MCB
Home Help [Feedback] [For Subscribers] [Archive] [Search] [Contents]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrowReprints and Permissions
Right arrow Copyright Information
Right arrow Books from ASM Press
Right arrow MicrobeWorld
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Savory, J. G. A.
Right arrow Articles by Lefebvre, Y. A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Savory, J. G. A.
Right arrow Articles by Lefebvre, Y. A.

 Previous Article  |  Next Article 

Molecular and Cellular Biology, February 1999, p. 1025-1037, Vol. 19, No. 2
0270-7306/99/$04.00+0
Copyright © 1999, American Society for Microbiology. All rights reserved.

Discrimination between NL1- and NL2-Mediated Nuclear Localization of the Glucocorticoid Receptor

Joanne G. A. Savory,1 Brian Hsu,1 Ian R. Laquian,1 Ward Giffin,2 Terry Reich,2 Robert J. G. Haché,2,3,* and Yvonne A. Lefebvre2,3,*

Departments of Medicine2 and Biochemistry, Immunology and Microbiology3 and Graduate Program in Biochemistry,1 The Loeb Health Research Institute at the Ottawa Hospital, University of Ottawa, Ottawa, Ontario, Canada

Received 16 June 1998/Returned for modification 4 August 1998/Accepted 3 November 1998


    ABSTRACT
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

Glucocorticoid receptor (GR) cycles between a free liganded form that is localized to the nucleus and a heat shock protein (hsp)-immunophilin-complexed, unliganded form that is usually localized to the cytoplasm but that can also be nuclear. In addition, rapid nucleocytoplasmic exchange or shuttling of the receptor underlies its localization. Nuclear import of liganded GR is mediated through a well-characterized sequence, NL1, adjacent to the receptor DNA binding domain and a second, uncharacterized motif, NL2, that overlaps with the ligand binding domain. In this study we report that rapid nuclear import (half-life [t1/2] of 4 to 6 min) of agonist- and antagonist-treated GR and the localization of unliganded, hsp-associated GRs to the nucleus in G0 are mediated through NL1 and correlate with the binding of GR to pendulin/importin alpha . By contrast, NL2-mediated nuclear transfer of GR occurred more slowly (t1/2 = 45 min to 1 h), was agonist specific, and appeared to be independent of binding to importin alpha . Together, these results suggest that NL2 mediates the nuclear import of GR through an alternative nuclear import pathway. Nuclear export of GR was inhibited by leptomycin B, suggesting that the transfer of GR to the cytoplasm is mediated through the CRM1-dependent pathway. Inhibition of GR nuclear export by leptomycin B enhanced the nuclear localization of both unliganded, wild-type GR and hormone-treated NL1- GR. These results highlight that the subcellular localization of both liganded and unliganded GRs is determined, at least in part, by a flexible equilibrium between the rates of nuclear import and export.


    INTRODUCTION
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

The predominant pathway for the nuclear import of transcription factors and other nuclear regulatory proteins originates with the interaction of importin alpha -like proteins (also called karyopherin alpha , Rch1/hSRPalpha , hSRP1/NPI-1, and pendulin/OHO31) with specific nuclear localization sequences (NLSs), which contain closely spaced arrangements of five to eight basic amino acids (31, 62, 64). For DNA sequence-specific transcription factors, NLSs generally colocalize with their DNA binding domains (DBDs), which appears to reflect a coevolutionary selective pressure to ensure that proteins that bind DNA are able to access the nucleus (52). Nuclear export, by contrast, occurs through alternative pathways, which for many proteins involves the binding of CRM1 (exportin 1) to hydrophobic nuclear export sequences (26, 90).

However, some transcription factors, including the glucocorticoid hormone receptor (GR), contain additional NLSs that occur in other regions of the proteins (69, 89, 95, 99). In at least some instances, the presence of these additional NLSs has been found to reflect a requirement for specialized or tightly regulated nuclear localization of the protein. For example, the nuclear localization potential of one of the two NLSs in the adenovirus E1A protein is active only during early development (92), while two of the three c-abl NLSs promote nuclear localization of c-abl only in certain cell types (97, 99). Therefore, it seems that some NLSs can function to dictate the nuclear localization of a protein under highly selective physiological conditions.

Steroid receptors are shuttling proteins that actively exchange or shuttle between the nucleus and cytoplasm (16, 34, 55). They each contain a basic-type NLS, NL1, that overlaps with and extends C-terminally from the receptor DBD (52). At least for GR, estrogen receptor (ER), and progesterone receptor (PR), this NLS is comprised of three components (35, 93, 108). A core basic sequence adjacent to the DBD is required for NLS function, while two smaller clusters of basic amino acids at the C terminus of the DBD appear to contribute to increasing the strength of the NLS and thus the efficiency with which these receptors are imported into the nucleus (93, 108). By contrast, how steroid receptors are exported from the nucleus and the identity of their NESs remain to be determined.

In the absence of ligand, steroid receptors are associated into high-molecular-weight complexes that include heat shock proteins (hsps) and immunophilins (72). For GR, association with the hsps appears to be a prerequisite for ligand binding and prevents the interaction of GR with DNA in vivo (5, 9, 71). Glucocorticoids are homeostatic steroids that play key roles in regulating stress responses, the production of surfactants in the lung, and controlling the immune system (13). With the notable exception of the immune system, the major cellular targets of glucocorticoids occur primarily in the G0 and G1 phases of the cell cycle (2, 42).

