ϫ94 bp, which encompasses the P1 site at position Ϫ70 (39). To demonstrate conclusively that the P1 site mediates p53-induced transactivation of the BdkrB2 gene, we compared the effect of exogenous p53 on the wild type BdkrB2-CAT reporter plasmid (Ϫ1184 to ϩ55 bp, designated P2/P1) with that of a Ϫ1184/ϩ55 BdkrB2-CAT plasmid lacking the P1 site (designated P2/⌬P1). The results showed that elimination of P1 completely abrogates p53-mediated activation (Fig. 1C). Furthermore, elimination of P1 abolishes p53-mediated activation of the Ϫ635/ϩ55-CAT promoter construct, which lacks the repressor P2 element (Fig. 1C). We have previously shown that a DNA-binding mutant p53 cannot activate BdkrB2 transcription (39). Based on these results, we conclude that p53 mediates direct activation of the BdkrB2 promoter via the P1 p53-binding site.
To further define the role of P1, we inserted one copy of the P1 sequence upstream of the TATA box in the TATA-Luciferase reporter construct (P1/TATA-Luc). As shown in Fig. 1D, p53 causes a powerful dose-dependent activation of the heterologous promoter construct. This activation is not observed in cells transfected with pCMV-empty vector alone. In addition, as shown below, P1 is capable of transactivation when substituted for P2, although the magnitude of P1-mediated activation is higher when P1 is located closer to the basal promoter region (compare P2/P1 with P1/⌬P1 in Fig. 2B). These results confirm and extend the notion that P1 is a bona fide positive p53response element.
The P2 Element Mediates Sequence-and Context-dependent Repression-To determine the role of P2, we generated several mutant versions of the original P2/P1 construct, which include single or double deletions (⌬P2/P1, P2/⌬P1, and ⌬P2/⌬P1), a swap (P1/P2), a swap with deletion (⌬P2/P2 and P1/⌬P1), and insertion with substitution (P2/P2). Deletion of the P2 site (⌬P2/P1) doubles P1-mediated activation (Fig. 2, A and B). Although the P1/⌬P1 construct is responsive to p53, the P1/P2 construct is not, confirming that P2 strongly represses P1mediated activation. However, because P2/P1 is more responsive to p53 than P1/P2, we conclude that the location of the two elements is an important determinant of P1 and P2 function. Relocation of P2 from its natural position at nt Ϫ707 to nt Ϫ358 did not affect p53-mediated responses (compare P2/P1 and Ϫ358P2/P1, Fig. 2B), suggesting that P2-mediated repression is independent of the distance (spacing) between the P1 and P2 sites. Also, the constructs in which P1 was substituted by P2 (⌬P2/P2 and P2/P2) are highly repressed by p53 as compared with ⌬P2/P1 and P2/P1 (Fig. 2, A and B). Based on these findings, we conclude that the presence of P2 dictates transcriptional repression in response to p53 in the BdkrB2 promoter. Interestingly, the double deletion construct (⌬P2/⌬P1) is responsive to p53-mediated activation (about 40% of P2/P1 activity) (Fig. 2B). Because 5Ј-end truncation of ⌬P2/⌬P1 from Ϫ1184 to Ϫ827 (by NcoI digestion) eliminates p53-mediated activation (Fig. 2B), we conclude that an additional positive p53-responsive element resides in the region between nt Ϫ1184 and Ϫ827. The function of this upstream element becomes To determine if p53-mediated repression via P2 is an intrinsic property of the P2 DNA sequence, we first replaced the P2 sequence in the P2/P1 and P2/⌬P1 plasmids with a consensus p53 sequence (p53-CS) (AGGCATGTCTAGGCATGTCT) to yield the plasmids CS/P1 and CS/⌬P1. Fig. 3A shows that replacement of P2 with p53-CS converts p53-induced repression to activation. The transcriptional activation by CS and P1 tended to be additive, because deletion of P1 reduces the magnitude of p53-mediated activation (p ϭ 0.06). In addition to P2 deletion or replacement, we evaluated the effects of P2 mutagenesis (double or triple mutations) on p53-mediated repression. Unexpectedly, a mutant P2 oligoduplex probe exhibited a higher binding activity to recombinant p53 than the wild-type P2 (Fig. 3B). To determine whether P2 interactions with p53 can be demonstrated in a more physiological setting, we incubated wild-type and triple-mutant P2 with nuclear extracts derived from newborn rat kidneys (Fig. 3C). As demonstrated here and previously (39,40), the newborn rat kidney contains higher amounts of p53 than the adult kidney (Fig. 3D). Similar to the EMSA results using recombinant p53, the binding of the mutant P2 probe to nuclear extracts was higher than that of the wild-type probe (Fig. 3C). The presence of p53 in the nuclear extracts is demonstrated by the supershift induced by the p53 antibodies (Fig. 3C). Transient transfection assays showed that the double and triple mutant P2 constructs are much more responsive to p53 than the wild-type P2/P1 construct (Fig. 3E). Collectively, P2 deletion, replacement, and mutagenesis provide firm evidence that P2-mediated repression is sequencespecific.
