The c-Jun δ-Domain Inhibits Neuroendocrine Promoter Activity in a DNA Sequence- and Pituitary-specific Manner

The transcription and transformation activity of c-Jun is governed by a 27-amino acid regulatory motif, labeled the δ-domain, which is deleted in v-Jun. We have previously shown that c-Jun is a potent inhibitor of the rat prolactin (rPRL) promoter activity induced by either oncogenic Ras or phorbol esters. Here, we have characterized the structural and cell-specific requirements for this c-Jun inhibitory response, and we show that this c-Jun inhibitory response mapped to the rPRL footprint II repressor site, was pituitary-specific and required the c-Jun δ-domain. Moreover, alteration of any one of these features (e.g., cis-element, trans-factor, or cell-specific background) switched c-Jun to a transcriptional activator of the rPRL promoter. In HeLa nonpituitary cells, c-Jun alone activated the rPRL promoter via the most proximal GHF-1/Pit-1 binding site, footprint I, and synergized with GHF-1. Finally, recombinant GHF-1 interacted directly with c-Jun but not c-Fos proteins. These data provide important fundamental insights into the molecular mechanisms by which the c-Jun δ-domain functions as a modulatory switch and further imply that the functional role of c-Jun is dictated by cell-specific influences and the δ-domain motif.

The transcription and transformation activity of c-Jun is governed by a 27-amino acid regulatory motif, labeled the ␦-domain, which is deleted in v-Jun. We have previously shown that c-Jun is a potent inhibitor of the rat prolactin (rPRL) promoter activity induced by either oncogenic Ras or phorbol esters. Here, we have characterized the structural and cell-specific requirements for this c-Jun inhibitory response, and we show that this c-Jun inhibitory response mapped to the rPRL footprint II repressor site, was pituitary-specific and required the c-Jun ␦-domain. Moreover, alteration of any one of these features (e.g., cis-element, trans-factor, or cell-specific background) switched c-Jun to a transcriptional activator of the rPRL promoter. In HeLa nonpituitary cells, c-Jun alone activated the rPRL promoter via the most proximal GHF-1/Pit-1 binding site, footprint I, and synergized with GHF-1. Finally, recombinant GHF-1 interacted directly with c-Jun but not c-Fos proteins. These data provide important fundamental insights into the molecular mechanisms by which the c-Jun ␦-domain functions as a modulatory switch and further imply that the functional role of c-Jun is dictated by cell-specific influences and the ␦-domain motif.
c-Jun is a member of the BZip family of transcription factors, which are characterized by a basic DNA binding domain and a leucine zipper protein dimerization domain (1,2). c-jun was first described as the cellular homologue of the v-jun oncogene, and subsequently, many other BZip family members have been identified, including several Jun-like and Fos-like proteins (3)(4)(5). c-Jun and c-Fos can combine to form homo-and heterodimers, with Jun-Jun homodimers and Jun-Fos heterodimers (also referred to as an AP-1 complex), binding with high affinity to the consensus DNA binding site, TGACTCA, whereas Fos-Fos homodimers are unable to stably bind to this sequence (1, 6 -8). Once bound to DNA, their effects are mediated through amino-terminal transactivation domains (9). In the cell, c-Jun serves as a downstream target for several signaling pathways, including 12-O-tetradecanoyl phorbol 13-acetate (TPA), 1 a phorbol ester that activates protein kinase C (3,10) and oncogenic Ha-Ras (11,12). TPA mediates its transcriptional effects through a TPA-responsive element, which is equivalent to the AP-1 site noted above (3,4,13,14). Stimulation of Ha-Ras leads to activation of Jun kinase, resulting in the phosphorylation of amino-terminal serines 63 and 73 in the transcription activation domain of c-Jun, thereby enhancing its transcription potency (15)(16)(17)(18). Also present in the amino terminus of c-Jun, but deleted in v-Jun, is a region known as the ␦-domain (19). This 27-amino acid region is important for cellspecific modulatory effects on transcription and transformation potency of c-Jun versus v-Jun, thought to be mediated via protein-protein interactions (9, 20 -23). One such ␦-domainspecific interaction is with Jun kinase, which binds to the amino-terminal half of the ␦-domain spanning amino acids 34 -45 and phosphorylates c-Jun but not v-Jun (9, 18, 24 -26).
