Stabilization of the Unliganded Glucocorticoid Receptor by TSG101* □ S

The glucocorticoid receptor (GR) has been shown to undergo hormone-dependent down-regulation via transcriptional, post-transcriptional, and posttranslational mechanisms. However, the mechanisms involved in modulating GR levels in the absence of hormone remain enigmatic. Here we demonstrate that TSG101, a previously identified GR-interacting protein, stabilizes the hypophosphorylated form of GR in the absence of ligand. We found that a non-phosphorylated version of GR (S203A/S211A) showed enhanced interaction with TSG101 as compared with the wild type GR, suggesting that TSG101 interacts more favorably with GR when it is not phosphorylated. A significant accumulation of GR S203A/S211A protein is detected in the absence of ligand when TSG101 is overexpressed, whereas no increase in the wild type phosphorylated GR or phosphomimetic GR S203E/S211E was observed in mammalian cells. In contrast, down-regulation of TSG101 expression by siRNA renders the hypophosphorylated form of GR un-stable. We further show that TSG101 stabilizes GR by impeding its degradation by the proteasome and extend-ing receptor half-life. Thus, in absence of a ligand, TSG101 binds GR and protects the non-phosphorylated receptor from degradation. The glucocorticoid

The glucocorticoid receptor (GR) 1 is a ligand-dependent transcriptional regulatory protein and a member of the steroid receptor superfamily. GR contains an N-terminal transcriptional activation domain (activation function 1; AF-1), a central zinc finger DNA binding domain, and a C-terminal region responsible for hormone binding and transcriptional regulation via activation function 2 (AF-2) (1, 2). The inactive unliganded form of GR is largely cytoplasmic and is associated with the heat shock protein 90-based chaperone complex (3). Upon glucocorticoid binding, GR undergoes a conformational change, dissociates from the chaperone complex, and translocates to the nucleus where it activates and represses gene transcription (4).
GR is phosphorylated in the absence of hormone, and additional phosphorylation events occur in conjunction with agonist binding (5)(6)(7)(8). We have developed antibodies that specifically recognize GR phosphorylated at major hormone-dependent sites, serine 203 (Ser-203) and serine 211 (Ser-211) (9). Using a battery of agonists and antagonists in combination with kinetic studies we demonstrated that phosphorylation of human GR is dynamic. In the absence of hormone only a small portion of the receptors are phosphorylated at Ser-211, and this fraction increases dramatically upon agonist binding, whereas under basal conditions a large receptor pool is phosphorylated at Ser-203, which further increases in the presence of hormone (9,10). Our more recent findings suggest that in the absence of ligand GR is being constantly phosphorylated and dephosphorylated at Ser-203 and Ser-211. 2 Thus, in the absence of ligand, different subpopulations of phosphorylated GR are present in vivo.
The hormone-activated GR regulates the transcription of target genes using specific receptor domains (e.g. AF-1 and AF-2) (12) that interact with distinct transcriptional regulatory factors. The unliganded GR uses the C-terminal domain to recruit chaperones that confer high affinity ligand binding to the receptor (13,14). Conceivably, the unliganded receptor employs additional domains and cofactors to control its localization or stability. To discover cofactors that interact with the N-terminal region of GR, we performed a yeast two-hybrid screen using this domain as bait and identified the tumor susceptibility gene 101 (TSG101) (15). TSG101 was originally identified in a screen for tumor suppressors (16). It has been implicated in the regulation of cellular proliferation; mouse homozygous tsg101 Ϫ/Ϫ embryos die during early embryogenesis, and TSG101-deficient embryos display decreased rates of proliferation, which correlate with the accumulation of p53 and p21 (17). TSG101 transcripts with different coding regions, attributed to alternative splicing, have been found in different carcinomas (18,19).
