Interplay between TAp73 Protein and Selected Activator Protein-1 (AP-1) Family Members Promotes AP-1 Target Gene Activation and Cellular Growth*

Background: TAp73, which is overexpressed in cancers, activates AP-1 target genes. Results: TAp73 binds to c-Jun on the chromatin around TRE sites on AP-1 target promoters, leading to recruitment of other AP-1 family members. Conclusion: Interaction of TAp73 with selected AP-1 members enhances target gene activation and cellular growth. Significance: c-Jun-dependent cooperativity between TAp73 and selected AP-1 members contributes to cellular growth.

Members of the p53 tumor suppressor family including p63 and p73 are crucial for the maintenance of balance between cell survival and death in physiological and pathological conditions (1)(2)(3). They are thus activated by both stress signals and growth factors, suggesting pleiotropic roles in a multitude of cellular processes (4 -6). Being transcription factors, transactivation of relevant target genes has been shown to be essential for their functions (7,8). Consistently, in a high percentage of cancers, p53 is mutated in the DNA-binding domain, thereby impairing its transactivation function (9). However, p73 is rarely mutated but is in fact overexpressed in cancers (10,11), highlighting potentially disparate functions among these similar proteins. Interestingly, p73-null mice do not develop tumors at an accelerated rate as do the p53-null mice (12,13) and instead show several developmental defects of the central nervous system, thus indicating that p73 has evolved to perform functions beyond tumor suppression (14). p73 exists in two major forms. The first major form is the full-length TAp73, which is structurally and functionally homologous to p53 but appears also to possess several properties distinct from p53 that can be associated with different promoter specificity of the two related proteins (7,8,15). The presence of the second internal promoter leads to the generation of the second major form, the ⌬Np73 (16). The latter lacks the amino-terminal transactivation domain and hence acts as a dominant negative inhibitor of both TAp73 and p53, giving rise to the notion that TAp73 is a tumor suppressor like p53, whereas ⌬Np73 is an oncogene (17). Recent generation of isoform-specific knock-out mice has been instrumental in clarifying some of the roles of TAp73 and ⌬Np73 (13,18). Absence of ⌬Np73 was shown to lead to neurodegeneration and to sensitize cells to DNA damage-inducing agents, suggesting that it might play a role in chemoresistance (18). Conversely, absence of TAp73 leads to infertility and hippocampal dysgenesis (13). Loss of TAp73 also leads to genomic instability, although this was suggested to occur in a cell type-specific manner, indicating that cellular context can dictate the biological outcome of p73 activity.
Although the overexpression of ⌬Np73 in several cancers is expected, significant data also exist demonstrating the overexpression of TAp73 in various cancers (19 -21). Importantly, several lines of evidence indicate that TAp73 expression can be associated with promoting cellular survival and antiapoptosis under certain conditions. For example, we have previously shown that TAp73 can induce the expression of the antiapoptotic caspase 2S, leading to resistance to cell death in neuroblastoma cells (22). Also, expression of TAp73 in conjunction with p53 can suppress p53-mediated telomerase activation via activation of HDM2 (23). TAp73 was also found to bind to p53-responsive elements in promoters of cell cycle progression genes, causing aberrant activation of cell proliferation (24), and recently, TAp73 was shown to regulate the pentose phosphate pathway to promote cellular proliferation (25). Finally, several proto-oncogenes have also been shown to induce and activate TAp73 isoforms (5). One such proto-oncogene is the activator protein (AP) 2 -1 family member c-Jun, which has been shown to stabilize TAp73 (26).
The mammalian AP-1 proteins belong to the Jun (c-Jun, JunB, and JunD), Fos (c-Fos, FosB, Fra1, and Fra2), and closely related activating factor (ATF-2, ATF-3, and B-ATF) families. These proteins can homo-or heterodimerize and can regulate embryonic development, cell survival, or death by means of inducing and/or repressing tumor suppressor genes or cell cycle regulatory genes (27,28). For instance, c-Jun is required for proliferation of fibroblasts and other cultured cells (29) and has been demonstrated to also inhibit p53-induced growth arrest upon ultraviolet (UV) irradiation (30). In contrast, dominant negative c-Jun mutant could protect sympathetic neurons against NGF withdrawal-induced apoptosis (31). Moreover, accumulating evidence points to differing combinations of AP-1 dimers having different biological functions as has been suggested in the antagonistic regulation of peroxisome proliferator-activated receptor ␥, thereby affecting obesity and hepatic functions (32). Thus, the cell type-and context-dependent signals could alter AP-1 dimer composition and thus dictate the eventual cellular fate.
Previously, we have shown that co-expression of TAp73 with c-Jun stabilizes TAp73 and leads to increased cellular survival via the up-regulation of cyclin D1 at both the RNA and protein levels (26,33). This up-regulation is dependent on c-Jun and is required for TAp73-induced cell proliferation. Furthermore, induction of another AP-1 target gene (collagenase/MMP-1) upon treatment with the tumor-promoting phorbol ester (12-O-tetradecanoylphorbol-13-acetate) was impaired in p73-null mouse embryonic fibroblasts (MEFs). These data suggested that TAp73 plays a vital role in activation of endogenous AP-1 target genes, thereby contributing to cellular survival. To further understand the details of how TAp73 acts as a growth promoter, we analyzed the mechanistic basis of AP-1 target gene activation by TAp73 and the composition of AP-1 family members involved in this process.
