A Nuclear Action of the Eukaryotic Cochaperone Rap46 in Downregulation of Glucocorticoid Receptor Activity

Simian kidney COS-7 cells and human mammary MCF-7 cells were cultured in DME supplemented with 10% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C in an atmosphere of 5% CO₂. For experiments involving the use of the human mineralocorticoid receptor (MR), the cells were kept in DME containing 3% charcoal treated FCS to remove residual steroids.

The expression vector containing the wild-type human GR and MR have been described by Giguère et al. 1986 and Arriza et al. 1987. The recombinants GMM, GGM, and GGΔ have been described by Arriza 1991. The plasmid pcDNA3RAP46 was constructed by insertion of the coding sequence of human RAP46 (Zeiner and Gehring 1995) into the plasmid pcDNA3. The expression plasmid pT2 contains sequences encoding two copies of the hemagglutinin influenza (HA)-tag 5′-AA GCT TCC ACC ATG ATC TTT TAC CCA TAC GAT GTT CCT GAC TAT GCG GGC TAT CCC TAT GAC GTC CCG GAC TAT GCA GGA TCT ACT CGA GAG GAT CCG GTA CCT ATC TAG A-3′ cloned between the HindIII and XbaI sites of the plasmid pcDNA3 (Invitrogen Corp.). The constructs pT2RAP46 and pT2RAP46ΔC47 were constructed by inserting blunt-ended BamHI/XbaI fragments from pcDNA3RAP46 and pcDNA3RAP46ΔC47 (lacking the last 47 COOH-terminal amino acids) into XhoI and XbaI sites of pT2 after a fill-in reaction with DNA polymerase I (Klenow fragment). The constructs pT2RAP46Δ40 and pT2RAP46Δ70, containing deletion mutants of the first 40 and 70 NH₂-terminal amino acids of RAP46, were constructed by PCR amplification with XhoI and BamHI ends and insertion into the corresponding sites of pT2. The construct p46N69GFP, encoding the first 69 NH₂-terminal amino acids of RAP46 fused to the NH₂ terminus of the green fluorescence protein (GFP) sequence, was obtained by subcloning a HaeIII/PvuII fragment from pT2RAP46 into the blunt-ended HindIII site of pcDNA3.1hGFP. The pcDNA3.1hGFP construct contains sequences encoding the humanized GFP (CLONTECH Laboratories, Inc.) between the HindIII and XbaI sites of pcDNA3.1 (Invitrogen Corp.). The reporter plasmid pGL3MMTV encodes the firefly luciferase gene under the control of the mouse mammary tumor virus (MMTV) long terminal repeat region (−241 to +137), cloned as a BamHI/BglII fragment from the plasmid pHCwt CAT (Kaspar et al. 1993). The reporter plasmid −517/+63 Coll-Luc already has been described by Heck et al. 1997 and Schneikert et al. 1996. The control plasmid for transfection pTKRenilla Luc was obtained from Promega and the expression vectors GR–GFP and GFP–MR have been described by Carey et al. 1996 and Fejes-Tóth et al. 1998. The GFP plasmids GRΔ491-515GFP and GR1-515GFP were constructed by exchanging the GR sequence in the construct GR–GFP (Carey et al. 1996) with the GRΔ491-515 and GR1-515 sequences (Hollenberg et al. 1987).

100,000 MCF-7 cells were transiently transfected by the calcium phosphate coprecipitation procedure in 3.4-cm-diam plates with 2 μg reporter plasmid, 0.1 μg pTKRenilla Luc, and 2 μg expression vector. After 5 h of incubation with DNA and 2 min shock with 10% glycerol in PBS, the cells were further incubated for 36 h. Thereafter, they were harvested and cellular extracts were prepared for luciferase assay as described by Schneikert et al. 1996. Renilla luciferase assay was performed with the dual luciferase reporter assay system from Promega, according to the manufacturer's instructions. COS-7 cells were transfected by electroporation as previously described by Klocker et al. 1992 with 16.5 μg DNA containing 2 μg reporter plasmid, 0.5 μg pTKRenilla Luc, 7 μg wild-type or mutant RAP46, and 7 μg receptor expression vector.