The predicted TSG101 protein structure reveals a C-terminal coiled coil domain, a central proline-rich segment, and an N-terminal ubiquitin-conjugating E2 variant (UEV) domain that is catalytically inactive, because it lacks a cysteine residue required for thioester bond formation with ubiquitin. The coiled coil domain of TSG101 interacts with GR AF-1, and the overexpression of TSG101 inhibits GR transcriptional enhancement (15). Interestingly, the inhibitory effect of TSG101 on transcriptional activity extends to other members of the nuclear hormone receptor family (20,21). In the case of the androgen receptor, inhibition of transcriptional activation was proposed to result from an interaction of TSG101 with the coactivator p300 (20). Paradoxically for a transcriptional regulatory factor, the subcellular location of the TSG101 protein is largely non-nuclear, suggesting that TSG101 performs additional functions outside the nucleus (22).
Because TSG101 is homologous to ubiquitin-conjugating (E2) enzymes, but is catalytically inactive, it has been suggested that TSG101 functions as a dominant negative regulator of ubiquitination (23,24). In support of this idea, recent studies reported the existence of a regulatory loop between TSG101 and MDM2, an E3 ubiquitin ligase; the TSG101 UEV domain binds MDM2 and prevents polyubiquitination and degradation of MDM2, which in turn leads to down-regulation of p53, an MDM2 target (25). Thus, TSG101 affects ubiquitin-dependent proteolysis (26).
TSG101 has also been implicated in the sorting and/or trafficking of ubiquitinated proteins. TSG101 interacts with hVSP28, a protein involved in vesicular trafficking (27) and appears to be recruited to early endosomes (28). In addition, TSG101, through its N-terminal UEV domain, binds the HIV gag protein and leads to viral release (29,30).
Although it is well established that TSG101 interacts with the GR N terminus and reduces receptor transcriptional activation when overexpressed, it is unclear whether this is the only role TSG101 plays in modulating GR function. TSG101 is localized outside the nucleus and functions, in part, as a regulator of protein stability. Given that GR is confined to the cytoplasm in the absence of hormone in a hypophosphorylated state, we examined whether GR phosphorylation could modulate GR-TSG101 interaction and whether TSG101 affects GR stability in the absence of ligand.

MATERIALS AND METHODS
Plasmids-The pCMV-hGR expression plasmid was used to produce an HA-tagged version of the human GR as well as the HA-tagged forms of the GR phosphorylation site-specific mutations (7). The pCMV-LacZ constitutively expresses ␤-galactosidase and is used as a marker for transfection efficiency. The untagged human TSG101 as well as myctagged human TSG101 expression vectors have been described previously (16,25).
Modified Yeast Two-hybrid Assay-The yeast two-hybrid assay was performed as described previously (15). The yeast two-hybrid "bait" proteins, 1) rat GR 107-237 wild type, 2) rat GR 107-237 30IIB, and 3) rat GR 107-237 S224A/S232A are in pJG4/5, which contains the GAL1-10 promoter and expresses HA-tagged, B42 activation domain fusion proteins when grown in the presence of galactose. The TSG101 "prey" protein is constitutively expressed from the alcohol dehydrogenase promoter as a LexA DNA binding domain fusion protein to TSG101 residues 183-381 in pEG202. Bait and prey vectors, along with a ␤-galactosidase reporter gene with a single LexA-operator (pJK103), were transformed into yeast strain EGY 188. For quantitative liquid ␤-galactosidase assays, yeast were grown in selective liquid medium containing 2% glucose for 12 h, pelleted, washed once in H 2 O, normalized according to cell number, and resuspended to an optical density (A 600 ) of 0.15 in 2% galactose, 1% raffinose medium. ␤-Galactosidase assays were performed 12 h later as described previously (31).
Cell Culture and Transient Transfections-HeLa cells were cultured in Dulbecco's modified Eagle's Medium (DMEM) (Cellgro) supplemented with 10% fetal bovine serum (FBS; HyClone), 2 mM L-glutamine (Cellgro), and 10 units/ml each penicillin and streptomycin (Cellgro). Cells were seeded on 6-well plates at a density of 1.5 ϫ 10 5 using antibiotic-and phenol red-free DMEM supplemented with 10% charcoal-stripped FBS and 2 mM L-glutamine 24 h prior to the transfection. Transfections were performed in phenol red-free DMEM without serum using the cationic polymer Exgen 500 (MBI Fermentas) according to manufacturer's recommendation. Total amount of DNA transfected is held constant in each transfection using an equivalent empty vector. Four hours post-transfection, an equal volume of phenol red-free DMEM supplemented with 20% charcoal stripped-FBS and 4 mM Lglutamine was added to the transfection medium. Cells were treated with 100 nM dexamethasone or ethanol vehicle for 16 -18 h, and whole cell lysates were prepared. For receptor half-life experiments, HeLa cells were transfected, and 18-h post transfection the cells were treated with 1 M cycloheximide (Sigma) for 2, 4, and 8 h, and whole cell lysates were prepared.