For modified chromatin immunoprecipitation (ChIP-IP), transfected cells were incubated in serum-free medium overnight for serum starvation, and 20% serum medium was added for 4 h before harvesting. Transfected cells were also incubated in normal medium overnight and exposed to UV irradiation (60 J) for 4 h before harvesting.
siRNA and Transfections-siRNAs for human c-Jun and Fra1 were purchased from Santa Cruz Biotechnology (sc-29223 and sc-35405). siRNAs for control (scrambled), human c-Fos, and human Fra2 (TTCTCCGAACGTGTCACGT, AGGAGAA-TCCGAAGGGAAA, and GCGCTGTAGTGGTGAAACA, respectively) were synthesized. siRNAs were transfected using Transmessenger (Qiagen) following the manufacturer's protocol. 24 h after siRNA transfection, the indicated plasmids were transfected using Lipofectamine Plus reagent as mentioned above.
Luciferase Assays-H1299 cells were seeded in 6-well plates and transiently transfected with appropriate plasmids (0.3-0.5 g) and ␤-galactosidase gene (50 ng) for normalization. Cells were washed and lysed in luciferase lysis buffer 24 h post-transfection, and luciferase assays were performed as described (23,26).
The ChIP protocol was modified as follows for ChIP-IP. 24 h after induction or transfection, cells (10-cm plate) were fixed with 1% formaldehyde for 10 min at room temperature. Cells were then washed twice with ice-cold PBS and lysed for 20 min using 4 ml of 0.25% Triton X-100 lysis buffer (0.25% Triton X-100, 20 mM EDTA, 10 mM Tris-HCl, pH 8.1) with protease inhibitors at 4°C. Samples were centrifuged at 750 ϫ g at 4°C for 5 min. The pellets were lysed in 400 l of 1% SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8.1) with protease inhibitors, and the extracts were sonicated on ice with 10-s pulses for 4 min each. Samples were centrifuged at 13,000 rpm at 4°C for 30 min, and supernatant was collected. Samples were diluted to 2 ml in ChIP dilution buffer (1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 150 mM NaCl) with protease inhibitors. 40 l of the diluted sample was kept aside as the input fraction before preclearing with mouse IgG-agarose beads (10 l/sample) for 4 h at 4°C. The supernatant was incubated with anti-FLAG-agarose beads (15 l/sample) (M2 affin-ity gel, Sigma) for 2 h at 4°C. Immune complexes were washed one time with LS buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 150 mM NaCl), one time with HS buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 500 mM NaCl), one time with 0.25 M LiCl buffer (250 mM LiCl, 1% deoxycholate, 1 mM EDTA, 1% Nonidet P-40, 10 mM Tris-HCl, pH 8), and two times with Tris-EDTA buffer. DNA⅐protein complex was eluted two times with 40 l of elution buffer (1% SDS, 0.1 M NaHCO 3 ). Cross-links were reversed by addition of 200 mM NaCl at 65°C for 4 h followed by DNase I treatment (7.5 Kunitz units/sample; Qiagen) at 37°C for 2 h. Protein concentration was measured, and samples were lysed in SDS sample buffer.
Colony Formation Assays-For colony formation assays, H1299 cells were transfected with the indicated plasmids and grown in appropriate selection medium (G418) for between 10 and 14 days. Surviving colonies were stained with crystal violet solution (Merck) as described previously (33). Quantifications were done using MetaMorph Offline (Version 7.8.0; Molecular Devices, LLC), which quantifies live cells by percent colony area.

Some Fos Family Members
Cooperate with TAp73 to Potentiate AP-1 Target Gene Activation-We have shown previously that TAp73 is capable of transactivating AP-1 target genes such as cyclin D1 and collagenase in a manner dependent on the expression of c-Jun (33). Moreover, TAp73 was able to synergize with c-Jun to further potentiate the activation of AP-1 target genes ( Fig. 1) unlike p53 (33). To gain further mechanistic insights into TAp73-mediated AP-1 activation, we first evaluated the role of the functional domains of c-Jun that are required for cooperation with TAp73. c-Jun deletion mutants that lack the leucine zipper domain that is required for homoand heterodimerization (c-Jun DM ), the delta domain that is required for docking of Jun N-terminal kinases (c-Jun D ), or the N-terminal transactivation domain (c-Jun TAM67 ) were used. The schematic shows the regions deleted, and the Western blot shows expression of all the c-Jun mutant constructs (all in 3xFLAG CMV vector). Unlike wild-type c-Jun, c-Jun DM or c-Jun TAM67 was unable to cooperate with TAp73 to activate both the human collagenase and the cyclin D1 promoter-luciferase constructs (Fig. 1). In contrast, c-Jun D mutant alone was able to activate both promoters even though phosphorylation at serine 63 was not detected, suggesting that JNK-mediated phosphorylation is not important. Additionally, TAp73 was able to cooperate with the c-Jun D mutant, suggesting that activation by JNK was not a prerequisite for the cooperativity, consistent with our previous findings that TAp73 is capable of cooperating with c-Jun in a JNK-independent manner (33). Furthermore, co-expression with TAp73 led to the phosphorylation of the c-Jun D mutant, confirming JNK-independent phosphorylation of c-Jun in the presence of TAp73 as suggested earlier (33).