To examine whether P2 function as a repressor can be maintained outside the context of the BdkrB2 promoter, we inserted three tandem P2 sites upstream of the 13 p53-binding sites in the PG13-Luc construct. Transient transfection in HeLa cells revealed that the P2(x3)-PG13-Luc construct had a higher basal activity than PG13-Luc, and p53 activated both constructs equally well (data not shown). Moreover, insertion of a single P2 site upstream of a TATA-Luc minimal promoter resulted in a strong dose-dependent p53-mediated activation rather than repression (Fig. 4). Therefore, the role of P2 as a repressor is promoter-dependent. A summary of the results of P2 and P1 mutagenesis is presented in Table II.
Role of Histone Acetylase and Deacetylase-Recruitment of HDACs to the promoter by p53 has been shown to mediate p53 repression of certain promoters and could account for P2-mediated repression (33). Also, HDACs may theoretically deacetylate p53. To test these possibilities, we first treated HeLa cells with trichostatin A (TSA), a selective HDAC inhibitor, at doses shown previously to reverse p53-mediated transcriptional repression. We then performed transient transfection assays with the p53 and the promoter-reporter constructs. As shown in Fig. 6A, treatment with TSA has no appreciable effect on P2-mediated repression of the P2/P1 and P2/⌬P1 constructs, indicating that p53-mediated repression is probably exerted by a mechanism that is independent of HDAC activity. In the absence of p53, TSA stimulated slightly the basal activity of the promoterless CAT construct (Fig. 6A).
We previously reported that p53-mediated activation of the P2/P1 construct is further enhanced in the presence of cotransfected CBP-expressing vector (39). We surmised that if P2 interferes with P1 function by competition for the available intracellular pool of CBP/p300, constructs lacking P2 but having P1 should respond more vigorously to CBP/p300 (and p53) than those with an intact P2 site. CBP enhanced p53-mediated activation by 40% above that of p53 alone. However, the magnitude of activation of P2/P1 and ⌬P2/P1 was not different (Fig. 6B).