Immunoprecipitation-For immunoprecipitation of transfected GR, HeLa cells (10-cm dish) expressing the indicated receptor derivative or vector alone were lysed in 200 l of 50 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, and protease inhibitor mixture (Roche Applied Science). The supernatant was incubated at 4°C with 10 g of HA monoclonal antibody for 5 h. A 50% slurry of protein A-Sepharose beads (20 l) were added to the reaction for an additional 2 h. The supernatant was removed, and beads were washed four times in 0.5ϫ lysis buffer. The pelleted beads were resuspended in 1ϫ SDS sample buffer and boiled for 5 min prior to loading onto a gel. For immunoprecipitation of endogenous GR, a 10-cm dish of HeLa cells that was 90% confluent was lysed in 500 l of 50 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, and protease inhibitors. Protein A-Sepharose beads (100 l) were coupled to GR Ser(P)-203 antibody, using the cross-linker dimethyl pimelimidate as per the manufactures instructions (Pierce). For the control reaction, beads were coupled to a nonspecific rabbit polyclonal IgG antibody. The beads were added to the lysates and incubated overnight at 4°C. The protein concentration in the unbound fraction from the supernatant was determined using the Bio-Rad protein assay kit, and equal amounts of protein were loaded on the gel. The bound fraction in the beads was washed three times in 0.5ϫ lysis buffer and once with PBS. The beads were resuspended in 50 l of 1ϫ SDS sample buffer, boiled, and cooled, and a portion of the sample was run on a gel. RNA Interference-A small interfering RNA (siRNA) duplex specific for the human TSG101 was designed as described previously (32) and was synthesized by Dharmacon. The nonspecific siRNA duplex used as a negative control targets luciferase (5Ј-CUU ACG CUG AGU ACU UCG AdTdT; Dharmacon). siRNA transfections were carried out by seeding HeLa cells in 6-well plates at a density of 1.5 ϫ 10 5 in antibioticand phenol red-free DMEM supplemented with 10% charcoal stripped-FBS and 2 mM L-glutamine 24 h prior to the transfection. Transfections were performed using Oligofectamine reagent (Invitrogen) according to the manufacturer's recommendations. 300 pmol of siRNA duplexes were transfected in Opti-MEM with pCMV-hGR wild type or the indicated phosphorylation site derivatives (0.25 g), along with an equal amount of pCMV-LacZ expression vector as an internal control for transfection efficiency. Four hours post-transfection, an equal volume of phenol red-free DMEM supplemented with 20% charcoal stripped-FBS and 4 mM L-glutamine was added to the transfection medium. Cell lysates were typically analyzed 20 h post-transfection. For the proteasome inhibitor treatment, the medium was changed 16 h post-transfection and replaced with phenol red-free DMEM, 10% charcoal stripped-FBS, 2 mM glutamine, and either 2 M MG132 (Calbiochem) in Me 2 SO or Me 2 SO alone. Whole cell lysates were prepared 4 h post-treatment.