COS-7 cells were transiently transfected with either empty expression vectors or expression vectors encoding GR, RAP46, and RAP46Δ70. After 36 h incubation in DME containing 3% charcoal treated FCS, the cells were trypsinized and resuspended in the same medium. 200,000 cells in 100 μl medium were incubated for 90 min at room temperature with different concentrations (0–100 nM) of [³H]dexamethasone (81.0 Ci/mmol; Nycomed Amersham Inc.) in the absence and presence of a 1,000-fold M excess of unlabeled dexamethasone. After incubation, the cells were washed three times with 1 ml PBS and recovered by centrifugation (3,800 g for 3 min). The cell pellet was then lysed in 5 ml scintillation fluid and the radioactivity determined by liquid scintillation counting.

Immunofluorescence experiments were performed as previously described by Herscovics et al. 1994 and the photographs were taken with an LSM 410 invert Zeiss confocal microscope.

Electrophoretic mobility shift and immunoblots assays were performed as described previously by Gast et al. 1995.

RAP46 is a protein initially isolated by virtue of its association with the GR in an interaction-screening assay (Zeiner and Gehring 1995). Since then, association of RAP46 and the GR has been shown in vitro in a glutathione S–transferase pull-down assay (Kullmann et al. 1998). To determine whether these two proteins associate in vivo, we performed immunofluorescence experiments with RAP46 and the GR in the absence and presence of hormone. To demonstrate the specificity of these interactions, control experiments were carried out with the MR, a structural and functional homologue of the GR (Arriza et al. 1987).

When expressed in COS-7 cells, most of the RAP46 protein resided in the cytoplasm as determined by laser confocal microscopy. Antibodies that recognize COOH-terminal epitopes or the HA-tag on this protein clearly showed cytoplasmic localization of RAP46 (Fig. 1A 1 and C 1; and results not shown). In certain regions of the cytoplasm, RAP46 colocalized with the unliganded GR or MR (Fig. 1, see the yellow color in A 2 and C 2). In the absence of hormone, the MR was identified in the cytoplasm, as well as in the nucleus (Fig. 1C 3), in agreement with the results of Fejes-Tóth et al. 1998. In the presence of ligand when the GR and MR are translocated from cytoplasm into the nucleus, RAP46 was transported with the GR (Fig. 1, see yellow staining in the nucleus of B 2), but not with the MR (Fig. 1D 2), showing a specific in vivo association of GR and RAP46. Note that cells containing only RAP46, but not the GR, do not translocate into the nucleus in the presence of dexamethasone (Fig. 1, compare A 1 with B 1).

Through its association with the GR and MR in the cytoplasm, RAP46 may alter the ligand binding properties of these receptors in line with its cochaperone activity (Stuart et al. 1998). We therefore analyzed the hormone binding properties of these receptors in the presence of RAP46 in whole cells and in cytosol preparations. These studies revealed only a slight change in the ligand binding activity of the receptors in the presence of RAP46. For example, Scatchard plot analysis showed an insignificant change in the dissociation constant (K_d) of the receptor for dexamethasone from 20 to 17 nM in the presence of RAP46 (Fig. 2, right). A slight (20%) reduction in the maximum hormone binding capacity (B_max) of the receptor was also detected (Fig. 2, left). This downregulation of the B_max is minimal compared with our previous report of a strong RAP46-mediated inhibition of transactivation by the GR (Kullmann et al. 1998). Thus, RAP46 must exert its negative regulatory function at other stages in the action of the GR.

In the liganded state, when both the GR and MR are in the nucleus, RAP46 downregulated the transactivation function of only the GR, but not that of the MR (Fig. 3a and Fig. B; note that GGG and MMM refer to GR and MR). This repression was independent of whether dexamethasone, cortisol, or aldosterone was the activating ligand (Fig. 3 B). The lack of RAP46 effect on transactivation by the MR is possibly due to the different cellular localizations of these two proteins in the presence of ligand (Fig. 1D 2; note nuclear and cytoplasmic localization of MR and RAP46). We therefore repeated the transfection experiments with the MR and the RAP46 isoform BAG-1L, which is constitutively localized in the nucleus (Froesch et al. 1998; our unpublished results). In this study, transactivation by the MR was not repressed by BAG-1L, although the response of the GR was inhibited (results not shown; see Fig. 7). This indicates a fundamental difference in the action of RAP46 towards the MR and the GR.