The distribution of GR in the cell is distinguished from that of ER and PR in at least two ways. First, in most circumstances, the hormone-free, hsp-complexed form of GR is localized to the cytoplasm (69, 83), while ER (77) and PR (74) are constitutively nuclear. Upon hormone treatment, the hsp-immunophilin complex dissociates and the liganded GR is rapidly transferred to the nucleus, where it remains localized as long as ligand is present (55, 83, 107). Somewhat surprisingly, however, following the withdrawal of steroid treatment, GR rapidly reassociates into the hsp-immunophilin ligand binding complexes, but it transfers back to the cytoplasm only slowly, over 12 to 24 h (36, 78, 83). Indeed, we have recently shown that following the withdrawal of the hormone antagonist RU486, GR rapidly regains hormone responsiveness but persists in the nucleus for an indefinite period (36, 83). Notably, this persistence in nuclear localization was not due to a gross defect in nuclear export, as hormone-withdrawn, hsp-associated GRs transferred rapidly between heterokaryon nuclei (36). Moreover, in some instances (particularly when GR is overexpressed), unliganded, complexed, hormone-responsive GRs can be localized partially or even primarily to the nucleus (57, 84). How this accomplished is not understood, but it does not appear to involve a change in the association of GR with the hsps (84). These results highlight a need to reevaluate the previous conclusion that localization of GR to the cytoplasm reflects a unique aspect of the GR-hsp complex that is absent from PR- and ER-hsp complexes (47, 71, 94).

Second, in addition to NL1, the nuclear import of GR appears to be facilitated by a second nuclear localization signal, NL2, that occurs in the 225-amino-acid GR ligand binding domain (LBD) but is absent from ER and PR (69). The identity of NL2 is not obvious, as the LBD of GR does not contain a basic motif suggestive of an NLS. Further, the complex nature of GR-hsp association and sensitivity of ligand binding to changes within the LBD have precluded localization of NL2 by deletion mutagenesis.

While several recent studies have examined the characteristics and properties of the GR NL1 (8, 83, 93), there is little detailed information on the NL2-mediated nuclear import of GR, particularly in the context of the full-length receptor. Early studies demonstrated that NL2 can function efficiently under certain conditions to direct the nuclear localization of the GR LBD and proteins fused to the LBD (69). By contrast, other reports have suggested that NL2 is an inefficient NLS (8, 51).

In the present study, we have examined the determinants required for the nuclear import and maintenance of unliganded, hsp-associated GRs and have dissected the properties of NL2-dependent GR trafficking in fibroblasts synchronized to G0. Our results indicate that nucleocytoplasmic trafficking of unliganded GR reflects the binding of NL1 in hsp-complexed receptors to importin alpha  and the export of GR from the nucleus through the CRM1-dependent pathway. Further, our results indicate that NL2 is an agonist-specific NLS that is likely dependent upon the positioning of the C-terminal alpha -helix of the GR LBD. By contrast to NL1, in G0 NL2 was a weak nuclear import signal that mediated the slow and incomplete transfer of GR to the nucleus. The inability of NL1- GR to bind pendulin/importin alpha  suggested that NL1 and NL2 mediate the nuclear import of GR through separate pathways.


    MATERIALS AND METHODS
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

Plasmids. The plasmids p6RGR, pRSV-beta gal, pMTG-GR, pMMTVCAT (contains mouse mammary tumor virus [MMTV] long terminal repeat sequences -631/+105), pGST-pendulin, pACT-pendulin, pGEMEXLEF-1, pGEMEX-TCF-1, pRESThnRNP A1, and pRSETCBP80 have been described previously (10, 49, 67, 73, 75). p6RGRNL1- was created by mutating amino acids 513KKK515 of wild-type (WT) rat GR to 513NNN515 by site-directed mutagenesis. To mutate 513KKK515 of the NL1 of WT GR to 513NNN515, a BspEI/PstI fragment encoding amino acids 391 to 524 of rat GR was subcloned into pBluescript (Strategene, La Jolla, Calif.). The mutagenesis of NL1 was performed by using the Sculptor kit from Amersham Life Science Inc. (Arlington Heights, Ill.) according to the manufacturer's instructions. The final product was transformed into Escherichia coli DH5alpha , and the NL1 mutation was confirmed by DNA sequencing. A simian virus 40 (SV40) origin of replication was inserted into the p6RGR and p6RGRNL1- vectors by excising the SV40 origin of replication from pRShGRalpha (8) as an ~300-bp NdeI fragment, which was ligated into the NaeI sites of p6RGR and p6RGRNL1-. pMTG-GRNL1- was created by subcloning the MscI/BamHI fragment from p6RGRNL1- into pMTG (73) digested with SmaI and BamHI.

Two constructs that express the amino terminus of either WT GR (pGFPGRN525) or GRNL1- (pGFPGRN524NL1-) fused to green fluorescence protein (GFP) were made. pGFPGRN525 was generated by removing an MscI/BamHI fragment corresponding to amino acids 22 to 525 of GR from GRN525 (30) and inserting it into the SmaI and BamHI sites of pEGFP-C1 (Clontech, Palo Alto, Calif.). pGFPGRN524NL1- was made by first inserting a linker containing a stop codon into the PstI site at amino acid 524 of p6RGRNL1-. An MscI/BamHI fragment was then excised from this construct and inserted into SmaI- and BamHI-digested pEGFP-C1.

To generate pGALGR540C, pGALGR505C, and pGALGR505CNL1-, rat GR LBD (amino acids 540 to 795) (540C) or the LBD with the hinge region (amino acids 505 to 795) (505C) containing either the WT or mutated NL1 sequence was amplified by PCR. The PCR products were then directly fused to the GAL4 DBD in pAS2 (Clontech) to generate the fusion proteins. The pTLGR and pTLGRNL1- expression vectors used for in vitro translation were created by cloning the 2.4-kb BamHI GR fragments from p6RGR and p6RGRNL1- into the pTL1 vector (32). pSP6luciferase was from Promega. All constructs created by PCR amplification of DNA fragments were verified by DNA sequencing.