Additionally, although c-Jun typically activates gene transcription, examples have accumulated documenting that it can also inhibit gene expression. A direct mode of inhibitory action was shown in the c-fos gene, where the AP-1 complex binds to the c-fos promoter to down-regulate c-fos gene expression (27). Alternatively, the AP-1 complex has been shown to bind to a TPA-responsive element that overlaps a critical retinoic acid response element/vitamin D response element required for osteocalcin promoter activity, and thus sterically interferes with retinoic acid receptor/vitamin D receptor binding (28). In certain cases, c-Jun heterodimerizes with other activators, either via its leucine zipper motif, e.g. with the glucocorticoid receptor (29), or via its amino-terminal domain, e.g. with MyoD (30). Indeed, the ability of c-Jun to inhibit the muscle-specific creatine kinase gene (30), cardiac-specific atrial natriuretic factor gene (31), or the liver-specific ␣-fetoprotein gene (32) was mapped to the amino-terminal domain of c-Jun. In this respect, it is striking that in each case whereby c-Jun inhibits a highly specialized, tissue-specific gene, it does so via its amino-terminal domain, usually requiring just the first 87 amino acids, including the ␦-domain.
One of the first indications that the ␦-domain serves a regulatory function was noted in cell-free transcription studies. Recombinant c-Jun proteins containing the ␦-domain weakly activated an AP-1-driven promoter in Jun/Fos-depleted HeLa cell extracts, whereas v-Jun or amino-terminal truncated c-Jun proteins, devoid of the ␦-domain, were very active in this in vitro transcription assay (21). Importantly, all of these recombinant Jun proteins bound equally well to the TPA-responsive DNA element. Using in vivo gene transfer methods, it was shown that in all cell lines tested, a c-Jun fusion protein with the DNA binding domain of E2, a transcription factor from bovine papilloma virus, activated the appropriate promoterreporter plasmid (9). Deletion of the ␦-domain resulted in an enhancement of Jun's activation effect in HeLa cells (9). However, not all tested cells revealed this differential effect of c-Jun versus v-Jun. For example, the transcriptional activity of c-Jun and v-Jun were the same in REF, SL2, and F9 cells, whereas c-Jun was shown to be a better transactivator than v-Jun in CEF and HepG2 cells (9,20,22). While the notion that the ␦-domain operates as a critical negative regulatory domain stems from the observation that simply removing the ␦-domain from c-Jun results in its oncogenic activation, it has become increasingly clear that transformation and transcription potency of Jun proteins are not directly correlated and may even be inversely related (22). Despite this important progress in elucidating the structural features of Jun proteins, the precise rules by which the ␦-domain functions remain unknown.
The prolactin and growth hormone (GH) genes are two ancestrally related genes whose expression is restricted to the lactotroph and somatotroph cells of the anterior pituitary, respectively (33,34). Both the ontogeny of these pituitary cells and the expression of these two pituitary-specific genes are regulated by the POU homeodomain transcription factor, GHF-1/Pit-1 (33,35,36). Significant insights into basal and hormone-activated PRL and GH gene expression have been provided by GH 4 rat pituitary tumor cells, which are a clonal cell line that maintains cell type-specific functions and hormonal responses (33,34,(37)(38)(39). Previous experiments in this system have demonstrated that c-Jun does not function as a downstream target for either oncogenic V12 Ras-or TPA-mediated activation of the rPRL promoter, but instead c-Jun inhibits both of these signal transduction pathways (40,41). Yet this inhibitory effect of c-Jun on the V12 Ras-and TPA-mediated activation of the rPRL promoter is promoter-specific and not GH 4 cell-specific, since we demonstrated that c-Jun enhances V12 Ras stimulation of the AP-1-dependent Ϫ73ColCAT promoter-reporter construct (40). These results suggest that c-Jun is capable of serving multiple functions within these cells and that some of these functions are promoter-specific. Thus, the rPRL promoter and GH 4 pituitary cells provide an important model system in which to elucidate the molecular mechanisms by which c-Jun mediates promoter-and cell-specific effects. The goal of the studies presented here was to dissect the mechanism of c-Jun inhibition of the rPRL promoter in GH 4 neuroendocrine cells. Using transient transfection studies we showed that c-Jun selectively inhibits basal rPRL promoter via the amino-terminal c-Jun ␦-domain and that this inhibition required the rPRL promoter FP II repressor-binding site and pituitary-specific influences. Moreover, eliminating any one of these elements switched Jun function on the rPRL promoter to an activator. These data provide critical and novel insights into the regulatory functions of the c-Jun ␦-domain and further imply that the precise functional role of c-Jun is dictated by the potential interaction of cell-specific factors with the ␦-domain motif.
Plasmid DNAs were purified either by alkaline-SDS extraction followed by cesium chloride density gradient centrifugation (50) or according to the Qiagen Mega protocol (Qiagen Inc., Chatsworth, CA). Plasmids were quantitated by both absorbance at 260 nm and by comparison with DNA standards on agarose electrophoresis (50). No significant difference was observed in transfection results using plasmids prepared by the two different purification methods.
