Maf transcriptionally activates the mouse p53 promoter and causes a p53-dependent cell death.

An increase in the level of the tumor suppressor protein p53 can induce cell cycle arrest or cell death. Although mechanisms for regulating the life span of p53 have been described, there is growing evidence that transcriptional regulation of the p53 gene contributes significantly to controlling p53 protein levels and therefore the fate of a cell. However, the signal transduction pathways that lead to transcriptional activation of the p53 gene are poorly understood. The oncoprotein v-Maf and its cellular counterparts belong to the large combinatorially complex basic leucine zipper family of transcription factors, which include the AP1 family. To date few cellular targets of c-Maf have been identified. It is demonstrated here that v-Maf can bind as a homodimer to a variant Maf recognition element located between -66 and -54 upstream in the mouse p53 promoter. V-Maf and its cellular counterparts are shown to activate p53 expression through this site. The ability of v-Maf to activate p53 expression is modulated by AP1 family members. In addition, overexpression of v-Maf in primary cells leads to a p53-dependent cell death. Thus, Maf and members of the AP1 family are able to regulate p53 expression through this site in the p53 promoter.

The v-maf oncogene was identified as the transforming gene of the avian retrovirus AS42 and was first isolated from a spontaneous musculoaponeurotic fibrosarcoma of chicken (1). Inoculation of this virus into newborn chickens induced tumors that were pathologically indistinguishable from the original tumor and when introduced into chicken embryo fibroblasts led to cellular transformation (1,2). In human carcinogenesis, c-MAF was shown to be overexpressed in 25% of multiple myelomas tested due to the translocation of c-MAF to the IgH locus (3).
A cellular homologue of v-maf, c-MAF type I, was identified in chickens. Comparison of the amino acid sequences revealed that the only differences between chicken c-MAF and v-Maf are the substitution at position 257 of a methionine residue for valine and the presence of the viral Gag sequence at the amino terminus (4). Comparison of the mouse, rat, and chicken c-Maf sequences shows them to be very highly conserved at the primary amino acid level.
c-Maf was found to be but one member of an extended multigene family, members of which all share homology in a basic leucine zipper (b-Zip) 1 motif, and hence form a distinct subgroup of the b-Zip family of transcription factors (5). Members of the Maf family are themselves divided into two subgroups as follows: the large Maf proteins c-Maf (4), MafB (6), and neural retina-specific leucine zipper (7), and the small Maf proteins MafG (8), MafF, and MafK (9).
Consistent with being a member of the b-Zip family of transcription factors, c-Maf contains a typical acidic type transactivation domain in the amino terminus, and a carboxyl terminus consisting of a basic DNA binding domain and an adjacent leucine-rich region. This domain shares 20 -30% homology with those of other b-Zip proteins (4). The leucine motif is somewhat atypical for b-Zip proteins as position 5 in the heptad repeat of hydrophobic residues is occupied by a tyrosine rather than a leucine residue. Maf is able to form homodimers through the leucine zipper motif, bind DNA, and transactivate promoter/ reporter constructs in a sequence-specific manner (4,10). Interestingly, the ability to form homodimers appears to be critical for v-Maf to transform primary chicken embryo fibroblasts (4).
There are two known DNA consensus sequences bound by Maf homodimers, a 13-base pair 12-O-tetradecanoate-13-acetate-responsive element (TRE) type Maf recognition element (MARE) and a 14-base pair cAMP-responsive element (CRE) type MARE (10). The former MARE contains an AP1 (TGCT-GACTCAGCA)-binding site, and the latter harbors a CRE (TGCTGACGTCAGCA) bound by members of CREB/ATF family of transcription factors (10). The various dimers formed among Maf family members and other b-Zip proteins have been shown to bind to MAREs and MARE-like sequences with varying affinities and with varying transactivation potentials (10,11), creating a combinatorially complex and sensitive network of b-Zip transcription factors. c-MAF is widely expressed to modest levels in adult and embryonic tissues and highly expressed in some developing skeletal tissues (6). Currently, there are few cellular genes known to be transcriptionally regulated by Maf. One is the interleukin-4 (IL-4) gene (12). IL-4 is known to affect T helper cell subset ratios. The expression of c-Maf in the immune system has been shown to be specific to the T helper 2 (Th2) subset of T helper cells. Maf binds to a MARE-like element in the IL-4 promoter, resulting in increased expression and secretion of IL-4 by Th2 cells, causing the preferential differentia-tion of naive T cells into Th2 cells (12). Another known target gene for c-Maf is the L7 gene, which is expressed in all adult cerebellar Purkinje cells as well as in cells in certain functional domains of the developing nervous system. Mouse c-MAF was shown to activate transcription from two sites within the L7 promoter in mouse cerebellar cells (13). In addition, induction of lens differentiation in the chicken was shown to be triggered by a lens-specific Maf that regulates multiple genes expressed in the lens (14).