Western Blot Analysis-Cells were typically lysed directly on the plate in 100 l of 1ϫ reporter lysis buffer (Promega). The whole cell extract was normalized for total protein using the Bio-Rad protein assay kit and boiled for 5 min in SDS sample buffer. Equal amounts of protein were fractionated by 9% SDS-polyacrylamide gel electrophoresis. Gel-fractionated proteins were then transferred to Immobilon membrane (Millipore) and probed with an anti-HA mouse monoclonal antibody (12CA5, Roche Applied Science), anti-TSG101 mouse monoclonal antibody (A410, Abcam), anti-␤-galactosidase rabbit polyclonal antibody (CR7001RP1, Cortex Biochem), and an anti-tubulin mouse monoclonal antibody (TU27, Covance). GR phosphorylation site-specific antisera to Ser(P)-203 and Ser(P)-211 as well as the antibody N499 to total GR have been described previously (7). The blots were developed using horseradish peroxidase-coupled goat anti-rabbit or anti-mouse antibodies and the ECL substrate as per the manufacturer's instructions (Amersham Biosciences). The same membrane was used for each experiment. After probing with a particular antibody, membranes were stripped in 62.5 mM Tris, pH 8.6, 100 mM ␤-mercaptoethanol, and 2% SDS at 50°C for 10 min twice, blocked for 2 h, and reprobed with the next antibody. Quantitative analysis of immunoblots was performed using the NIH Image software package (version 1.62).

TSG101 Interacts
Preferentially with the Non-phosphorylated Form of GR-TSG101 was previously identified in a yeast two-hybrid screen as an interacting cofactor with the wild type GR AF-1, a region that contains two of the major hormone-dependent phosphorylation sites in GR (8,9,33). Because phosphorylation of many transcription factors is often a regulator of cofactor recruitment (34 -36), we investigated whether GR phosphorylation affects its interaction with TSG101. To test this, a GR with the hormone-dependent phosphorylation sites Ser-203 and Ser-211 converted to alanine, thereby mimicking the non-phosphorylated state of the receptor, was tested for interaction with TSG101 using the yeast two-hybrid system. It has been demonstrated previously that the same residues on GR are phosphorylated in both yeast and mammalian cells (37). As shown previously, TSG101 interacts strongly with the wild type GR and shows reduced affinity for the receptor mutant 30IIB, which harbors three point mutations in AF-1 (15). In contrast, GR S203A and S211A increase interaction with TSG101 as compared with wild type GR (Fig. 1A). Immunoblot analysis demonstrates that this enhanced interaction is not because of differences in receptor expression between the wild type and the mutant GR derivatives (Fig. 1B). Therefore, TSG101 interacts with the wild type GR and more favorably with the hypophosphorylated form of GR.
Expression and Phosphorylation of Endogenous and Transfected GR in HeLa Cells-Studies from our and other laboratories suggest that GR phosphorylation in mammalian cells is heterogeneous (9,38,39). To examine the heterogeneity of receptor phosphorylation in HeLa cells, we first determined the phosphorylation of endogenous GR using GR phosphorylation site-specific antisera. In agreement with our previous studies on GR phosphorylation in U2OS-hGR, A459 cells, and mouse embryonic fibroblasts (7,9), GR displayed a significant basal level of phosphorylation at Ser-203 but not at Ser-211, and the phosphorylation of Ser-211, and to a lesser extent Ser-203, increased with dexamethasone treatment ( Fig. 2A).
Next, to examine the amount of non-phosphorylated GR, lysates from HeLa cells were depleted of the GR phospho-203 isoform using the Ser(P)-203 antibody linked to protein A beads. The bound material as well as the unbound fraction was analyzed by immunoblotting with GR phospho-specific antisera and with an antibody that recognizes GR independent of its phosphorylation state as a measure of total GR. We found that the Ser(P)-203-bound fraction showed strong immunoreactivity to total GR, Ser(P)-203, and to a lesser extent Ser(P)-211 antibodies, which resembles the pattern observed with whole cell lysates (Fig. 2B). In contrast, in the unbound fraction, GR was detectable only with an antibody against total GR, and is not recognized by the phospho-specific Ser(P)-203 and Ser(P)-211 antisera above background levels (Fig. 2C), even with extended exposures of the immunoblot (not shown), suggesting that a small fraction of GR molecules in HeLa cells are not phosphorylated at either at Ser-203 or Ser-211. It is noteworthy that TSG101 coimmunoprecipitated with endogenous GR from HeLa cells (Fig. 2B) and was also present in the unbound fraction with the unphosphorylated GR (Fig. 2C). Unfortunately, the lack of an antibody that recognizes exclusively the non-phosphorylated, rather than total GR precludes direct analysis of this interesting receptor pool. We also do not yet have a method to enrich the endogenous non-phosphorylated GR in vivo through pharmacological manipulation such as kinase inhibitors or phosphatase activators. Therefore, we have elected to use the GR S203A/ S211A mutation as a mimetic for non-phosphorylated GR in our subsequent experiments.