Cell Culture, Electroporation, and Luciferase Assay-GH 4 T2 rat pituitary tumor cells and HeLa human cervical carcinoma cells were grown in 5% CO 2 at 37°C in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal calf serum (HyClone, Logan, UT) and 50 g/ml of penicillin and streptomycin (Life Technologies). Cells were transfected via electroporation, keeping the total amount of DNA constant with pGEM7 or pGEM4 DNA, and pRSV␤-globin was used to control for nonspecific effects of RSV expression vectors (37). Electroporations were performed in triplicate for each condition within a single experiment, and experiments were repeated using different plasmid preparations of each construct. Cells were harvested at 24 h after transfection unless otherwise stated, cell extracts were prepared, and luciferase assays were performed as described previously (37). Luciferase light units of the control value were set to 1, and the data was expressed as -fold stimulation relative to control. All data was expressed as the mean Ϯ S.E. for replicated experiments. Since c-Jun expression modulated the activity of each of the various viral promoters typically used to drive a ␤-galactosidase reporter, those ␤-galactosidase vectors could not be used as internal controls for transfection efficiency (see Fig. 2). Previously, we have found that by repeating the various transfections multiple times and applying statistical analysis to the resultant data, we are able to achieve consistency of agreement that is equal to or better than using an internal control reporter vector (37-39, 42, 43).
Western Blot Analysis-GH 4 cells transiently transfected with the rPRL promoter-reporter and various effector plasmids were harvested with phosphate-buffered saline with 3 mM EDTA. GH 4 cells were lysed by sonication using five 10-s pulses on ice in 300 l of lysis buffer containing 20 mM HEPES, pH 7.9, 0.42 M KCl, 1.5 mM MgCl 2 , 25% glycerol, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.15 mM spermine, 0.5 mM spermidine, 0.5% Triton X-100 and 0.45 mg/ml Boehringer Mannheim protease inhibitor set. After sonication, cell extracts were centrifuged at 10,000 ϫ g for 5 min at 4°C. HeLa cells transiently transfected with the rPRL promoter reporter and various effector plasmids were harvested with phosphate-buffered saline containing 3 mM EDTA. Cells were lysed with three sequential freeze-thaw cycles in 100 mM potassium phosphate, pH 7.8, and 1 mM dithiothreitol. Vortexing ensured cell lysis. The cells were spun down at 10,000 ϫ g for 5 min at 4°C to pellet unlysed cells and cell debris. The protein concentration of the supernatant was determined using the Bio-Rad protein assay. Equal amounts of total cellular protein (100 g) were resolved on an SDS-10% polyacrylamide gel and transferred to nitrocellulose in 192 mM glycine, 25 mM Tris, 10% methanol at 100 mA for 16 h. Membranes were blocked overnight with 7.5% nonfat dried milk in 20 mM Tris-Cl (pH 7.4), 150 mM NaCl, and 0.2% Tween 20.
The membranes were then probed with a rabbit polyclonal c-Jun antibody directed against amino acids 95-105 (Santa Cruz Biotech; Santa Cruz, CA), a rabbit polyclonal GHF-1 antibody directed against amino acids 214 -230 (BabCO; Richmond, CA), or a mouse monoclonal actin antibody, clone C4 (Boehringer Mannheim) in a 1:1000 dilution in blocking buffer with 1% dried milk. The membranes were extensively washed and developed with a 1:5000 dilution of goat anti-rabbit or goat anti-mouse antibodies linked to horseradish peroxidase (Life Technologies), using an enhanced chemiluminescence kit from Amersham Life Sciences Inc. Between probes with different antibodies, the nitrocellulose membranes were stripped by incubation at 50°C for 30 min in a solution of 0.7% ␤-mercaptoethanol, 2% (w/v) SDS, 62.5 mM Tris, pH 6.8. After stripping, the membranes were reblocked overnight, and the membrane was then reprobed as described above.
In Vitro Protein Binding Studies-The protein A/GHF-1 fusion (pA/ GHF-1) vector was constructed by filling in, with Klenow polymerase, the NcoI to NotI fragment of the GHF-1 clone, SK-9 (47), in which the ATG codon of GHF-1 was modified to an NcoI site and the NotI site is downstream of the stop codon. The blunt-ended fragment was inserted into the SmaI site of the pA vector, Rit 32 (a modification of the Rit 2 vector) (51). The amino-terminal start codon of GHF-1 is thus fused, in frame, to the carboxyl tail of protein A. The pA and pA/GHF-1 vectors were transformed into N4830 -1 bacteria (Pharmacia Biotech Inc.) grown in 500 ml of Luria broth (LB) at 30°C until the A 600 ϭ 0.6 and induced with 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside and increasing the temperature to 42°C by the addition of 500 ml of LB at 55°C; the N4830 -1 bacteria used to grow the pA and pA/GHF-1 constructs contain a temperature-sensitive cI repressor which, with the Lac repressor, regulates the Rit32 promoter. The pA-and pA/GHF-1expressing cells were collected 45 min after induction, since pA/GHF-1 expression was quite toxic to the cells.