Although the inability of the c-Jun TAM67 to cooperate with TAp73 was expected due to the lack of the transactivation domain, the defect due to the c-Jun DM mutant suggested the requirement of c-Jun dimerization for cooperation with TAp73. Because the dimerization domain is an important component for c-Jun to homo-or heterodimerize with its family members, we assessed the role of other AP-1 family members to cooperate with TAp73. We found that like c-Jun, expression of c-Fos could synergize with TAp73 to activate the collagenase promoter-luciferase construct to significant levels ( Fig. 2A). Other than c-Fos, Fra1, FosB, and JunB could also cooperate with TAp73 albeit to lesser but significant levels ( Fig. 2A, upper panel). Fra2 and JunD were unable to cooperate at all with TAp73. Similar results were obtained with cyclin D1 promoterluciferase construct ( Fig. 2A, lower panel). This suggested selectivity among AP-1 family members in their ability to cooperate with TAp73 to activate specific target genes. To determine whether specific AP-1 dimer pairs are therefore better able to cooperate with TAp73, we evaluated the effects of some specific single chain AP-1 dimers, which have been used to demonstrate AP-1 member-pairing specificity (37,40). Surprisingly, although c-Jun homodimers did not cooperate as well with TAp73, the c-Jun/c-Fos and c-Jun/Fra1 dimers showed a high level of synergy in activating the collagenase promoter-luciferase construct (Fig. 2B). Like c-Jun homodimers, c-Jun/Fra2 and c-Jun/ATF2 heterodimers did not synergize well with TAp73. Because JunB and FosB could also cooperate with TAp73 ( Fig.  2A), we further assessed whether c-Jun/FosB, JunB/JunB, or JunB/c-Jun dimers could synergize with TAp73 (Fig. 2C). c-Jun/FosB dimers were capable of activating the collagenase promoter-luciferase construct and cooperated marginally with TAp73. However, similar to c-Jun homodimers, JunB homodimers could not cooperate with TAp73, whereas JunB/ c-Jun cooperated with TAp73 although less efficiently as compared with c-Jun/c-Fos dimers probably due to promoter specificity of the AP-1 members as has been described (32,40). These data, therefore, suggest that heterodimerization of c-Jun with certain AP-1 family members, specifically the Fos family members (c-Fos, Fra1, and FosB but not Fra2), is important for cooperation with TAp73 to mediate AP-1 target gene transactivation.
c-Fos and Fra1 Are Required for TAp73-mediated AP-1 Target Gene Activation in a c-Jun-dependent Manner-We next assessed whether the strong cooperation of the other AP-1 members with TAp73 is dependent on c-Jun. Therefore, we overexpressed c-Fos, Fra1, and other AP-1 members and assessed the synergistic effects with TAp73 on the collagenase promoter-luciferase construct in p53 Ϫ/Ϫ c-Jun Ϫ/Ϫ and p53 Ϫ/Ϫ MEFs (Fig. 3A). TAp73 strongly synergized with c-Jun, c-Fos, and Fra1 in p53 Ϫ/Ϫ MEFs, suggesting that this cooperation can be broadly recapitulated independently of the cell type used, although there were subtle differences as in the case with FosB, which did not show as robust a cooperation, and JunB, which did not synergize at all. Nonetheless, Fra2 and JunD did not cooperate with TAp73 as noted earlier. Interestingly, the synergistic effect between TAp73 and c-Fos/Fra1 was decreased significantly in p53 Ϫ/Ϫ c-Jun Ϫ/Ϫ MEFs, indicating that the cooperation of TAp73 with c-Fos and Fra1 was dependent on c-Jun. Even though FosB did not synergize robustly with TAp73 in the murine cell line, there was a decrease in cooperation in the p53 Ϫ/Ϫ c-Jun Ϫ/Ϫ MEFs, again indicating dependence on c-Jun. However, the residual activation of the collagenase promoter-luciferase construct by c-Fos and Fra1 in p53 Ϫ/Ϫ c-Jun Ϫ/Ϫ MEFs indicates potential compensation by other Jun members. Similar results were obtained with cyclin D1 promoter-luciferase construct (data not shown). Because c-Fos and

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
Fra1 cooperated with TAp73 in a c-Jun-dependent manner consistently in both human and murine cell lines, we further investigated whether abrogation of c-Fos or Fra1 expression would affect TAp73-mediated AP-1 target gene activation. siRNA-mediated knockdown of c-Fos or Fra1 expression indeed significantly reduced TAp73-mediated activation of collagenase and cyclin D1 promoter-luciferase activities, whereas knockdown of Fra2 expression did not have any effect (Fig. 3B). Together, the data indicate that TAp73 cooperates with specific AP-1 family members like c-Fos and Fra1 in a c-Jun-dependent manner to mediate AP-1 target gene activation.