To further ensure that this GR phosphorylation site mutant reflects the non-phosphorylated state, wild type GR or GR S203A/S211A derivatives were expressed in HeLa cells in the absence of hormone and immunoprecipitated from lysates using the common HA-epitope resident on each protein (Fig. 2D). The immunoprecipitates were analyzed by immunoblotting with HA and the GR phospho-specific antisera. As expected, Ser(P)-203 antibody recognized the wild type GR but did not recognize the GR S203A/S211A mutant, despite the fact that the HA antibody immunoprecipitated an equivalent amount of GR from each derivative (Fig. 2D). The immunoprecipitation results confirm that the GR S203A/S211A mutation is characteristic of the non-phosphorylated form of the receptor.
TSG101 Overexpression Stabilizes the Hypophosphorylated Form of GR-TSG101 represses the hormone-dependent GR transcriptional activity under conditions when GR is hyperphosphorylated (9,15). Given that TSG101 is mainly localized in the cytoplasm (22) and appears to associate more avidly with the GR serine-to-alanine mutant that mimics the hypophosphorylated form of the receptor (Fig. 1), we hypothesized that TSG101 might also affect GR in the cytoplasm and regulate receptor steady state levels, as has been shown for other TSG101-interacting proteins such as MDM2 and p21 (25,40).
GR has been previously shown to undergo ligand-dependent down-regulation to limit the hormone response (11). However, the mechanisms that preserve the level of the unliganded GR protein are not well understood. To assess the effect of TSG101 on GR protein levels, HeLa cells were transiently transfected with expression vectors for an HA-tagged wild type human GR (CMV-GR wt), or its phosphorylation-deficient or mimetic derivatives (CMV-GR S203A/S211A or CMV-GR S203E/S211E), along with either an untagged version of TSG101 or an empty expression vector. Cells also received a ␤-galactosidase expression vector (CMV-LacZ) as an internal control for transfection efficiency, to normalize GR protein levels to that of ␤-galactosidase. Cells were treated with vehicle or with 100 nM dexamethasone, and the steady state levels of GR and ␤-galactosidase were determined by immunoblot analysis. Both GR wild type (Fig. 3A, lanes 1 and 3) and GR S203E/S211E (Fig. 3A,  lanes 9 and 11) protein levels are comparable in the absence or in the presence of TSG101 overexpression and display the characteristic GR down-regulation in the presence of dexamethasone. In contrast, GR S203A/S211A shows a significant increase in the steady state GR protein level when TSG101 is overexpressed, most notably in the absence of ligand (Fig. 3A,  lanes 5 and 7).
We quantitated the amount of GR protein relative to the ␤-galactosidase control in the absence of hormone from four independent experiments. Little change in protein levels is noted for the wild type GR, the double phospho-mimetic mutant GR S203E/S211E or the single GR S211A substitution regardless of TSG101 level, whereas a small increase in steady state level was observed for the GR S203A mutant in the presence of overexpressed TSG101 (Fig. 3B). Importantly, a  3. TSG101 overexpression stabilizes the non-phosphorylatable form of GR in the absence of ligand. A, HeLa cells were transfected using Exgen 500 with 0.25 g of an HA-tagged-wild type GR (WT) or the indicated GR phosphorylation site mutants expression constructs, 0.25 g of a ␤-galactosidase expression plasmid (LacZ), along with 1.0 g of TSG101 or corresponding empty vector as indicated. Twelve hours after the transfection, cells were treated with ethanol vehicle (Ϫ) or with 100 nM dexamethasone (Dex) (ϩ) for 16 h. Whole cell extracts were prepared from transfected cells and the expression of GR and ␤-galactosidase was analyzed by immunoblotting using anti-HA and anti-␤galactosidase antibodies, respectively. B, the relative amount of GR immunoreactivity for each derivative was quantitated using NIH Image relative to ␤-galactosidase expression levels in presence and absence of TSG101. A ratio Ͼ1 suggests that TSG101 is stabilizing GR, whereas a ratio of Ͻ1 shows that TSG101 is destabilizing GR. A ratio of 1 means that there is no effect of TSG101 on GR protein levels. Shown are the average and S.D. from four independent experiments. C, TSG101 protein is not affected by co-expression with GR. HeLa cells were transfected as above except that a c-myc-tagged TSG101 construct was used. significant 2.9 Ϯ 0.5-fold increase in the steady state level of the GR S203A/S211A derivative in response to TSG101 was observed (Fig. 3B). We conclude that the increase in GR protein level most likely results from stabilization of the non-phosphorylated form of the receptor by TSG101.