Bacterial cell pellets were resuspended with 7-8 ml of TST (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20, 1 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, and 5 g/ml each of the protease inhibitors antipain, chymostatin, leupeptin, and pepstatin A) and lysed by sonication, and the debris was pelleted for 20 min at 15,000 rpm in an SS34 rotor. The pA-and pA/GHF-1-containing lysates were then passed over 0.5 ml of IgG-agarose (Sigma) columns, washed twice with 5 ml of TST and then twice with 5 ml of buffer A (25 mM Hepes, pH 7.9, 80 mM KCl, 6 mM MgCl 2 , 10% glycerol, 1 mM dithiothreitol, and protease inhibitors listed above). The beads were then resuspended with 1.5 ml of buffer A and stored at 4°C. Efficient production of either fusion protein was assessed by boiling approximately 10 l of packed beads in SDS sample loading buffer, loading onto an SDS-10% polyacrylamide gel and staining with Coomassie Blue.
Reticulocyte lysates (Promega) were programmed with RNA transcribed by T7 polymerase, resulting in either human c-Jun, rat c-Fos, or rat GHF-1 protein labeled with [ 35 S]Met (DuPont NEN). Efficiency of labeling was assessed by electrophoresing 1 l of each sample on an SDS protein gel followed by autoradiography. Roughly equivalent amounts of each radiolabeled protein (2-5 l of each programmed reticulocyte lysate) were incubated with about 5 l of packed pA or pA/GHF-1 beads in 100 l of buffer A containing 0.05% Nonidet P-40 and 200 g/ml ethidium bromide (52) for 1.5 h at 4°C with gentle rocking in an Eppendorf tube. Unprogrammed reticulocyte lysate was added to each incubation so that all incubations were done with the same amount of lysate. Following incubation, the beads were pelleted, the supernatant removed, and the pellet was washed six additional times each with 170 l of buffer A, 0.1% Nonidet P-40. To each pellet or input sample was added 15 l of SDS gel loading buffer. The samples were boiled and loaded onto an SDS 3% stacking, 10% resolving polyacrylamide gel, the signal was enhanced with "Amplify" (Amersham), and the gel was dried and visualized by autoradiography.

RESULTS
c-Jun Selectively Inhibits the rPRL Promoter in a Dose-responsive Fashion-Since c-Jun has previously been shown to inhibit basal rPRL promoter activity (41), we first tested whether this effect was dose-responsive on the rPRL promoter in GH 4 cells. As shown in Fig. 1A, c-Jun produced a dose-dependent inhibition of the Ϫ425-base pair rPRL promoter reaching 58% inhibition at a dose of 20 g of pRSVc-jun. Although the inhibitory effect of c-Jun is modest, it is very reproducible and statistically significant (p Ͻ 0.05). Also, similar doses of pRSVc-jun inhibited a Ϫ2.5 kilobase pair rPRL promoter construct, containing the distal enhancer (data not shown). The maximal dose of pRSVc-jun, 20 g, was then used in all further studies. The inhibition of the Ϫ425 rPRL promoter by c-Jun reached 50% by 12 h post-transfection and remained constant from 12 to 24 h post-transfection, with maximal inhibition reaching 77% at 36 h post-transfection (data not shown). In order to ensure that the transiently transfected c-Jun expression vector resulted in detectable levels of c-Jun protein in GH 4 cells, we performed Western blot analysis. As shown in Fig. 1B, control (lane 1) and mock-transfected (lane 2) GH 4 cells did not appear to have detectable levels endogenous c-Jun, whereas transfection of pRSVc-jun resulted in readily detectable levels of c-Jun protein (lane 3). The same blot was reprobed with an anti-GHF-1 antibody in order to verify that the proteins in the cellular extract loaded in lanes 1-3 were equivalent and intact. The results show that the levels of endogenous GHF-1 protein were equivalent (Fig. 1B, lanes 4 -6). Also, c-Jun expression had no effect on GHF-1 protein levels, indicating that the negative effect of c-Jun on the rPRL promoter was not due to a decrease in endogenous GHF-1 protein levels.