Cell culture and transfections. Cos7 cells (ATCC CRL 1651) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Transient transfection of cDNA expression plasmids was performed by the manufacturer's protocol with Lipofectamine (Gibco BRL) (15 µl per 6-cm-diameter dish) with an incubation time of 8 to 10 h. Transfections were stopped by adding serum to 10%.

Indirect immunofluorescence. Cos7 cells were transfected with 0.5 µg of cDNA expression plasmids as described above. At 16 h posttransfection, the medium was replaced with phenol red-free DMEM containing 10% charcoal-stripped FBS. Six hours later, the cells were plated onto poly-L-lysine-coated coverslips and incubated overnight in phenol-free DMEM containing 10% charcoal-stripped FBS. Cells were synchronized in G0 by incubation for a further 21 h in serum-free medium. For induction experiments, dexamethasone (Dex), cortisol, and RU486 were added to final concentrations of 10-6 M in serum-free medium. To monitor redistribution of GR to the cytoplasm following hormone withdrawal, cells were pretreated with cortisol (10-6 M) for 6 h, and withdrawal was initiated by three washes with phosphate-buffered saline and two with serum-free medium, followed by incubation in serum-free medium supplemented with bovine serum albumin (BSA) to 5%. In many experiments, 40 to 1,000 µM leptomycin B (Novartis Preclinical Research, Basel, Switzerland) was added to the culture as described for the individual experiments. Indirect immunofluorescence was carried out exactly as described previously (83) with primary murine BuGR2 antibody (Affinity BioReagents, Inc.) on a Zeiss Axiophot photomicroscope. Quantification was performed with double-blind encryption. For each time point, at least 250 stained cells were counted for each sample. Each experiment was repeated two to four times over a period of several months. Immunofluorescent cells were classified into five categories (N, N>C, N=C, C>N, and C), depending upon the localization of GR. Briefly, for cells classified as N, the immunofluorescent signal from GR was localized entirely to the nucleus, with no detectable staining in the cytoplasm. For cells classified as N>C, immunofluorescence was predominantly nuclear, with some fluorescence detectable in the cytoplasm. Cells with GR evenly distributed throughout the cell were classified as N=C, while cells with GR predominantly in the cytoplasm were characterized as C>N. Finally, cells in which no GR immunofluorescence was detectable in the cell nucleus were scored as C. The validity of this scoring system for monitoring the nucleocytoplasmic trafficking of steroid receptors has been previously demonstrated (83, 108). For WT GR, assessment of nucleocytoplasmic trafficking by this technique yields results that closely parallel those obtained by other techniques (43, 63, 81).

Transient-transfection analysis of reporter gene activation. For chloramphenicol acetyltransferase (CAT) assay, Cos7 cells were transfected as described above with pRSV-beta gal (200 ng), an MMTV CAT reporter gene (pMMTVCAT) (200 ng), and either p6RGR or p6RGRNL1- (30 ng). At 24 h after transfection, cells were transferred to serum-free medium and incubated for a further 16 h. The cells were treated with hormone (10-6 M Dex) for 24 h, and CAT assays were performed by a standard protocol. beta -Galactosidase assays of the same samples were used to normalize results for variations in transfection efficiency. Each experiment was performed in duplicate three times. In the figures, the error bars reflect the standard errors of the means from all repetitions. The levels of expression of the GR constructs were verified by Western analysis with BuGR2 as described previously (83). Chemiluminescent signals were quantified by using a CH screen on a Bio-Rad GS-525 phosphorimager.

Analysis of GR-hsp interactions by sucrose density gradient centrifugation. Whole-cell extracts of Cos7 cells transfected with either pMTG-GR or pMTG-GRNL1- prepared before or after incubation with 10-6 M Dex were run on separate 15 to 30% linear sucrose gradients for 16 h at 368,000 × g at 4°C. Fractions (300 µl) were collected, immunoprecipitated with the anti-Myc antibody 9E10, and fractionated through sodium dodecyl sulfate (SDS)-8% polyacrylamide gels. Western immunoblotting was performed with 9E10 as the primary antibody. The chemiluminescent signals from the GRs were quantified by densitometry. Markers for the gradients were aldolase (7.3S) and BSA (4.6S).

Two-hybrid analysis in yeast. The yeast strain Y190 was grown in yeast extract-peptone-dextrose. Transformation was carried out by the lithium acetate method with plasmid DNA. Yeast colonies transformed with fusion constructs were grown in synthetic medium lacking leucine or tryptophan or both. Transformed yeast cells were selected and cultured overnight in the absence of hormone. The yeast cultures were then subcultured (1:10) in fresh selective medium that contained either ethanol or 1 µM deacylcortivasol (DAC) (28) and grown for a further 16 h. The optical density at 600 nm (OD600) was determined, and the cultures were then assayed for beta -galactosidase activity.

Assays for beta -galactosidase activity were performed essentially as described previously (86). The yeast cells were resuspended in 100 µl of 1× Z buffer (10 mM KCl, 1 mM Mg2SO4, 50 mM beta -mercaptoethanol, 100 mM NaHPO4, pH 7.0) and extracted with chloroform (50 µl). Following addition of 700 µl of 2-mg/ml o-nitrophenyl-beta -D-galactopyranoside (ONPG) in Z buffer, the tubes were incubated at 30°C until a yellow color developed. The reaction was stopped by the addition of 500 µl of 1 M Na2CO3. beta -Galactosidase activity was calculated from the OD420 as (1,000 · OD420)/(t · v · OD600), where t is the reaction time (minutes) and v is the initial culture volume (milliliters).

