Glucocorticoid Signaling Is Perturbed by the Atypical Orphan Receptor and Corepressor SHP*

SHP (NROB2) is an atypical orphan nuclear receptor that lacks a DNA-binding domain but contains a putative ligand-binding domain. Previous studies have revealed that SHP interacts with a variety of nuclear receptors and inhibits their transcriptional activity, thereby acting as a corepressor. In this report we identify the glucocorticoid receptor (GR) as a novel downstream target receptor for SHP inhibition. SHP potently inhibits dexamethasone-induced transcriptional GR activity in mammalian cells, and the inhibition involves a functional second NR-box within SHP. Interestingly, this motif shows a high homology with the NR-box in the glucocorticoid and cAMP-inducible GR coactivator PGC-1, indicating similar binding specificity and shared target receptors. We show that SHP antagonizes PGC-1 coactivation and, in addition, we identify the PGC- 1-regulated phospho(enol)pyruvate carboxykinase (PEPCK) promoter as a novel target promoter for SHP inhibition. This implies a physiologically relevant role for SHP in modulating hepatic glucocorticoid action. Furthermore, when coexpressing green fluorescent protein-tagged GR together with SHP, an intranuclear redistribution of GR was observed. As inhibition-deficient SHP mutants were unable to induce this redistribution, intranuclear tethering of target receptors may represent yet another, previously uncovered, aspect of SHP inhibition.

The nuclear receptor superfamily consists of structurally and functionally related transcription factors, which in mammals includes ϳ50 distinct proteins (1). They are key regulators of cellular processes including cell growth, differentiation, development, and homeostasis (2,3). Transcriptional activation by typical nuclear receptors involves binding of the cognate ligand, dimerization, binding to target DNA-sequences, and modulation of gene transcription via interactions with chromatin components and with the basal transcriptional machinery (4). A broad range of coregulatory factors, defined as non DNAbinding proteins or protein complexes, which associate with DNA-bound nuclear receptors, has been isolated, acting either as coactivators or coinhibitors of gene expression (5,6). Many of these coregulators interact ligand-dependently with the C-terminal ligand binding and transcriptional activation domain (LBD/AF-2) 1 of nuclear receptors. In agreement with the structural conservation of the interaction surface on the receptors, short conserved leucine-rich interaction motifs (LXXLL, NRbox) have been identified in most AF-2 binding coregulators, while the interaction of corepressors with unliganded receptors is mediated by related but distinct hydrophobic helical motifs (5,6).
SHP is an atypical orphan nuclear receptor lacking a DNAbinding domain consisting only of a putative LBD. SHP is known to interact with and inhibit the transcriptional activity of various nuclear receptors such as ER␣ and -␤, RXR, HNF-4 and LRH-1 (7)(8)(9)(10)(11)(12)(13). SHP plays a physiological role in regulating bile acid homeostasis in the liver by inhibiting the expression of cholesterol 7␣-hydroxylase (CYP7A1), the rate-limiting enzyme in the liver bile acid biosynthesis pathway (12,13). This was also recently shown in mice with a targeted disruption of the SHP gene, although additional pathways were shown to be involved in the repression of CYP7A1 (14,15). In addition, broader function have been suggested for SHP in liver, regulating the activity of additional target genes (16,17). Furthermore, the tissue distribution of SHP indicates a possible role in other tissues, including pancreas, heart, kidney, and intestine (10, 18 -20). Mechanistically, SHP interacts with the target nuclear receptor LBD/AF-2 domains via LXXLL-related motifs, thereby competing with coactivators for binding to the receptors (8). The physical interaction between SHP and the receptor did not appear to be sufficient for inhibition, indicating participation of additional repressor mechanisms, for example antagonism of coactivator function via EID1 proteins (27).
The glucocorticoid receptor (GR) is highly expressed in liver and plays an important role in control of glucose metabolism via regulation of expression of rate-limiting enzymes in gluconeogenesis, such as phospho(enol)pyruvate carboxykinase (PEPCK) (21). GR also plays an important role in pancreatic ␤-cells where different studies have shown that overexpression of GR or treatment with dexamethasone leads to an inhibition of insulin secretion (22,23). Taken together, glucocorticoids and GR are important players in regulating blood glucose levels in mammals. The fact that SHP is highly expressed both in liver and pancreas indicates a possible involvement of SHP in glucose homeostasis and points at GR as a putative target receptor for SHP.