FIG. 3. P2-mediated repression is dictated by its sequence. A,
HeLa cells were transfected with various BdkrB2 promoter constructs (1.5 g) with or without 10 ng of pCMV-p53. Replacement of P2 in the P2/P1 or P2/⌬P1 constructs with a consensus p53 sequence (CS) converts p53-induced repression to activation. B, EMSA using P2-wt or P2-tplmut oligoprobes (40,000 cpm) and recombinant p53 in the amounts indicated. C, EMSA using P2-wt or P2-tplmut oligoprobes incubated with nuclear extracts (5 g of protein) derived from newborn or adult rat kidneys. Anti-p53 antibodies (a mixture of phospho-Ser-20 and -Ser-15 and acetylated p53 antibodies, 2.5 l each) were added 30 min before addition of probe. D, immunoblot of newborn (NB) and adult (AD) kidney nuclear extracts (15 g of protein/lane) for Ser-20 phospho-p53. E, HeLa cells were transfected with the indicated promoter constructs (1.2 g each) with or without 10 ng of pCMV-p53. Cell extracts were harvested at 24 h and assayed for CAT activity.  Differential Binding of p53 to P1 and P2-Differences in the binding affinity of the P1 and P2 elements to p53 might account, at least partly, for their differential roles as activator and repressor, respectively. We therefore compared the binding activity of recombinant p53 to P1 and P2 using EMSA (Fig. 7A). The protein⅐DNA complex is efficiently competed by unlabeled P1 or P2 in a dose-dependent manner. As judged from the amount of competitor oligoduplex required to displace p53 binding, the relative affinity of p53 for P1 is ϳ30% higher than P2 (Fig. 7B). We next tested the binding of radiolabeled P1 and P2 to newborn rat kidney nuclear extracts. The protein⅐DNA complexes produced using P1 were more abundant and migrated more slowly than those of P2. These qualitative and quantitative differences prompted further investigations into whether other nuclear DNA-binding proteins share with p53 the ability to bind to P2. A transcription factor search revealed that the P2 sequence has putative overlapping binding sites for NF-B and Kruppel-like factor-4 (formerly known as GKLF) (Fig. 8A). Competitive EMSA utilizing the P2 probe and unlabeled p53-CS, P1, NF-B, AP-1, KLF-4, and cAMP-response element (CRE) oligoduplexes were performed (Fig. 8B). The P2 probe produced three specific DNA⅐protein complexes (C1-C3). C1 is competed completely by unlabeled p53 CS. Unlabeled P1 also competes well against P2 for binding. Interestingly, C1 is reduced equally well by unlabeled KLF-4 and to a lesser extent by CRE and Jun/AP-1 but not by NF-B oligoduplex. C2 is competed only weakly by p53 CS, P1, and other oligoduplexes. C3 is completely eliminated by unlabeled KLF-4 oligoduplex and decreased by unlabeled NF-B and AP-1 duplexes. These findings indicate that P2 is a potential binding site for multiple transcription factors, including p53, NF-B, and KLF-4 and possibly others, and suggest that p53 competes with NF-B and KLF-4 for binding to the P2 sequence.

DISCUSSION
The present study demonstrates that the transcription factor p53 can simultaneously activate and repress a target gene via sequence-specific binding to two different cis-response elements. In the case of the rat BdkrB2 promoter, the end-result is transcriptional activation as the P1 element greatly overpowers the P2 repressor element. Multiple lines of evidence suggest that P1 is a bona fide positive p53-response element: 1) the minimal BdkrB2 promoter (Ϫ38/ϩ55) cannot respond to p53; in contrast, a Ϫ94/ϩ55 promoter construct containing the P1 p53-binding site (Ϫ50 to Ϫ70) is highly p53-responsive (22,39); 2) deletion of P1 abrogates p53-mediated activation; 3) relocation of P1 (e.g. P1/⌬P1) maintains p53-mediated activation; and 4) P1 mediates activation of a heterologous minimal promoter-reporter construct. Conversely, several observations support the notion that P2 is a sequence-specific repressor element: 1) interference with P2 function by either truncation, deletion, or mutagenesis up-regulates p53-mediated activation via P1; 2) P2 represses transcription when substituted for P1; 3) substitution of P2 by a consensus p53-binding sequence converts the repression to activation; and 4) P2 maintains its repressor activity within the context of the BdkrB2 promoter irrespective of P2-P1 spacing. However, once taken out of its native context and placed in front of a foreign promoter, P2 behaves as an activator. Thus, the role of P2 is dictated by both its own sequence and the inherent properties of the surrounding DNA elements.
The mechanisms of p53-mediated transcriptional repression are complex and not fully understood but include sequence FIG. 5. Mutagenesis of the p53 minimal repression domain (p53MRD) has no effect on P2-mediated repression. A, HeLa cells were co-transfected with pCMV-p53 or pCMV-M340P/F341D (0, 10, 50, and 100 ng) and 1.0 g of pSV-lacZ. Cell extracts were harvested 48 h later, normalized for protein concentrations, and assayed for ␤-galactosidase activity. B, HeLa cells were co-transfected with pCMV-p53 or pCMV-M340P/F341D (50 ng each) and the indicated promoter constructs (1.5 g each). Cell extracts were harvested 48 h later and assayed for CAT activity.