GR–MR chimeric constructs were therefore used to identify the regions of the GR necessary for repression by RAP46 (Fig. 3 A). Two chimeric GR–MR constructs (GMM and GGM), as well as a truncated GR construct (GGΔ), were used in these analyses. Transactivation by the chimeric MR receptor (GMM) containing the NH₂-terminal sequence of the GR (Arriza 1991) in the presence of either aldosterone, dexamethasone, or cortisol was not repressed by RAP46 (Fig. 3 B). This may be due to the absence of the hinge region (amino acids 491–515) of the GR that we previously identified as an interaction domain for RAP46 and demonstrated its necessity for RAP46-mediated inhibition of transactivation by the GR (Kullmann et al. 1998). The presence of this domain in the chimeric construct GGM still did not allow RAP46 to repress the transactivation function of this mutant receptor (Fig. 3 C), demonstrating the need for the presence of the hormone binding domain (HBD) of the GR. This finding was further confirmed by the use of the truncated receptor (GGΔ), which contains the interaction domain, but lacks the HBD of the GR. In cotransfection experiments, this mutant GR constitutively transactivated a glucocorticoid responsive gene construct, but its activity was not repressed by RAP46 (Fig. 3 C). Further control experiments with chimeric receptors MGG and MMG were performed. Transactivation by MGG containing both the hinge region and HBD of the GR was repressed by RAP46, but not transactivation by MMG containing only the HBD of the GR, but not the hinge region of the GR (results not shown). These results, together with our previous findings on the hinge region (Kullmann et al. 1998), demonstrate that this region, together with the HBD of the GR, are both required for RAP46-mediated inhibition of transactivation by the GR.

To further determine the role of the HBD in the repression of the transactivating function of the GR by RAP46, we asked whether this region is involved in the GR-mediated recruitment of RAP46 into the nucleus. In laser confocal immunofluorescence experiments, we showed that the GGΔ mutant receptor lacking the HBD is permanently localized in the nucleus in the absence and presence of hormone, as already reported (Hollenberg et al. 1987). However, in >75% of the cells, RAP46 was colocalized with the receptor in the nucleus (Fig. 4 A). Thus, the inability of RAP46 to repress transactivation by GGΔ is not due to the preclusion of this protein from the nucleus. Rather, it must be due to the absence of an important function in the HBD required for the negative action of RAP46.

To find out how RAP46 is recruited into the nucleus by the GR, we used the mutant GR with a deletion of the hinge region (amino acids 491–515) in the confocal immunofluorescence experiments. In this study, both RAP46 and the mutant receptor were localized in the cytoplasm in the absence of hormone. But, in the presence of dexamethasone, the ability of the GR to recruit RAP46 was severely compromised (Fig. 4 B). Only ∼50% of the cells containing the mutant receptor and RAP46 showed colocalization of the two proteins in the nucleus (Fig. 4 B). In the remaining cases, RAP46 was localized in the cytoplasm while the liganded receptor was translocated into the nucleus (Fig. 4 B). This demonstrates that the hinge region of the GR is indeed involved in the transport of RAP46 into the nucleus, although this process could be aided by other regions of the receptor. These regions are most likely contained in the HBD since its deletion, as shown above, slightly decreased nuclear transport of RAP46.

As RAP46 does not drastically alter the hormone binding properties of the GR in the cytoplasm, and it is translocated into the nucleus by the GR, its major site of action must be in the nucleus. Previously, we have shown that it inhibits the DNA binding activity of the receptor (Kullmann et al. 1998), but the mechanism involved is not known. We therefore deleted the NH₂-terminal sequences of this protein encompassing the sequence motif [EEX₄]₈ (see underlined sequences in Fig. 5 A) to determine the effect of the truncated proteins on DNA binding by the GR and GR-mediated transactivation at the MMTV promoter.

While RAP46 inhibited DNA binding by the GR, mutants with truncation of the first 40 or 70 NH₂-terminal amino acids, eliminating five or all eight of the repeat-motif, partially or completely abolished this negative effect of RAP46 on DNA binding, as well as transactivation by the GR (Fig. 5 B). Control immunoblot experiments with extracts of the transfected cells showed that the GR and RAP46 constructs were adequately expressed in this experiment (Fig. 5 B). Note that the apparent complete inhibition of DNA binding by the GR in Fig. 5 B is due to the very short autoradiographic exposure time of the gel. Longer exposures showed that DNA binding was drastically repressed, but not completely abolished (results not shown).