A data base search of sequence segments differing from the MARE by fewer than 5 base pairs revealed several other possible target genes for Maf, including the placental-type glutathione S-transferase gene and the proto-oncogenes pim-1 and c-erbB, although none of these has been confirmed (10). Another possible gene target for Maf, which was also identified by the data base search, is the p53 tumor suppressor gene (10). Between Ϫ66 and Ϫ54 (all numbering is relative to the start site of transcription) in the mouse p53 promoter is a sequence that differs from the TRE-type MARE consensus sequence by only 2 base pairs. This region is completely conserved between the mouse and human p53 promoters (Fig. 1, see Refs. 15 and 16). The AP1 site within the p53 MARE (which differs by 1 base pair to the AP1 consensus sequence) has been designated the p53 factor 1 (PF1) site (17). This site is important in regulating p53 promoter activity (18,19) and has been shown to bind several activities, some of which contain the AP1 components, c-Fos and c-Jun (18,19).
These observations led us to test the possibility that Maf could transcriptionally regulate the p53 gene. We demonstrate that v-Maf binds to the MARE-like site in the mouse p53 promoter and transactivates the p53 gene in a sequence-specific manner. Results show that overexpression of v-Maf causes cells to die in a p53-dependent manner. Furthermore, the ability of v-Maf to activate p53 expression is modulated by members of the AP1 family. Therefore, we conclude that a cellular target of the Maf transcription factor is the p53 gene and suggest that Maf has the potential to act as a "tumor suppressor" since it is able to induce cell death through its control of p53 expression.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmid pCAT3M contains the bacterial chloramphenicol acetyltransferase (CAT) gene but no eukaryotic promoter sequences upstream of this gene (20). pAACAT contains Ϫ224 to ϩ116 of the mouse p53 promoter blunt-end cloned in front of the CAT gene in pCAT3M (15).
pEF/luc expresses the firefly luciferase gene under control of the EF promoter and was created by cloning the firefly luciferase gene into the BssHII site of pEFBssHII.
In the cytomegalovirus (CMV)-based plasmids, transcription is con-trolled by the immediate early enhancer-promoter of human CMV. pCMVE1b58 kDa expresses the adenovirus 2 E1b 58-kDa protein (21). pCMVjun expresses rat c-Jun, and pCMVfos expresses c-Fos (22). The construction of pCMVmfra2, which expresses mouse FRA-2, is described in McHenry et al. (23). pCMVneo was used for selecting cells in the colony formation assay as it contains the Tn5 gene that encodes resistance to the neomycin family of antibiotics (24). pBR322 was used as the control plasmid in these assays. pCMVE1a (25) contains a genomic fragment of adenovirus early region 1 that encodes all the E1a proteins. pCMV280-390 encodes only the carboxyl-terminal region of p53 from amino acids 280 to 390 (26).
Creation of pAAmut-In order to create the reporter plasmid, pAAmut, that contains 4-base pair substitutions in the p53 MARE-like site present within pAACAT, the technique of inverse polymerase chain reaction was used (27).
The amplified linear DNA was agarose gel-purified, T 4 polynucleotide kinase-treated, and then a portion was self-ligated in 15 mM Tris-HCl, pH 7.8, 5 mM MgCl 2 , 5 mM DTT, 0.25 mM ATP, 30 mM KCl, 1 mM hexamine cobalt chloride, and 8 units of T 4 DNA ligase at 14°C for 16 h and then used to transform competent Escherichia coli DH5 ␣ cells. The substitutions within pAAmut were confirmed by sequencing.
In Vitro Transcription and Translation-pRAM-GEM is a subclone of the v-maf gene based on the pGEM-4 vector, which was used to in vitro translate (IVT) v-Maf PT (4). The structure and construction of the deletion and point mutants of the v-maf PT gene used are based on pRAM-GEM and have been described previously (4).
The plasmids pSP6jun and pSP6fos (kindly provided by Dr. Donna Cohen, JCSMR, Canberra, Australia) were used to IVT rat c-Jun and c-Fos proteins, respectively.
All plasmids were transcribed and translated in vitro using the TNT-coupled Wheat Germ Extract kit (Promega).
Oligomers-The sequence of Oligomer M, which contains the mouse p53 MARE and flanking p53 promoter sequences, is shown in Fig. 2a. Also shown in Fig. 2a is the sequence of Oligomer Mmut that contains a mutant MARE. The sequence of Oligomer 11, which contains a MARE known to bind both the v-Maf PT and c-Jun homodimers and the c-Fos/c-Jun heterodimer (10), is 5Ј-GGGAG CTCGG AATTG ATGAC TCATC ATTAC TC-3Ј.
