RAP46 Is a Negative Regulator of Glucocorticoid Receptor Action and Hormone-induced Apoptosis*

RAP46 was first identified by its ability to bind the glucocorticoid receptor. It has since been reported to bind several cellular proteins, including the anti-apoptotic protein Bcl-2, but the biological significance of these interactions is unknown. Here we show that RAP46 binds the hinge region of the glucocorticoid receptor and inhibits DNA binding and transactivation by the receptor. We further show that overexpression of RAP46 in mouse thymoma S49.1 cells inhibits glucocorticoid-induced apoptosis. Conversely, glucocorticoid-induced apoptosis and transactivation were enhanced after treating S49.1 cells with the immunosuppressant rapamycin, which down-regulates cellular levels of BAG-1, the mouse homolog of RAP46. The effect of rapamycin can, however, be overcome by overexpression of RAP46. These results together identify RAP46 as a protein that controls glucocorticoid-induced apoptosis through its negative regulatory action on the transactivation property of the glucocorticoid receptor.

RAP46 was first cloned from a human cDNA expression library by virtue of its association with the glucocorticoid receptor (GR) 1 (1). Since then, it has been shown to bind several other proteins, although the functional significance of these interactions remains to be identified (2). A murine protein, BAG-1, with a high degree of homology to RAP46 was isolated independently by interaction cloning with the anti-apoptotic protein Bcl-2 (3) and as an interacting partner of the intracellular domain of the hepatocyte growth factor receptor (4). This clone lacked 55 amino acid residues at its N-terminal region, which made it arguable whether it was a homolog of RAP46. Meanwhile, a human BAG-1 protein with N-terminal sequences homologous to RAP46 has been cloned (5). More recently, further N-terminal sequences from the murine BAG-1 protein have been isolated by the 5Ј-rapid amplification of cDNA ends technique, indicating that larger transcripts of BAG-1 exist (6). From these studies, it was proposed that RAP46 is an in-frame initiation codon from an even larger mRNA, whereas the cloned murine BAG-1 protein is a partial sequence (6). Nevertheless, BAG-1 has been shown to be a multifunctional protein. (i) It binds the catalytic domain of the serine/threonine-specific protein kinase Raf-1 and activates this kinase in vitro (7). (ii) It also binds to the plasma membrane-associated receptors for hepatic growth factor and platelet-derived growth factor, thereby enhancing their ability to protect cells from apoptosis (4). (iii) BAG-1 further interacts with Bcl-2 and potentiates the anti-apoptotic function of this protein (3).
Recently, both RAP46 and BAG-1 were shown to function as molecular modulators of the chaperones hsp70 and hsc70 (2,8). They bound and interfered with the ability of these proteins to refold unfolded proteins (2,8). Thus, BAG-1 and RAP46 may be novel chaperone regulatory proteins linking signal transduction pathways with the cellular apoptotic process and steroid hormone action. We therefore examined the role of RAP46 in transactivation by the GR and in glucocorticoid-induced apoptosis.
The GR belongs to a class of ligand-binding transcription factors that play diverse roles in development, differentiation, and cellular proliferation (9). Members of this class contain an N-terminal modulator domain, a centrally located DNA-binding domain (DBD), and a hinge region that separates this domain from a carboxyl-terminal hormone-binding domain (9). Transcriptional regulation by these receptors requires additional regulators termed coactivators and corepressors (10,11). Coactivators bind mainly to a region known as AF-2 in the hormone-binding domain and enhance ligand-activated transcriptional activity of the receptors (10,11). Corepressors like N-CoR/RIP13 (nuclear receptor co-repressor/retinoid x receptor-␣ interacting protein 13) (12, 13), SMRT/TRAC (silencing mediator (co-repressor) for retinoid and thyroid hormone receptors/thyroid retinoic acid receptor-associated co-repressor) (14,15), and TRUP (thyroid hormone receptor uncoupling protein) (16) interact with the hinge region of members of the thyroid and retinoic acid receptor family and inhibit their activity in the absence of hormone (12,14). The GR also associates with a number of cofactors, including the Ada adaptor complex (17), 14-3-3 (18), calreticulin (19), and RAP46 (1), but not all of these interactions have been functionally analyzed.
