Protein-Protein Interactions Are Implied in Glucocorticoid Receptor Mutant 465*-mediated Cell Death*

Previously we have shown that ICR-27, a clone of glucocorticoid-resistant human leukemic T cells, showed rapid cell loss upon transient transfection with plasmids expressing certain fragments of the human glucocorticoid receptor lacking the ligand binding domain. An extreme example was the frameshift GR mutant 465*, mutated after amino acid 465. This generated a novel 21-amino acid “tail,” beginning within the second zinc finger of the human glucocorticoid receptor DNA binding domain, a region required for ICR-27 cell death caused by hologlucocorticoid receptor plus hormone. The cell loss mediated by 465* was faster but quantitatively equivalent to that caused by hologlucocorticoid receptor plus hormone. We are therefore investigating the mechanism of action of 465*. We overexpressed 465* with or without a glutathione S-transferase tag fused to its N terminus and tested its ability to affect glucocorticoid response element (GRE)-driven reactions in vitro. Partially purified 465* showed little binding to a consensus GRE, caused virtually no stimulation of transcription from a GRE, and failed to inhibit GR-driven transcription. However, GST-465* “trapped” several proteins from ICR-27 cell extracts, including c-Jun; recombinant c-Jun also was bound in vitro. In co-transfection assays of CV-1 cells, 465* expression reduced phorbol ester (12-O-tetradecanoylphorbol-13-acetate) transcriptional activation from a promoter containing multiple AP-1 sites. Further studies proved the repressive activity of 465* was c-Jun-specific and not due to squelching artifacts. The data suggest that interaction of 465* with other proteins, such as c-Jun, might be responsible for its cell killing function.

Glucocorticoids exert their effects on target cells by binding to an intracellular GR 1 (1). The GR is required for glucocorticoid-evoked apoptosis of lymphoid cells (2)(3). Biochemical, im-munological, and genetic analyses of the GR imply that the protein folds into three distinct functional domains and several subdomains (4 -5). The N-terminal portion of the hGR, containing the major transactivation activity, is referred to as 1 (amino acids 77-262; Refs. 4 and 6). A central 66-residue DNA binding domain (DBD, amino acids 420 -486) comprises two zinc finger motifs and is followed by a "hinge region." The binding of two zinc atoms by two clusters of four cysteines helps stabilize a system of loops and helices responsible for sitespecific DNA binding and homodimerization (4,(7)(8). The steroid binding domain (amino acids 556 -777) lies at the C terminus. This region overlaps another transactivation domain, 2 (9 -12). In the course of transfection experiments aimed at mapping the regions of the hGR required for its lethal effect, we discovered that certain fragments of the receptor could reduce the number of viable test cells independently of added steroid (13). All of these hGR mutants lacked a large portion of their C termini, such that the entire LBD was missing. The hinge region of the hGR was removed to varying degrees in these mutants with no effect on their relative potencies in our assay. These mutants had been shown to have little or no transactivation activity of a GRE-driven reporter system, tested in cotransfection assays (14,15). An extreme case was mutant 465*, the product of a frameshift mutation at amino acid 465, within the second zinc finger of DBD. This results in retention of the amino acids that define the more amino-terminal zinc binding sequence but loss of part of the second "zinc finger" sequence from the wild type hGR DBD. Nevertheless, in our assay system, 465* was fully as active in reducing cell numbers as other C-terminal truncated hGR mutants that contained the entire DBD along with varying amounts of the hinge region. Therefore, we have focused our attention on the cell loss caused by transfection of 465*, further exploring possible mechanisms involved in the cell death process. ICR-27 cells were used to investigate the lethal effect of 465*. ICR-27 is a clone of cells selected for resistance to a high dose dexamethasone (Dex) from the Dex-sensitive cell line CEM-C7 (16). CEM-C7 cells contain two alleles for the hGR, one wild type and the other containing a point mutation (17). In ICR-27 cells, a deletion of the wild type gene has left only the ineffective mutant GR (17)(18)(19). Whole cell competitive ligand binding assays show few glucocorticoid binding sites in ICR-27 cells. However, transfection with an expression vector containing the whole coding sequence of the hGR gene restores glucocorticoid responsiveness to ICR-27 cells (13,20). When plasmids expressing 465* are introduced into ICR-27 cells, a rapid reduction in viable cell numbers is seen, which does not require added steroid.