MATERIALS AND METHODS
Plasmids-All plasmids were generated using standard cloning procedures and verified by DNA sequencing. The point mutations in pSG5rSHPmt1 and rSHPmt2 were introduced into the SHP sequence by PCR-mediated mutagenesis using primers containing the mutation, and the inserts were cloned into the EcoRI/BamHI site of pSG5. The PCR-generated fragment of rSHP⌬H12 (amino acids 1-245) was cloned into the EcoRI site of pSG5. pSG5rSHPwt and pSG5rSHPmt1.2 have been described previously (8). To generate pSG5rHNF-4, HNF-4 was recloned from pLEN4SrHNF-4 (24) into the BamHI site of pSG5. GFP-hGR (GFP fused to the N terminus of GR) was constructed by inserting a BamHI fragment of human GR (amino acids 1-777) into the BglII site of pEGFPC2 (Clontech Laboratories, Inc.) The PGC1A-pSV-Sport (25), pcmvhGR, and the MMTV-LUC reporter plasmid (26) have been described previously. pGL3rPEPCK-Luc (-489 to ϩ73) was generously provided by Dr. Jörg Leers.
Mammalian Cell Transfections-293 human embryo kidney cells and COS-7 monkey kidney cells were maintained as previously described (7,27). FaoII rat liver hepatoma cells were maintained in RPMI medium supplemented with 10% calf serum, penicillin (100 g/ml), and streptomycin (100 g/ml). Transfections were performed using Lipofectin (Invitrogen) as previously described (7), using phenol-free medium for 293 cells in the absence or presence of 1 M dexamethasone. 24 h after transfection, cells were harvested and luciferase activities were measured. For COS-7 cells, transfections were performed using 0.5 g of MMTV-Luc reporter, 1 ng (PGC-1 study) or 10 ng of pcmv5GR␣ together with noted amounts of pSG5SHP expression plasmid. For 293 cells, 0.1 g of pcmv5GR and 0.5 g of MMTV-Luc reporter was used, together with 0.5 g of pSG5SHPwt, pSG5SHPmt1, pSG5SHPmt2, pSG5SHPmt1.2 or pSG5SHP⌬H12. For FaoII cells 0.1 g of PEPCK-Luc reporter was used, together with 10 ng pcmv5GR, 0.1 g of pSG5HNF-4, and 0.5 g of pSG5SHPwt, pSG5SHPmt1.2, or pSG5SHP⌬H12. All transfections were performed in 35-mm-diameter plates, and pSG5 (empty vector) was added to equalize total transfected plasmid DNA concentrations.
Coimmunoprecipitation-COS-7 cells were transfected with the corresponding plasmids using DEAE dextran (Amersham Biosciences) in 150-mm-diameter plates and kept for 30 h in the presence of 1 M dexamethasone. After 30 h, whole cell extracts were prepared using a high-salt buffer containing 10 nM HEPES-KOH, pH 7.9, 0.4 M NaCl, 0.1 mM EDTA, 5% glycerol, and 1ϫ protease inhibitor mixture (Roche Molecular Biochemicals). Protein extracts were incubated for 2-3 h at were transfected with either wild-type SHP, full-length GR, or simultaneously with both constructs. Whole cell extracts were prepared and subjected to immunoprecipitation. After incubation, the samples were analyzed on a Western blot. The SHP protein was detected using a purified polyclonal SHP antibody, and the GR protein was detected using a GR-specific polyclonal antibody, sc-1003. The input represents ϳ10% of the whole cell extract cotransfected with both SHP and GR. ϩ4°C with 20 l of protein A/G-agarose matrix, 1 g of the GR-specific polyclonal antibody (sc-1003, Santa Cruz Biotechnology) in the presence of 1 M dexamethasone in IP-T150 buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.2% Nonidet P-40, 1 mM EDTA, and 10% glycerol). After incubation, the beads were washed three times using the IP-T150 buffer. Western analysis was performed using a purified polyclonal SHP antibody (1:1000) and the GR-specific polyclonal sc-1003 antibody (1:500).
