c-Jun Potentiates the Functional Interaction between the Amino and Carboxyl Termini of the Androgen Receptor*

The transactivation functions of the human androgen receptor (hAR) are regulated by several accessory factors that can be either positive or negative. One factor that has been previously shown to mediate hAR transactivation is the proto-oncoprotein c-Jun. The positive effect is a primary one, can be exerted by both endogenous and exogenous c-Jun, and requires multiple regions of c-Jun. However, the exact mechanism by which c-Jun exerts its enhancing function is unknown. In this study, we have used a mammalian two-hybrid system to ask if c-Jun influences the ligand-dependent amino- to carboxyl-terminal (N-to-C) interaction of hAR, which is thought to be responsible for the homodimerization of this receptor. Our results show that c-Jun enhances both hAR N-to-C terminal interaction and DNA binding in vitro . We have also tested a panel of c-Jun and c-Fos mutants for their activities on the N-to-C interaction, and the data demonstrate that the activities of these mutants parallel their activities on hAR transactivation. A mutation in the hAR activation function-2 (AF-2) ab-rogates N-to-C interaction, DNA binding, and transactivation, and these activities are not rescued by exogenous c-Jun. Interestingly, the p160 coactivator TIF2 can stimulate hAR N-to-C interaction, a finding consistent with the effect on hAR transactivation. These data strongly (cid:2) ), an an- drogen-response element. After the addition of an equal number of (cid:2) -galactosidase units of nuclear extract, the reactions were gently vor- texed and incubated for 15 min at 25 °C. The samples were run on a 6% polyacrylamide gel for 1.5 h at room temperature, after which the gel was dried and exposed to autoradiography. To do antibody supershifts, 1 (cid:1) l of either the anti-hAR antibody PA1–111A (Santa Cruz Biotech- nology) or the anti-Birch profilin antibody 4A6 (a kind gift of Martin Rothkegel) were added prior to addition of probe. various c-Jun proteins were expressed from three different promoters (SV40, Rous sarcoma virus ( and cytomegalovirus ( RSV promoter expression of mutants M8, M9, M14, Ala 63/73 , (cid:1) 194–223, 146–221, v-Jun, the CMV promoter the mutants

Transcriptional activation is an important step at which gene expression can be regulated. The nuclear receptors form the only family of transcriptional activators whose activity is ligand-dependent (reviewed in Refs. [1][2][3][4][5]. Upon binding lipophilic ligands, nuclear receptors are activated to turn on (and sometimes turn off) the transcription of specific responsive genes, thereby regulating gene expression. In this way, these small hydrophobic ligands, which include steroid and thyroid hormones and vitamins A and D, exert their effects on many biological processes, including development, differentiation, and homeostasis. The receptors that mediate these effects include the retinoic acid receptor (RAR 1 ), retinoid X receptor, thyroid hormone receptor, vitamin D receptor, the subfamily of steroid receptors (including the glucocorticoid receptor, progesterone receptor, androgen receptor (AR), and estrogen receptors), and numerous "orphan" receptors, ligands for which remain unidentified (6).
AR is the nuclear receptor that mediates the biological actions of androgens. Androgens are found in two physiologically active forms, testosterone and dihydrotestosterone (DHT), and are involved in male sexual differentiation and development (7)(8)(9). As a member of the nuclear receptor superfamily, AR consists of multiple functionally distinct domains, including domains involved in DNA or ligand binding, dimerization, and transcriptional activation specifically found in activation functions-1 (AF-1) and -2 (AF-2) (10). Perhaps unique among the nuclear receptors, AR and other steroid receptors require ligand binding for homodimerization, which itself is necessary for subsequent DNA binding and transcriptional activation (reviewed in Ref. 1). This sequential order of events is responsible for androgen signaling within a cell, and disruption of any one step interferes with the signaling pathway. AR homodimerization has been reported by several laboratories to result from an intra-or inter-molecular interaction between the amino and carboxyl termini of the receptor (11)(12)(13). This interaction was demonstrated using a mammalian two-hybrid system and provided evidence to Langley et al. (12) suggesting that AR binds to DNA as an anti-parallel dimer. Subsequent studies further characterized the amino-to carboxyl-terminal (N-to-C) interaction and showed that naturally occurring mutations in the ligand-binding domain (LBD) found in androgen insensitivity syndromes do not affect ligand binding but disrupt the N-to-C interaction (11). It has been suggested that the N-to-C interaction may also facilitate ligand retention by the AR (14). Mutational analysis of the AR amino and carboxyl termini has identified amino acids 3-36 and the activation function-2 (AF-2) core as being required for the N-to-C interaction (13,15). Further dissection of AR regions involved in the N-to-C interaction has led to the identification of two sequences similar to but distinct from LXXLL core sequence that are known to directly interact with the nuclear receptor AF-2. These AR sequences are FQNLF (amino acids [23][24][25][26][27] and WHTLF (amino acids 433-437), which were shown to bind to the receptor's AF-2 and a region outside AF-2, respectively (16). The AR AF-2 is not only the target of these amino-terminal regions but also provides a site of interaction for LXXLL-containing coactivators, including CREB-binding protein (CBP) (17,18) and the p160 coactivators steroid receptor coactivator-1 (SRC-1) (19) and transcription intermediary factor-2 (TIF2) (20). Interestingly, these same coactivators have been reported to facilitate the N-to-C interaction of AR and suggested in this way to mediate the transcriptional activity of this steroid receptor (15,20,21).