Our in vitro studies showed that 465* could not bind to a consensus GRE, nor could it cause the stimulation of transcription of a GRE-driven reporter. We therefore hypothesize that protein-protein interactions are involved in the lethal effects of 465* and other GR mutants lacking their LBDs. Our results suggest that such interactions, rather than direct binding of 465* to GRE, are likely to be important in the observed biological effects of the GR fragment. As an example, we show that the 465* fragment interacts with c-Jun and interferes with its site-specific regulation of transcription.

MATERIALS AND METHODS
Cells and Cell Culture-The CCRF-CEM cell line was established from the peripheral blood of a child with acute lymphoblastic leukemia. CEM-C7, a clonal glucocorticoid-sensitive cell line, was isolated from uncloned CCRF-CEM cells by Norman and Thompson (16). By high dose Dex selection, ICR-27 cells, which only contain an inactive mutant hGR, were isolated (17). They were cultured at 37°C in RPMI 1640 medium, pH 7.4, with 5% heat-inactivated fetal bovine serum (Atlanta Biologicals, Norcross, GA) in a humidified atmosphere of 95% air, 5% CO 2 . Cells were maintained in the early to mid-log phase, where they displayed viability higher than 96%. Spodoptera frugiperda (Sf 9) cells were grown at 27°C in TNM-FH medium supplemented with 5% fetal bovine serum as described (21)(22). CV-1 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, pH 7.4, at 37°C under the conditions described for ICR-27 cells.
Plasmids-The 5XTRE-TATA-CAT construct was kindly provided by Dr. H. J. Rahmsdorf (Institute of Genetics, Karlsruhe, Germany). It contains five copies of a TPA response element (TRE) in sequence, upstream of an artificial TATA box followed by the reporter chloramphenicol acetyltransferase (CAT) (23)(24). The recombinant plasmids pRShGR␣ and pRSh465* were provided by Dr. S. M. Hollenberg (The Salk Institute, San Diego, CA). The pMSG-CAT reporter plasmid (Pharmacia Biotech Inc.) contains the mouse mammary tumor virus long terminal repeat (MMTV LTR) driving the CAT reporter gene. The ERE-CAT promoter-reporter construct holds the CAT reporter gene under the control of an estrogen response element (ERE). Both the plasmid and the human estrogen receptor (hER) expression vector pRShER were provided by Dr. Ming-jer Tsai (Baylor College of Medicine, Houston, TX).
Construction of pAcG2T/hGR, pAcG2T/465*, and pAcG2T/500*-hGR expression vector pVL941/hGR-1 was obtained by adding a BamHI site in front of the hGR translation initiation site in pVL941/hGR (25). The coding region of the hGR was obtained from the vector pVL941/ hGR-1 by BamHI digestion, separation by electrophoresis in agarose, and recovery from the agarose gel. After treatment with T 4 polymerase to create blunt ends, the fragment was ligated to the commercial baculovirus transfer vector pAcG2T (Pharmingen, San Diego, CA) after it had been cut with restriction enzyme EcoRI and blunt-ended by T 4 polymerase. The resulting pAcG2T/hGR plasmid contained the hGR in the correct orientation and open reading frame downstream from GST to create a GST/hGR fusion gene. Similarly, pAcG2T/500* and pAcG2T/ 465* fusion protein vectors were created by use of convenient enzymes and ligations to yield vectors that expressed hybrid GST-500* and GST-465* proteins. 500*-(1-500) is also a C-terminal truncated hGR frameshift mutant, made by cutting the hGR cDNA with EcoRI, removing the resultant small fragment, and religating. The mutation occurs at amino acid 500 followed by an extension of 5 novel amino acids. Thus, 500* contains the entire hGR DBD plus the hinge region.
Generation of Recombinant Virus-Each of the vectors, pAcG2T, pAcG2T/465*, pAcG2T/500*, or pAcG2T/hGR was co-transfected into Sf9 cells along with wild type Baculogold linearized baculovirus DNA (Pharmingen) by the calcium phosphate precipitation method (26). The recombinant viruses, expressing GST, GST-465*, GST-500*, or GST-hGR were isolated by two rounds of plaque purification (21). The recombinant virus of 465* without the GST tag was generated the same way, using the baculovirus transfer vector pVL941/465*, containing the coding region of 465*, plus the wild type Baculogold linearized baculovirus.