In this study, we demonstrate that RAP46 interacts with the hinge region of the GR and down-regulates the transcriptional activity of this receptor. Overexpression of RAP46 inhibits glucocorticoid-mediated apoptosis in mouse thymoma cells, whereas down-regulation of the levels of the mouse homolog of RAP46 in the same cells enhances glucocorticoid-induced apoptosis. These findings identify RAP46 as a negative regulator that links the activity of the GR with the cellular apoptotic pathway.

MATERIALS AND METHODS
Plasmid Constructs-Wild-type and mutant GR vectors have been previously described by Hollenberg et al. (20), and the mutant A458T by Heck et al. (21). These constructs either were used as Rous sarcoma virus-based mammalian expression vectors (20,21) or were recloned into the plasmid pBAT (22) for in vitro transcription/translation reactions. The plasmid pHCwtCAT and pHCwtLUC constructs have been previously described (23,24). The recombinant plasmid GST-RAP46 was obtained by cloning the coding sequence of RAP46 in frame into the multiple cloning site of the vector pGEX-2T (Amersham Pharmacia Biotech). The constructs Gal4-DBD, Gal4-NFI/CTF1, and pHC8/17MX2 have been previously described (23). Gal4-RAP46 fusion protein-encoding plasmid was generated by cloning the RAP46 sequence in frame into pSG424 (the Gal4-DBD vector) (23). Androgen receptor expression vector has been previously described (25).
Cell Culture and Transfections-Human choriocarcinoma JEG-3 cells, human Jurkat cells, simian kidney COS-7 cells, and mouse thymoma S49.1 cells were all cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum at 37°C and in a 5% CO 2 atmosphere. All the culture media contained 100 units/ml penicillin and 100 g/ml streptomycin. Unless otherwise stated, transient transfection of JEG-3 and COS-7 cells was carried out by the calcium phosphate coprecipitation method. In this assay, the activity of the reporter gene was occasionally normalized by the inclusion of the plasmid pCH110 (Amersham Pharmacia Biotech) in the transfection mixture. This plasmid consists of a ␤-galactosidase coding sequence controlled by the SV40 promoter. Results obtained with this internal control were no different from those generated with equal amounts of cellular proteins. Cellular extracts for electrophoretic mobility shift assay (EMSA) were obtained from COS-7 cells transiently transfected with GR and RAP46 constructs with the use of electroporation as described previously (25). Stable and transient transfections in S49.1 cells were carried out by a DEAE-dextran method previously described for lymphoid cell lines (26). Briefly, 5 g of DNA/2 ϫ 10 6 cells was resuspended in 200 l of Tris-buffered saline (25 mM Tris-HCl (pH 7.4), 137 mM NaCl, 5 mM KCl, 0.7 mM CaCl 2 , 0.5 mM MgCl 2 , and 0.6 mM Na 2 HPO 4 ) containing 500 g/ml DEAE-dextran for 20 min at room temperature. The cells were then treated with 1% Me 2 SO for 3 min and thereafter washed twice with Tris-buffered saline and resuspended in culture medium. In some experiments, 1 g of Renilla luciferase expression vector was cotransfected to help quantify the efficiency of transfection. Chloramphenicol acetyltransferase and luciferase enzyme activity determinations were performed as described previously (27).
For stable transfection, the transfected S49.1 cells were selected for 3 weeks in medium supplemented with 1 g/ml puromycin. Puromycinresistant pools of cells were cloned by limiting dilution in 96-well plates and verified for expression of RAP46 by RT-PCR.
Northern Blot Analysis-Poly(A) ϩ RNA from ϳ10 7 cells was prepared and subjected to Northern blot analysis as described previously (28). The filters were hybridized with a randomly primed radioactively labeled 0.9-kilobase pair EcoRI fragment of RAP46 cDNA (1) and a 1.5-kilobase pair fragment of the human elongation factor 1␣ gene (29).