Electrophoretic Mobility Shift Assays (EMSAs)-Binding reactions were performed in a volume of 15 l containing 6 l of IVT product, 0.1-0.5 g of poly(dI⅐dC)⅐poly(dI⅐dC), 20 mM HEPES-KOH, pH 7.9, 1 mM EDTA, 20 mM KCl, 4 mM MgCl 2 , 5 mM DTT. Reactions were allowed to proceed for 15 min at room temperature. 1 ϫ 10 4 cpm of target oligomer, end-labeled with [ 32 P]dCTP, was added to the binding reaction and incubated at room temperature for a further 15 min. Following this, 1.5 l of 10ϫ DNA loading dye was added, and the binding reaction was immediately loaded onto a pre-electrophoresed 5% polyacrylamide gel. After electrophoresis, gels were fixed in 10% acetic acid for 10 min, dried for 30 min at 80°C, and exposed to Kodak X-Omat AR film at Ϫ70°C, usually for 18 h.
Cell Culture-The mouse NIH3T3 and human HeLa cell lines were maintained at 37°C, 10% CO 2 in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum. Rat embryo fibroblasts (REFs) were prepared from 15-to 17-day-old Wistar rat embryos as described previously (28) and routinely cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. REFs were used up to passage six and then replaced with new cells.
Mammalian Cell Transfection-For all cell types, 2.5 ϫ 10 5 cells were seeded into 35-mm dishes and transfected with FuGENE 6 (Roche Molecular Biochemicals) after 18 h. For each dish transfected the total amount of DNA was kept at 4 g (except for the transfection performed in Fig. 9c, where the total amount of DNA was 4.5 g). The amounts of reporter and expression plasmids used are indicated in the figure legends, and sonicated salmon sperm DNA was used as carrier DNA to keep the total amount of DNA constant. The ratio of DNA (g) to the volume of FuGENE 6 Reagent (l) used was kept at 2:3 for each

transfection.
CAT Assay-Sixty hours after transfection cells were washed twice in ice-cold phosphate-buffered saline and harvested by scraping into 1 ml of ice-cold phosphate-buffered saline, centrifuged, then resuspended in 100 l of 0.25 M Tris-HCl, pH 7.5. Extracts of transfected cells were then prepared by three rounds of freezing and thawing followed by centrifugation for 15 min at 12,000 ϫ g and 4°C to remove cellular debris. The supernatant was then heated to 65°C for 10 min to inactivate a CAT inhibitor (29). Cell lysates were then normalized for protein content using the BCA Protein Assay Reagent Kit (Pierce), and CAT activities were determined essentially as described in Sleigh (29).
Western Immunoblotting-Extracts of transfected cells were prepared as described above (see "CAT Assay"). Western immunoblot analysis was performed according to a standard procedure (30). Briefly, protein fractionation was performed by 10% SDS-polyacrylamide gel electrophoresis using 11 g of cell extract per lane. Extracts were diluted in loading buffer containing 40% glycerol, 4% SDS, 50 mM Tris-HCl, pH 6.8, 80 M DTT, 0.08% bromphenol blue, and heated to 100°C for 5 min before loading. Gels were transferred to polyvinylidene difluoride membrane (Millipore) in transfer buffer (192 mM glycine, 25 mM Tris, 0.05% SDS, 20% methanol). The filters were incubated for 1 h at room temperature in 1% non-fat dried milk and TBS buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl). After washing in TTBS buffer (20 mM Tris-HCl, pH 7.9, 0.5 M NaCl, 0.35% Tween), the filters were incubated for 1 h in TTBS buffer containing a 1:50 dilution of the commercial monoclonal p53 antibody DO-7 (Dako). After three washes, blots were incubated with anti-mouse antibodies conjugated with alkaline phosphatase for 1 h in TTBS. After further washes, detection of bound antibody was performed with the Immun-star Chemiluminescent Protein Detection System (Bio-Rad).
Colony Formation Assay-To investigate the survival and growth of cells, 1.5 ϫ 10 5 REFs were cotransfected, with the appropriate plasmid DNA, using FuGENE 6 Reagent (as described above). pCMVneo was included as one of the transfected plasmids at an amount of 0.2 g to selectively permit survival of the transfected cells in the presence of the neomycin analogue G418 (24). Forty eight hours post-transfection REFs were maintained in medium containing 200 g/ml G418 (Life Technologies, Inc.) for 21 days. Cells were then fixed and stained with 0.1% crystal violet, 20% ethanol. The extent of cell survival was reflected by the number of G418-resistant colonies present.