Expression of GST-465*, GST-500*, or GST-hGR-Sf9 cells were infected with the purified recombinant viruses, and 48 h after infection, the cells were harvested. Cytosolic fractions were prepared as described by Srinivasan and Thompson (25) and analyzed by immunoblot analysis for the expression of fusion proteins.
Purification of GST Fusion Proteins-Insect cell cytosol containing the expressed protein of interest was incubated with prewashed glutathione-Sepharose beads (Pharmacia) for 4 h at 4°C with continuous rocking. The beads were then washed 6 -8 times with cold phosphatebuffered isotonic saline, pH 7.4 (PBS). After elution in Tris buffer containing 20 mM glutathione (pH 8.0), the relevant protein was analyzed by electrophoresis on SDS-polyacrylamide gel stained with Coomassie Blue R-250. The beads with bound GST fusion protein were used for protein-protein interaction studies.
Gel Mobility Shift Assay (GMSA)-Deoxyribonucleotides containing consensus GREs 5Ј-AGGCTGTACAGGATGTTCTGC-3Ј and 5Ј-AG-GCAGAACATCTGTACAGC-3Ј were synthesized and annealed. The oligonucleotides were labeled with [ 32 P]dCTP by incubation with the Klenow fragment of DNase I at 37°C for 30 min. Affinity-purified GST-465* and GST-500* expressed in Sf9 cells were prepared as described above. Aliquots of 0.25 ng of [ 32 P]GRE probe were incubated with differing amounts of GST-465* or GST-500*, as indicated in the Fig. 2 legend, for 30 min at 25°C. After incubation, samples were subjected to electrophoresis and autoradiography as described by Srinivasan and Thompson (25).
In Vitro Transcription Assay-HeLa nuclear extracts were prepared and used in the in vitro transcription reaction essentially as described by Tsai et al. (27). Each 30-l reaction mixture contained 20 mM HEPES (pH 7.9), 8 -10% glycerol, 60 mM KCl, 6 mM MgCl 2 , 2 mM dithiothreitol, 0.5 mM CTP, 20 M UTP, 20 Ci of [␣-32 P]UTP (800 Ci/mmol), 0.3 mM 3Ј-O-methyl-GTP, 40 units of T1 RNase, 0.2 g of sonicated salmon sperm DNA, 5 l of HeLa nuclear extract (40 -60 g of protein), 10 ng of pADML200, 250 ng of pPRE2TATA, and various amounts of partially purified hGR and 465*, obtained by ammonium sulfate precipitation of extracts from appropriately infected Sf9 cells as described (26). Reactions were initiated by the addition of HeLa nuclear extract and incubated for 50 min at 30°C. After incubation, 25 g of tRNA and 120 l of 1.2 M ammonium acetate were added, and the reaction was terminated by extraction with phenol/CHCl 3 /isoamyl alcohol (25:24:1). The reaction products were precipitated with 70% ethanol at Ϫ70°C and dissolved in formamide buffer (98% formamide, 2% 0.5 M EDTA, pH 8.0, containing bromphenol blue and xylene cyanol). The reaction products were separated on a 4% polyacrylamide, 7 M urea gel by electrophoresis at 300 V for 4 h at 25°C. The gel was dried and exposed to x-ray film at Ϫ70°C for 24 h.
Metabolic Labeling of Cell Proteins-ICR-27 cells were washed twice with serum-free RPMI 1640 medium lacking methionine. After a further 1-h incubation at 37°C, the cells were collected by centrifugation at 100 -200 ϫ g for 5 min and cultured for 4 h at 1 ϫ 10 7 cells/ml in this same medium with the addition of 100 Ci/ml [ 35 S]methionine with a specific activity of 1261 Ci/mmol (ICN, Irvine, CA). Labeling was terminated by washing the cells twice in serum-free RPMI 1640 medium lacking methionine, collecting them by centrifugation at 100 -200 ϫ g for 5 min each at 25°C. Whole cell extracts were prepared (28) and precleared by incubation with GST immobilized on glutathione-Sepharose beads for 1 h at 4°C. After a centrifugation at 500 ϫ g at 4°C, the supernatant over the beads and debris were collected and incubated at 4°C for 2 h with GST or GST fusion protein adsorbed to glutathione-Sepharose beads. The unbound proteins were removed by washing the beads with PBS at 4°C and followed by centrifugation at 500 ϫ g. The bound proteins were eluted by adding the SDS sample buffer (0.12 M Tris, 4% SDS, 0.05% bromphenol blue, 20% glycerol) and immersing the tubes with the beads in a boiling water bath for 5 min. After centrifugation, the supernatant was collected and analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE).