Analysis of Intracellular Localization Using Confocal Microscopy-COS-7 cells were plated on coverslips in 6-well plates and transfected with 0.5 g of the GFP-GR plasmid and/or 1.0 g of all of the pSG5SHP constructs, respectively, using Lipofectin (Invitrogen) as previously described (7). Five hours following change of medium, 1 M dexamethasone was added. After an additional 3 h, cells were fixed with 3% paraformaldehyde in 5% sucrose/PBS for 20 min at room temperature. For indirect immunofluorescence, fixed cells were permeabilized with PBS/Tween (0.1%), blocked with 5% goat serum (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) in PBS/Tween for 1 h at room temperature, and incubated with purified polyclonal SHP antibody for 1 h at room temperature. After washing, cells were treated with Lissamine Rhodamine-conjugated AffiniPure goat anti-rabbit IgH (HϩL) (Jackson ImmunoResearch Laboratories, Inc.) for 1 h at room temperature, washed intensively, and fixed to slides using antiphotobleaching fluorsave (Callbiochem, La Jolla, CA). Subcellular localization was determined using a TCS SP Multiband Confocal Imaging System (Leica Corp., Deerfield, IL).

SHP Inhibits the Transcriptional Activity of GR-Previous
studies have demonstrated that SHP functions as a potent inhibitor of various nuclear receptors in mammalian cells. We therefore tested whether SHP could affect GR transcriptional activity under transient transfection conditions. We also used a variety of SHP mutations as tools to dissect the inhibitory mechanism ( Fig. 1A), both to investigate the necessity of the previously described LXXLL-related motifs within SHP and also to study the active repression mechanism by using a novel SHP mutation, SHP⌬H12. This mutant was made in view of previous reports showing that in RXR a deletion of helix 12 promotes the inhibition via stabilization of corepressor interactions (28,29). COS-7 cells were cotransfected with the MMTV-Luc reporter and full-length GR, together with increasing amounts of SHP. As shown in Fig. 1B, SHP inhibits dexamethasone-induced GR activity in a dose-dependent manner, leading to an almost complete inhibition. Similar results were observed with human SHP (data not shown). The different SHP mutants were cotransfected into 293 cells, together with the MMTV-Luc reporter and full-length GR. A control Western blot showed that the different SHP proteins were expressed at the same level (Fig. 1C). As shown in Fig. 1D, wild-type SHP inhibits GR activity, as previously seen in COS-7 cells. Interestingly, inhibition of GR activity seems to depend on only one functional NR-box, since mutation of box 1 did not influence the inhibition while mutation of box 2 completely disrupted the ability of SHP to inhibit GR. Not surprisingly, the double NR-box mutant, mt1.2, also lacked the ability to inhibit GR activity. In contrast to the effect of helix 12 deletion on RXR repression (28,29), deletion of helix 12 in SHP disrupted the inhibition ability, confirming a difference between SHP inhibitory mechanisms compared with other repressing receptors. Furthermore, SHP inhibition of GR was unaffected by trichostatin A, a specific histone deacetylase inhibitor (data not shown), consistent with previous observations (11). To elucidate the interaction between SHP and GR, coimmunoprecipitations were performed using COS-7 cells transfected with GR in the absence or presence of wild-type SHP. Proteins were subjected to immunoprecipitation using a GR-specific mouse monoclonal antibody. As shown in Fig. 1E, lane 2, GR was able to precipitate SHP in the presence of dexamethasone. In addition, GST-pulldowns using GST-SHP protein together with in vitro translated GR, showed a direct interaction between SHP and GR (data not shown). From these results we conclude that SHP can act as a potent inhibitor of GR activity. Furthermore, both a functional box 2 and an intact helix 12 of SHP appear to be necessary for inhibition of GR.