To determine whether the inhibition of the rPRL promoter was promoter-specific, the effect of c-Jun was examined on a variety of pituitary and nonpituitary promoters in GH 4 cells. As shown in Fig. 2, c-Jun inhibited the Ϫ425 rPRL promoter and the Ϫ2.5 kilobase pair rPRL promoter 54 and 34%, respectively. Likewise, the pituitary-specific growth hormone promoter was also inhibited by c-Jun to 53% of its basal activity. However, c-Jun was not a general inhibitor of pituitary promoters as evidenced by its 1.7-fold stimulation of the human glycoprotein ␣-subunit promoter. Also, c-Jun stimulated the c-fosTK promoter, containing a consensus AP-1 binding site, by 1.9-fold. The effects of c-Jun on several viral promoters in the GH 4 cells also varied, with the CMV promoter inhibited 45% by c-Jun, whereas the SV40 and RSV promoters were stimulated 2.1-and 2.4-fold, respectively. The effect of c-Jun on the promoterless pA 3 luc reporter vector, used as the parental reporter was minimal. These data indicate that the c-Jun inhibition was promoter-selective for rPRL, rGH, and CMV and that the neg- ative effect was not mediated by DNA sequences in the pA 3 luc vector background.
c-Jun Inhibition Maps to the FP II Region of the rPRL Promoter-In order to address the question of mechanism, we sought to determine which region of the rPRL promoter was important for mediating this c-Jun inhibition. In the rPRL promoter, footprints (FPs) I, III, and IV bind GHF-1/Pit-1, a pituitary-specific POU-homeodomain transactivator (36,53). FP II binds a ubiquitous repressor denoted as F2F, and the Ϫ117 to Ϫ80 BTE binds a ubiquitous basal transcription-activating factor, both of which have yet to be characterized (43). Using a series of 5Ј deletions and site-specific mutations of the rPRL promoter impinging on these various regulatory cis-acting sites (shown in Fig. 3A), we show that c-Jun equally inhibited rPRL promoter constructs with 5Ј end points of Ϫ425, Ϫ255, and Ϫ189 (Fig. 3B). However, the effect of c-Jun switched to an activating response with rPRL promoter 5Ј end points of Ϫ125, Ϫ54, and Ϫ36 (Fig. 3B). Of note, statistical analysis (Student's t test) revealed that the effects of c-Jun are statistically significant to p Ͻ 0.05 for the Ϫ425 and Ϫ125 end points and significant to p Ͻ 0.01 for the Ϫ255 and Ϫ189 end points, whereas the effects of c-Jun on the Ϫ54, Ϫ36, and promoterless constructs were not statistically significant. These data indicate that DNA sequences between Ϫ189 and Ϫ125 were responsible for the inhibitory effect of c-Jun, while sequences between Ϫ125 and Ϫ54 were responsible for c-Jun's activating effect. Moreover, neither the negative nor positive effects of c-Jun were due to a cryptic AP-1 site in the pA 3 luc background vector, since c-Jun had no effect on the pA 3 luc promoterless vector (Figs. 2 and 3B).
Within the Ϫ189 to Ϫ125 region of the rPRL promoter lie both FP III and FP II. In order to determine which of these two footprints was important for c-Jun action, we used a panel of site-specific mutants that featured a SalI linker sequence substituted for specific bases within the various footprints and a site-specific deletion of the BTE in a FP II mutant background (Fig. 3A). As shown in Fig. 3C, mutations in FPs I and III, the ⌬1 and ⌬1,3 mutants did not prevent the c-Jun-mediated inhibition of the rPRL promoter, excluding these sites as DNA targets of the c-Jun inhibitory effect. Indeed, these two mutant promoters were inhibited to a greater extent by c-Jun, perhaps because of the disruption of GHF-1 activating sites. However, mutation of FP II, in either the ⌬2 or ⌬2,D constructs, led to a loss of inhibition, thereby mapping the c-Jun effect to FP II of the rPRL promoter (Fig. 3C). These data are consistent with the 5Ј deletion data, showing that disruption of FP II in the Ϫ125 and shorter constructs eliminated the repressing effect of c-Jun (Fig. 3B).
Removal of the ␦-Domain Switches c-Jun from an Inhibitor to an Activator-In order to investigate potential structure-function correlations as they related to mechanism, we sought to determine which domains of the c-Jun protein were required to mediate the inhibition of the rPRL promoter. As shown schematically in Fig. 4A, the c-Jun protein consists of five major domains. The ␦-domain is important for protein-protein interactions, including the interaction with Jun kinase (18,54). Next, there are two functionally determined transactivation domains (9). Finally, there is a basic domain for DNA binding and a COOH-terminal leucine zipper domain, which is important for protein dimerization (1,2). Given the data indicating that the ␦-domain plays an important regulatory role, modulating c-Jun's transactivation potential in a cell-specific manner (9, 20 -23), we chose to use a v-Jun construct, which differs from c-Jun in that it lacks the NH 2 -terminal ␦-domain and contains three COOH-terminal point mutations (9,19).