Interaction of Recombinant c-Jun or Nuclear c-Jun with GST-465*-Purified recombinant c-Jun, homogeneous in that it gave only one band detected by Coomassie Blue R-250 staining on SDS-PAGE, was obtained from Promega (Madison, WI). Nuclear extracts from ICR-27 cells were prepared as described (29). The purified c-Jun and the ICR-27 nuclear extracts were incubated with GST or GST-465* immobilized on glutathione-Sepharose beads at 4°C for 2 h. The beads were washed and extracted as described above. The extracted proteins were separated by SDS-PAGE and transferred onto a polyvinylidene difluoride membrane, on which c-Jun was detected immunologically. The membrane was first blocked by incubation for 1 h at 25°C in PBS buffer containing 5% nonfat milk proteins and then incubated with rabbit polyclonal anti-c-Jun antibody (Santa Cruz Biotechnology) overnight at 4°C. After washing with PBS, the membrane was incubated for 1 h at 25°C with goat anti-rabbit horseradish peroxidase conjugate (Bio-Rad) and the appropriate substrates.
Transfection and Reporter Assay-Transfection of plasmid DNA into CV-1 cells was performed using the LipofectAMINE-mediated method as described by the manufacturer (Life Technologies, Inc.). The total amount of transfected DNA, containing 2 g of reporter plasmid, 1 g of a RSV-␤-galactosidase (internal control for transfection efficiency indication) expression plasmids, and various amounts of expression constructs indicated in the figure legends, was always adjusted to 8 g with vector alone. Transfected cells were treated with different drugs as indicated in the figure legends. Cells were collected by low speed centrifugation for CAT assays performed as described (30). All of the experiments were repeated 3-6 times. ␤-Galactosidase was determined spectrophotomerically by the hydrolysis of O-nitrophenol-␤-D-galactoside (31). Data concerning experimental variables were normalized to the levels of ␤-galactosidase expression.

Expression of Recombinant hGR-based Proteins in Sf9
Cells-The baculovirus system has been shown to be an effective vehicle for producing quantities of functional hGR for use in biochemical systems (22,25). Therefore, recombinant viruses were prepared that expressed GST alone or hybrid proteins with GST added to the amino terminus of the complete hGR (GST-hGR), or two hGR fragments 500*-(GST-500*) and 465*-(GST-465*). Mutant 500* consists of hGR amino acids 1-500 followed by an extension of 5 novel amino acids (indicated by the asterisk), hence it includes the entire, normal hGR DBD. Mutant 465* contains normal hGR amino acids 1-465, at which point the second zinc binding segment of the DBD is interrupted by a frameshift, resulting in an extension of 21 novel amino acids. Expression of each peptide was verified by examination of Sf9 extracts immunologically, after SDS-PAGE (Fig. 1). The blots show unique reactions with peptides of the predicted sizes. No cross-reacting proteins were detected in extracts from uninfected cells. Each peptide was also expressed without the GST tag.
GMSA detected binding of GST-500* but not GST-465* to a consensus GRE-The GR binds to the GRE directly and thereby modulates expression of certain genes, and the DBD of the GR is required for this direct GRE binding. Within the GR DBD, a helix stabilized by the coordinate binding of a zinc atom by 4 cysteines in the 5Ј region historically referred to as the first zinc finger is essential for GRE binding (32)(33). Since the sequence of the 465* suggests that its first zinc finger may be intact, it might still possess the ability to bind to GRE, although its second zinc finger sequence has been altered. To test for GRE binding, we compared interactions between the expressed GST-465* and GST-500* peptides and a consensus GRE, using the GMSA method. Recombinant GST-465* and GST-500* peptides were partially affinity-purified on glutathione-Sepharose beads. As shown in Fig. 2, only the GST-500* peptide caused a gel shift of the GRE oligomer (lanes 3 and 4). No gel shift of the GRE by peptide GST-465* was observed at the concentrations tested (lanes 1 and 2). The result with GST-500* indicated that the GST tag did not interfere with the GRE binding activity of 500*; hence, the failure to detect the interaction of GST-465* to the GRE is not likely to be due to interference by the GST-tag.