The two RAP46 NH₂-terminal mutants behaved as the wild-type RAP46 in several experiments. This was demonstrated by RAP46Δ70, which slightly decreased the maximum hormone binding capacity of the GR without significantly affecting the K_d for dexamethasone, as we have reported already (Fig. 2, compare K_d = 20 nM with 19 nM). In the presence of ligand, this mutant, as in the case of the wild-type RAP46, was also recruited into the nucleus by the GR (Fig. 5 C, compare A 1 with B 2). However, unlike the wild-type RAP46, it did not downregulate DNA binding by the GR, nor did it repress GR-mediated transactivation at the MMTV promoter (Fig. 5 B). Thus, COOH-terminal sequences of RAP46 are required by the GR for the recruitment of this protein into the nucleus.

To determine the sequences at the COOH terminus necessary for the recruitment, we used a mutant RAP46 lacking the last 47 amino acids known to bind hsp70/hsc70 in immunofluorescence experiments (Froesch et al. 1998). This mutant, unlike the NH₂-terminal deletion construct, was not transported into the nucleus (Fig. 5 D), indicating that the extreme COOH-terminal sequence of RAP46 is required to get this protein into the nucleus. Interestingly, deletion of this COOH-terminal sequence also abolished the repressive action of RAP46 on DNA binding by the GR (results not shown). This demonstrates the necessity for nuclear transport of RAP46 for its negative regulatory activity. These results, together with our findings from the experiments with the NH₂-terminal deletion mutant RAP46Δ70 (Fig. 5 C), demonstrate that although nuclear transport of RAP46 is necessary, it is not sufficient for inhibition of DNA binding by the GR. This function requires the NH₂ terminus of RAP46.

To determine the contribution of the NH₂ terminus of RAP46 to downregulation of GR activity, we fused the first 69 amino acids to GFP and asked whether such a construct was sufficient in repressing GR activity. In immunofluorescence experiments, this fusion protein was found both in the cytoplasm and the nucleus (results not shown). In transactivation studies carried out in GR- and RAP46-positive human mammary tumor MCF-7 cells (Cato et al. 1986; Yang et al. 1998), overexpression of RAP46 inhibited transactivation by the GR, as we had previously reported (Kullmann et al. 1998; Fig. 5 B). However, overexpression of GFP alone had no effect on GR response, whereas the N69GFP construct increased transactivation by the endogenous GR (Fig. 6 A). This shows a dominant–negative effect of the N69GFP protein on the repressive action of endogenous RAP46 in GR-mediated transactivation. This dominant–negative effect was further demonstrated in DNA binding studies involving the GR. In this assay, the N69GFP fusion protein, but not GFP alone, abolished the repressive action of RAP46 on DNA binding by the GR (Fig. 6 B, compare lanes 2–4 with lanes 5–7). These studies confirm that RAP46 uses its NH₂-terminal sequence for the repression of GR activity, possibly through interaction with other factors since its negative effect could be competed by an excess of the N69 sequence fused to GFP.

To further determine the significance of the [EEX₄]₈ motif in the negative action of RAP46, we examined the effect of the RAP46 isoform BAG-1L (which also carries this sequence motif) on DNA binding by the GR and transactivation by this receptor at the MMTV promoter. In these experiments, both RAP46 and BAG-1L negatively regulated the DNA binding and transactivation functions of the GR to the same extent (Fig. 7). BAG-1, the isoform of RAP46 that lacks a considerable portion of the repeat motif, had no effect, as we have already shown for its closely related homologue RAP46Δ40 (unpublished results; see also Fig. 5 B).