Although SRC-1, TIF2, and CBP target both the amino terminus and AF-2 of AR (15,(17)(18)(19)(20)(21), c-Jun, the dimerization partner of c-Fos (reviewed in Ref. 22) appears to function as an AR coactivator by acting only on the amino terminus (23). Indeed, the AR amino acids 503-555, which harbor an autonomous transactivation function, are both necessary and sufficient for the c-Jun-positive response (23). Earlier studies on the c-Jun enhancement of AR transactivation had shown that (i) the activity is independent of promoter-or cell-specific factors (24,25), (ii) both exogenous and endogenous c-Jun can carry out this activity (24), (iii) the c-Jun effect is primary (24), (iv) multiple regions of c-Jun are required for the activity (26), and (v) c-Fos dimerization with c-Jun blocks the latter's positive activity on hAR (27). All these results suggest an important and general effect by c-Jun on hAR-dependent transcription.
To determine the mechanism of action of c-Jun on AR, we initiated the studies presented here. These studies demonstrate that c-Jun, like CBP and p160 coactivators, can mediate the ligand-dependent N-to-C interaction of AR. Furthermore, c-Jun can enhance AR DNA binding, an important finding because AR dimerization is necessary for sequence-specific DNA binding. The specificity of this effect on AR N-to-C interaction was confirmed by utilizing a series of c-Jun and c-Fos mutants. This positive activity of c-Jun in both AR dimerization and DNA binding required a functional AF-2, demonstrating that the AR AF-2 is essential for the activity of c-Jun and p160 coactivators, because it is required for the N-to-C interaction.
To make hAR(⌬358 -555), hAR amino acids were PCR-amplified using the upstream oligo 5Ј-GATCGAGCTCCGGGACACTT-3Ј and the downstream oligo 5Ј-GATCAGATCTAAGCTTCATCTCCACAGATCAG-GCAGGTCTTGTCGCGACTCTGGTACGCAGC-3Ј. This PCR fragment was digested with SacI and BglII and inserted into hAR(AB)/pTL1, which was digested with the same enzymes. This new construct was subsequently digested with BamHI and HindIII and inserted into hARI, digested with the same enzymes. hAR(AB)/pTL1 was constructed by digesting hAR with BamHI and XhoI and inserting the restriction fragment into pTL1, digested with the same enzymes.
To make hAR(E896P), hAR amino acids 556 -918 were PCR-amplified using the upstream oligo 5Ј-GATCGAATTCATGAAGACCTGCCT-GATCTGT-3Ј and the downstream oligo 5Ј-GATCAGATCTCTCGAGT-CACTGGGTGTGGAAATAGATGGGCTTGACTTTCCCAGAAAGGAT-CTTGGGCACTTGCACAGAGATGATGGGTGCCATCATTTC-3Ј. This PCR fragment, containing the mutated codon that is underlined, was digested with HindIII and BglII and inserted into hAR, which was digested with the same enzymes, and replaced the wild-type region (amino acids 556 -918) in hAR with one that was mutated.
To make GAL-hAR(DE), the GAL4 DNA-binding domain (DBD) was first PCR-amplified using the upstream oligo 5Ј-GATCGAATTC-GATATCTAGAAGCTTCTATGAAGCTACTGTCT-3Ј and the downstream oligo 5Ј-GATCAGATCTATTCAGTCAGGGCCCGAGCTCGGAT-CCGGT-3Ј and inserted into pTL1, a derivative of pSG5, to make the plasmid pG4AB. Then, the hAR DE region was PCR-amplified using the upstream oligo 5Ј-GATCGGTACCCTCGAGTCTAGAACTCTGGGAGC-CCGGAAG-3Ј and the downstream oligo 5Ј-GATCGGATCCTCACTGG-GTGTGGAAATA-3Ј and inserted into pG4AB, yielding GAL-hAR(DE). GAL-hAR(DE/E896P) was in made the same way as GAL-hAR(DE), except that the template for PCR was hAR(E896P) instead of wild-type hAR. VP16-hAR(AB) was constructed by digesting the AB region out of hAR with EcoRI and XhoI and inserting into the EcoRI and SalI sites of pCMX-VP16-N (29). All of the above clones generated by PCR have been verified by DNA sequencing (available upon request).