In vitro binding to GST fusion proteins. pGST-pendulin and the glutathione S-transferase (GST) expression vector pGEX-3X were transformed into E. coli BL21(DES)/pLys, grown to an OD600 of 0.8, and then induced with 0.1 mM isopropyl-beta -D-thiogalactopyranoside (IPTG) overnight at room temperature. GST and the GST-pendulin fusion protein were prepared as described previously (76). WT GR, GRNL1-, human cap binding protein (CBP 80), heterogeneous nuclear RNP A1 (hnRNP A1), lymphoid enhancer factor 1 (LEF1), and T-cell factor 1 (TCF1) were in vitro translated in the presence of [35S]methionine by using the coupled transcription-translation TNT rabbit reticulocyte lysate. Transformation of in vitro-translated GRs was accomplished by treatment with Dex (10-6 M) for 2 h at 4°C (18). GR-pendulin binding assays were performed as previously described (75). When the binding to pendulin of liganded GR and the binding of other proteins were compared, the in vitro-translated proteins were added to 0.5 µg of either GST or GST-pendulin in TBST (10 mM Tris [pH 8.0], 200 mM NaCl, 0.2% Tween 20, 200 µM ethidium bromide, 100 µg of RNase A per ml, and 0.1% Nonidet P-40) for 30 min at room temperature. Following three washes with the same buffer, the proteins retained on the matrix were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Binding was compared to that of 10% of the in vitro-translated proteins added to the incubation mixtures.

For comparison of the binding of liganded and hsp-associated, unliganded GRs to GST-pendulin, GST-pendulin and GST beads preincubated in unprogrammed reticulocyte lysate were added directly to in vitro-translated WT GR following in vitro translation in the presence or absence of 10-6 M Dex. Binding was performed for 30 min at 30°C. Samples were washed three times with TBST supplemented with sodium molybdate to 20 mM. Half of the samples were resolved by SDS-PAGE and subjected to autoradiography, while the other half were assessed for the presence of hsp90 by Western immunoblotting with an hsp90 antibody (Stressgen Biotechnologies Corp., Victoria, Canada).


    RESULTS
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

Replacement of lysines 513 to 515 with asparagines inactivates the GR NL1. Redistribution of GR between the nucleus and cytoplasm upon hormone treatment and withdrawal is accomplished through a complex process that can take several hours to complete (36, 78, 83). The rapid changes in biosynthesis and degradation of GR that accompany the onset and withdrawal of hormone treatment (7, 12, 23) have the potential to obscure the intracellular trafficking of individual GR molecules. Previously we have demonstrated that the expression of rat GR in Cos7 cells by transient transfection, followed by culture in the absence of serum, allows for study of the nucleocytoplasmic trafficking of stably maintained pools of GRs in cells synchronized to G0 (36, 83). In cells manipulated in this manner, there is minimal new synthesis or degradation of GR for periods of up to 72 h following serum withdrawal.

In order to study the properties of NL2 in the nuclear import of a stably maintained population of full-length GR in Cos7 cells, it was first necessary to inactivate NL1. Replacement of lysine or arginine residues with asparagine has been demonstrated to inactivate simple basic NLSs (80). The GR NL1 is somewhat more complex than these simple NLSs, comprising a tripartite basic motif (Fig. 1A). The C-terminal cluster of basic amino acids in the GR NL1 (amino acids 510 to 517) comprise a core sequence that is required for NL1 function (8, 51, 69, 93). By contrast, the two smaller N-terminal clusters of basic residues in NL1 contribute to the efficiency of the nuclear uptake of GR but are not sufficient to direct the nuclear import of GR peptides (51, 93). Therefore, we reasoned that the conversion of lysines 513 to 515 in the GR NL1 core to asparagines (GRNL1-) might be sufficient to abrogate NL1 activity in full-length GR. These substitutions did not appear to affect the synthesis or stability of GRNL1-, since when transfected into Cos7 cells, WT GR and GRNL1- were expressed to similar levels (Fig. 1B).


View larger version (25K):
[in this window]
[in a new window]
  		FIG. 1.   Site-directed mutagensis of the core motif of NL1 of rat GR. (A) NL1 of rat GR overlaps with the receptor DBD and is composed of a tripartite basic motif (labeled 1 to 3) (93). By contrast, the smallest GR fragment know to contain NL2 extends from amino acid 547 to 795 and encompasses the entire GR LBD as well as the determinants for hsp association (amino acids 568 to 616). Site-directed mutagenesis was used to convert lysines 513 to 515 to asparagines (lowercase) in GRNL1-. (B) Western blot with GR-specific antibody BuGR2 of GR (lane 1) and GRNL1- (lane 2) immunoprecipitated from whole-cell extracts prepared 20 h following transient transfection of p6RGR and pGRNL1-.

To determine whether the three amino acid substitutions in GRNL1- eliminated the potential for NL1 to direct the nuclear uptake of GR, we first examined the effect of the same substitutions on the subcellular localization of truncated GRs in which the LBD had been deleted to remove NL2. GRN525 is a C-terminal deletion mutant of WT GR that lacks the ligand binding, NL2, and hsp association properties of WT GR. This GR fragment is constitutively localized to the nucleus under normal cell culture conditions in an NL1-dependent manner (69).