Apoptosis Measurements-The percentage of apoptotic cells was determined with the annexin V procedure (30) according to the manufacturer's instructions. Briefly, the cells were washed once with phosphate-buffered saline and annexin incubation buffer (10 mM HEPES/ NaOH (pH 7.4), 140 mM NaCl, and 5 mM CaCl 2 ). Thereafter, they were resuspended in 100 l of annexin incubation buffer containing 0.02 volume of fluorescein isothiocyanate-conjugated annexin V stock solution provided by the manufacturer and incubated for at least 15 min at room temperature. The cells were then analyzed by flow cytometry with a fluorescence-activated cell sorter (FACStar PLUS , Becton Dickinson) after diluting the samples to 500 l with annexin incubation buffer containing 1 g/ml propidium iodide. This latter treatment was important to distinguish cells that had lost membrane integrity. Only propidium iodide-negative cells were further analyzed.
Glutathione S-Transferase Pull-down Experiments-The production of GST and GST-RAP46 as well as the pull-down experiments were performed as described previously (27), with the exception that the in vitro translated products were made 3 M with urea and incubated for 30 min on ice. Thereafter, the concentration of urea was reduced to 1 M before binding to the glutathione-Sepharose beads. The urea treatment increased binding of the GR to RAP46. Identical results were obtained in the absence of urea, albeit with a lower binding efficiency.
EMSA and Immunoblotting-Preparation of whole cell extract and EMSA were performed as described previously (25).

RESULTS
To investigate how RAP46 influences GR activity, we examined its effect on transactivation by the receptor in a transient transfection assay. As recipient cells, receptor-negative JEG-3 cells that express moderate endogenous levels of RAP46 were transfected with GR and RAP46 expression vectors and an indicator construct. This construct consists of the mouse mammary tumor virus (MMTV) promoter driving the expression of a luciferase gene. As a negative control, an androgen receptor (AR) expression vector was used instead of the GR. The GR was activated by the synthetic glucocorticoid dexamethasone, and the AR by the androgen dihydrotestosterone.
Transactivation by the GR was inhibited by different amounts of transfected RAP46 (Fig. 1A, hatched bars) without any noticeable effect on the basal level of expression (data not shown). In contrast, RAP46 did not repress transactivation by the AR (Fig. 1A, open bars), indicating a specific negative regulatory action of this protein on GR function.
To determine whether RAP46 itself is transcriptionally active, we fused it to the DBD of the yeast transcription factor Gal4 and expressed the fusion protein with a GAL4 reporter gene in COS-7 cells. The activity of the GAL4 reporter gene was not affected, showing that RAP46 is transcriptionally inactive (data not shown). A different result was obtained when Gal4-RAP46 and the GR were expressed together with an MMTV indicator gene in which an NFI/CTF1-binding site next to the GR-binding sites had been replaced by Gal4-binding sites (23). In this case, Gal4-RAP46 compared with Gal4-DBD alone repressed the glucocorticoid response by 50% (Fig. 1B). In contrast, the transcription factor NFI/CTF1 linked to Gal4-DBD (Gal4-NFI/CTF1) enhanced glucocorticoid response (Fig. 1B), as we have previously reported (23). These results demonstrate that when physically close to the GR, RAP46 inhibits transactivation by the GR.
As transactivation is dependent on the ability of the GR to bind DNA, we investigated whether RAP46 interferes with this activity of the receptor. COS-7 cells were transfected with the GR by electroporation, and cellular extracts from the transfected cells were used for EMSA. These experiments showed that RAP46 drastically reduced the DNA binding activity of the GR (Fig. 2A, compare lanes 3 and 5). To clearly demonstrate the effect of RAP46, an anti-GR antibody that stabilizes GR-DNA interactions (21) was added to the reaction mixture. Even in the presence of this antibody, the negative effect of RAP on the DNA binding activity of the GR was still evident (Fig. 2A,  compare lanes 4 and 6). Extracts from cells transfected with the empty vector or with RAP46 alone did not bind DNA ( Fig. 2A,  lanes 1, 2, 9, and 10). Down-regulation of the DNA binding activity of the GR by RAP46 occurred in the absence of an altered receptor level as shown by immunoblots with the same cellular extracts used for the EMSA (Fig. 2B, compare lanes 3  and 4 with lanes 5 and 6). The increased level of the GR in lanes 7 and 8 arises from transfection of double the amount of the GR expression vector. Thus, negative regulation of DNA binding activity by RAP46 is one of the means used by this protein to inhibit transactivation by the GR.