The reporter plasmids have the gene for chloramphenicol acetyl transferase (CAT) driven by the MMTV or 17M-tk promoters. The AR-inducible reporter plasmid MMTV-CAT and the GAL4-inducible reporter 17M-tk-CAT have been previously described (25). Transfection FIG. 1. The hAR carboxyl terminus, but not a complete AB region, is required for the c-Jun positive effect. A, COS cells were transfected with 1 g of MMTV-CAT reporter plasmid and 1 g of each hAR expression plasmid, with (black bars) or without (gray bars) 1 g c-Jun as indicated. 100 nM DHT was used as indicated. Note that CAT activity is represented relative to activity of the first condition, which was transfection of only the reporter plasmid, and this was set to 1. In the diagram of hAR and mutants thereof, the positions of the bipartite AF-1 (amino acids 169 -182 and 293-357) and the AF-2 (amino acids 883-889) are shown as black boxes. The AF-1 has been divided into an AF-1a and AF-1b, according to Chamberlain et al. (38). B, expression of hAR truncation mutants in transfected cells. COS cells were transfected with 5 g of each of the hAR plasmids and subjected to Western blot analysis. efficiency was standardized by measuring the ␤-galactosidase activity, originating from the cotransfected plasmid pCH110.
Cell Transfections and CAT Assays-COS cells were grown in Dulbecco's modified Eagle's medium (Sigma Chemical Co.) and supplemented with 10% fetal bovine serum (HyClone Laboratories). Cells were grown in 60-mm dishes and transiently transfected using the calcium phosphate precipitation method (30). hAR was activated by the addition of 100 nM DHT. CAT assays were performed and standardized according to the measured ␤-galactosidase activity as previously described (30). For all transfections, we used different amounts of expression plasmid, 1 g of reporter plasmid (either MMTV-CAT or 17M-tk-CAT), 2 g of pCH110, and enough pTL1 to bring the final plasmid amount to 15 g per dish.
CAT assay results were quantified by densitometric scanning (420 oe scanner, PDI, Inc.) of autoradiograms of at least three repeats for each transfection, and, thus, each value represents the average of three to four repetitions plus the standard deviation.

SDS-Polyacrylamide Gel Electrophoresis and Western
Blot-To prepare cell extracts for Western blot analysis, transfected COS cells were boiled for 5-10 min in SDS sample buffer (63 mM Tris, pH 6.8; 20% glycerol; 2% SDS; 5% ␤-mercaptoethanol). The amount of extract used was standardized according to ␤-galactosidase activity. Proteins were separated by SDS-polyacrylamide gel electrophoresis and were electrotransferred onto nitrocellulose. After blocking with nonfat dry milk, the nitrocellulose blots were probed with either the anti-AR antibody PA1-111A (Affinity Bioreagents) or anti-c-Jun antibody KM-1 (Santa Cruz Biotechnology). The blots were developed using the ECL chemiluminescence detection kit from Amersham Pharmacia Biotech.
Gel Mobility Shift Assay-COS cells were grown in 100-mm dishes and transfected using the calcium phosphate precipitation method. Cells were treated with 100 nM DHT 24 h prior to harvesting. Cells were harvested in ice-cold phosphate-buffered saline and spun at 5000 rpm for 5 min. 10% of the cells was used to do a ␤-galactosidase assay for quantification of transfection efficiency. The remainder of the cells was resuspended in buffer I (10 mM Tris-HCl, pH 7.5; 10 mM NaCl; 5 mM MgCl 2 ) and incubated at 4°C for 5 min. Sucrose (0.3 M) was then added, and cells were lysed with a Dounce homogenizer. Nuclei were pelleted by centrifuging lysed cells at 2500 rpm (600 ϫ g) for 10 min. The nuclear pellet was washed once with buffer II (buffer I containing 0.3 M sucrose). Then, the nuclear pellet was resuspended in buffer III (50 mM Tris-HCl, pH 8; 150 mM NaCl; 5 mM EDTA; 0.1% Nonidet P-40) with protease inhibitors and incubated with shaking at 4°C for 30 min. The lysed nuclei were centrifuged at 15,000 rpm for 15 min, and the supernatant, constituting the nuclear extract, was saved.