RESULTS
v-Maf Binds to a Conserved DNA Sequence within the p53 Gene Promoter-To determine whether v-Maf binds to the MARE-like site within the mouse p53 promoter, EMSAs were performed where an end-labeled synthetic double-stranded oligonucleotide, Oligomer M (Fig. 2a), was mixed with IVT v-Maf PT. Oligomer M spans 43 base pairs of the mouse p53 promoter from Ϫ81 to Ϫ39, and this region encompasses the MARE-like site situated between Ϫ66 and Ϫ54. Whereas v-Maf PT is a v-Maf protein that lacks the first 18 amino acids of the amino terminus, however, it has been shown to be as active as fulllength v-Maf and c-MAF in DNA binding, transactivation and cell transformation abilities (4,10). Results (Fig. 2b) showed that IVT v-Maf PT bound to Oligomer M. Furthermore, four nucleotide substitutions within the MARE-like site of Oligomer M (Oligomer Mmut, Fig. 2a) abolished the ability of v-Maf PT to bind to this oligomer (Fig. 2b). Therefore, we conclude that v-Maf binds to the MARE-like site located from Ϫ66 to Ϫ54 in the mouse p53 promoter.
v-Maf Transactivates the Mouse p53 Promoter-To determine whether the binding of v-Maf to the p53 promoter results in transcriptional activation, the plasmid, v-mafPT/pEFB-ssHII, expressing v-Maf PT under control of the human EF promoter (31) was cotransfected into NIH3T3 cells with the reporter construct, pAACAT. pAACAT contains the region of the mouse p53 promoter from Ϫ224 to ϩ116 cloned in front of the reporter CAT gene. This 340-base pair region of the p53 promoter, contained within pAACAT, has been shown to be sufficient to drive maximal expression of the mouse p53 promoter (15,32). Results of an experiment (Fig. 3a) in which pAACAT activity was measured after titration of v-mafPT/ pEFBssHII showed that v-Maf PT transactivated the p53 promoter in a dose-dependent manner. At the ratio of 3:1 (v-mafPT/pEFBssHII:pAACAT), over six independent experiments, v-Maf PT increased p53 promoter activity between 3and 10-fold (data not shown), although typically activation was between 3-and 4-fold as shown in Fig. 3a. In contrast, expression of v-Maf PT had little effect on pCAT3M, a "promoter-less" CAT construct into which the 340-base pair p53 promoter fragment was cloned to create pAACAT. Therefore, these experiments demonstrate that v-Maf is able to activate p53 expression in at least three different cellular environments, although the level of activation varies.
To test if v-Maf requires the MARE-like site located in the mouse p53 promoter for transactivation, 4-base pair substitutions were introduced into the MARE-like site within the p53 promoter/reporter construct pAACAT. The 4-base pair substitutions are the same as those that abolished the binding of v-Maf PT to the MARE-like site within Oligomer M (Fig. 2b). This mutant construct called pAAmut was then tested for its response to v-Maf PT when cotransfected into NIH3T3 cells with v-mafPT/pEFBssHII. In this experiment v-Maf PT transactivated pAACAT about 3-fold, whereas it had no effect on pAAmut (Fig. 4). This activation of the p53 promoter is specific to v-Maf as expression of luciferase or the adenovirus 2 E1b 58-kDa protein has no effect on pAACAT or pAAmut activity (Fig. 4).
Consistent with the promoter/reporter experiments, overexpression of v-Maf results in an increase in p53 protein levels ( v-Maf PT. In comparison, no p53 protein was detectable in mock-transfected NIH3T3 cells or in cells that had been transfected with a construct expressing the v-Maf PT deletion mutant ND5. This mutant lacks the amino-terminal activation domain (its structure is shown in Fig. 6a) and as demonstrated below cannot transcriptionally activate the p53 promoter (Fig. 6c).
These data show that v-Maf transcriptionally activates the mouse p53 promoter and that this transactivation requires the MARE-like site located between Ϫ66 to Ϫ54 in the mouse p53 promoter. As a result of this transactivation, overexpression of v-Maf results in an increase in endogenous p53 protein levels.  Fig. 6a) were tested for their ability to bind to the p53 MARE and transcriptionally activate p53 expression.
These v-Maf PT mutants were translated in vitro, and EM-SAs were performed to test their ability to bind to Oligomer M. Fig. 6b shows that although v-Maf PT binds to its site within Oligomer M as demonstrated above (Fig. 2b), the v-Maf PT mutants CD3, CD4, and L2PL4P, which have previously been shown not to form homodimers (4), failed to bind to Oligomer  M. The v-Maf PT mutant CD3 lacks the last 45 amino acids of the carboxyl terminus, which includes the last three leucine residues of the leucine zipper motif, whereas CD4 lacks the last 115 amino acids of the carboxyl terminus and therefore does not contain either the leucine zipper or the basic DNA binding domain (4). The substitution mutant L2PL4P has the second and fourth leucine residues of the zipper motif each replaced with a proline (4).
Of the two mutants, ND5 and R22E, that retain the ability to homodimerize (4), only ND5 that lacks the first 239 amino acids, including the acidic activation domain, binds to Oligomer M (Fig. 6b). The mutant R22E, which contains a single amino acid substitution from arginine to glutamate within the basic (DNA binding) domain of v-Maf PT, fails to bind to the p53 MARE within Oligomer M (Fig. 6b). This substitution also prevents R22E from binding to the TRE-type MARE consensus sequence (10). Fig. 6b also demonstrates that the complexes formed are specific to the MARE as no complexes are observed with Oligomer Mmut, which contains the mutant p53 MARE.