TRE/AP-1 Site Is Required for the Efficient Transactivation of AP-1 Promoters by TAp73-TAp73 requires AP-1 proteins such as c-Jun, c-Fos, and Fra1 and their dimerization for trans-activation of AP-1 targets, suggesting that this phenomenon occurs via the AP-1 binding site on the promoters. Among the AP-1 gene promoters used, human cyclin D1 promoter has one AP-1 site, whereas human collagenase promoter has two AP-1 sites that are required for both constitutive and inducible expression (41)(42)(43). We found that TAp73 could only partially activate the cyclin D1 promoter lacking the AP-1 site (⌬AP-1), indicating that the AP-1 site is important for this process (Fig.  4A). However, deletion of the CRE (⌬CRE) also partially inhibited TAp73-mediated activation, indicating that TAp73 can act via other elements or signaling pathways. Consistently, deletion of both the AP-1 site and the CRE (⌬AP-1-⌬CRE) completely abrogated TAp73-mediated activation of the cyclin D1 promoter. To further confirm the requirement of AP-1 site, we analyzed the two AP-1 sites in the human collagenase pro- moter. To determine whether either of the AP-1 sites are necessary for TAp73 activity, either the promoter-proximal AP-1 site (denoted as 2) or the promoter-distal AP-1 site (denoted as 3) were mutated (Fig. 4B). Mutation of the promoter-proximal AP-1 site completely abrogated activation of the promoter by both c-Jun and p73. However, mutation of the promoter-distal AP-1 site only partially inhibited activation by c-Jun and TAp73. This suggested that the promoter-proximal AP-1 site in the collagenase promoter is required for activation by both c-Jun and TAp73 and that not all AP-1 sites are utilized in the same manner by these transcription factors. Mutation of both AP-1 sites completely abolished both c-Jun-and TAp73-mediated activation. As expected, the promoter-proximal AP-1 site in the human collagenase promoter is also important in cooperation with c-Jun and other AP-1 members as mutation of this site almost completely inhibited synergistic activation by TAp73 and the AP-1 proteins (c-Jun, c-Fos, and Fra1) (Fig. 4C).
Because the promoter-proximal AP-1 site was more predominantly utilized by TAp73, we tested the ⌬Col promoter that mainly harbors only the promoter-proximal AP-1 site besides the NFB and STAT3 transcription factor binding sites. Activation of the minimal promoter was at least 10-fold higher than the full-length collagenase promoter, confirming that TAp73 can strongly activate the collagenase promoter through this AP-1 site (Fig. 4D). Furthermore, strong activation of the minimal promoter by TAp73 suggests that there may be certain repressive elements in the full-length promoter. Mutation of the AP-1 site on the minimal promoter completely abrogated activation by c-Jun and significantly inhibited activation by TAp73, suggesting that although the AP-1 family members are crucial other factors may play a role in TAp73-mediated activation (Fig. 4E). This correlates well with the results obtained from the cyclin D1 promoter mutants (Fig. 4A). To evaluate this possibility, we tested whether the other response elements in the minimal promoter (i.e. NFB/p65 and STAT3) could be activated by the appropriate transcription factors and whether they could synergize with TAp73. However, neither STAT3 nor p65 were found to activate the full-length or minimal collagenase promoter (Fig. 4F), and neither synergized with TAp73 (data not shown). These data together demonstrate the importance of the AP-1 binding sites on the target promoters for transactivation by TAp73 with AP-1 family members.
TAp73 Binds at or Near the AP-1 Site to Activate Target Genes-TAp73 has been shown to bind to the promoter of its canonical target genes such as mdm2 to transactivate their expression. However, our previous gel shift studies indicated that TAp73 was unable to bind to the exact short canonical TRE sites of AP-1 target genes (33). Nonetheless, given that TAp73 was able to cooperate with AP-1 members in a TRE-dependent manner, we re-evaluated whether the DNA binding capacity of

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
TAp73 is required for this process by using a DNA-binding domain-defective mutant, TAp73 R292H (44). As expected, TAp73 R292H was defective in the transactivation of the mdm2 promoter (Fig. 5A). However and surprisingly, this mutant was also defective in the activation of the collagenase promoter (Fig.  5A), thus indicating that DNA binding is a prerequisite for AP-1 target gene activation. Therefore, we explored whether TAp73 can directly bind to the AP-1 target gene promoters at or near the canonical AP-1 site. Chromatin immunoprecipitation assays in cells transfected with TAp73 showed that TAp73 binds at or near the AP-1 sites in cyclin D1 and cyclin A1 regulatory regions (Fig. 5B) but not to an unrelated site as assayed with an internal region in the cyclin D1 gene (Fig. 5B). However, side-by-side comparison revealed that the binding of TAp73 to these sites on AP-1 target genes is much weaker compared with its binding efficiency at its canonical binding site on the mdm2

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
promoter (Fig. 5C). Nonetheless, these data highlight the ability of TAp73 to bind to DNA for the transactivation of AP-1 target genes.