Gel mobility shift assays were performed with nuclear extracts containing the same amount of ␤-galactosidase activity. These reactions were performed in a final volume of 20 l in DNA binding buffer (10 mM Tris, pH 8; 0.1 mM EDTA; 4 mM dithiothreitol), which also containing 1 g of poly(dI-dC), 100 mM KCl, and 150,000 cpm of 32 P-labeled probe (5Ј-GATCCAAAGTCAGAACACAGTGTTCTGATCAAAGA-3Ј), an androgen-response element. After the addition of an equal number of ␤-galactosidase units of nuclear extract, the reactions were gently vortexed and incubated for 15 min at 25°C. The samples were run on a 6% polyacrylamide gel for 1.5 h at room temperature, after which the gel was dried and exposed to autoradiography. To do antibody supershifts, 1 l of either the anti-hAR antibody PA1-111A (Santa Cruz Biotechnology) or the anti-Birch profilin antibody 4A6 (a kind gift of Martin Rothkegel) were added prior to addition of probe.

RESULTS
The c-Jun-positive Activity on hAR Requires the Carboxyl Terminus of the Receptor but Not a Complete AB Region-As a member of the nuclear receptor superfamily, hAR is a modular protein, consisting of an amino-terminal AB region, which contains AF-1, and a carboxyl-terminal E region, which harbors both the LBD and AF-2 (Fig. 1). The AF-1 has been further divided into the functions AF-1a (amino acids 169 -182) and AF-1b (amino acids 293-357) (31). We have previously shown that the hAR AB region and, specifically, amino acids 503-555 within this region are essential and sufficient for c-Jun enhancement of GAL-hAR(AB) transcriptional activity (23). To study the importance of these amino acids in the full-length receptor, hAR proteins lacking different parts of the AB region were generated and studied in transient transfection experiments in COS cells using the androgen-responsive reporter MMTV-CAT and the internal control pCH110 (24) (Fig. 1). When hAR(⌬504 -555) was compared with wild-type hAR, the two proteins exhibited the same transcriptional activity. Surprisingly, the mutant protein also responded to cotransfected c-Jun as strongly as the wild-type counterpart, clearly demonstrating that, in the context of full-length protein, amino acids 504 -555 are dispensable for c-Jun enhancement of hAR transactivation (Fig. 1A). A larger deletion of the AB region, encompassing amino acids 358 -555, which eliminates part of the AF-1b, greatly compromised hAR transcriptional activity, demonstrating the importance of the AF-1 (Fig. 1A). However, this mutated hAR, hAR(⌬358 -555), responded strongly to transfected c-Jun, further arguing that the carboxyl terminus of the hAR AB region is not required for c-Jun activity (Fig. 1A). In contrast, truncation of amino acids 804 -918 completely blocked hAR transactivation in either the absence or presence of c-Jun, suggesting that the carboxyl terminus of the fulllength hAR is required for c-Jun activity (Fig. 1A). Note that the absence of activity in hAR(⌬804 -918) is not due to the absence of protein expression (Fig. 1B). Together, these data show that c-Jun requires the carboxyl terminus of the fulllength hAR, but not a complete AB region, for its positive activity.
A Functional AF-2 Is Necessary for the Positive Effect on hAR by Either c-Jun or TIF2-Because amino acids 804 -918 contain the hAR AF-2 (32), it is possible that hAR(⌬804 -918) is unable to activate transcription due to the absence of the AF-2. 100 nM DHT was used as indicated. Note that in both cases CAT activity is represented relative to activity of the first condition, which was transfection of only the reporter plasmid, and this was set to 1.

c-Jun Potentiates hAR N-to-C Interaction
To directly test this hypothesis, we generated a mutant of hAR with a single amino acid change in the AF-2 (E896P) ( Fig. 2A) that has previously been demonstrated to disrupt AF-2 function in RAR (32). hAR(E896P) was transcriptionally silent, either in the absence or presence of transfected c-Jun, whereas wild-type hAR exhibited the expected strong activity (Fig. 2C). As a control for this mutant's activity, we used TIF2, a coactivator that has previously been shown to require a functional AF-2 (15,20). As expected, hAR(E896P) was unable to respond to transfected TIF2 (Fig. 2D), the same as with c-Jun. To confirm that the lack of activity was not caused by a lack of expression of hAR(E896P), Western blot analysis was performed. As shown in Fig. 2B, hAR(E896P) was expressed almost as well as wild-type hAR. These results strongly suggest that c-Jun, like TIF2, requires a functional AF-2 to act as a coactivator in hAR-dependent transactivation.