465* Caused Minimal Increase of GRE-driven Transcription in Vitro-
The GMSA data suggested that, at best, 465* bound weakly to the GRE. To test the effect of 465* peptide on GREspecific transcription, we performed in vitro transcription assays with crude 465* or hGR preparations from infected Sf9 cells. The salt-activated hGR in these preparations enhanced transcription of the 360-nucleotide product specifically driven by the tandem GREs in the template (Fig. 3, compare control  lanes 1 and 2 with lane 3). Preparations of peptide 465* increased the specific transcript only slightly at best (Fig. 3, lane  4). When up to 48-fold excess recombinant 465* was added to the reaction driven by hGR, there was no alteration in the specific transcription initiated by hGR alone (data not shown). Thus, Sf9 extracts containing 465* peptide did not interfere with hGR-mediated transcription in vitro. The reduction in the nonspecific transcript in hGR-mediated transcription compared with basal or 465* transcription (indicated by an asterisk in Fig. 3, lane 3) implies that hGR can stabilize the RNA polymerase preinitiation complex at a specific location to force the transcription to start from that site (nucleotide 360). However, since Sf9 cell extract and 465* peptide-containing cell extract did not have the factors that can stabilize the RNA polymerase preinitiation complex, the transcription started strongly from other sites rather than the specific sites, resulting in higher quantities of the nonspecific transcript (lanes 1 and 4, indicated by an asterisk).
Lack of Effect of 465* on Transcription from MMTV-CAT-Clearly, 465* was a poor transcription activator from the GRE in vitro; but to test its activity via the GRE, we examined its transactivation function in CV-1 cells. GRE-mediated transactivation was determined by use of a MMTV-CAT reporter construct containing the natural GREs of the MMTV LTR controlling a CAT gene reporter. Consistent with the in vitro transcription data, 465* exhibited minimal transactivation potency and did not alter induction of the MMTV-CAT reporter driven by the holo-GR (Fig. 4).
Peptides GST-465* and GST-500* Selectively Interacted with Proteins from ICR-27 Cells-Since GRE binding by the 465* peptide seemed unlikely to be involved in 465*-mediated ICR-27 cell death, we investigated the possible involvement of protein-protein interactions. A GST "trap" system was applied to see whether particular cell proteins interacted with GST, GST-465*, and GST-500*. Whole cell extracts of metabolically 35 S-labeled proteins from ICR-27 cells (Fig. 5, lane 4) were incubated with the GST fusion proteins immobilized on glutathione-Sepharose beads. The bound proteins were eluted and resolved by electrophoresis. As shown in Fig. 5, comparison of the pattern of labeled proteins in unfractionated cell extracts with those eluted from GST-500* and GST-465* beads shows that a subset of proteins selectively bound to the GR fragments. The GST control beads did not adsorb these proteins. Four major protein bands of molecular mass of 39, 46, 50, and 62 kDa were associated with both GST-465* and GST-500*, but not with GST per se (lanes 1, 2, and 3). These results suggest that GST-465* and GST-500* selectively trapped certain proteins from ICR-27 cells.
Both GST-465* and GST-500* Peptide Interacted with c-Jun in Vitro-The experiments above show that 465* has lost much of the GRE binding activity of the full GR DBD, although GST-465* interacts with certain proteins from ICR-27 cells in a fashion similar to the GST-500* fragment, which contains a complete DBD. Thus, protein-protein interactions might be important for the lethal function of such truncated forms of the GR, lacking their LBDs. We used c-Jun as a specific model to test this hypothesis, since c-Jun is well known to have interactions with holo-GR (14 -15) and is implicated in cell death processes (34). Purified recombinant c-Jun was added to the GST-fusion proteins immobilized on glutathione-Sepharose beads. After extensive washing of the beads, the bound proteins were eluted in SDS sample buffer and resolved by 10% SDS-PAGE (Fig. 6A). The results show that GST alone failed to retain the c-Jun (data not shown), but GST-465*, GST-500*, and GST-hGR effectively bound a protein the size of recombinant c-Jun (Fig. 6A, compare the recombinant c-Jun in lane  cJun with lanes 3, 6, and 9). No c-Jun was found in the supernatant fractions from the GST-465*, GST-500*, and GST-hGR

FIG. 3. 465* causes minimal increase in GRE-driven transcription in vitro.