Addition of GFP to the N terminus of GRN525 (Fig. 2A), increases the size of the GR peptide to above 60 kDa, which prevented the entry of these peptides into the nucleus by passive diffusion (17, 66). Expression of GFPGRN525 by transient transfection led to the production of a protein that was localized to the nucleus (Fig. 2B). By contrast, the GFPGRN524NL1- peptide was localized to the cytoplasm (Fig. 2B). The same result was observed whether the localization of the GRN525 peptides was determined by direct examination of GFP fluorescence or by indirect immunofluorescence with the anti-GR antibody BuGR2 (40). We conclude from this data that alteration of Lys 513 to 515 abrogated the nuclear localization activity of the GR NL1 in Cos7 cells.


View larger version (55K):
[in this window]
[in a new window]
  		FIG. 2.   Conversion of Lys 513 to 515 to Asn localizes the N terminus of GR to the cytoplasm. (A) Schematic presentation of GFPGRN525 and GFPGR524NL1-. The asterisk indicates the position of the Lys-Asn substitution. (B) Immunofluorescent photomicrographs of cells, illustrating the subcellular localization of GFPGRN525 and GFPGR524NL1- expressed by transient transfection in Cos7 cells. The photomicrographs displayed are illustrative of the immunofluorescence pattern observed in over 300 cells examined in each duplicate of three independent experiments. Overall, GFPGRN525 was at least mostly nuclear in 98% ± 1.5% of transfected cells, while GFPGR524NL1- was nuclear or partially nuclear in 2% ± 1.5% of cells.

NL2-mediated nuclear uptake of GRNL1- in G0 differs markedly from nuclear transfer of WT GR. Overexpression of GR can promote the partial transfer of unliganded, hsp-associated receptor to the nucleus (57, 84). How this is accomplished is not understood. Lipofectamine-mediated transfection of Cos7 cells with GR expression plasmids that replicate from SV40 replication origins resulted in a level of GR expression at which the WT receptor was partially localized to the nucleus (28% ± 9% N + N>C, 18% ± 3% N=C) prior to exposure to ligand (Fig. 3A and B, t = 0 h).


View larger version (21K):
[in this window]
[in a new window]
  		FIG. 3.   Site-directed mutagenesis of NL1 decreased the nuclear localization of GR in the absence of hormone and in response to Dex and RU486. Changes in the subcellular distribution of WT GR (bullet ) and GRNL1- () following treatment with Dex (A and C) or RU486 (B and D) are plotted as a function of time. (A and B) Percentage of cells with mostly nuclear GR (N + N>C) following Dex or RU486 treatment; (C and D) percentages of cells with completely nuclear staining (N) over the same period. Note that time is plotted on a logarithmic scale.

By contrast, GRNL1- expressed from the same vector at similar levels (39) was completely localized to the cytoplasm in the absence of hormone treatment (Fig. 3A and B). This result establishes the cytoplasmic localization of GRNL1- in the absence of steroid but also provides the first direct evidence that the accumulation of overexpressed WT GR in the nucleus is dependent upon NL1.

Treatment with the steroid agonist Dex or the steroid antagonist RU486 induced the transfer of the remaining cytoplasmic WT GR to the nucleus within 10 (Dex) to 30 (RU486) min (Fig. 3C and D). The slower initial nuclear transfer observed for RU486-treated GRs (Fig. 3D) is consistent with the decreased rate of transformation of hsp-associated GRs by this antagonist (19, 56, 58). Both results are exactly consistent with previous reports for the rate of the ligand-dependent nuclear uptake of WT GR (36, 61, 83).

By contrast, the nuclear transfer of GRNL1- was slow, inefficient, and agonist specific. Significant nuclear transfer of GRNL1- was not observed until 30 min following Dex treatment, with equilibrium attained after 2 h (Fig. 3C). Further, at equilibrium, GRNL1- was mostly nuclear in only 55 to 60% of the cells. In over 90% of the remaining cells, GR was equally distributed between the nucleus and cytoplasm (40). However, once attained, the equilibrium distribution of GRNL1- in the presence of Dex was maintained for at least 24 h.

Interestingly, NL2 appeared to be agonist specific, as treatment of GRNL1--expressing cells with RU486 was almost completely ineffective in promoting the transfer of GRNL1- to the nucleus (Fig. 3D). Fewer than 15% of the transfected cells displayed significant accumulation of GRNL1- in the nucleus even 10 h after RU486 treatment. This result suggested that NL2 activity was not simply dependent on the exposure of the LBD through dissociation of the hsps from GR but rather appeared to be specifically dependent upon the tertiary LBD structure resulting from the binding of hormone agonists (1, 6).

However, another potential explanation for the striking differences in the transfer of GRNL1- to the nucleus following treatment with Dex or RU486 is that GRNL1- had a decreased ability to dissociate from the hsps following ligand binding. To examine this possibility, we compared the sedimentation behavior of WT GR and GRNL1- on sucrose gradients before and after hormone treatment of the Cos7 cells (Fig. 4).


View larger version (19K):
[in this window]
[in a new window]
  		FIG. 4.   GRNL1- is rapidly dissociated from the Hsp-immunophilin protein complex following treatment with Dex or RU486. (A) Sucrose gradient sedimentation profile (10 to 35% linear sucrose gradients) of WT GR in whole-cell extracts prepared from transiently transfected control Cos7 cells (open circle ) or cells treated with Dex for 2 h (bullet ). (B and C) Sucrose gradient sedimentation profile of GRNL1- in whole-cell extracts prepared from transiently transfected control cells () or cells treated with Dex (B) or RU486 (C) for either 15 min () or 2 h (black-triangle). The GRs were immunoprecipitated from gradient fractions with BuGR2, subjected to SDS-10% PAGE, and identified by Western blotting. The signals were quantified by densitometry and phosphoimager analysis. The positions of the aldolase and BSA size markers are indicated.