c-Jun Potentiates the in Vivo Interaction between the Amino and Carboxyl Termini of hAR-Previous work has demonstrated an in vivo interaction between the AB and E regions of AR (11)(12)(13), which is stimulated by the coactivators CBP, SRC-1, and TIF-2 (15,20,21). To determine whether c-Jun has the same effect on hAR, we established a mammalian twohybrid system using the expression plasmids GAL-hAR(DE) and VP16-hAR(AB) (Fig. 3A) and the GAL4-responsive reporter plasmid 17M-tk-CAT (23) in transient transfection experiments in COS cells. Like the aforementioned coactivators, transfected c-Jun was indeed able to stimulate the ligand-dependent interaction between the amino and carboxyl termini of hAR (Fig. 3B). To analyze the specificity of this c-Jun effect, we utilized a series of c-Jun mutants previously tested for coactivation of hAR (26). In this earlier study (47), we showed that multiple regions of c-Jun are involved in mediating hAR trans-activation, including the bZIP region and a portion of the amino-terminal activation functions. When these c-Jun mutants were tested for mediating hAR(DE)/hAR(AB) interaction, their activities matched what was seen for hAR transactivation (Fig.  3C). Specifically, c-Jun mutant proteins defective in either bZIP function (A 265 3 D In265, ⌬287-332, M14) or activation function (⌬1-245 and ⌬146 -221) are equally ineffective in supporting hAR transactivation or N-to-C interaction, whereas mutants M8, M9, Ala 63/73 , and ⌬194 -223 are able to stimulate both hAR activities (Fig. 3C). These results strongly argue that c-Jun may be influencing hAR transcriptional activity by mediating its ability to form dimers via N-to-C interactions.
The data in Fig. 2 show that a functional AF-2 is necessary for hAR transactivation in either the absence or presence of transfected c-Jun. Because previous work has shown that the AF-2 is required for hAR N-to-C interaction (11,13), we analyzed the possibility that our AF-2 mutant, hAR(E896P) is unable to interact with the AB region, using VP16-hAR(AB) and GAL-hAR(DE/E896P) (Fig. 4A). Although GAL-hAR(DE/ E896P) is expressed in COS cells to the same extent as GAL-hAR(DE), 2 the mutant is incapable of interacting with the AB region (Fig. 4B). Furthermore, although transfected c-Jun potentiated the AB interaction with DE, there is no c-Jun effect observed with the mutant DE (Fig. 4B). These data strongly suggest that a functional AF-2 is required for hAR N-to-C interaction and c-Jun potentiation of this interaction.

c-Fos Represses and TIF2 Stimulates the in Vivo Interaction between the Amino and Carboxyl Termini of hAR-
Although c-Jun stimulates hAR transactivation, its dimeriza-2 A. Bubulya, S.-Y. Chen, and L. Shemshedini, unpublished data.

FIG. 3. c-Jun enhances the N-to-C interaction of hAR.
A, schematic diagram of hAR fusion proteins used in mammalian two-hybrid system to measure the hAR N-to-C interaction. Note that the hAR(DE) is encoded by amino acids 624 -918 and hAR(AB) by amino acids 1-555. B, c-Jun potentiates the hAR N-to-C interaction. COS cells were transfected with 1 g of 17M-tk-CAT reporter plasmid and 5 g of GAL-hAR(DE), with or without 3 g of VP16-hAR(AB) and 1 g of c-Jun, as indicated. Note that gray bars are without and black bars with transfected c-Jun. 100 nM DHT was used as indicated. C, the activities of c-Jun mutants in hAR N-to-C interaction closely parallel their activities in hAR transactivation. COS cells were transfected with 1 g of MMTV-CAT reporter, 1 g of hAR, and 1 g of c-Jun plasmid (for transactivation) or 1 g of 17M-tk-CAT, 5 g of GAL-hAR(DE), 3 g of VP16-hAR(AB), and 1 g of c-Jun plasmid (for N-to-C interaction). The various c-Jun proteins were expressed from three different promoters (SV40, Rous sarcoma virus (RSV), and cytomegalovirus (CMV)). The RSV promoter drives the expression of mutants M8, M9, M14, Ala 63/73 , ⌬194 -223, ⌬146 -221, and v-Jun, whereas the CMV promoter does the same for mutants A 265 3 D In265, ⌬287-331, ⌬1-245, JunB, and JunD. 100 nM DHT was used in all cases. Note that, for both transactivation and N-to-C interaction, CAT activity is represented relative to activity of the first condition, which was activity without transfected c-Jun, and this was set to 1.