To test whether the v-Maf PT mutants were able to transcriptionally activate the p53 promoter, the mutants under control of the EF promoter were separately cotransfected into NIH3T3 cells along with the p53 promoter/reporter construct, pAACAT (Fig. 6c). The deletion mutants CD3 and CD4, which lack all or part of the leucine zipper and cannot bind to the p53 MARE, failed to activate p53 expression. Although the deletion mutant ND5 is able to bind to the p53 MARE, it lacks the activation domain and therefore had little effect on expression from pAACAT (Fig. 6c). Western immunoblot analysis showed that ND5 is expressed at least as well as v-Maf PT in NIH3T3 cells (data not shown).
Although the deletion mutants that failed to bind to the p53 promoter failed to activate p53 expression, both substitution mutants R22E and L2PL4P increased pAACAT activity (Fig.  6c). This result was unexpected as neither of these mutants is able to bind to the p53 MARE (Fig. 6b). One possible explanation is that these proteins are able to prevent a negative regulatory factor from binding to the p53 MARE by forming nonfunctional protein complexes with this factor off the promoter, thereby resulting in the activation of p53 expression.
Cellular Homologues of v-Maf Transactivate the p53 Promoter-The chicken contains two cellular homologues of v-Maf, c-MAF type I and c-MAF type II. These proteins are generated by alternate splicing. 2 Whereas both mRNAs are 3.2 kilobases in length, they produce two different sized proteins. c-MAF type I is transcribed from three exons and has an additional 10 amino acids at the extreme carboxyl terminus compared with c-MAF type II. Both c-MAF proteins also differ to v-Maf due to a valine to methionine substitution at codon 257 (4). In addition to c-Maf, there is a large family of Maf-related proteins (reviewed in Refs. 33 and 34). One family member, MAFB, is related by sequence and general domain structure to v-Maf; however, unlike v-Maf, MAFB fails to dimerize with c-Jun, is a less efficient transactivator, and exhibits weaker transforming ability than v-Maf (31).
To test whether the ability to transcriptionally activate p53 expression is restricted to v-Maf, or it is a more general prop-2 M. Nishizawa, unpublished data. pEFBssHII, CD4/pEFBssHII, ND5/pEFBssHII, R22E/pEFBssHII, or L2PL4P/pEFBssHII. Sixty hours post-transfection cells were harvested and assayed for CAT activity. The promoter-less CAT construct pCAT3M (1 g) was also cotransfected with 3 g of either carrier DNA (mock) or v-mafPT/pEFBssHII as a control.  Fig. 7 demonstrates that both c-MAF isoforms, as well as MAFB, can transactivate the p53 promoter. Type II appears to be a more potent activator than v-Maf PT or type I, whereas MAFB is the least potent. This difference in the transactivation ability of the cellular Maf proteins is not due to differences in the level of protein expression in NIH3T3 cells, as all of the Maf proteins were detectable to similar levels by Western immunoblotting (data not shown). This finding regarding MAFB is in agreement with a previous study that showed MAFB to be a less potent activator, compared with v-Maf PT, of a promoter construct containing multiple copies of a MARE (31). Thus the ability to transcriptionally activate p53 expression appears to be a general property of the cellular Maf proteins tested. However, the effect of the small Maf proteins on the ability of the larger Maf family members to activate p53 expression remains to be tested.
v-Maf and c-Maf Cause an Induction of Cell Death-The transcriptional activation of the p53 promoter by the Maf proteins suggests that one physiological role of c-Maf is to regulate the expression of the p53 gene. If this is the case, it seems plausible that Maf might cause growth arrest or apoptosis upon overexpression (35), properties that are associated with increased levels of the tumor suppressor protein p53 (36). To test this possibility a colony formation assay described in Baker et al. (37) was carried out. Early passage REFs were transfected with various Maf expression constructs (listed in Table I) along with a plasmid, pCMVneo, expressing the gene that confers resistance to the neomycin/kanamycin family of antibiotics. After transfection, REFs were then treated with the neomycin analogue G418 for 21 days at which time the antibiotic-resistant (G418 R ) colonies were fixed and stained with crystal violet. If Maf causes cell death, then there will be few (if any) surviving colonies compared with the control (pBR322). The results of two independent transfection experiments are summarized in Table I, and a third set of results is shown in Fig. 8. These data show that expression of v-Maf PT (construct v-mafPT/pEFB-ssHII) or c-MAF type I (construct c-MAF type I/pEFBssHII) inhibited colony formation, generating only a few or no colonies, compared with control cultures transfected with pBR322 (Table I and Fig. 8). By contrast, cultures transfected with pCMVE1a, the construct that expresses the apoptosis-inducing agent adenovirus E1a (38 -40), contained no G418 R colonies nor indeed was there any evidence of surviving cells (Table I and Fig. 8). The v-Maf PT substitution mutant L2PL4P (construct L2PL4P/pEFBssHII), shown to activate p53 expression (Fig. 6c), also causes a reduction in colony number compared with the pBR322 control (Table I). However, expression of the deletion mutants CD4 (construct CD4/pEFBssHII) and ND5 (construct ND5/pEFBssHII), which cannot activate p53 expression (Fig. 6c), do not cause a reduction in colony number compared with the pBR322 control (Table I and Fig. 8). These data imply that the ability of Maf to reduce the colony number is due to its ability to transcriptionally activate p53 expression.