Since the nuclear action of the GR involves both positive and negative regulation of gene expression (Beato et al. 1995; Cato and Wade 1996; McEwan et al. 1997), we examined the effect of RAP46 and BAG-1L on the negative regulatory activity of the GR. This was achieved by measuring the ability of the GR to repress the expression of a human collagenase I gene construct induced by the phorbol ester 12-O-tetradecanoyl phorbol 13-acetate (TPA). This study showed that transrepression was unaffected by RAP46, whereas BAG-1L only slightly (30%) abrogated this function of the GR (Fig. 7). These effects remained unchanged, even when different amounts of the two constructs were transfected (results not shown). These findings show that the transactivating and transrepressing functions of the GR are regulated differently by RAP46. They further demonstrate that the major nuclear action of RAP46, and its isoform BAG-1L, on GR activity is the repression of the transactivation function of this receptor.

hsp70 and hsp90 provide the GR with the proper conformation for ligand binding and, therefore, play an important role in the ligand-mediated activation of this receptor (Hutchinson et al., 1994; Dittmar and Pratt 1997; Dittmar et al. 1998). In this study we have shown that RAP46, a protein that binds hsp70 to inhibit its ability to refold denatured proteins (Gebauer et al. 1997, Gebauer et al. 1998), downregulates transactivation by the GR. This negative regulation does not involve inhibition of the ability of the receptor to bind hormone. In hormone binding studies, RAP46 did not alter the K_d of the receptor for its ligand. We therefore concluded that the negative regulation of GR activity by RAP46 must occur through other mechanisms.

We demonstrated by immunofluorescence experiments that in the presence of ligand, RAP46 is recruited from the cytoplasm into the nucleus by the GR, but not by the MR. As RAP46 binds hsp70/hsc70, the nuclear transfer could occur via this heat shock protein, especially since it is reported to be transported into the nucleus by both the GR and MR (Diehl and Schmidt 1993; Srinivasan et al. 1994; Bruner et al. 1997). However, if RAP46 were to be transported via its interaction with hsp70/hsc70, it would be expected to be recruited into the nucleus by both receptors. Since only the GR, but not the MR, recruited RAP46 into the nucleus, it is very likely that the nuclear transport is mediated by a specific association of RAP46 with the GR.

Previously, we have shown in glutathione S–transferase pull-down experiments that the hinge region of the GR (amino acids 491–515) is necessary for the association of RAP46 with the GR. In those experiments, we showed that deletion of this region abolished RAP46-mediated downregulation of transactivation by the GR (Kullmann et al. 1998). In the present study, we have shown that deletion of the hinge region partially decreased the ability of RAP46 to be transported into the nucleus by the GR. This means that the hinge region has at least two functions. First, it contributes to nuclear recruitment of RAP46 and second, it also plays a role in inhibition of GR-mediated transactivation. As the hinge region of the GR differs considerably in sequence from that of the MR, it could explain why RAP46 is not translocated into the nucleus by the MR and why transactivation by this receptor is not repressed by RAP46.

In our study to find out which region of RAP46 is required for recruitment into the nucleus, we used a mutant that lacks the last 47 COOH-terminal amino acids known to bind hsp70/hsc70 (Froesch et al. 1998). This protein was not translocated into the nucleus by the GR, implicating the hsp70/hsc70 binding site in the nuclear transport of RAP46. Thus, binding of hsp70/hsc70 to RAP46 may be involved in the nuclear transport and the heat shock protein must interact differently with the GR and MR to account for the differences observed in the recruitment of RAP46 into the nucleus by these two receptors. Alternatively, hsp70/hsc70 may not be necessary at all, but the same sequence used for binding the heat shock protein may be involved in a direct interaction with the GR. This would mean that RAP46 must compete with hsp70/hsc70 for binding to the GR. Our experiments so far do not distinguish between these possibilities.

We have shown that, although nuclear recruitment of RAP46 is necessary, it is not sufficient for negative regulation of GR activity. A RAP46 protein lacking the first 70 amino acid residues was transported into the nucleus by the GR, but nevertheless failed to inhibit transactivation by the receptor. This indicates that the inhibitory function of RAP46 requires its NH₂-terminal sequences. This region of RAP46 is possibly involved in interaction with other proteins, since a fusion protein containing this segment exerted a strong dominant–negative function and relieved the downregulatory action of RAP46 on transactivation by the GR. The first NH₂-terminal 70 amino acids of RAP46 contain a sequence motif [EEX₄]₈ that may contribute to the downregulation of transactivation by the GR since BAG-1L which contains this motif, but not BAG-1, also repressed GR-mediated transactivation. BAG-1, on the other hand contains two copies, rather than the eight copies of the [EEX₄] motif. Thus, whatever the function of this sequence, several copies, rather than a duplication, are necessary for the manifestation of its repressive function. The [EEX₄]₈ sequence is highly enriched in threonine and serine residues (27% of the total sequence). Our preliminary in vivo labeling studies showed that RAP46 is a phosphoprotein, with almost all its phosphorylated residues localized at the [EEX₄]₈ sequence in the first NH₂-terminal 70 amino acid residues (Schneikert, J., and A.C.B. Cato, unpublished). This finding is rather striking and highly suggestive of a role of phosphorylation in the action of the NH₂ terminus of RAP46.