The negative effect of RAP46 on transactivation and the DNA binding activity of the GR possibly occurs through an interaction of this protein with the GR. To determine this, we performed GST pull-down assays in which we used the wildtype GR and deletion mutants lacking the N-terminal transactivation region (⌬9 -385) (20), the DBD (⌬428 -490), or the hinge region (⌬490 -515) (20). The receptor constructs were labeled in vitro by translation and allowed to interact with GST or GST-RAP46 proteins immobilized on glutathione-Sepharose beads. In these experiments, the wild-type GR and the two mutants ⌬9 -385 and ⌬428 -490 preferentially interacted with GST-RAP46 as opposed to GST (Fig. 3A, lanes 6 -11), but not the hinge region deletion ⌬490 -515 (Fig. 3A, lanes 12 and 13). We therefore concluded that the hinge region (amino acids 490 -515) of the GR is the site of interaction with RAP46. These GST pull-down results were corroborated by results of transfection experiments in which deletion mutants and a point mutant of the GR were cotransfected with the MMTV luciferase indicator gene into JEG-3 cells. RAP46 down-regulated the activity of the wild-type GR and all the mutants, with the exception of the hinge region mutant ⌬490 -515 (Fig. 3B), despite the fact that the wild-type and mutant receptors were all expressed at identical levels (data not shown). Thus, the GST pull-down and transfection experiments together demonstrate the contribution of the hinge region of the GR to the RAP46mediated negative regulation of glucocorticoid action.
Since the functional activity of the GR is also required for glucocorticoid-mediated apoptosis, we investigated the effect of RAP46 on this process. To this end, we overexpressed RAP46 by stable transfection into thymoma S49.1 cells, and positive clones were analyzed for their ability to undergo apoptosis upon glucocorticoid treatment. All the positive clones expressed RAP46 as well as the GR, but were resistant to GR-induced apoptosis. These results have been demonstrated in Fig. 4 with a representative RAP46 expression clone (clone 1). The level of RAP46 expressed in the transfected clones was so low that it could only be detected by RT-PCR. In Fig. 4A, these results are shown with mRNAs derived from clone 1 as well as from a clone containing an empty expression vector and, as positive control, with mRNA from glucocorticoid-resistant Jurkat cells (31). Overexpression of RAP46 was detected in clone 1 and in the Jurkat cells, but not in the S49.1 cells containing the empty vector (Fig. 4A, lanes 1-3). Clone 1 also expressed the GR as demonstrated by an immunoblot assay (Fig. 4B). Nevertheless, upon treatment with dexamethasone, it did not undergo apoptosis as determined by flow cytometric measurements (Fig.  4C). This observation was also made with all the other RAP46expressing cells and the Jurkat cells (data not shown). Note that a smear was obtained in the RT-PCR with mRNA from mouse S49.1 cells that contain BAG-1 but no RAP46 sequences (Fig. 4A, lane 1). This is most likely due to a nonspecific amplification reaction. Since the RAP46 3Ј-primer used is homologous to sequences in BAG-1, amplification products may be obtained if the RAP46 5Ј-primer hybridizes nonspecifically to the BAG-1 sequence. Interestingly, Jurkat cells that are resistant to glucocorticoids express a relatively high level of RAP46. These cells probably express several isoforms of this gene since the RT-PCR products, compared with those obtained with the RAP46-transfected S49.1 cells, showed fragments with retarded electrophoretic mobility (Fig. 4A, compare lanes 2 and 3  with lanes 5 and 6).