WT GR and GRNL1- sedimented at 8S on sucrose gradients when extracted from untreated cells (Fig. 4), reflecting the association of both factors into the ligand binding hsp-GR complexes (54). Treatment of Cos7 cells expressing WT GR with Dex resulted in the rapid and complete dissociation of the GR-hsp complex to free GR, which sedimented at 4S on the sucrose gradient (Fig. 4A). A similar result is also obtained following RU486 treatment (83). GRNL1- also rapidly dissociated the hsps in response to both Dex and RU486. Within 15 min of treatment with either Dex or RU486, a time prior to the transfer of significant GRNL1- to the nucleus in response to Dex, GRNL1- sedimented almost entirely at 4S (Fig. 4B and C). The same result was obtained 2 h following ligand treatment. These results indicated that it was extremely unlikely that differences in the nuclear transfer of GRNL1- and WT GR in response to Dex and RU486 were due to a defect in receptor transformation upon exposure to ligand.

The three lysine-asparagine substitutions in GRNL1- were made immediately adjacent to the DBD of GR. We have previously demonstrated that the ability of GR to bind DNA is an important determinant for complete localization of the liganded receptor to the nucleus (83). However, recombinant WT and NL- GR DBD peptides expressed and purified from bacteria as His-tagged fusion proteins exhibited the same DNA binding properties (39). Therefore, the decreased nuclear transfer of GRNL1- in response to Dex also was unlikely to have reflected a significantly decreased affinity of GRNL1- for DNA.

The difference in the NL2-dependent nuclear uptake of GR in response to Dex and RU486 indicated that the activation of NL2 was not merely dependent upon the dissociation of the hsp-immunophilin complex from GR but was specifically dependent upon the conformation of the LBD that was induced by ligand. The minimal domain of GR required for high-affinity ligand binding extends from amino acid 550 to 795 (82, 104). Ligand binding to nuclear receptors induces a change in LBD conformation that is marked by the repositioning of the C-terminal alpha -helix over the LBD core. However, hormone antagonists induce a different change in conformation than hormone agonists. Similar conformational changes are postulated for GR agonist- and antagonist-bound GR (103). For some nuclear receptors, the C-terminal alpha -helix can be removed from the receptor without completely compromising ligand binding. Whether this can be accomplished for GR is unclear, as different results have been reported (82, 109).

Nonetheless, to attempt to determine whether the C-terminal alpha -helix of GR was required for NL2, we expressed GR and GRNL1- constructs truncated through the C-terminal alpha -helix of the GR LBD to amino acid 781, which have been reported to maintain a low level of hormone binding (82). However, consistent with the results of Zhang et al. (109), GR781 and GR781/NL1- were completely unable to respond to hormone, as they remained completely 8S in response to 10 µM Dex and hormone treatment had no effect on their subcellular localization (53).

The activation of reporter gene transcription is proportional to the nuclear occupancy of GR. Restricting the access of transcription factors to the nucleus is an effective way to control their ability to regulate transcription (37, 70, 96). To compare the transcriptional regulatory potentials of WT GR and GRNL1-, we examined the expression of a cotransfected CAT reporter gene whose transcription was dependent on the promoter-proximal steroid-regulatory region of MMTV (Fig. 5). To prevent the possibility of transcriptional squelching in this experiment as a result of high levels of expression of the GRs, the GR expression plasmids employed in this experiment were transfected in reduced quantity and lacked the SV40 replication origin. Under these conditions, treatment of the cells transfected with WT GR for 24 h with Dex resulted in a typically strong induction of expression of the CAT reporter gene. By contrast, over the same period, reporter gene expression in cells transfected with GRNL1- was three- to fourfold lower. Thus, the reduction in CAT activity was exactly consistent with reduced nuclear transfer of GRNL1-.


View larger version (18K):
[in this window]
[in a new window]
  		FIG. 5.   Transcriptional activation of an MMTV CAT reporter gene by GRNL1- reflects its decreased nuclear occupancy. Cells transfected with equal amounts of WT GR and GRNL1- expression plasmids were assayed for activation of a cotransfected MMTV CAT reporter gene 24 h following the addition of 2 × 10-6 M Dex (h+) or vehicle (h-) to cells withdrawn from serum for 16 h. CAT activity is expressed relative to the response obtained for WT GR. Error bars represent the standard errors of the means from three experiments performed in duplicate.

NL1 mediates the prolonged nuclear retention of GR following hormone withdrawal. Another intriguing feature of the nucleocytoplasmic trafficking of GR is that, under most circumstances, it redistributes only slowly over a 12- to 24-h period to the cytoplasm following hormone withdrawal (36, 55, 83). By contrast, the reassociation of GRs with hsps is complete within 1 to 2 h, and the receptor continues to transfer efficiently between heterokaryon nuclei thereafter, suggesting that the recycled receptors are rapidly exported from the nucleus (36, 60, 61). To examine whether the extended nuclear occupancy of GR following hormone withdrawal exhibits the same NL1 dependence as the localization of overexpressed, naive receptors to the nucleus, we compared the redistributions of WT GR and GRNL1- to the cytoplasm upon withdrawal of the natural glucocorticoid cortisol (Fig. 6). In the overexpression system employed for these experiments, the redistribution of WT GR from the nucleus occurred even more slowly than reported previously for cells with lower levels of receptor, with GR remaining mostly nuclear in upwards of 80% of the cells 24 h following the withdrawal of cortisol. By contrast, GRNL1- rapidly redistributed to the cytoplasm following hormone withdrawal and had completely returned to the cytoplasm by 2 h after withdrawal. This result indicated that the prolonged maintenance of GR in the nucleus following hormone withdrawal required an intact NL1.