Cotransfection of the construct pCMV280-290, which expresses the carboxyl terminus of the p53 protein (p53CT; amino acids 280 -390), with the v-Maf PT expression construct (v-mafPT/pEFBssHII) or the L2PL4P expression construct (L2PL4P/pEFBssHII) rescued cells from death, as the colony number was similar to the pBR322 control (Table I and Fig. 8). The carboxyl-terminal region of p53 is able to inhibit p53-dependent apoptosis in a dominant negative manner by interacting with wild-type p53 and preventing its ability to induce apoptosis (41). Consistent with this, v-Maf was unable to induce cell death in the human IIIcF/c cell line (data not shown), which is null for p53 (41). Therefore, overexpression of a transcriptionally competent Maf leads to a p53-dependent cell death.
Members of the AP1 Family Also Regulate p53 Expression-Maf has been shown to be able to dimerize with other members of the extensive b-Zip family of transcription factors (10,11,33,42); therefore, it is likely that these b-Zip proteins will influence the ability of Maf to regulate p53 expression. The PF1 site (the variant AP1 site) located within the p53 MARE is bound by several DNA binding activities, some of which have been shown to contain the AP1 components, c-Fos and c-Jun (18,19). Studies using antisense oligonucleotides directed against c-jun or c-fos show these b-Zip proteins are required for full p53 expression (19), and it appears this regulation occurs through the PF1 site (17)(18)(19). Since Maf can form heterodimeric complexes with c-Fos and c-Jun (5, 10), which affect the transactivation potential of Maf by forming heterodimers with distinct DNA binding specificities (11), the following experiments attempt to understand the contribution these AP1 family members make in the regulation of p53 expression by Maf.
To test whether v-Maf PT and c-Jun or c-Fos dimers have the FIG. 7. Cellular Maf proteins activate p53 expression. NIH3T3 cells were cotransfected with the mouse p53 promoter/reporter construct pAACAT (1 g) and either 3 g of carrier DNA (mock) or the Maf expression constructs v-mafPT/pEFBssHII, c-MAF type I/pEFBssHII, c-MAF type II/pEFBssHII, or MAFB/pEFBssHII. Sixty hours posttransfection cells were harvested and assayed for CAT activity. The promoter-less CAT construct pCAT3M (1 g) was also cotransfected with 3 g of either carrier DNA (mock) or v-mafPT/pEFBssHII as a control. ability to bind to the p53 MARE, these proteins were cotranslated in vitro and subjected to EMSA. Initially, to confirm dimer formation between these proteins, an oligomer (designated Oligomer 11; see "Experimental Procedures") containing a MARE known to bind both the v-Maf PT and c-Jun homodimers and the c-Fos/c-Jun heterodimer was used as a probe in the EMSA (Fig. 9a). The two v-Maf proteins PT and ND5 are able to homodimerize and bind to the MARE in Oligomer 11 (10). When these two v-Maf proteins of different lengths were cotranslated, an additional retarded complex of intermediate mobility was observed, confirming homodimer formation of v-Maf (Fig. 9a). When c-Jun was cotranslated with ND5, the intermediate complex represents the ND5/c-Jun heterodimer. Fig. 9a also demonstrates the ability of the c-Jun homodimer to bind to the MARE contained within Oligomer 11. Although c-Fos does not homodimerize (43, data not shown), the cotranslation of ND5 and c-Fos also resulted in a complex with the size expected for the ND5/c-Fos heterodimer. Cotranslation of c-Fos and c-Jun yields a complex that represents the AP1 activity (Fig. 9a). These results are in agreement with similar experiments performed by Kataoka et al. (10). The EMSA in Fig. 9a was repeated with the probe Oligomer M (which contains the p53 MARE and flanking promoter sequences, Fig. 2a), instead of Oligomer 11, to test the DNA binding specificity of the dimers for the p53 MARE. Fig. 9b demonstrates that the v-Maf PT (and ND5) homodimers are able to bind to the p53 MARE. Interestingly, in contrast to Oligomer 11 (Fig. 9a), the ND5/c-Jun and ND5/c-Fos heterodimers and the c-Jun homodimer were unable to bind to the p53 MARE (Fig. 9b). This is demonstrated by the lack of the slower migrating complexes that correspond to these dimers in the appropriate lanes in Fig. 9b compared with Fig. 9a. Cotranslation of c-Fos and c-Jun yielded a weak complex with Oligomer M (Fig. 9b) compared with Oligomer 11 (Fig. 9a), which corresponds to the AP1 activity. The differences in the DNA binding specificities of these dimers for the two oligomers is likely to be due to the 4-base pair differences between the MARE in Oligomer 11 and the p53 MARE in Oligomer M.