Surprisingly, pRSVv-jun had the opposite effect of c-Jun on the intact Ϫ425 rPRL promoter in GH 4 cells, producing a 12.9-fold promoter stimulation, whereas c-Jun resulted in a 42% inhibition (Fig. 4B). These data demonstrate that loss of the ␦-domain provides a functional switch, turning v-Jun into an activator. To address whether equal amounts of transfected plasmids produced equivalent amounts of expressed Jun protein, we performed Western blot analysis using a Jun antibody capable of recognizing both c-Jun and v-Jun. As shown in Fig.  4C, c-Jun was actually produced in slightly greater amounts than v-Jun, and the levels of endogenous GHF-1 in these transfected cells were unaffected by c-Jun and mildly stimulated by v-Jun. These data show that differences in c-Jun or v-Jun protein production are not responsible for the specific effects of each Jun construct.

The Absence of Pituitary-specific Factor(s) Switches c-Jun from an Inhibitor to an Activator of the rPRL Promoter-Since
the ␦-domain of c-Jun has been implicated in cell-specific modulation of both transcription and transformation (9, 20 -23) and we documented that deletion of the ␦-domain modulates the inhibitory function of c-Jun in GH 4 pituitary cells (Fig. 4B), we next sought to determine the functional role of c-Jun and v-Jun on the rPRL promoter in nonpituitary HeLa human cervical carcinoma cells. To address these points, we used a nonpituitary gene transfer reconstitution system, whereby the ability of co-transfected effector expression vectors (e.g. GHF-1) to reconstitute the activity of the rPRL promoter in HeLa cells, which otherwise fail to express endogenous PRL or GHF-1, can be determined (45). As shown in Fig. 5A, transfected GHF-1 alone stimulated rPRL promoter activity by 21.3-fold, verifying that the reconstitution approach is functional. Surprisingly, co-transfection with c-Jun alone did not inhibit the rPRL promoter, as described previously in GH 4  tors were co-transfected into HeLa cells, a synergistic activation of the rPRL promoter of 131-fold was observed (Fig. 5A). By contrast, v-Jun alone stimulated the rPRL promoter only slightly (1.5-fold), compared with its effects in GH 4 cells (12.8fold; Fig. 4B). Nevertheless, v-Jun also synergized with GHF-1 to stimulate the rPRL promoter 43.5-fold. Western blot analysis, internally controlled for actin protein levels, demonstrated that the transfected c-Jun and v-Jun expression vectors resulted in comparable protein levels in HeLa cells (Fig. 5B). These data demonstrate that in a nonpituitary cell background, the effect of c-Jun on the rPRL promoter is switched to that of an activator, whereas v-Jun functions as an activator in both pituitary and nonpituitary cells, underscoring that the functional effect of the ␦-domain is cell type-specific. Furthermore, these data show that c-Jun is unlikely to inhibit rPRL promoter activity in GH 4 cells by interfering with GHF-1.
Having shown that c-Jun alone activated rPRL promoter activity and synergized with GHF-1 to further enhance rPRL promoter activity in the HeLa nonpituitary cell system, we next sought to map the cis-acting element mediating the c-Jun activation of the rPRL promoter in HeLa cells. As shown in Fig.  6, the activating effect of c-Jun on the intact Ϫ425 rPRL promoter (3.5-fold) was lost upon site-specific mutation of FP I, either alone (1.2-fold) or in combination with a FP III mutation (1.2-fold); whereas mutation of FP II had no effect on the ability of c-Jun to activate rPRL promoter (3.8-fold) in HeLa cells. These results are in agreement with the data from GH 4 cells shown in Fig. 3B, indicating that the activating effects of c-Jun co-localize to DNA sequences containing the most proximal GHF-1 binding site. Since c-Jun synergized with GHF-1 in HeLa cells (Fig. 5A), we wanted to ascertain whether these two proteins could physically interact.