View larger version (13K):
[in this window]
[in a new window]
  		FIG. 6.   GRNL1- redistributes rapidly to the cytoplasm following hormone withdrawal. Cells expressing WT GR (bullet ) and GRNL1- () were treated with cortisol for 6 h. Hormone withdrawal was initiated by thoroughly washing the cells as described in Materials and Methods and incubating the cells thereafter in serum-free medium containing 5% BSA. The percentage of cells in which GR was mostly nuclear (N + N>C) is plotted as a function of time. Note that time is displayed on a logarithmic scale.

Inhibition of CRM1-mediated nuclear exports promotes the nuclear localization of WT unliganded GR and Dex-treated GRNL1-. GR is a shuttling protein that trafficks between the nucleus and cytoplasm when liganded and when targeted to the nucleus in the absence of ligand by a exogenous nuclear retention signal (36). However, the mechanism whereby GR is exported from the nucleus has not been determined.

To begin to examine the molecular basis for the nuclear export of GR, we treated cells expressing WT and NL- GRs with leptomycin B, a specific inhibitor of the CRM1 nuclear export pathway (102). For WT GR, treatment with 200 nM leptomycin B in the culture medium for 1 h promoted a striking shift in unliganded GRs to the nucleus (Fig. 7A). This result suggests that the export of GR from the nucleus was accomplished at least in part through the CRM1 pathway. Increasing the time of leptomycin treatment to 2 and 4 h, which is comparable to that used in previous studies with leptomycin B (105), induced a much greater transfer of unliganded GR to the nucleus (53). However, incubation of serum-starved Cos7 cells with leptomycin B for 4 h also raised the potential of effects mediated through cytotoxicity. Therefore, in this and subsequent experiments, leptomycin B treatments were limited to 1 h, a treatment from which the cells recovered without apparent ill effects.


View larger version (22K):
[in this window]
[in a new window]
  		FIG. 7.   Leptomycin B promotes the nuclear localization of unliganded WT GR and GRNL1-. Changes in the subcellular distribution of WT GR (A) and GRNL1- (B) following treatment for 1 h with Dex or with Dex and leptomycin B (LMB) are shown. The localization of GR was scored by indirect immunofluorescence. The plots illustrate the percentages of cells in which GR was mostly nuclear (N + N>C), evenly distributed throughout the cell (N+C), and primarily cytoplasmic (N<C). The error bars represent the standard errors of the means from three independent experiments performed in duplicate.

By contrast to WT GR, GRNL1-1 remained entirely cytoplasmic following leptomycin B treatment, further supporting the conclusion that the nuclear import of unliganded GR was entirely NL1 dependent (Fig. 7B). In addition, this result also clearly indicated that the effects of leptomycin B on GR localization were NLS dependent.

Addition of leptomycin B to GRNL1--expressing cells together with Dex promoted a shift in GRNL1- towards the nucleus that was very similar to that observed with the WT, unliganded receptor (Fig. 7B). This result indicated that the substitution that inactivated NL1 appeared to have little, if any, effect on the nuclear export of GR through the CRM1 pathway. By contrast, the addition of leptomycin B to cells treated with RU486 was ineffective in promoting nuclear uptake of GRNL1- (53), lending further support to the proposal that NL2 was agonist specific.

Mutation of NL1 abrogates the association of liganded GR with pendulin in a yeast two-hybrid system. Pendulin is a murine homologue of the human importin alpha  proteins, which are known to directly recognize basic NLSs and which are presumed (76), but have not yet been confirmed, to mediate the nuclear import of GR by NL1. By contrast, the region of GR containing NL2 lacks an obvious basic motif. This, together with the markedly reduced rate of nuclear transfer of GRNL1-, suggested that NL2 may mediate the nuclear uptake of GR by an alternative mechanism that does not involve importin alpha -like proteins.

In a first experiment to examine the potential for the recognition of NL1 and NL2 by pendulin, we tested the ability of yeast GAL4-DBD-GR-LBD fusion proteins to associate with pendulin in yeast following hormone treatment (Fig. 8). In the yeast two-hybrid protein-protein binding assay, expression of a beta -galactosidase gene dependent upon GAL4 DNA binding sites is activated when a fusion protein containing the DNA-bound GAL4 DBD comes into contact with a second fusion protein containing the GAL4 activation domain (24).


View larger version (20K):
[in this window]
[in a new window]
  		FIG. 8.   Replacement of Lys 513 to 515 by Asn abrogates the binding of the GR C terminus to pendulin in yeast. (A) Results of a two-hybrid assay in yeast reflecting the activation of a GAL4 response element beta -galactosidase reporter gene following the expression of GAL4-DBD-GR-LBD fusion proteins in the presence (bars 1, 2, and 9 to 14) or absence (bars 2 to 8) of murine pendulin fused to a GAL4 activation domain. The GAL-DBD-GR-LBD fusion proteins expressed were GALGR505C, GALGR540C, and GALGR505CNL1-. One micromolar DAC was added as indicated. The error bars represent the standard errors of the means from three experiments performed in duplicate. (B) Western blot indicating the levels of expression of GALGR505C (lanes 1 and 2), GALGR540C (lanes 3 and 4), and GALGR505CNL1- (lanes 5 and 6) in the presence or absence of 1 µM DAC.