To exclude the possibility that the lack of stabilization in wild type GR or certain phosphorylation site mutants results from differences in TSG101 expression, transfections were carried out as before but using a c-myc tagged form of TSG101 (25). This derivative is larger than the native TSG101 and therefore can be distinguished from the endogenous protein.
Equal amounts of total protein were loaded on SDS-PAGE, and GR, ␤-galactosidase, and both endogenous and c-myc tagged forms of TSG101 were analyzed by immunoblotting (Fig. 3C). Similar levels of the myc-TSG101 protein were coexpressed with each GR derivative, and as before, only GR S203A/S211A protein levels were substantially increased in the presence of myc-TSG101 (Fig. 3C).
Cellular Depletion of TSG101 Protein Results in Decreased GR Protein Levels-Having demonstrated that the overexpression of TSG101 increases GR stability, we next asked whether a decrease in TSG101 concentration would reduce GR stability. To this end, we depleted HeLa cells of endogenous TSG101 using siRNA and determined the level of GR protein in the absence of ligand. HeLa cells were cotransfected with either a specific TSG101 siRNA duplex or a nonspecific luciferase siRNA duplex, along with expression vectors for HA-tagged GR derivatives and ␤-galactosidase. As assessed by immunoblotting, a significant decrease in TSG101 protein level is observed in the presence of TSG101 siRNA (Fig. 4A, lanes 1 and 2). When TSG101 expression is reduced, steady state levels of wild type GR, GR S203A, GR S211A, and GR S203E/S211E are largely unaffected (Fig. 4A). Importantly, TSG101 siRNA results in a significant decrease in the steady state protein level of the non-phosphorylated GR S203A/S211A mutant (Fig. 4A,  lanes 3 and 4). Quantitation of three independent experiments indicate that the wild type GR levels were reduced on average by 20%, whereas the non-phosphorylatable form of GR was consistently reduced by Ͼ90% when TSG101 was depleted. A slight reduction of endogenous GR is also observed when TSG101 is depleted (Fig. 4B, lanes 11-14), and a small increase in endogenous GR is seen when TSG101 is overexpressed (Fig.  4B, lanes 14 and 15), consistent with a small fraction of nonphosphorylated GR in HeLa cells. Taken together, these results suggest that TSG101 preferentially stabilizes the non-phosphorylated form of the receptor in the absence of ligand.
Inhibition of the Proteasome Partially Restores the Stability of the Hypophosphorylated GR when TSG101 Is Reduced-We next examined whether the decrease in protein levels of GR S203A/S211A in the absence of TSG101 is because of degradation by the proteasome. We depleted HeLa cells of endogenous TSG101 using siRNA in the presence or in the absence of the proteasome inhibitor MG132 (Fig. 5). HeLa cells were cotransfected with either a TSG101 or luciferase siRNA duplex together with expression vectors for GR S203A/S211A and ␤-galactosidase. Cells were then treated with either MG132 or the Me 2 SO vehicle for 4 h. TSG101 levels were reduced in the presence of the siRNA for TSG101 whether the cells are treated with MG132 or not (Fig. 5, second panel, lanes 1, 2, and 4). However, in the presence of MG132 the TSG101-dependent reduction in the steady state level GR was partially suppressed (Fig. 5, top panel, lanes 2 and 4). This effect is not a reflection of variation in transfection efficiency or loading as indicated by level of ␤-galactosidase and tubulin, respectively (Fig. 5, bottom  two panels).