Although the only dimers tested above that were capable of binding to the p53 MARE are the v-Maf homodimer and the AP1 (c-Fos/c-Jun) heterodimer, the ability of these AP1 members to regulate p53 promoter activity when expressed in the cell was investigated. NIH3T3 cells were cotransfected with the p53 promoter/reporter construct pAACAT and constructs expressing either c-Jun (pCMVjun), c-Fos (pCMVfos), FRA-2 (pC-MVmfra2), or v-Maf PT (v-mafPT/pEFBssHII). As shown above, expression of v-Maf PT activates p53 promoter activity 3-fold, and the three AP1 members also influenced pAACAT activity when expressed individually (Fig. 9c). Expression of c-Fos transactivates pAACAT 7-fold. In contrast to the repression by Schreiber et al. (19), expression of c-Jun is able to transactivate p53 expression 2-fold. This is repeatable, and the level of repression/activation appears to be dependent on the cell type and the amount of pCMVjun transfected. 3 The Fosrelated antigen, Fra-2, that can bind to the PF1 site (19) was also tested for its ability to regulate p53 promoter activity. FRA-2 has been shown to act as negative regulator of AP1 activity (44) and represses pAACAT activity 2-fold (Fig. 9c).
To determine the ability of these AP1 proteins to modulate the activity of Maf with regard to p53 expression, these proteins were expressed in combination with v-Maf PT and tested for their ability to regulate pAACAT activity (Fig. 9c). When v-Maf PT and c-Jun are coexpressed, activity from pAACAT is increased approximately 7-fold, even though the v-Maf/c-Jun heterodimer is unable to bind to the MARE present in the p53 promoter (Fig. 9b). This, along with the finding that levels of p53 mRNA decrease after treatment of cells with c-jun antisense oligonucleotides (18), supports the idea that regulation of p53 expression by c-Jun may be more complex than just its role as a repressor of p53 promoter activity (19).
Compared with c-Fos alone, coexpression of v-Maf PT and c-Fos transactivates the p53 promoter 4-fold, suggesting that expression of v-Maf PT antagonizes the ability of c-Fos to activate p53 expression. The formation of v-Maf/c-Fos dimers that are unable to bind to the p53 MARE (Fig. 9b) may prevent c-Fos from interacting with another dimeric partner to activate p53 expression.
When v-Maf PT and FRA-2 are coexpressed there is no effect on the level of basal pAACAT activity. Therefore, FRA-2 suppresses the 3-fold transactivation seen when v-Maf PT is expressed alone, demonstrating that the activity of v-Maf, like c-Jun (44), is repressed by Fra-2.
Transcriptional regulation of p53 by these b-Zip proteins requires the MARE-like site within the promoter. When the p53 promoter/reporter construct pAAmut, which has a mutated MARE that abolishes the binding of the v-Maf PT homodimer (Fig. 2), replaces pAACAT in the transfection assay, expression of these b-Zip proteins has no effect on p53 promoter activity (Fig. 9c). The exception is c-Fos, which is still able to activate expression from pAAmut 3-fold; however, this is less than the 7-fold activation of the wild-type p53 promoter (pAACAT) by c-Fos. This is likely to be due to the heterodimer involved in the activation by c-Fos having a higher DNA binding specificity for the mutant p53 MARE.
From these results it appears that p53 promoter activity is controlled through the MARE by multiple b-Zip proteins, including the Maf and AP1 families of transcription factors. Whether p53 expression is positively or negatively regulated is dependent on the combination of these b-Zip proteins present. DISCUSSION The p53 protein appears to be a key regulator of the fate of a cell (45). It is well established that an increase in p53 protein levels, in response to cellular stress, can lead to either cell cycle arrest or apoptosis (36). This accumulation of p53 in response to stress, such as DNA damage, is achieved through posttranscriptional mechanisms (46), as well as through transcrip-3 T. Hale, unpublished data. tional activation of the p53 gene (47,48). Conversely, repression of p53 transcription was shown to promote cellular proliferation (19). In addition, expression rates of mutant and wild-type p53 in tumors can be altered by deregulated p53 transcription (49,50). Therefore, identifying the factors that tightly control p53 expression is of importance in understanding the pathways that determine cell proliferation and cell death.
The mouse p53 promoter contains a site located between Ϫ66 and Ϫ54 that differs by only 2 base pairs to the TRE-type MARE consensus sequence (Fig. 1). We show here that the oncoprotein v-Maf can bind as a homodimer to this MARE-like sequence (Fig. 2b) and that this site is essential for v-Maf to activate p53 expression (Fig. 4). Furthermore, the ability to regulate p53 expression is not confined to v-Maf. The two cellular homologues of v-Maf in chicken, c-MAF type I and type II as well as MAFB, another large Maf family member (31), can activate p53 expression, although to varying degrees (Fig. 7).