c-Jun Binds to GHF-1 in Vitro-To directly determine whether c-Jun and GHF-1 physically interact, we performed in vitro binding assays whereby the ability of either recombinant protein A alone or a protein A-GHF-1 fusion protein prebound to IgG beads to pull down radiolabeled c-Fos, c-Jun, or GHF-1 proteins was assessed (Fig. 7). Twenty percent of the radiolabeled protein input of rat c-Fos, human c-Jun, and rat GHF-1 is shown in lanes 1, 4, and 7, respectively. Rat c-Fos showed no interaction with either the protein A alone or protein A-GHF-1 beads (lanes 2-3). By contrast, c-Jun showed a low level of interaction with the protein A beads (lane 5), whereas there was significant binding of c-Jun to protein A-GHF-1 (lane 6). Since GHF-1 is known to homodimerize (36), we showed that radiolabeled GHF-1 interacted with the protein A-GHF-1 beads, as a positive control for the binding assay (lane 9) but that labeled GHF-1 binds to protein A beads only minimally Ϫ189 and Ϫ125 base pairs. The indicated rPRL promoter deletions (5 g) were transiently transfected into GH 4 cells with either 20 g of pRSV␤-globin or pRSVc-jun. -Fold activation was determined relative to the basal activity of each individual promoter construct. Data are expressed as mean -fold for four transfections done in triplicate. Statistical analysis (student t test) reveal that the effects of c-Jun are significant to p Ͻ 0.05 for the Ϫ425 and Ϫ125 end points and to p Ͻ 0.01 for the Ϫ255 and Ϫ189 end points. Additionally, the effects of c-Jun on the Ϫ54, Ϫ36, and promoterless constructs were not statistically significant. C, c-Jun repression of the rPRL promoter maps to the FP II region. The indicated rPRL site-specific mutants (5 g) were transiently transfected into GH 4 cells with either 20 g of pRSV␤-globin or pRSVc-jun. -Fold activation was determined relative to the basal activity of each individual promoter construct. Data are expressed as mean -fold Ϯ S.E. for three transfections done in triplicate.
(lane 8). The inability of labeled c-Fos to bind to protein A-GHF-1 beads indicates that there is specificity to the GHF-1c-Jun interaction.

DISCUSSION
Although our understanding of c-Jun function has progressed quite rapidly, some of the original observations relating to the differential and cell-specific effects of c-Jun versus v-Jun in transcription and transformation assays have remained unexplained. In this paper, we show that c-Jun inhibition of rPRL promoter basal activity requires the c-Jun ␦-domain, the FP II site, and the pituitary-specific cell type. Alteration of any one of these features switches the inhibitory effect of Jun to that of an activator. Moreover, since the putative DNA binding sites for both c-Jun and v-Jun proteins are the same, the striking differences in their effects on the rPRL promoter strongly suggest that c-Jun and v-Jun mediate their differential effects upon the same promoter in the same cell type due to a mechanism independent of DNA binding specificities. Taken together, our data are most consistent with the model that the structural difference between these two Jun isoforms, specifically the ␦-domain, dictates a second level of transcription control by governing factor-factor interactions. The information gained provides critical mechanistic insights into the molecular code by which Jun proteins regulate gene expression.
The absence of the ␦-domain in v-Jun renders this protein unable to bind or to be phosphorylated by Jun kinases (JNKs), yet v-Jun is typically a much more potent oncogene, raising the interesting possibility that phosphorylation is required primarily to "inactivate" the ␦-domain in c-Jun (25,55). From these data, it is tempting to speculate that catalytically inactive JNK, which binds tightly to the ␦-domain, functions as the putative repressor, possibly by masking the amino-terminal transactivation domain and/or influencing c-Jun's ubiquitination and degradation, and that JNK activation causes its release from c-Jun (18, 24 -26). Nevertheless, several lines of evidence argue against this hypothesis: 1) recent studies have shown that only a small fraction (5-10%) of c-Jun in quiescent cells is bound with JNK (25); 2) the region of the ␦-domain required for JNK binding (amino-terminal) is distinct from that required for repression (carboxyl-terminal) (18,23,25); 3) the JNK pathway is primarily an apoptotic one and not a transforming one (56); and 4) the difference in c-Jun versus v-Jun protein levels is small (23). Thus, ␦-domain-dependent mechanisms, other than JNK binding or ubiquitination control, must exist and are yet to be elucidated (57).