Subcellular Distribution of Wild-type and Mutated Forms of SHP-To further investigate the mechanism behind the inability of the different SHP mutants to inhibit GR activity, we decided to study their intracellular localization in mammalian cells. While the existence of localization signals in many cases explains the subcellular distribution of proteins, no conserved localization signals have been identified in SHP, although previous results revealed a nuclear distribution of a GFP-SHP fusion protein (7). Recent publications have shown a connection between nuclear localization and protein function for different cofactors, including the corepressors SMRT (30) and RIP140 (31). To elucidate the localization of SHP, COS-7 cells were transiently transfected with the different SHP constructs, and 8 h posttransfection, the cells were fixed and stained with a purified polyclonal SHP antibody. As seen in Fig. 2, the different SHP proteins appear to have different subcellular distributions. Wild-type SHP was found predominantly in the nucleus showing a distinct punctuate distribution. Similar, but presumably not identical, nuclear dot patterns have been reported not only for additional corepressors including RIP140 (31) and SMRT (30) but also for the coactivator GRIP-1 (32). The SHP typical dot pattern was found in over 80% of the cells examined, and the remaining 20% of the cells showed either exclusive cytoplasmatic staining or a mixture of both nuclear and cytoplasmic staining (data not shown). No staining was visible in non-transfected cells. As shown in Fig. 2A, the SHPmt1.2 mutant was also predominantly (74%) localized to the nucleus, showing the same punctuate pattern as the wild-type SHP. This indicates that nuclear localization, including the dot formation, is independent of nuclear receptor binding and that nuclear localization alone not is sufficient for the inhibitory effect. In comparison, the SHP⌬H12 mutant was exclusively localized in the cytoplasm, either unable to localize to the nucleus or actively exported out from the nucleus.

Redistribution of GR within Nuclear Compartments by Wildtype SHP but Not by the Inhibition-deficient SHP Mutants-
The distinct distribution pattern of both SHP (Fig. 2) and GR (33,34) enabled us to investigate the combined effect of both proteins regarding their cellular distribution. To study the localization of GR, a GFP-GR construct was prepared, using the full-length human GR. This construct enables transcriptional activation in transient transfections using a GRE-Luc reporter in the presence of dexamethasone (data not shown). GFP-GR was transfected into COS-7 cells, and 5 h posttransfection dexamethasone was added to selected wells. The cells were left for an additional 3 h, fixed, and visualized in the confocal microscope. As previously reported (33,34), GFP-GR, in the absence of ligand, is uniformly distributed in the cytoplasm (Fig. 3Aa) but is translocated into the nucleus in the presence of dexamethasone (Fig. 3Ab). To determine the effect of SHP on GFP-GR distribution, COS-7 cells were cotransfected with GFP-GR together with wild-type SHP, and the cells were treated as described above. After fixation, the cells were stained with a purified polyclonal SHP antibody and visualized in the confocal microscope. No effect of the antibody-staining procedure was seen on cells only expressing GFP-GR (data not shown). In cells containing both GFP-GR and SHP, in the absence of dexamethasone, both proteins showed their typical individual pattern, see Fig. 3B, a and b. This is consistent with the idea that SHP only interacts with ligand-activated GR. Addition of dexamethasone leads to a relocalization of GFP-GR into the nucleus as expected, but the presence of SHP seems to redistribute GFP-GR from the rather diffuse nuclear distribution to a specific dot pattern, which was never observed in the absence of SHP, see Fig. 3Bd. The dot pattern of GFP-GR shows an overlapping distribution with the dot pattern of SHP (Fig. 3Bf), consistent with a possible physical in vivo interaction between the two proteins. Apparently, SHP distribution determines the localization of the GR/SHP complex in the presence of dexamethasone. The colocalization of SHP and GR was seen in ϳ64% of the cotransfected cells. In the remaining cells, the same dot pattern was seen for GR, but SHP was instead localized to the cytoplasm. If the redistribution of GR into the specific dot pattern seen in Fig. 3B, d-f is associated with inhibition of transcription, then a SHP mutant lacking the inhibition ability may not affect the distribution of GR to the same extent as wild-type SHP. Cotransfections were performed using the GFP-GR construct together with either of the SHP mutants, and the cells were treated as in the former experiment. As shown in Fig. 3, C and D, none of the repressiondeficient mutants was able to affect the GFP-GR distribution. Instead, GFP-GR in the presence of dexamethasone was localized to the nucleus into its characteristic diffuse pattern, and the different SHP mutants showed their original pattern. In the absence of dexamethasone, GFP-GR is distributed all over the cytoplasm, with no obvious effect of coexpression with the different SHP mutants (data not shown). These results indicate (i) that functional NR-boxes are necessary for the relocalization of GR since the SHPmt1.2 despite its punctuate pattern is unable to induce relocalization of GR, and (ii) that loss of nuclear distribution of the SHP⌬H12 mutant affects the ability to inhibit GR function. In conclusion, we suggest that the intranuclear redistribution of ligand-activated GR by SHP could be linked to inhibitory mechanisms and that the inability of the SHP mutants to alter the intracellular localization of GR could be a reason for their inability to inhibit GR activity.