Inhibition of GR-mediated transactivation by the corepressor calreticulin has been shown to be dependent on inhibition of DNA binding brought about by the interaction of this protein with sequences in the DNA binding domain of the receptor (Burns et al. 1994). Distinct sequences in the DNA binding domain of the GR are not recognized by RAP46 for the inhibition of transactivation by the GR. The site of interaction of RAP46 on the GR is the hinge region (Kullmann et al. 1998; present results). However, the HBD of the GR also contributes to the negative action of RAP46. From our studies, we conclude that the negative action of RAP46 is effected through a prior nuclear transport of this protein via a process that requires the COOH-terminal region (the hinge region and, to some extent, the HBD) of the GR, as well as the last 47 COOH-terminal amino acids of RAP46. In the nucleus, RAP46 inhibits DNA binding by the receptor through a mechanism that is not quite clear at the moment. The fact that the HBD of the GR binds hsp70/hsc70, which is also recognized by RAP46, and all three proteins are found in the nucleus, argues in favor of at least a tertiary protein complex in the negative regulatory action of RAP46. Other proteins may be part of this complex as well, especially protein binding to the NH₂-terminal region of RAP46.

We have demonstrated that the inhibitory action of RAP46 is specific for those actions of the GR requiring DNA binding. Thus, transactivation, which is dependent on DNA binding by the GR is repressed, but not GR-mediated repression of the activity of the transcription factor AP-1, which does not require DNA binding by the receptor (Heck et al. 1994). We cannot, however, rule out the possibility that the use of TPA for the activation of AP-1 in our experiments disrupts the RAP46 complex necessary for inhibiting GR action.

The biological significance of the negative regulatory action of RAP46 are manifold. We previously have shown that RAP46-mediated inhibition of transactivation by the GR correlates with inhibition of glucocorticoid induced apoptosis (Kullmann et al. 1998). We also observed that T cells that are resistant to glucocorticoid-induced apoptosis express relatively high levels of RAP46 (Kullmann et al. 1998). Conversely, conditions that downregulate the endogenous levels of RAP46, such as exposure of thymoma cells to the immunosuppressant rapamycin, enhanced the transactivation potential of the GR and reduced GR-mediated apoptosis (Kullmann et al. 1998). In the present study we have shown, by the use of a dominant–negative mutant construct, that the transactivation function of the GR in the human mammary carcinoma MCF-7 cells is negatively regulated by endogenous RAP46. As neoplastic cells generally express relatively high levels of RAP46 and BAG-1L (Takayama et al. 1998; Yang et al. 1998), our results on the inhibition of GR-mediated transactivation by these proteins may provide mechanistic explanations for the reported cases of glucocorticoid resistance in neoplastic transformation. This point would have to be taken into consideration in studies on glucocorticoid resistance in disorders such as acute lymphocytic and nonlymphocytic leukemia (Crabtree et al. 1981; Wells et al. 1981).

We thank G. Rechkemmer for allowing us to use the confocal laser microscope at the Bundesforschungsanstalt für Ernährung Karlsruhe and we are also grateful to T. Stefany of the European Molecular Biology Laboratories (Heidelberg) for teaching us how to use this microscope. We thank B. Groner for the gift of anti-GR antibody, J. Reed for the human BAG-1L construct, as well as M. Zeiner and U. Gehring for the RAP46ΔC47 construct. We also acknowledge with thanks the gift of GR–GFP and GFP–MR constructs from I.G. Macara and D. Pearce.

This work was supported in part by a grant to J. Schneikert from the Training and Mobility of Researchers program of the European Community.