c-Jun Potentiates hAR N-to-C Interaction
tion partner c-Fos inhibits this activity (27). Utilizing several mutants of c-Fos, we have earlier suggested that heterodimerization of c-Fos with c-Jun blocks c-Jun's ability to enhance hAR-induced transcription (27). To determine if c-Fos would have the same negative effect on the c-Jun enhancement of hAR N-to-C interaction, transient transfections were performed measuring hAR transactivation and N-to-C interaction (Fig. 5). Wild-type c-Fos was able to block hAR N-to-C interaction, just as well as it can block hAR transactivation. The c-Fos mutants had similar activities in both hAR transactivation and N-to-C interaction, with mutant proteins containing a bZIP region having wild-type negative activity while those lacking a bZIP having significantly reduced activity. Thus, c-Fos, like c-Jun, displays the same activity on the hAR transactivation and N-to-C interaction functions.
Similar to c-Jun but unlike c-Fos, TIF-2 has been shown to mediate hAR activity (15,20). Therefore, not surprisingly, TIF-2 had a weak, but positive and reproducible, effect on hAR N-to-C interaction (Fig. 6). However, the TIF-2 activity was significantly weaker than that of c-Jun, such that a high amount (5 g) of TIF-2 expression plasmid was required to observe the effect (Fig. 6). Interestingly, elevating the concentration of c-Jun to 5 g resulted in attenuated levels of activity (Fig. 6), suggesting squelching of some accessory factor(s) (33,34).
c-Jun Potentiates hAR DNA Binding-Our transfection results above show c-Jun stimulates the DHT-dependent interaction between the hAR AB and DE regions (see Fig. 3). If this N-to-C interaction is required for hAR dimerization and subsequent DNA binding, then enhancement of this interaction should lead to elevated hAR DNA binding. To analyze this, we measured the in vitro DNA binding ability of hAR transiently expressed in COS cells in the absence or presence of transfected c-Jun (Fig. 7A). An extract from cells transfected with hAR exhibited weak, but significant, DNA binding (compare lane 1 with 5). The presence of hAR was confirmed by a supershift with an anti-hAR antibody (lane 3) but no supershift with a heterologous antibody (lane 4). Importantly, cell extract containing transfected hAR and c-Jun exhibited strongly enhanced hAR DNA binding (lane 2). The c-Jun effect cannot be explained by enhanced levels of hAR protein, because c-Jun had little, if any, influence on hAR expression (Fig. 7C). These results are consistent with the c-Jun activity on hAR N-to-C interaction and argue that c-Jun mediates hAR transactivation by enhancing receptor DNA binding through a positive effect on receptor homodimerization.
Mutations of AF-2 have been shown in previous studies (11,15,20) and in this study to disrupt hAR N-to-C interaction and, therefore, transactivation. To directly test the hypothesis that disruption of the N-to-C interaction will lead to reduced or no DNA binding, the hAR AF-2 mutant E896P was analyzed for in vitro DNA binding. Transfecting cells with calcium phosphate, as above, resulted in no detectable DNA binding by hAR(E896P). 3 To determine if increased expression might result in detectable binding, we used another transfection reagent (LipofectAMINE 2000, CLONTECH), which has been determined to increase protein expression levels from transfected COS cells by over 10-fold. 3 Even under these conditions, the E896P mutant exhibited no detectable DNA binding with or without transfected c-Jun (Fig. 7B, compare lanes 4 and 5 with 1). In contrast, wild-type hAR has high binding activity (compare lane 2 with 1). Interestingly, transfected c-Jun had no effect on hAR DNA binding when high levels of the receptor were expressed (compare lane 3 with 2). Note that the absence of binding activity by hAR(E896P) is not due to reduced protein expression (Fig. 7D). These results clearly demonstrate that a functional AF-2 is required for hAR DNA binding and provide a direct correlation between hAR N-to-C interaction and DNA binding.