To exclude the possibility that the increase in GR S203A/ S211A levels in the presence of MG132 is because of a variation in the efficiency of TSG101 silencing by siRNA, we quantitated both GR and TSG101 levels and determined the expression of GR relative to TSG101. Plotting the normalized values indicates that the -fold destabilization of GR S203A/S211A relative to TSG101 is lower in presence of the proteasome inhibitor MG132 (Fig. 5B). Therefore, it appears that TSG101 stabilizes the non-phosphorylated GR S203A/S211A protein by, in part, impeding its degradation by the proteasome.
TSG101 Overexpression Affects Receptor Half-life-Because TSG101 stabilizes the non-phosphorylated form of GR by preventing its proteasomal degradation, we next assessed whether the half-life of GR S203A/S211A was affected by TSG101. To evaluate the effect of TSG101 on receptor protein half-life, we determined the level of GR in cells treated with the protein synthesis inhibitor cycloheximide in the absence and presence of overexpressed TSG101 (Fig. 6A). As expected, we found an ϳ3-fold increase in steady state levels of GR S203A/S211A in the presence of TSG101 prior to cycloheximide treatment. In the absence of overexpressed TSG101, the half-life of transfected GR was 2.5 h (Fig. 6B). When TSG101 was overexpressed, the GR half-life increased to 3.5 h (Fig. 6B), suggesting that TSG101 extends GR half-life, consistent with the protective effect of TSG101 that renders GR resistant to degradation. DISCUSSION Our experiments demonstrate a relationship between the stability of the unliganded hypophosphorylated GR and TSG101. We found that TSG101 interacts with wild type phosphorylated GR and that this interaction is enhanced by the GR S203A/S211A double mutation in yeast. We have also demonstrated that endogenous TSG101 and GR interact in coimmunoprecipitation experiments. Although a small pool of hypophosphorylated GR is present in HeLa cells, because of the lack of an antibody that exclusively recognizes this non-phosphorylated form of GR, it is difficult to study this receptor population directly. Therefore, GR serine-to-alanine mutations were used to mimic a homogeneously non-phosphorylated pop- FIG. 4. Cellular depletion of endogenous TSG101 protein results in decreased GR stability in the absence of ligand. A, depletion of TSG101 affects GR stability. HeLa cells were transfected with 0.25 g of wild type (WT) GR or GR phosphorylation site mutants, 0.25 g of a ␤-galactosidase (LacZ) expression plasmid, along with 300 pmol of a TSG101-specific siRNA (ϩ) or a nonspecific luciferase siRNA (Ϫ) as a control. Twenty hours after the transfection, whole cell extracts were prepared, and the expression of GR, TSG101, LacZ, and tubulin was analyzed by immunoblotting. B, TSG101 affects endogenous GR levels. HeLa cells were transfected with 300 pmol of a TSG101-specific siRNA (ϩ) or a nonspecific luciferase siRNA (Ϫ) as a control, or with 1.0 g of myc-TSG101 (ϩ) or empty vector (Ϫ) as indicated. Expression of endogenous GR, TSG101, and tubulin, which serves as a loading control, was analyzed by immunoblotting. Shown is a representative experiment that was repeated twice with similar results. ulation of GR in vivo. This approach revealed a significant accumulation of GR S203A/S211A in the absence of ligand when TSG101 is overexpressed, whereas the TSG101-specific siRNA had the opposite effect, rendering the non-phosphorylated form of GR unstable. The endogenous wild type GR is also partially destabilized when TSG101 expression is reduced, although this effect is not as apparent as only a small fraction of GR is non-phosphorylated in HeLa cells. From our findings, we propose a model whereby in the absence of ligand, TSG101 is recruited to GR and protects the hypophosphorylated receptor from degradation, akin to MDM2 and p21 (Fig. 7). Our findings further suggest that GR phosphorylation or TSG101 binding to the non-phosphorylated GR are "fail-safe" mechanisms that operate to prevent the ligand-free form of the receptor from being degraded.