The ability of hGR and 465* to enhance transcription in vitro from pPRE2TATA, a GRE-driven G-free cassette reporter gene system, was determined. Cytosol from uninfected Sf9 cells and a GREless basal promoter-driven construct pLovTATA, were used as negative controls (lanes 1 and 2). Correctly initiated transcript (360 nucleotides) was observed when pPRE2TATA was incubated with partially purified, salt-activated recombinant hGR (lane 3). Recombinant 465* peptide was incubated with pPRE2TATA, with the products shown (lane 4). An AdMLTATA-driven G-free cassette reporter containing a shorter G-free cassette (190 nucleotides) was used as an internal control in all lanes. M denotes size markers.

FIG. 4. Induction from MMTV LTR-CAT is seen by hGR plus
Dex but not by 465*. CV-1 cells were co-transfected with 2 g of the MMTV-CAT reporter plasmid pMSG-CAT, 1 g of RSV-␤-galactosidase as the internal control, and 2 g of pRShGR␣, pRSh465*, or the combination of both as indicated. After transfection, cells as indicated were treated with Dex (10 Ϫ7 M) for 48 h and harvested. Cell lysates were analyzed for CAT and ␤-galactosidase activity. After normalizing to ␤-galactosidase expression, the activity of CAT in the absence of Dex was assigned a value of 1, and the average -fold increase of the CAT activity after various treatments was plotted. Data are expressed as means Ϯ S.D. (n ϭ 3). beads (Fig. 6A, lanes 2, 5, and 8). Immunoblot data confirmed binding of GST-465*, GST-500*, and GST-hGR to recombinant c-Jun in vitro (Fig. 6B). This strongly suggested that 465* retained c-Jun binding activity akin to that of holo-GR or the 500* fragment, although the GRE binding property of 465* had been largely lost.
GST-465* Beads Trapped Native c-Jun in Extracts from ICR-27 Cells-The 465* peptide interacted with recombinant c-Jun in vitro, but whether it could trap native cellular c-Jun remained to be seen. To examine this question, GST-465* bound to beads was incubated with nuclear proteins extracted from ICR-27 cells. After exposure to the beads, no c-Jun was found in the supernatant over the beads or in the washes (Fig.  7B, lanes 1-4). The control beads, with GST only bound to them, failed to retain cellular c-Jun (Fig. 7A, lane 4). However, extracting the GST-465* beads after the incubation and washes showed that c-Jun had bound to them (Fig. 7B, lane 5). The results suggested that GST-465* can bind not only recombinant c-Jun presented in relatively high concentration but also the relatively rare native c-Jun in the mix of proteins extracted from ICR-27 cells. GST itself failed to interact with c-Jun.
GST-465* Did Not Interact with c-Myc in Vitro-We have shown that down-regulation of c-Myc precedes and closely correlates with glucocorticoid-evoked apoptosis of CEM cells (35)(36). We therefore also tested for interaction between the GST-465* peptide and c-Myc, using the GST-trap assay. The results (Fig. 7C) demonstrated that c-Myc in ICR-27 nuclear extracts was not retained on GST or GST-465* beads. All of the c-Myc applied in the extracts appeared in the supernatant fraction over both GST and GST-465* immobilized on beads (Fig. 7C,   lanes 2 and 5). 465* Peptide Repressed c-Jun-mediated Activation-GR can repress AP-1 activity through direct interaction with certain members of the AP-1 family, such as c-Jun (15). We have shown a direct interaction between 465* and c-Jun. To see whether 465* could alter the function of c-Jun in vivo, transient cotransfections of plasmids expressing 465*, c-Jun, and a TREdriven reporter gene were performed to test the ability of 465* to repress AP-1 activity. A 5XTRE-TATA-CAT construct, containing five copies of a TPA response element, was used as the AP-1-dependent reporter. We co-transfected pRShGR␣ or pRSh465* expression vector with the reporter plasmid 5XTRE-TATA-CAT into GR-negative CV-1 cells. That the cells contained active c-Jun was shown by stimulating AP-1 activity with TPA. As shown in Fig. 8, basal AP-1 activity, as measured by the activity of the CAT reporter, was very low in nonactivated CV-1 cells, but CAT activity was strongly induced upon treatment with TPA. The addition of Dex after transfection of the plasmid expressing hGR resulted in a potent inhibition of this induction. Transfection of the plasmid expressing 465* alone, without Dex, led to a partial inhibition. When we transfected CV-1 cells with increasing amounts of the pRShGR␣ or  1, 4, and 7); supernatant solution over beads that had been incubated with c- Jun (lanes 2, 5, and 8); and eluate of beads incubated with c- Jun (lanes 3, 6, and 9). Control c-Jun lanes are labeled as such. M denotes molecular size markers. All samples were analyzed by electrophoresis on two 10% SDS-polyacrylamide gels. Gel A was stained with Coomassie Blue R-250. The peptides from gel B were transferred to a polyvinylidene difluoride membrane with subsequent immunochemical analysis using a specific, polyclonal anti-c-Jun serum for the primary reaction. The arrow with c-Jun in A indicates the position of peptides that were retained on the beads and that ran with molecular mass identical to that of c-Jun.