DISCUSSION
Fusion of AR regions to the DBD of the yeast transcription factor GAL4 (35) has been used to identify and delineate the AFs of this receptor and to determine that AF-1 contains significantly more inherent transcriptional activity than AF-2 (31, 36 -38). In an earlier study, we used the same approach with the AB and E regions of hAR and determined that the c-Junpositive effect on hAR is targeted to the AB region (23). Truncation analysis of the hAR AB region led to the identification of amino acids 503-555 as necessary for the c-Jun activity and harboring an autonomous and c-Jun-responsive transcriptional activation function (23). In this report, we studied the importance of these amino acids in the full-length receptor. Deletion of amino acids 504 -555 was found to have no effect on hAR transactivation either in the absence or presence of transfected c-Jun. Although this result is seemingly surprising, it is consistent with published data (31,38) showing that the size and location of the amino-terminal AFs in hAR are variable, depending on whether the LBD is present or not. Another interesting aspect to come out of this present study is that, although deletion of amino acids 358 -555 completely abolishes hAR transactivation, significant activity is recovered when c-Jun is cotransfected. These data suggest that hAR amino acids 1-357 are sufficient for c-Jun potentiation of hAR transactivation when the LBD is present. In contrast, deletion of only amino acids 503-555 is sufficient for eliminating the c-Jun effect on the AB region when the LBD is absent (23). All these results taken together demonstrate that the transcriptional activity of the AR AB region is complex and argue that the AR has the capacity to use different parts of this AB region for transacti-3 C. J. Fisher and L. Shemshedini, unpublished data. vation, depending on the presence or absence of the LBD, the concentration of c-Jun and/or other coactivators, and specific type of androgen-responsive promoter (31).
Although the LBD found in the E region is clearly playing an important regulatory role in the transcriptional activity of the AB region, a demonstration of significant transcriptional activity residing within the E region, and specifically in the AF-2, remains elusive. It now appears that the main function of the AF-2 is to interact with the amino terminus and allow hAR dimers to form. This N-to-C interaction has been demonstrated for both AR (11)(12)(13) and estrogen receptors (39). Although most mutations in the AR ligand-binding domain (LBD) lead to a loss of ligand binding, Langley et al. (11) identified two mutations within the AR LBD, V889M and R752Q, that disrupt the N-to-C interaction without altering androgen binding affinity. Because both mutations are found in androgen insensitivity (reviewed in Ref. 40), these results argue that this medical condition may arise from a defect in the N-to-C interaction of AR. In view of this, the c-Jun-positive effect on hAR N-to-C interaction is likely to be physiologically relevant and responsible for the c-Jun enhancement of hAR transactivation. This is supported by our data with c-Jun and c-Fos mutants. Our earlier results showed that multiple regions of c-Jun, including the amino terminus and bZIP regions, were critical for the c-Jun effect on hAR transactivation (26). It now appears that these same regions are important for the c-Jun potentiation of hAR N-to-C interaction, because c-Jun mutants have the same activity in both hAR transactivation and N-to-C interaction. Similarly, the activities of c-Fos mutants in hAR N-to-C interaction parallel their activities in hAR transactivation, with mutants having a functional bZIP region repressing c-Jun en-hancement of both hAR activities and those lacking this region having no effect. These results with c-Jun and c-Fos mutants strongly argue that c-Jun enhances hAR transactivation by mediating receptor dimerization and that c-Fos can interfere with this c-Jun activity.
Homodimerization of AR and other steroid receptors is required for these transcription factors to bind DNA in a sequence-specific manner (reviewed in Ref. 1). Therefore, a factor that mediates this dimerization would be expected to enhance DNA binding. Our data with c-Jun support this, because this proto-oncoprotein is able to significantly enhance hAR DNA binding when it is cotransfected with the receptor. The hAR N-to-C interaction is dependent on a functional AF-2, which appears to be directly involved in associating with the amino terminus. As others have shown previously (11) and we show here, mutations that disrupt AF-2 function interfere with the N-to-C interaction. However, in this study we go a step further to provide a direct link between hAR N-to-C interaction and DNA binding, because the AF-2 mutation E896P disrupts both N-to-C interaction and in vitro DNA binding. Moreover, hAR (E896P) is completely inactive in transcription, even with an androgen concentration (100 nM) above the physiological range. This differs from an earlier study in which other mutations in the AR LBD were determined to also disrupt both hAR N-to-C interaction and transactivation, but the effect on transactivation could be overcome with 100 nM DHT (11). It appears that that residue E896, which is highly conserved among nuclear receptors (32), is critical for hAR transactivation because of its possible role in the N-to-C interaction. In view of this and the c-Jun enhancement of hAR N-to-C interaction, it was not surprising to discover that c-Jun was unable to up-regulate DNA binding, N-to-C interaction, or transactivation by the E896P mutant.