Elegant work from the Cidlowski laboratory (41,42) has demonstrated that hormone-induced down-regulation of GR is a means of limiting the hormone response and is dependent, in part, upon receptor phosphorylation. For example, ligandbound GR mutated at multiple potential phosphorylation sites has a longer half-life than its wild type counterpart (43). Our findings also indicate that TSG101 overexpression can extend GR half-life in the absence of ligand and is consistent with the protective effect from degradation TSG101 exerts on the nonphosphorylated GR. Our findings suggest that protein factors, such as TSG101, can recognize and shield hypophosphorylated GR from degradation by the proteasome and extend GR half-life.
Our findings may also be relevant to the repressive effect of TSG101 on GR transcriptional activity. TSG101 has been shown to interact with p300, which in principle could affect receptor transcriptional regulatory properties. Alternatively, changes in receptor ubiquitination status as a result of TSG101 binding might indirectly affect receptor transcriptional activity. Increasing evidence suggests a link between ubiquitination and transcriptional regulation. For example, inhibition of the proteasome affects transcriptional activation by androgen receptor (44), GR (11,45), and estrogen receptor (46). Burgdorf et al. (47) recently reported that overexpression of TSG101 led to the accumulation of a transcriptionally active monoubiquiti- FIG. 7. TSG101 stabilizes the ligand-free and hypophosphorylated form of GR. A model whereby TSG101 associates with the ligandfree GR and protects it from degradation, whereas the unphosphorylated, non-TSG101 bound form of the receptor is degraded. We posit that GR phosphorylation can also stabilize the receptor independent of TSG101 binding. Our findings suggest that GR phosphorylation and TSG101 binding are redundant mechanisms that operate to prevent the unliganded non-phosphorylated receptor from being degraded. This may represent a fail-safe mechanism to maintain GR levels in vivo and whole cell lysates were prepared after 0, 2, 4, and 8 h. The expression of GR, TSG101, LacZ, and tubulin was analyzed by immunoblotting. B, GR immunoreactivity was quantitated relative to tubulin expression in presence or absence of TSG101 as in Fig. 3. nated form of AR. They proposed that TSG101 binds the monoubiquitinated receptor and locks it in that state until transcription starts in the nucleus, or alternatively, that TSG101 binds the monoubiquitinated receptor in the cytoplasm and protects it from degradation. It is also possible that TSG101 would recognize the hypophosphorylated form of GR not engaged in transcription and would protect it from degradation. It will be interesting to determine the modifications that distinguish between such pools.
It is known that cellular sensitivity to glucocorticoids depends on the number of GRs in the cell (42, 48 -51). Although GR mRNA levels differ among tissues because of cell typespecific differences in promoter activity or mRNA stability, our findings demonstrate that steady state protein levels of the unliganded receptor can also be modulated post-transcriptionally by TSG101. This could provide an additional means of controlling receptor levels in vivo. It will be important to assess GR and TSG101 protein levels in normal tissues and during pathophysiological processes. For instance, GR protein levels are reduced in prostate cancer (52), and as GR activation is typically anti-proliferative, it has been proposed that the loss of GR would lift a restriction on cell growth and promote hyperplasia and oncogenesis. It is interesting to note that TSG101 transcripts also show frequent deletions within the coding regions in prostate cancer (19). This may contribute to a reduction of GR levels, a hypothesis we are currently testing.
In addition, given that that HIV binds TSG101 and utilizes it for viral budding, we would anticipate that the steady state GR levels would be reduced in HIV-infected macrophages or T-cells producing virus, by virtue of HIV sequestering TSG101 away from GR, thereby rendering TSG101 rate-limiting in the cell. Indeed, a chronically HIV-infected T-cell line contains fewer GRs and is resistant to glucocorticoid-induced cell death (53,54). This may represent a strategy used by the virus to circumvent host defense mechanisms controlled by GR, such as expression of toll-like receptors (55). Whether this reduction in GR number is a function of TSG101 remains to be explored.
In summary, our results suggest that TSG101 associates with GR and stabilizes the unliganded hypophosphorylated receptor by impeding its degradation by the proteasome and prolonging its half-life. We conclude that it is the balance between TSG101 and GR phosphorylation that helps determine the intracellular concentration of GR.