The inability of RAP46-expressing cells to respond to glucocorticoid-induced apoptosis correlated with the inhibition of glucocorticoid-mediated cell cycle arrest by RAP46 (data not shown) and with the inhibition of GR-induced transactivation. Clone 1 and two other RAP46-expressing clones, but not the clone with the empty expression vector, failed to show a dexamethasone-induced transactivation of the MMTV promoter construct in a transient transfection assay (Fig. 4D). A cotransfected Renilla luciferase construct demonstrated that all the clones examined were transfectable (Fig. 4D, lower panel). This rules out differences in transfection efficiency as the cause for the lack of GR-induced transactivation in the RAP46-expressing cells. Thus, overexpression of RAP46 inhibits glucocorticoid-induced apoptosis and transactivation. The complete block of transactivation by the GR in the S49.1 cells by overexpression of RAP46 (Fig. 4D) strongly contrasts with the 80% inhibition in JEG-3 cells (Fig. 1A). The reason for this difference is unknown at the moment.
To further confirm the inverse correlation between increased expression of RAP46 and a reduced activity of the GR, we decreased the level of the mouse homolog of RAP46 (BAG-1) in S49.1 cells and expected an increase in GR-induced apoptosis and transactivation. This was done by treating the S49.1 cells with the immunosuppressant rapamycin, which is known to decrease BAG-1 levels (32). This treatment decreased the Bag-1 mRNA level in the S49.1 cells by 40% (Fig. 5A) and BAG-1 protein levels to the same extent (data not shown). At the same time, it produced a slight increase in GR-mediated transactivation (Fig. 5B), in agreement with reports of other investigators (33)(34)(35). Rapamycin treatment also enhanced the apoptotic signal of dexamethasone, although it had no effect on its own (Fig. 5C), as previously reported by Ishizuka et al. (33).
To confirm that the down-regulation of BAG-1 expression by rapamycin was directly responsible for the increase in glucocorticoid-mediated apoptosis and transactivation, we repeated the

FIG. 4. Overexpression of RAP46 protects S49.1 cells against dexamethasone-induced apoptosis and abolishes GR transactivation function.
A, shown is a 2% agarose gel containing the reaction products of a RT-PCR amplification with RAP46-and glyceraldehyde-3-phosphate dehydrogenase (GAPDH)-specific primers. The templates used were cDNAs synthesized from mRNAs of S49.1 cells transfected with an empty expression vector or with the RAP46 expression vector. As a control, cDNAs from Jurkat cells was also used. B, shown are the results from immunoblot assay of cellular extracts from untransfected S49.1 cells or S49.1 cells transfected with an empty or RAP46 expression vector and probed with the anti-GR antibody PA1-512. C, S49.1 cells stably transfected with an empty expression vector or with RAP46 were treated without hormone or with 10 Ϫ7 M dexamethasone for 48 and 70 h and assayed by flow cytometric analysis after staining by the annexin V-fluorescein isothiocyanate method. The panels indicate the percentage of cells that had undergone apoptosis. D, mouse S49.1 cells containing an empty expression vector or RAP46 expression vectors (clones 1-3) were transiently transfected with an MMTV luciferase reporter construct together with an expression vector coding for Renilla luciferase for determination of the efficiency of transfection. Sixteen hours later, the cells were treated without or with 10 Ϫ7 M dexamethasone (Dex) for 24 h. The cells were harvested, and cellular extracts were prepared for luciferase assay. Nine-tenths of the extracts were assayed for firefly luciferase activity (upper panel), and the other one-tenth for Renilla luciferase activity (lower panel). apoptosis experiments with S49.1 cells stably transfected with RAP46. In these experiments, we hoped that the overexpressed RAP46 would overcome the effect of rapamycin. As we expected, apoptosis in these cells was no longer induced by dexamethasone even in the absence of rapamycin (Fig. 6). Similarly, the cells were also resistant to transactivation by the GR (data not shown). These results together prove a negative regulatory effect of RAP46 and endogenous BAG-1 on GR-induced apoptosis and transactivation. DISCUSSION We have shown in this work that RAP46 represses transactivation by the GR. This was demonstrated by transient transfection experiments in human choriocarcinoma JEG-3 cells and simian COS-7 cells. Inhibition of GR transactivation activity was also observed in mouse thymoma S49.1 cells stably transfected with RAP46. The negative effect of RAP46 correlated with inhibition of GR-induced apoptosis. Conversely, rapamycin-mediated down-regulation of BAG-1, the mouse homolog of RAP46, enhanced transactivation by the GR and glucocorticoid-induced apoptosis.