TAp73 Interacts with c-Jun at the Chromatin-To better understand the synergy between TAp73 and c-Jun, we examined the interaction between them given that TAp73 is able to bind to AP-1 target gene regulatory regions. Cells were transfected with FLAG-tagged c-Jun and various forms of untagged TAp73 (i.e. transactivation domain-deleted ⌬Np73 and TAp73 R292H ). As reported previously (26), we were unable to observe binding between the various p73 forms and c-Jun in co-immunoprecipitation experiments by immunoprecipitating c-Jun (data not shown) or in reverse co-immunoprecipitation experiments where FLAG-tagged p73 forms were immunoprecipitated (Fig. 6A), indicating that TAp73 and c-Jun do not interact under normal conditions in the soluble cellular fractions. Nonetheless, because both c-Jun and TAp73 bind to or near the AP-1 site on the target promoters, we hypothesized that c-Jun and TAp73 may interact at the level of chromatin. To test this, we isolated chromatin-bound FLAG-tagged c-Jun from transfected cells using a modified chromatin immunoprecipitation protocol and found that c-Jun can indeed interact with TAp73, ⌬Np73, and even the DNA binding mutant TAp73 R292H (Fig. 6B). Reverse immunoprecipitation experiments using chromatin-bound FLAG-tagged p73 forms confirmed that this interaction indeed occurred at the level of chro- Both TAp73 and c-Jun are activated and up-regulated during both growth factor stimulation and genotoxic stress (5,6). Therefore, we assessed the status of interaction between TAp73 and c-Jun during stress conditions such as UV irradiation as well as during prosurvival condition-associated growth factor signaling. Chromatin-bound FLAG-tagged c-Jun was isolated from transfected cells, and co-immunoprecipitation revealed the association with TAp73 both under serum-starved or serum-stimulated conditions (Fig. 6D) with the interaction being enhanced during serum stimulation, suggesting that prosurvival stimulation strengthens the association between TAp73 and c-Jun at the chromatin level. In contrast, their interaction at the chromatin level was significantly reduced upon UV treatment, implying that proapoptotic stress signals can lead to dissociation of TAp73 and c-Jun (Fig. 6E). Consistently, activation of the collagenase or cyclin D1 promoter-luciferase construct by TAp73 and c-Jun was reduced upon UV irradiation (Fig. 6F), although activation of the mdm2 promoter-lucif-erase construct was not affected (data not shown), suggesting that disruption of the interaction between TAp73 and c-Jun during the stress response leads to compromised activation of AP-1 target genes.
Stresses like UV irradiation and stimulation with growth factors are known to induce phosphorylation of c-Jun (45,46). Although JNK-dependent phosphorylation of c-Jun was not required for activation of the AP-1 target promoters as well as for cooperation with TAp73, TAp73 could induce phosphorylation of the c-Jun D mutant (Fig. 1). We therefore wanted to assess whether c-Jun phosphorylation is required for binding to TAp73. To this end, we utilized the FLAG-tagged c-Jun AA mutant that cannot be phosphorylated at serines 63 and 73. Isolation of chromatin-bound FLAG-tagged c-Jun AA mutant or wild-type c-Jun and co-immunoprecipitation showed that the c-Jun AA mutant was able to bind to TAp73 at the chromatin level (Fig. 6G), highlighting that phosphorylation of c-Jun was not necessary for binding to TAp73.
To delineate the region of p73 required for binding to c-Jun, we utilized several deletion mutants of p73 and assessed their FIGURE 6. p73 interacts with c-Jun on the chromatin. A, H1299 cells were transfected with FLAG-tagged TAp73␤, ⌬Np73␤, or TAp73␤ R292H in the presence or absence of c-Jun, and whole cell lysates were used for IPs with anti-FLAG antibody beads followed by immunoblotting with the indicated antibodies (upper panels). B, H1299 cells were transfected with TAp73␤, ⌬Np73␤, or TAp73␤ R292H in the presence or absence of FLAG-tagged c-Jun, and nuclear (Nuc) lysates were used for the modified ChIP protocol as mentioned under "Experimental Procedures" for IP with anti-FLAG antibody followed by immunoblotting with the indicated antibodies (upper panels). # indicates a nonspecific band. C, H1299 cells were transfected with FLAG-tagged TAp73␤, ⌬Np73␤, or TAp73␤ R292H in the presence or absence of c-Jun, and nuclear lysates were used for the modified ChIP for IP with anti-FLAG antibody followed by immunoblotting with the indicated antibodies (upper panels). D, H1299 cells were transfected with TAp73␤ in the presence or absence of FLAG-tagged c-Jun, left in serum-free medium overnight and fed with 20% serum-containing medium (ϩserum) or otherwise (Ϫserum) for 4 h, and nuclear lysates were collected and used for the modified ChIP for IP with anti-FLAG antibody followed by immunoblotting with the indicated antibodies (upper panels). # indicates a nonspecific band. E and F, H1299 cells were transfected with TAp73␤ in the presence or absence of FLAG-tagged c-Jun, left untreated or treated with UV irradiation (60 J). Nuclear lysates were collected 4 h later and used for the modified ChIP for IP with anti-FLAG antibody followed by immunoblotting with the indicated antibodies (upper panels) (E). # indicates a nonspecific band. Activation of collagenase promoter-luciferase (Col Luc) and cyclin D1 promoter-luciferase (CycD1 Luc) in the untreated (UnT) or UV-treated condition in concurrent experiments is shown (F). Data show the mean of three to five experiments, and error bars indicate S.D. ***, p Յ 0.001 comparing c-Jun in combination with TAp73␤ in untreated versus UV-treated condition. G, H1299 cells were transfected with TAp73␤ in the presence or absence of FLAG-tagged c-Jun or c-Jun AA mutant, and nuclear lysates were used for the modified ChIP with anti-FLAG antibody followed by immunoblotting with the indicated antibodies (upper panels). The lower panels show straight immunoblot data from the lysates without IP in all cases. Experiments were repeated at least two times, and representative results are shown. Vec, vector.