FIG. 6. Effect of trichostatin A (TSA) or CBP on P2-mediated repression.
A, HeLa cells were treated with TSA (100 nM for 18 h) then co-transfected with pCMV-p53 (10 ng) and the indicated BdkrB2 constructs or the promoterless construct, pCAT3Basic. B, co-transfection of pRSV-CBP (50 ng) and pCMV-p53 (10 ng) activated transcription of P2/P1 and ⌬P2/P1 to a similar extent relative to p53 alone. A value of 1.0 was assigned to CAT activity in response to p53 alone (dashed line).
nonspecific and sequence-specific mechanisms (26 -38). Nonspecific repression includes binding to and interference with basal transcription factors, binding to an activator (e.g. SP-1 or AP-1), or competition for a co-activator. Sequence-specific repression includes direct DNA binding to a p53 response element, recruitment of a co-repressor (e.g. msin3⅐HDAC complex) and competition for binding with other transcription factors with overlapping binding sites. Our results indicate that TSA, a selective HDAC inhibitor, had no effect on P2-mediated repression. Thus, a direct role for HDAC in histone or p53 deacetylation does not appear to be a mechanism. This is not surprising, however, because not all p53-repressed genes are TSA-sensitive. We also tested the possibility that P1 and P2 compete for a common co-activator, such as CBP/p300. However, forced CBP expression in the absence of P2 had no additional influence on p53-mediated activation. This finding suggests that competition for CBP is not a contributing factor.
We further considered whether p53-mediated repression via P2 is determined by structural elements in the p53 protein itself. Hong et al. (41) have identified a minimal repression domain (p53MRD) that inhibits p53-mediated transcriptional activation. The MRD maps to an 8-amino acid (339 -346) segment of the C-terminal of native p53. The function of this domain is selective for p53, because it did not affect VP16mediated activation. Also, mutations of two amino acids, M340P and F341D, were sufficient to prevent p53-mediated repression of the thymidine kinase promoter. We introduced p53M340P/F341D into HeLa cells and determined its effects on transcription from the P2/P1 construct and its mutant derivatives. We reasoned that, if the p53MRD contributes to P2mediated repression, then the mutant p53 should not have such an effect. The results showed that P2-mediated repression is maintained in the presence of p53M340P/F341D. Interestingly, the mutant p53 construct exhibited a weaker transcriptional activity than native p53, despite similar transfection efficiency. These results suggest that p53-mediated repression via P2 is not dependent on the p53MRD. Although the MRD is the site of E6 binding by the papilloma viral protein and HeLa FIG. 7. Binding characteristics of P1 and P2 to p53. A, electrophoretic mobility shift assay (EMSA). Double-stranded 32 P-labeled P1 (5Ј-GGAGGTGCCCAGGAGAGTGA-3Ј) and P2 (5Ј-ACTCTTGCCTGGTCTTCCCT-3Ј) oligoduplexes (40,000 cpm/reaction) were incubated with 500 ng of recombinant p53. Increasing amounts of cold competitor oligoduplex (50-to 150-fold) were added to the reaction mixture. B, densitometric analysis of the band intensities in A. C, EMSA utilizing radiolabeled p53 consensus sequence (CS), P1 or P2 (20,000 cpm/reaction), incubated with 5 g of nuclear proteins extracted from newborn rat kidneys. Increasing amounts of cold competitor oligoduplex (50-to 100-fold) were added to the reaction mixture. Arrows point to specific DNA⅐protein complexes. cells express the E6 protein, impaired E6-p53 interactions cannot explain the loss of repression mediated by mutant p53MRD. p53MRD-mediated repression may be more relevant to non-sequence-specific repression seen with overexpression of p53 than to sequence-specific repression. For example, p53mediated repression of the survivin gene is not affected by mutation or deletion within the C-terminal domain, including the MRD (36). In the case of the P2 element, our results clearly show that p53 acts in a sequence-specific manner, and maintained p53mRD function is therefore consistent with the survivin example.