Potentiation of glucocorticoid-induced apoptosis by rapamycin may occur through a number of processes since this drug interferes with several signal transduction pathways (for review, see Ref. 36). It was therefore necessary for us to demonstrate that the effect of this drug on glucocorticoid-induced apoptosis is linked directly with down-regulation of BAG-1 levels. We achieved this in experiments in which we analyzed the effect of rapamycin in S49.1 cells that overexpress RAP46. We showed that RAP46 not only abolished the effect of rapamycin on dexamethasone-mediated responses, but also inhibited the response mediated by dexamethasone alone. This suggests either that the level of the transfected RAP46 far exceeded the amount needed to overcome the repressed endogenous level of BAG-1 or, alternatively, that RAP46 may be functionally more potent than BAG-1. To distinguish between these two possibilities would require a direct comparison of the function of RAP46 and the full-length murine BAG-1 sequence in GR-mediated transactivation and GR-induced apoptosis.
Contradictory reports exist on how the GR contributes to apoptosis. SRG3, a member of the SW1⅐SNF complex of coactivators of the GR (37), has been reported to be necessary for glucocorticoid-induced apoptosis (38), implying that the transactivation function of the receptor is necessary for apoptosis. On the other hand, transrepressing activity of the GR has also been shown to be essential for glucocorticoid-induced apoptosis (31,39). In one study, the GR mutant LS-7, which transrepresses but does not transactivate, was used to prove that transrepression is essential for the apoptotic process (31). However, recent results show that the LS-7 GR mutant is not totally defective in transactivation (24). This makes it difficult to assess the exact functional requirement of the GR for apoptosis. In our study, although we showed that inhibition of GR-mediated apoptosis and transactivation are linked, we cannot rule out the involvement of the transrepressive function of the receptor in the apoptotic process.
RAP46 represses transactivation by the GR through interaction with the hinge region of the receptor. It is interesting to note that the hinge region is the site of interaction of other repressors of nuclear receptors such as N-CoR/RIP13 (12), SMRT/TRAC (14), and TRUP (16). However, no sequence homology exists between the hinge region bound by these corepressors and the region on the GR bound by RAP46. Furthermore, these cofactors differ from RAP46 in other aspects. N-CoR and SMRT interact with Sin3A/B and histone deacetylase 1 in the absence of ligand, causing histone deacetylation and transcriptional repression (40 -42). In the presence of hormone, they dissociate and allow other factors that cause histone acetylation to interact with the receptors to enhance transcription. In the case of RAP46, repression takes place in the presence of ligand, making it unlikely that recruitment of Sin3A/B and/or histone deacetylase 1 is involved. However, it has been shown that RAP46 associates with several proteins in an indirect manner via the molecular chaperones hsp70/hsc70 (2,8). As these proteins interact with the nonactivated and activated forms of the GR (43), hsp70-GR interactions may play an important role in the negative regulation of the activity of the GR by RAP46.
hsp70 also interacts with other steroid receptors, suggesting that RAP46 may regulate the activity of these receptors as well. Our studies show that down-regulation of GR activity by RAP46 is specific for this receptor as no negative effect of this protein was observed on transactivation by the AR. What then is the biological significance of the negative regulation of GR action by RAP46? In a number of developmental processes, the activity of glucocorticoids needs to be carefully controlled, and this may require the action of RAP46. For example, in Xenopus, ectopic expression and activation of the GR lead to inhibition of early differentiation of the embryo (44), whereas mice that do not express the GR at all die perinatally (45). These results demonstrate the importance of controlled expression and action of the GR during development. It is therefore at this stage that we expect RAP46 to exert its major influence on GR activity.
Control of GR action may not be restricted only to early development. GR-induced lymphocytolysis is a well known example of apoptosis that is likely to be regulated by RAP46 in adult organisms. In certain cultured cell lines, BAG-1, the mouse homolog of RAP46, is negatively regulated by glucocorticoids (46). Although this is not the case in our S49.1 cells, our finding that RAP46 down-regulates the activity of the GR in a number of cells implies the involvement of BAG-1/RAP46 in a cell type-specific feedback control of GR action.