pRSh465*, the results demonstrated a correlation between the amount of pRShGR␣ or pRSh465* used and the extent of repression of TPA-induced AP-1 activity (Fig. 9).
Levels of 465* Were Lower than Those of hGR following Transfection of CV-1 Cells-The maximum extent of repression by transfecting equal amounts of plasmid DNA was greater for holo-GR than for 465*. We suspected that this might be due to the lower levels of mutant 465* peptide being expressed and retained. To test this idea, we transfected two groups of CV-1 cells with equal amounts of each expression plasmid under the same condition timing as in the foregoing experiments. Whole cell extracts containing equal amounts of total protein were then tested by immunoblotting with the antiserum against hGR as used in Fig. 1. This antiserum recognizes an epitope found identically in both proteins. The results (Fig. 10) showed that considerably less 465* peptide than hGR was present in the transfected cells.
465* Repression of AP-1 Activity Was c-Jun-specific-Expression of 465* reduced transcriptional activation by TPA from the 5XTRE-TATA-CAT. If 465* interfered with c-Jun activity, the inhibitory activity of 465* should be counteracted by excess c-Jun protein. Fig. 11 showed that co-transfection of increasing amounts of c-Jun expression plasmid indeed prevented the repression of AP-1 activity by a constant amount of hGR plus Dex or by 465* alone. To further verify that the repression of AP-1 activity by 465* was not due to squelching artifacts (37-38), we transfected CV-1 cells with a heterologous reporter gene construct ERE-CAT, an ER-dependent reporter gene system (Fig. 12). The hER expression vector transfected into CV-1 cells strongly induced reporter activity upon the addition of 17-␤-estradiol. However, 465* alone did not activate ERE-CAT, nor did it significantly inhibit the ERE-CAT activity induced by ER plus hormone. This suggests that 465* is not a generic transcriptional repressor and that its repressive function can only be exerted on certain factors, such as c-Jun. DISCUSSION When plasmids expressing fragments of the hGR that lack the LBD but retain the DBD are transiently transfected into any of several lymphoid cell lines, they cause cell loss (13), which appears to be apoptotic. The most extreme example of these truncated hGR fragments is 465*, in which a frameshift mutation interrupts the second zinc-binding region of the hGR DBD, replaces the ensuing 21 amino acids with a novel sequence, and then terminates the protein. We surveyed the lethal effect of 465* in a series of glucocorticoid-resistant cell lines from T cell leukemias, B cell myelomas, and myeloid leukemias (39). Transfection of 465* caused cell loss only in the T cell and B cell lines and not in the myeloid cell lines tested. This raises the possibility of therapeutic applications against glucocorticoid-resistant lymphoid malignancies for agents based on 465*. In the work presented here, we have pursued several possible mechanisms for the activity of 465*.