An important finding of this study is that P2 mediates repression of BdkrB2 promoter regardless of whether or not P1 is present and irrespective of the spacing between P1 and P2. Also, because replacement of P2 with P1 or the p53 consensus sequence strongly activates transcription and mutagenesis of P2 increases its p53-binding activity and enhances transcription, we suggest that P2-mediated repression of the BdkrB2 promoter is sequence-dependent. Finally, because P2 represses constructs lacking P1, P2-mediated repression cannot simply be attributed to interference with P1 function. However, because P2 activates, rather than represses, a minimal TATAreporter construct, we conclude that P2-mediated repression is context-dependent and suggest further that sequences within and surrounding the P2 site in the BdkrB2 promoter play a crucial role in the determination of P2 function as a repressor.
The present studies have also allowed us to identify a potential third p53-response element in the BdkrB2 promoter. This is because the promoter construct lacking both the P2 and P1 sites, i.e. ⌬P2/⌬P1, is activated by p53 (about 40% of P2/P1 activity). Because the Ϫ635/⌬P1 construct is not activated by p53, we hypothesized that one or more additional p53-responsive elements reside further upstream of nucleotide Ϫ635 of the BdkrB2 promoter. Indeed, deletion of the upstream region extending from nt Ϫ1184 to Ϫ827 in the ⌬P2/⌬P1 construct abrogated p53-mediated activation. Additional studies will need to be performed to fully characterize this new p53-responsive element.
It is important to point out additional intrinsic similarities and differences in the functional behavior of P1 and P2. For example, in gel shift assays, P1 and P2 bind recombinant p53 with different affinities (P1 Ͼ P2). Also, when tissue nuclear extracts are utilized, the intensity and migration of protein⅐P1 complexes are strikingly different from those of protein⅐P2; the P1 complexes exhibit higher affinity and slower mobility than P2 complexes. Therefore, the conformation and/or composition of the P1 and P2 DNA⅐protein complexes are likely different. In this regard, it is tempting to speculate that binding of p53 to P1 favors DNA looping promoting nucleation of the preinitiation complex or that P1 recruits p53 along with a co-activator, whereas P2 does not. Another possibility that is suggested by the EMSA results is that p53 competes with other transcription factors (e.g. NF-B and KLF-4) for binding to the P2 element, which may contribute to P2-mediated repression.
The physiological relevance of p53-mediated regulation of the BdkrB2 gene is just beginning to be elucidated. For example, during kidney development, p53 co-localizes with B 2 R in renal epithelial cells undergoing terminal differentiation, and nuclear p53 levels are much higher in developing than mature kidneys as are B 2 R levels (22,40). Further, p53-deficient pups on C57BL6 background manifest ectopic B 2 R expression in the kidney (22). Thus, p53 may regulate the developmental spatiotemporal expression of the BdkrB2 gene. It is not clear, however, how the interplay between P1 and P2 modulates these processes. Whereas P1 is evolutionarily conserved between rat, mouse, and human, P2 is not. This raises the intriguing possibility that the BdkrB2 gene has acquired enhanced responsiveness to p53 during evolution. Perhaps P2 serves to restrict P1-mediated activation of the BdkrB2 gene during terminal differentiation.
In summary, the rat BdkrB2 gene represents a rare example of a mammalian gene that is reciprocally regulated by a single transcription factor via sequence-specific DNA binding. Although promoters with more than one functional p53-binding site have been described (e.g. p21 and cyclin G) (42), the presence of two functionally divergent p53-binding elements within the same promoter is a novel finding. This should offer a useful system to investigate how p53 activates or represses gene transcription in the context of the endogenous promoter. More importantly, this system may allow us to determine whether the divergent p53-mediated transcriptional responses are stimulus-dependent (e.g. DNA damage versus development).