Functional domain analysis of the v-Maf protein demonstrated that the leucine zipper motif, and hence the ability to dimerize, is a requirement for binding to the p53 MARE. This DNA binding (Fig. 6) is required for activation of p53 expression, since the deletion mutants of v-Maf PT, CD3, and CD4 that lack either part or all of the leucine zipper motif and therefore cannot form homodimers (4) failed to bind to the p53 MARE and activate p53 expression (Fig. 6). The v-Maf PT deletion mutant ND5 that lacks the acidic activation domain, although able to bind to the p53 MARE, was unable to activate p53 expression (Fig. 6). Therefore, in addition to DNA binding, the acidic activation domain of v-Maf is required to activate p53 expression. These results are in agreement with previous activation studies that used a different promoter construct (10,11). However, the v-Maf PT substitution mutants R22E and L2PL4P were both able to activate p53 expression. This is in direct contrast to a previous study that showed these mutants were unable to activate a promoter construct containing three copies of the consensus MARE (11). Whereas R22E and L2PL4P cannot bind to the p53 MARE (Fig. 6b), one possible explanation for the ability of these mutants to activate p53 expression (Fig. 6c) is that they prevent an endogenous repressor from binding to the MARE, thus allowing activation of p53 expression to occur. This raises the possibility that other b-Zip proteins are involved in regulating p53 expression through this MARE.
Maf is a member of a large group of b-Zip proteins that form a complex regulatory network, as members can form homo-and heterodimers with one another to regulate transcription in a positive or negative manner through variant MAREs (33). The MARE-like sequence in the mouse p53 promoter contains the PF1 site, which is a variant AP1 site (17). We demonstrate that three AP1 components, c-Fos, c-Jun, and Fra-2, can also regu-  Fig. 6a. b, the ability of Maf heterodimers to bind to the p53 MARE. V-Maf PT, ND5, c-Jun, or c-Fos (combinations indicated at the top) were translated in vitro and then assayed for the ability to bind to Oligomer M, which contains the p53 MARE (Fig. 2a) in an EMSA. Lane P refers to the reaction containing only Oligomer M; lane L represents the reaction containing only wheat germ extract. The retarded bands corresponding to the various dimers are indicated. c, AP1 family members regulate p53 promoter activity. The mouse p53 promoter/reporter construct pAACAT (0.5 g) or the construct pAAmut (which contains the p53 promoter with a mutated MARE) was cotransfected into NIH3T3 cells, with either 4 g of carrier DNA (mock) or 2 g of either v-mafPT/pEFBssHII (expresses v-Maf PT), pCMVjun (expresses c-Jun), pCMVfos (expresses c-Fos), or pCMVmfra2 (expresses FRA-2) alone or in the combinations listed above. Sixty hours post-transfection cells were harvested and assayed for CAT activity. late p53 expression through the p53 MARE (Fig. 9c). In addition, when these AP1 proteins are coexpressed with v-Maf, they have the ability to modulate the transactivation potential of one another with regard to the p53 MARE (Fig. 9c). These results support the idea that the MARE-like site in the p53 promoter is utilized by several members of the b-Zip family that are part of an extensive regulatory network, their interactions in response to cellular signals fine-tune p53 expression in different cellular environments. It is likely that several b-Zip proteins capable of binding to the p53 MARE are yet to be identified, given the number of different binding activities that have been identified that can bind to this site (17)(18)(19) and the demonstration that v-Maf/c-Jun and v-Maf/c-Fos heterodimers as well as c-Jun homodimers are unable to bind to the p53 MARE (Fig. 9b).
The AP1 complex has been implicated in both the positive and negative modulation of apoptosis (51,52). Expression of c-Fos is required for apoptosis in several different cell types (53). Thus, the ability of c-Fos and other AP1 members to modulate apoptotic pathways may in part be due to the ability of these complexes to transcriptionally regulate p53 expression in a positive or negative manner through the MARE present in the p53 promoter (Fig. 9). Results here suggest that Maf is also involved in regulating cell death ( Table I). Overexpression of v-Maf was shown to cause a p53-dependent cell death, as blocking of p53 function by the carboxyl-terminal oligomerization domain of p53 leads to recovery of colony formation (Fig. 8). Presumably this is due to Maf transcriptionally activating the p53 promoter, since the ability to cause cell death is restricted to the Maf proteins shown to activate p53 expression (Table I and Fig. 6). Maf family members have recently been shown to be important transcriptional regulators of cellular differentiation and morphogenesis (33). Given the important role transcriptional control of the p53 gene plays in differentiation and development (50,54), it is likely then that Maf is involved in regulating p53 function during these important developmental processes.
These findings support the recent reports (19,47,48) that demonstrate regulation of p53 transcription, like stabilization of the p53 protein (46), is important in controlling p53 function and therefore the fate of a cell.