SHP Antagonizes PGC-1 Coactivation of GR in Mammalian Cells-As shown in Fig. 1, a functional NR-box 2 of SHP seems to be necessary for inhibition of GR. Intriguingly, when searching databases for SHP box 2-homologous sequences, the closest motif in any mammalian protein found was the NR-box in the nuclear receptor coactivator PGC-1 (Fig. 4A). PGC-1 was first identified as a coactivator for PPAR␥ and thyroid receptor (25), linked to adaptive thermogenesis. Interestingly, PGC-1 has lately been shown to play an important role in up-regulating gluconeogenesis in liver by acting as a coactivator for GR and HNF-4 (35,36). PGC-1 mRNA was induced by glucocorticoids and cAMP, by fasting, or in other states of relative insulin deficiency. PGC-1 contains only one NR-box that is involved in the ligand-dependent interaction and coactivation of these receptors. Since SHP box 2 and the PGC-1 motif show such a high homology, we asked whether SHP would antagonize the ability of PGC-1 to potentiate the ligand-induced activity of GR. COS-7 cells were used and, as shown in Fig. 4B, GR-dependent transcriptional activity was significantly elevated when coexpressing PGC-1. Adding increasing amounts of SHP to the system not only abrogated PGC-1 activation, but also inhibited GR activity. Furthermore, by the addition of increased amounts of PGC-1, it was possible to overcome the inhibitory effect of SHP, even though the elevation of the GR activity seemed to be weaker (Fig. 4C). These results demonstrate that PGC-1 and SHP antagonize each other presumably by competing for binding to the GR AF-2 domain.
SHP Inhibits the Transactivation of the PEPCK Promoter by GR and HNF-4 -The antagonizing effect of SHP on PGC-1 coactivation shown here, together with the fact that PGC-1 acts as a coactivator for GR and HNF-4 in vivo, thereby up-regulating the expression of gluconeogenic enzymes such as PEPCK (35,36), made us curious about the effect of SHP on the activity of the PEPCK promoter. PEPCK catalyzes the rate-controlling step in hepatic gluconeogenesis and is mainly regulated at the level of transcription. We examined the inhibitory effect of SHP on the PEPCK promoter using FaoII rat hepatoma cells (Fig.  5). GR alone had no effect on the activation of the PEPCK promoter-reporter construct, although a slight induction was seen by addition of dexamethasone (see lanes 1, 3, and 4). This verifies the necessity of additional factors except GR for full dexamethasone-mediated transcriptional response of PEPCK (37). Addition of HNF-4 had no effect in the absence of dexamethasone (lane 2). In contrast, the PEPCK promoter was synergistically activated by dexamethasone in the presence of both GR and HNF-4 (compare lanes 2 and 5). Interestingly, when wild-type SHP was cotransfected there was a clear inhibition of the activated PEPCK promoter (compare lanes 5 and 6). In comparison, the NR-box mutant, SHPmt1.2, was unable to inhibit the activated PEPCK promoter (lane 7), implying that the inhibition was a direct effect mediated via GR and HNF-4. Furthermore, also the SHP⌬H12 deletion was unable to inhibit the PEPCK promoter (lane 8). In summary, although the native PEPCK promoter exhibits a more complex regulation, involving both GR and HNF-4, SHP appear to inhibit this promoter via a similar mechanism as described for the GR-dependent promoters.