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
binding to c-Jun at the chromatin level. Absence of the aminoterminal domain did not affect binding to c-Jun (Figs. 6, B and  C, and 7A). Similarly, interaction with c-Jun was unaffected by the absence of the extreme carboxyl-terminal domain (i.e. TAp73 1-399 ) (Fig. 7A). However, the deletion mutant partially lacking the DNA-binding domain and the rest of the carboxylterminal domain (i.e. TAp73 58 -249 ) was unable to bind c-Jun, whereas the carboxyl-terminal domain alone (i.e. TAp73 355-474 ) could bind to c-Jun. These data thus indicate that the region between amino acids 355 and 399 of TAp73 is important for binding to c-Jun at the level of the chromatin. The deletion mutants of TAp73 were also tested for their effect on cellular survival by colony formation assay (Fig. 7B). Overexpression of TAp73 and TAp73 1-399 led to a potent inhibition of cellular growth, whereas the DNA binding mutant TAp73 R292H did not. Similar to TAp73 R292H , p73 58 -249 and p73 355-474 did not induce growth arrest. However, p73 355-474 expression margin-ally but consistently led to better survival than p73 58 -249 probably because of the ability of the former to interact with c-Jun.

Fos Family Members Are Recruited to the AP-1 Promoters in a c-Jun-dependent Manner upon TAp73 Expression and Cooperate with TAp73 to Potentiate Cellular
Growth-One of the major questions involving TAp73 and AP-1 members is how does TAp73 activate or potentiate the AP-1 response? To address this, we examined whether TAp73 can transcriptionally up-regulate the AP-1 members. Inducible expression of TAp73 in p53-null Saos2 or in H1299 cells induced AP-1 target genes such as cyclin D1 at the transcriptional level but did not regulate the levels of c-Jun, c-Fos, Fra1, and Fra2, suggesting that TAp73 does not regulate the AP-1 proteins (Fig. 8A). We therefore examined whether TAp73 induction would lead to the recruitment of AP-1 members to the AP-1 site on the target promoters. Chromatin immunoprecipitation assays done in Saos2-TAp73 inducible cells showed that upon induction of TAp73 there was an increase in c-Fos and Fra1 but not c-Jun, FosB, or Fra2 recruitment at the AP-1 sites on the endogenous cyclin D1 promoter (Fig. 8B). This was specific to the AP-1 target cyclin D1 promoter and not the canonical p53 target mdm2 promoter on which only TAp73 was bound. The TAp73mediated recruitment of c-Fos and Fra1 to the AP-1 site was dependent on c-Jun because depletion of c-Jun led to a marked decrease in the recruitment of c-Fos and Fra1 onto the cyclin D1 promoter (Fig. 8C), indicating that TAp73 indeed promotes the recruitment of AP-1 member in a c-Jun-dependent manner to the AP-1 target gene promoters.
Analysis of the functional relevance of the interaction between p73 and the AP-1 members by colony formation assay showed that although TAp73 induction alone led to decreased colony growth in Saos2 cells as expected there was a significant increase in cellular survival as determined by colony formation in the presence of c-Jun, c-Fos, and Fra1 (Fig. 8D, colony numbers indicated in parentheses). However, this was not the case in the presence of JunB, JunD, and FosB, which were also not recruited to AP-1 target promoters ( Fig. 8B and data not shown) and which do not generally synergize as well as c-Fos or Fra1 with TAp73 to transactivate AP-1 target genes. Surprisingly, Fra2 also showed a cooperative effect with TAp73 on cellular growth, although it does not synergize with p73 in activating AP-1 target genes, nor is it required for p73-mediated target gene activation. These data therefore highlight the biological effects of cooperation between TAp73 and the selected AP-1 members on cellular growth.