The AR N-to-C interaction is mediated not only by c-Jun but also by several nuclear receptor coactivators, including SRC-1, TIF2, and CBP (15,20,21). All three coactivators have been shown to interact with the AR AF-2, and SRC-1 was recently reported to also interact with the AF-1 (41). Because our data suggest that c-Jun is acting via the AB region (23), it is possible that c-Jun-and AF-2-interacting coactivators function together by targeting different regions of the receptor. Our previous data with c-Jun and TIF2 support this hypothesis, because these two proteins have additive effects in hAR transactivation (23). The role of SRC-1 or CBP has yet to be fully examined, although it is interesting to note that both of these proteins have been shown to interact with and mediate the transcriptional properties of c-Jun on AP-1-responsive promoters (42,43). Preliminary data suggest that neither CBP nor the related FIG. 6. Both c-Jun and TIF2 act on the hAR N-to-C interaction. COS cells were transfected with 1 g of 17M-tk-CAT, 5 g of GAL-hAR(DE), 3 g of VP16-hAR(AB), and 1, 3, or 5 g of either c-Jun or TIF2. 100 nM DHT was used in all cases. Note that CAT activity is represented relative to first condition, which was transfected GAL-hAR(DE) and VP16-hAR(AB), and this was set to 1.
FIG. 5. c-Fos inhibits the N-to-C interaction of hAR. COS cells were transfected with 1 g of MMTV-CAT reporter, 1 g of hAR, and 1 g of c-Jun (for transactivation) or 1 g of 17M-tk-CAT, 5 g of GAL-hAR(DE), 3 g of VP16-hAR(AB), and 1 g of c-Jun (for N-to-C interaction) with or without 3 g of c-Fos expression plasmid. 100 nM DHT was used in all cases. Note that, for both transactivation and N-to-C interaction, CAT activity is represented relative to first condition, which was transfected c-Jun together with either hAR (transactivation) or GAL-hAR(DE) and VP16-hAR(AB) (N-to-C interaction), and this was set to 1.
c-Jun Potentiates hAR N-to-C Interaction protein p300 significantly affect the c-Jun coactivation functions on hAR. 2 How could c-Jun mediate the hAR N-to-C interaction? c-Jun may be acting as a bridging factor between the hAR amino and carboxyl termini, as it is has been suggested for p160 coactivators (19). Support for this mechanism of c-Jun action comes from Sato et al. (44), who have reported that c-Jun can physically associate with hAR. Using a similar immunoprecipitation approach in transient transfection studies, we have preliminary data to support this finding. 3 However, this possible c-Jun⅐hAR complex appears to be very weak, because only a small fraction of c-Jun is found to interact with hAR 3 . Furthermore, our finding in this report that c-Jun can increase the intensity of the hAR⅐DNA complex without affecting its mobility in in vitro DNA binding studies demonstrates that c-Jun is not found in hAR when this receptor is bound to DNA. With respect to other AR coactivators, even those shown to physically interact with AR, there is no evidence reported to suggest that the interaction occurs on DNA (17)(18)(19)(20)(21)(45)(46)(47)(48)(49)(50). Thus, it is likely that proteins that enhance hAR N-to-C interaction, and therefore are presumed to potentiate DNA binding, perform this activity by engaging in short-lived protein⅐protein interactions with the receptor. Irrespective of what the precise mechanism of c-Jun action is, it is clear from all of the data accumulated thus far that the AR N-to-C interaction is a target of several different kinds of proteins that mediate hAR transactivation. Interestingly, this interaction is also affected by proteins that inhibit hAR activity. In this study, we provide evidence that c-Fos represses hAR activity by blocking the c-Jun potentiation of hAR N-to-C interaction. We have also obtained data to suggest that p53 exerts its negative effect on the expression of prostate-specific antigen by inhibiting hAR N-to-C interaction and DNA binding (51). All these data are consistent with the hypothesis that accessory factors can up-or downregulate AR transcriptional activity by influencing the N-to-C interaction, which directly controls the transcriptional properties of this receptor. FIG. 7. c-Jun enhances hAR DNA binding by wild-type hAR but not by the AF-2 mutant E896P. COS cells were transfected with 5 g of either hAR or hAR(E896P) with or without 5 g of c-Jun using either calcium phosphate (A) or Li-pofectAMINE 2000 (B) and treated with 100 nM DHT. Nuclear extracts were tested for hAR DNA binding using a gel mobility shift assay. Antibody supershift analysis was done with either an anti-hAR or anti-Birch antibody as a control. The same nuclear extracts were subjected to Western blot analysis using an anti-hAR antibody (C and D). Note that S is specific binding, SS is supershift, NS is nonspecific binding, and FP is free probe.