DISCUSSION
The transcriptional activity of GR appears to be regulated by various coactivators such as GRIP1 and PGC-1, interacting with the AF-2 of GR in a ligand-dependent manner (38,39). These interactions are mediated via short LXXLL motifs within the coactivators. We have previously shown that two highly related LXXLL motifs within SHP are responsible for the interaction with the ERs and that SHP competes with coactivators for binding to the receptor, the first step of SHP inhibition (8). When investigating the necessity of the two functional SHP motifs for GR inhibition we could now clearly show that a disruption of motif 2, but not of motif 1, was sufficient to completely abolish the ability of SHP to inhibit GR activation. Searching databases for homologous motifs revealed that the closest motif in any mammalian protein is the LXXLL motif of the GR coactivator PGC-1. The homology extends beyond the leucine core to adjacent residues, suggesting similar binding specificity and shared target receptors. Furthermore, both SHP and PGC-1 motifs fall into the LXXLL peptide class III (40). The ability of SHP to antagonize PGC-1 coactivation indicates competition for binding to GR and thus supports the functional relevance of the LXXLL-related SHP motif 2 for receptor interactions. Similar binding specificity for this intriguing pair of inducible coregulators, the corepressor SHP and the coactivator PGC-1, might give rise to functional antagonism on additional relevant target receptors such as HNF4 (35) and ER␣ (41) including their regulated genes.
When investigating wild-type SHP and the various inhibition-deficient mutants we observed a clear variation with respect to their subcellular distribution. The reason for this variation is currently unknown as no nuclear export or import signal has been identified in SHP. Perhaps putative SHP corepressors have an important role in maintaining the nuclear localization of wild-type SHP and SHPmt1.2 but not of SHP⌬H12 (27). In addition, the distribution of wild-type SHP in both cytoplasm and nucleus could indicate a shuttling potential and reveal a possible, novel regulation mechanism of SHP activity, as previously being described for DAX-1 function and thought to be connected with RNA binding (42).
The relocalization of GR in the presence of SHP, from a rather diffuse nuclear distribution to a specific dot pattern that overlaps with the SHP dots, indicates a specific in vivo interaction between the two proteins. Notably, a similar phenomenon has been described for RAR␣, which in the absence of ligand is redistributed from a diffuse nuclear pattern into distinct dots, overlapping with the corepressor SMRT (30). Although the functional relevance of the SMRT-RAR or the SHP-GR dots is currently unknown, they could possibly be linked to the inhibition mechanism. For example, SHP could tether GR to subnuclear locations where GR would no longer be accessible for the transcriptional machinery. Alternatively, these locations could represent assembly sites for functional corepressor complexes. In addition, because we have observed that a functional LXXLL-related motif 2 is required for agonistdependent GR relocalization by SHP in vivo, it is likely that this is due to direct physical interactions between the GR AF-2 and SHP motif 2.
The ability of SHP to inhibit dexamethasone-induced activation of the PEPCK promoter, in addition to the fact that SHP and GR are coexpressed in liver, argues for a physiological role of SHP in glucocorticoid signaling. Our finding that SHP antagonizes the activity of PGC-1 on physiologically relevant target receptors and target genes suggests a structural and functional relationship between the corepressor SHP and the coactivator PGC-1 in glucose homeostasis. Notably, the involvement of SHP in glucose homeostasis has already been implicated since SHP was known to potently inhibit HNF-4 activity (11) via direct NR-box-dependent binding to this constitutively active orphan receptor. 2 Further support comes from recently identified natural mutations in the shp coding sequence that might be linked to mild or moderate obesity (43). Functional analysis has shown that what all these mutations have in common is that they reduce the ability of SHP to inhibit HNF-4 transactivation. In view of our present study, it will be very interesting to investigate the subcellular localization of these mutants and their potential to inhibit GR activity.
In conjunction with our findings and those of others we want to propose a model where elevated SHP levels would antagonize PGC-1-mediated coactivation of GR and HNF-4 in the liver, leading to a decrease in the expression of rate-limiting enzymes in gluconeogenesis such as PEPCK. Presumably, this would counteract the stimulatory effects of PGC-1 and thereby lead to a decreased glucose release. The level of GR activity modulation, i.e. the ratio between SHP inhibition versus PGC-1 coactivation, will presumably depend on the relative concentration of SHP in the nucleus and at the promoter regions, which could be regulated by metabolically induced expression or buy ligand/modification-dependent conformational changes affecting the intracellular localization and the interactions of SHP with nuclear target receptors and cofactors. Therefore, future studies have to define the physiological circumstances, natural or pharmaceutical, that will affect the activity of SHP, either at the transcriptional level or at the posttranslational level.