Discussion
The results presented here provide mechanistic insights into the role of TAp73 in transactivating AP-1 target genes and consequently promoting cellular growth. We provide evidence demonstrating that 1) TAp73 is able to cooperate with other selected AP-1 family members such as c-Fos and Fra1 besides c-Jun in transactivating AP-1 target genes and promoting cellular growth; 2) TAp73 is capable of binding to the AP-1 target promoters on or near the canonical AP-1 binding sites, a requirement for TAp73-mediated target gene activation; 3) TAp73 associates with c-Jun in the context of the chromatin; 4) its expression leads to recruitment of c-Fos and Fra1 onto the

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
AP-1 target gene promoters; and 5) all these effects are dependent on the presence of c-Jun. Thus, TAp73 works in a complex with c-Jun and selected AP-1 family members to transactivate AP-1 target gene expression.
Although TAp73 has been demonstrated to be able to induce apoptosis and functions to suppress tumorigenesis, it is also overexpressed in many cancers. Our previous work had suggested a role for TAp73 in supporting cellular growth through the activation of AP-1 target genes such as cyclin D1 in a c-Jun-dependent manner (33). This role of TAp73 in promoting cellular growth has now been confirmed with the newly identified role for TAp73 in activating the pentose phosphate pathway and thus cellular proliferation (25). Hence, it is apparent that TAp73 can indeed promote cellular survival in a context-dependent manner especially in cancers where it is overexpressed. One context in which the growth-supporting role of TAp73 is manifested is when it was co-expressed with the AP-1 member c-Jun, leading to the potentiation of the expression of AP-1 FIGURE 8. TAp73 expression leads to the recruitment of AP-1 family members to target gene promoters in a c-Jun-dependent manner and potentiates cellular growth. A, RNA was extracted from Saos2 or H1299-TAp73␤ inducible cells after induction with doxycycline (Dox). Semiquantitative PCR for target genes and AP-1 members was performed. B, Saos-TAp73␤ inducible cells were induced with doxycycline and subjected to chromatin immunoprecipitation with the control IgG or with antibodies against p73 or specific AP-1 members followed by semiquantitative PCR amplification of target regions on cyclin D1 or mdm2 promoters. PCR primer set hCyD1 ϩve covers the region around the AP-1 site on the cyclin D1 promoter (as in Fig. 5B), whereas hCyD1 Ϫve is the negative control. PCR primer set hdm2 ϩve is at the canonical binding site for TAp73 on the mdm2 promoter, whereas hdm2 Ϫve is the negative control. C, similar experiments were carried out in Saos-TAp73␤ inducible cells either in the absence or presence of c-Jun shRNA (Sh). Chromatin immunoprecipitation was carried out with the indicated antibodies followed by real time PCR analysis with the PCR primer set hCyD1 ϩve covering the AP-1 site on the cyclin D1 promoter. Data show the mean of experiments, and error bars indicate S.D. D, Saos-TAp73␤ inducible cells were stably transfected with the different AP-1 members, TAp73␤ expression was induced by doxycycline, and cell survival was determined by growth of cellular colonies over 10 -14 days in culture. Representative pictures are shown from three independent experiments for all cases. Numbers in parentheses indicate percent growth area. E, working model. c-Jun is required for TAp73␤-mediated AP-1 target gene activation in a manner dependent on the canonical AP-1 sites and the presence of Fos family members (c-Fos and Fra1) that dimerize with c-Jun. TAp73␤ does not transcriptionally activate AP-1 members but enhances c-Jun phosphorylation (33) and leads to the recruitment of Fos family members to the AP-1 target gene promoters. TAp73␤ binds to the promoters near the AP-1 site and interacts with c-Jun through its C-terminal region at the level of chromatin. This results in activation of AP-1 target genes and potentiation of cellular growth.

TAp73 Cooperates with AP-1 Members to Promote Cell Growth
targets such as cyclin D1, although further details had not been previously elucidated. Hence, we embarked on exploring the mechanistic details, and our studies have now revealed that other selected AP-1 members like c-Fos and Fra1 are also able to cooperate with TAp73 in a c-Jun-dependent manner.
AP-1 family proteins are transcription factors involved in a plethora of cellular process including apoptosis and growth promotion (28). Animal models have shown a tumor-promoting role for c-Jun, c-Fos, and Fra1. For instance, c-Jun and c-Fos cooperate to enhance c-Fos-mediated osteosarcoma formation (47), and c-Fos was shown to be required for development of 12-O-tetradecanoylphorbol-13-acetate-induced malignant skin tumors (48). Moreover, like c-Jun and c-Fos, Fra1 is also overexpressed in tumors (49), and in response to Ras expression, Fra1 was reported to cooperate with the Erk pathway and stabilize c-Jun by heterodimerizing with it (50). Thus, ample evidence exists for the promotion of tumor formation by these AP-1 factors, which are able to cooperate with TAp73 in promoting cellular growth. Incidentally, as overexpression of TAp73 has been noted in a large number of cancers (10,11), it is tempting to speculate that the ability of TAp73 to promote cellular growth might be in these contexts due to co-expression with AP-1 proteins in vivo; this requires future investigation.