Using a cellular promoter in a homologous and highly specialized neuroendocrine cell line rather than a viral or artificial heterologous promoter in a fibroblast cell, we show that the ␦-domain maintains cell-specific effects. Of note, the effect of the c-Jun ␦-domain in our system was to inhibit rPRL promoter activity rather than to mediate a partial activation or simply produce no effect, as reported previously (9,21). If c-Jun were to bind a putative pituitary-specific repressor via the ␦-domain and thus recruit this repressor to an AP-1 site by c-Jun binding to DNA, then c-Jun should function as an intrinsic inhibitor on all AP-1-containing promoters in the GH 4 cell type. However, our current and published data show that c-Jun activates the AP-1-containing Fos (Fig. 2) and collagenase promoters (40) in these GH 4 cells. Additionally, the rPRL promoter does not contain a canonical AP-1 site (41). With regard to mechanism, if c-Jun were to titrate a putative cell-specific co-activator, then this co-activator would have to display specificity for the rPRL and rGH genes, since c-Jun inhibits both promoters (Fig. 2) and v-Jun activates both promoters ( Fig. 4 and data not shown). In this respect, we initially surmised that c-Jun might be interfering with the function of Pit-1/GHF-1 by binding to this factor, thereby inhibiting both rPRL and rGH promoter activities. Surprisingly, our data show that c-Jun does interact with Pit-1/GHF-1 directly (Fig. 7) but that such an interaction results in activation of the rPRL promoter in a HeLa cell reconstitution assay (Fig. 5) via the most proximal Pit-1/GHF-1 binding site (Fig. 6). By contrast, mapping of the cis-acting element mediating c-Jun's inhibitory response co-localized the Jun responsive element to the FP II site, previously identified as a binding site for the putative repressor, F2F (43). Since F2F (43) and the c-Jun inhibitor both require an intact FP II site, the formal possibility remains that these two proteins might belong to the same family of transcription factors. Nevertheless, it is highly unlikely that they will be the same factor. Indeed, there are several lines of evidence that show that the F2F repressor functions and is expressed in a variety of nonpituitary cell lines, including HeLa and Rat 2 cells (43), whereas the putative target of c-Jun functions in a GH 4 pituitary-and FP II-specific manner ( Fig. 3 and 5). It is the apparent absence of the c-Jun inhibitor in HeLa nonpituitary cells that allows c-Jun to switch function and become an activator of the rPRL promoter. These data imply that the interaction between the c-Jun ␦-domain and the pituitary-specific inhibitor is dominant, and abrogation of this interaction is required in GH 4 pituitary cells in order to unmask the recessive and activating effects of c-Jun, which are mediated by FP I and GHF-1.
Based on the results presented here, we have formulated a model, presented in Fig. 8, which incorporates all of these features. In this model, we propose that c-Jun stabilizes the binding of a pituitary cell-specific repressor protein to FP II, by a protein-protein interaction mechanism that would require the ␦-domain of c-Jun. Thus, deletion of the ␦-domain, elimination of the pituitary-specific inhibitor, or site-specific mutation of FP II, would nullify this dominant inhibitory effect and unmask the recessive activating effect of the GHF-1⅐c-Jun complex acting via FP I. Consistent with this model is the observation that both c-Jun and v-Jun functionally interact with GHF-1 to cooperatively activate the rPRL promoter via FP I (Figs. 5A and 6) in HeLa nonpituitary cells devoid of the pituitary-specific repressor. These results indicate that a region distinct from the ␦-domain of c-Jun interacts with GHF-1, since both c-Jun and v-Jun synergize with GHF-1 (Fig. 5A). Nevertheless, the ␦-domain does appear to contribute to the GHF-1⅐c-Jun interaction, since c-Jun cooperates more efficiently with GHF-1 than does v-Jun (Fig. 5A). However, if c-Jun is able to interact with GHF-1 to stimulate rPRL promoter activity, why is this interaction recessive to the c-Jun-pituitary-specific F2F interaction in GH 4 cells? We have previously shown that the FP II sequence functions as an inhibitory element only when juxtaposed to the vicinal BTE (43). Moreover, site-specific mutation of the BTE renders the rPRL promoter devoid of basal activity, despite intact GHF-1 binding sites (43,44). Thus, if c-Jun enhances the inhibitory effect of the FP II site and if this effect is transduced to the BTE, as shown previously, then the functional interaction between c-Jun and the pituitary-specific FP II-binding factor should dominate over the c-Jun⅐GHF-1 response. Finally, implicit in this model is that DNA binding of either c-Jun or v-Jun to the rPRL promoter is not necessary, but instead c-Jun and v-Jun would mediate their effects through protein-protein interactions with other factors whose presence is dictated by the developmental state of the cell. This notion would be consistent with the lack of a canonical AP-1 site in the rPRL promoter.
Although the ␦-domain definitely affects c-Jun activity, the cumulated data indicate that its function is much more complicated than a simple interaction of the ␦-domain with a putative cell-specific repressor, as initially postulated (9,23,48). In this respect, the ability of c-Jun to induce differentiation in F9 teratocarcinoma cells (11) and to inhibit the basal activity of highly specialized promoters (30 -32), suggest that the putative effects of the ␦-domain may be governed by multiple regulatory influences, including the differentiated state of the cell, environmental cues, signaling events, repressors, co-activators, and ubiquitination machinery. Indeed, as if to verify this point, several reports have shown that the c-Jun amino-terminal transactivation domain, including the ␦-domain, functionally (and in some cases physically) interacts with 1) certain transcription factors, such as MyoD, myogenin, steroid receptors, and STAT3␤, to either repress or activate transcription of specific target genes, and 2) JNK (18, 24, 29 -32, 57). These data suggest that the ␦-domain may contain several functional faces, one interacting with JNK and a separate face that might interact with other proteins, some of which may be cell-specific.