The cell kill mediated by 465* occurs rapidly compared with that following parallel transfections of holo-GR and treatment with agonist hormone (13). Loss of cells after 465* transfection starts 6 -12 h later and usually peaks by 18 -36 h. When holo-GR is transfected, no apoptosis occurs unless ligand is added, in which case cell loss begins about 24 h later and continues for 2-3 days, recapitulating the time course of apoptosis in glucocorticoid-sensitive CEM cells containing native hGR. The rapid cell death kinetics following 465* transfection suggested that the mechanism was unlikely to be through the classic GRE-driven glucocorticoid response pathway of these cells. However, in natural thymocytes, GR-steroid complexes evoke apoptosis much more swiftly (40), so it was incumbent to consider the classic GRE-driven pathways as a possible mechanism. Further studies showed that 465* peptide had little affinity for a consensus GRE, as assayed by GMSA. In addition, recombinant 465* peptide did not activate a GRE-driven transcription in the cell-free system, and even in excess, 465* peptide added to the in vitro transcription system did not alter the activation of the target gene caused by hGR alone. Furthermore, in CV-1 cells, 465* did not activate the MMTV-CAT, a GRE-dependent reporter gene construct, and did not interfere with holo-GR induction. All these data strongly suggested that the classic GRE-driven transcriptional activation was not involved in the death of cells caused by 465*. This led us to investigate the possibility that a repressive function of 465* might cause cell lysis.
It has been shown that holo-GR interacts with various transcription factors and regulatory proteins (41-49) through imprecisely defined domains. The 465* peptide may recognize certain transcription factors in a novel way or factors that are normally recognized by the holo-GR. Of the transcription factors known to interact with the holo-GR, c-Jun has been implicated in GR-caused apoptosis, specifically in CEM cells (34). We therefore used c-Jun as a model system to test the hypothesis that the 465* peptide could interfere with the function of important cellular proteins. The results from several groups suggested that the GR DBD was indispensable for GR-c-Jun interaction (24, 50 -51), and, in some cases at least, direct DNA binding by the GR was not necessary for the GR to repress AP-1 activity. Further study suggested that repression of AP-1 activity and transactivation functions of the GR were separable entities and that the repression could be a function of GR monomers (24). Although 465* has lost part of the second zinc finger of the DBD, it retains the ability to interact with c-Jun in vitro and to inhibit AP-1 activity from an AP-1-dependent reporter gene. The data showed less inhibitory effect for a given amount of transfected expression vector carrying the 465* gene than for one carrying the holo-GR, but this result does not measure the amount of each expressed protein. To better compare the inhibitory potency between 465* and holo-GR, we examined the relative level of immunoreactive 465* or hGR found in equivalent whole cell extracts from CV-1 cells transfected with equal amount of each expression plasmid. The results demonstrated that approximately twice as much hGR protein as 465* had been expressed. Although 465* and hGR were expressed from the same promoter, the stability of these two proteins may not be the same. The lower amount of expressed 465* protein in CV-1 cells may explain the lesser inhibitory effect of transfection with pRSh465*. Nevertheless, our data clearly demonstrate that transfection of the 465* construct altered AP-1-dependent transcription. The interference was AP-1-specific, since increased expression of c-Jun counteracted the repressive effect caused by 465*. Also, 465* did not alter either GRE-or ERE-driven reporter gene activity. This indicated that 465* did not block the activity of general transcription machinery.
The similar kinetics of cell death caused by 465* and several other GR mutants that lack the LBD but possess the DBD intact suggest that protein-protein interactions are more likely to be involved in the cell death process they evoke than is their direct binding to GRE sites in the genome. All of these mutants have little or no transactivation activity in co-transfection assays (14 -15). We chose mutant 500* as an example of such mutants for comparison with 465*. Mutant 500* is as potent as 465* in affecting cell lysis, but 500* codes for a fully intact DBD. The 500* peptide bound a GRE, while 465* did not. On the other hand, by using the GST "trap" system, we found that both 465* and 500* selectively interacted with certain factors from the ICR-27 cells. It is possible that 465* and 500* interacted with the same proteins. We further demonstrated that both 465* and 500* interacted with recombinant c-Jun in vitro. By using c-Jun as a model system, we have shown that 465* not only physically interacted with c-Jun but also altered c-Jun's function. Thus, our data are consistent with the hypothesis that 465* and, by extension, other GR mutants lacking their LBDs, exert their lethal effect through binding to and altering the activity of certain transcription factors or other proteins essential for cell viability. For example, repression of potential "survival genes" could be mediated through transcriptional interference between 465* and transcription factors such as AP-1. Alternatively, 465* and similar constructs could directly interfere with some "vitality factors" or could activate lethal caspase cascades directly or indirectly (52)(53). These possibilities are being investigated.
From the evidence described in this paper, we propose that protein-protein interactions rather than direct binding to GRE are important in the lethal effects of 465*. The observed biological cell kill by 465* could be mediated by interference with certain transcription factors or other regulatory proteins required for cell survival.