The cooperativity on target gene transactivation is consistent with the ability of TAp73 to bind to the chromatin near or at the AP-1 sites as assessed by chromatin immunoprecipitation assays. Although direct DNA binding was earlier ruled out by gel shift analysis using consensus AP-1 sites (33), the current data are consistent with TAp73 being bound to multiple sites on the genome at or close to AP-1 sites as also established by an earlier genome-wide chromatin immunoprecipitation study (51). Interestingly, the TAp73 R292H mutant that is incapable of binding to DNA is also defective in its ability to activate AP-1 target genes, further suggesting that the DNA binding ability of TAp73 is necessary for AP-1 target gene transactivation probably through c-Jun, which is required for the process. In this context, it is also to be noted that TAp73 was found to bind to c-Jun only in the context of the chromatin. Although it had been difficult to detect reciprocal interactions between p73 and c-Jun previously using cytosolic fractions (Fig. 6A), binding of TAp73 and c-Jun at or near the AP-1 sites suggests that TAp73 and c-Jun are in a complex at the level of chromatin.
Although the AP-1 sites are critical for TAp73-mediated transactivation of AP-1 target genes, other regulatory regions also appear to play a role. For example, the CRE sites on the cyclin D1 promoter appear to have a contributory role in TAp73-mediated activation. p300 and its homologue CREbinding protein are known to bind to the CRE sites at the promoter of their target genes, interact with several transcription factors, and activate transcription (52). p300 has also been shown to bind to TAp73 and positively regulate its transcriptional activity (53). Thus, the possibility exists that p300 binds to the CRE site on the cyclin D1 promoter and enhances TAp73 activity at the AP-1 site. Conversely, TAp73 could influence the activity of the p300⅐CRE-binding protein complex, thereby activating the cyclin D1 promoter. Nonetheless, these cofactors appear to be specific for the individual target genes as CREs are absent on the collagenase promoter.
Similarly, although c-Jun is required for TAp73-mediated AP-1 target gene activation, it is also apparent that this occurs with other partners, especially c-Fos and Fra1, which are recruited to AP-1 target promoters in a c-Jun-dependent manner for the activation of targets, suggesting that TAp73 activation could lead to changes that result in this recruitment. Interestingly, although JunB alone did not cooperate with TAp73 in MEFs (Fig. 3A), it is noteworthy that JunB can substitute for c-Jun in mouse development and cell proliferation (54). Thus, JunB could possibly substitute for c-Jun in c-Jun Ϫ/Ϫ MEFs, leading to some level of cooperation between c-Fos/Fra1 and TAp73 even in the absence of c-Jun as noted earlier (Fig. 3A). Surprisingly, however, although FosB and JunB did synergize to some extent with TAp73 in activating target genes, they were not recruited to the promoter upon TAp73 induction. This could possibly be due to context-specific effects as noted earlier. Thus, FosB and JunB might function differently compared with c-Fos and Fra1 perhaps by enhancing the activity of TAp73 on target gene promoters in contrast to c-Fos and Fra1, which are recruited to the promoters and thereby synergize with TAp73 in many different cell types.
In addition, we had previously found c-Jun to be phosphorylated in a JNK-independent manner by TAp73 upon its induction and that this phosphorylation was critical for the manifestation of the ability of TAp73 to transactivate AP-1 target genes (33). Consistently, the c-Jun D mutant was phosphorylated in the presence of TAp73 and could cooperate with it ( Fig. 1), but this phosphorylation is not required for binding at the chromatin level (Fig. 6G). Although the identity and mechanisms of regulators of TAp73-mediated c-Jun phosphorylation have yet to be unraveled, it is emerging that TAp73, which is overexpressed in cancers, causes c-Jun phosphorylation and thus could result in the recruitment of specific AP-1 members that would altogether cooperate in the activation of AP-1 target genes (Fig. 8E). Consistently, these AP-1 members were also found to be able to cooperate with TAp73 to promote cellular growth, highlighting the biological significance of these findings. These data are also conceptually similar to recent findings that AP-1 target genes are also similarly co-activated by other transcription factors such as FOXK2 in concert with AP-1 family members (55). In this context, it is noteworthy that treatment with DNA-damaging/genotoxic agents like UV irradiation led to a decrease in the interaction between c-Jun and TAp73 at the chromatin as well as activation of AP-1 target genes, and contrastingly, growth factor stimulation led to an increased association between them. This suggests that, during stress conditions, these two proteins may be less prone to interact and activate cell survival genes. Instead, TAp73 probably dissociates from c-Jun to induce apoptotic genes; this requires further investigation.
Taken together, the results presented demonstrate that TAp73 cooperates with several AP-1 members that dimerize to regulate AP-1 target gene transactivation and thus cellular growth. Thus, co-activation of TAp73 in the presence of AP-1 members would therefore provide a context for the manifestation of the survival properties of TAp73, which would otherwise lead to inhibition of cellular growth. Hence, these data suggest that inhibition of TAp73/AP-1 cooperation may be an