GR and the GR LBD fused to the yeast GAL4 DBD or activation domain associates in yeast with hsps into complexes analogous to those that occur in mammalian cells (11, 28, 86). These complexes are fully responsive to the synthetic glucocorticoid agonist DAC (28). However, pendulin activates beta -galactosidase transcription strongly when expressed in yeast fused to the GAL4 DBD (85). Because this strong activation function could mask a two-hybrid interaction with GR, we were limited in these experiments to working with pendulin fused to the GAL4 activation domain and with the GR LBD fused to the GAL4 DBD. This in turn limited us to examining the interaction between pendulin and liganded GR, as the GAL4-DBD-GR-LBD fusion protein-hsp complexes could not be expected to bind to the beta -galactosidase promoter. Nuclear localization of all constructs was constitutive and mediated by yeast NLSs in the GAL4 portions of the fusion proteins and thus independent of the GR NL1 and NL2 sequences. Fortuitously, the GAL4 NLSs do not interact with the murine pendulin construct (75, 76).

In a control experiment, coexpression of the GAL4 DBD together with a pendulin-GAL4 activation domain fusion protein was unable to induce beta -galactosidase activity in Y190 cells (Fig. 8A, bars 1 and 2), reiterating previous results that pendulin is unable to interact with the GAL4 DBD or NLS in this system (76).

Three GAL4-DBD-GR-LBD fusion proteins (GALGR505C, GALGR540C, and GALGR505CNL1-) were also unable to activate transcription of the reporter gene appreciably, even following DAC treatment (Fig. 8A, bars 3 to 8). The lack of beta -galactosidase activity in the presence of DAC was consistent with the weak activity reported for the GR AF2 region in yeast (28).

However, coexpression of GALGR505C, which contained the GR NL1, with the pendulin-GAL4 activation domain fusion protein resulted in a strong, DAC-dependent induction in beta -galactosidase activity (Fig. 8A, bars 9 and 10), indicating that this construct associated with pendulin in the presence of ligand. By contrast, GALGR540C, in which the GR NL1 had been completely deleted but which contained a functional LBD (69), failed to associate with pendulin (bars 11 and 12), despite being expressed at a higher level than GALGR505C (Fig. 8B). Similarly, GALGR505CNL1-, which contained the same three Lys-Asn substitutions as GRNL1-, was also unable to induce significant reporter gene activity when coexpressed with the pendulin fusion protein (Fig. 8A, bars 13 and 14). These results demonstrate that the GR NL1 sequence can associate with pendulin in vivo and that the lysine-asparagine substitutions in GR505CNL1- abrogated this interaction. NL2, however, was unable to associate with pendulin in this assay.

Association of hsp-complexed GR with pendulin in vitro. The dependence on NL1 for the binding of pendulin to full-length GR was investigated further in a GST pulldown assay (Fig. 9). In the first instance, we compared the ability of liganded GR and GRNL1- to bind specifically to GST-pendulin with the binding of several other proteins whose pendulin/importin alpha  binding characteristics have been characterized in detail (49, 75, 98). As shown in Fig. 9A, WT, liganded GR bound to pendulin (lane 8) in a manner that was highly similar to the binding of CBP 80 (lane 9) and LEF1 (lane 11). By contrast, GRNL1- failed to bind GST-pendulin and thus behaved similarly to hnRNP A1 (lane 4) and TCF1 (lane 12), two nuclear proteins that have been previously reported to be unable to bind appreciably to pendulin/importin alpha  (49, 75).


View larger version (31K):
[in this window]
[in a new window]
  		FIG. 9.   Binding of liganded and unliganded, hsp-associated GRs to GST-pendulin is sensitive to the replacement of Lys 513 to 515 by Asn. (A) SDS-polyacrylamide gels comparing the binding of 35S-labeled WT GR, GRNL1-, CBP 80, hnRNP A1, LEF1, and TCF1 to GST-pendulin (lanes 7 to 12) and GST (lanes 13 to 18). Ten percent of the in vitro-translated proteins are shown in lanes 1 to 6. (B) SDS-polyacrylamide gel comparing the binding of in vitro-translated Dex-treated and unliganded WT GR to GST-pendulin (lanes 3 and 4) and GST (lanes 5 and 6). (C) Western blot of one-half of the samples shown in panel B probed with an hsp90 antibody to assess the coprecipitation of hsp90 with GST-pendulin-bound GR (lanes 3 and 4). The slight band in lane 4 represents spillage of some of the sample in lane 3.

Over the past several years, rabbit reticulocyte lysate has been demonstrated to be an in vitro system in which GR and other steroid hormone receptors are assembled into hsp-immunophilin complexes that accurately reflect the complexed state of GR in the cell (14, 21, 46, 87, 88). Indeed, this system has been used in several studies to assess the pathway through which the assembly of GR with hsps takes place (20, 22, 45, 48, 91). Therefore, this system also offered an opportunity to determine whether unliganded, hsp-complexed GR could also associate with pendulin through the GR NL1.

GR associates stably with hsps in reticulocyte lysate at 30°C (14, 46, 47, 87, 88). However, it can transform or dissociate the hsps spontaneously if diluted. Therefore, to test the binding of native, hsp-complexed GR to pendulin, we resuspended the GST-pendulin beads in undiluted, in vitro-translated, unliganded WT GR and in Dex-liganded WT GR. Following the equilibration of binding, the glutathione-Sepharose beads were washed extensively and GR binding was assessed by SDS-PAGE analysis of the material associated with the beads.

Both GR forms bound strongly to GST-pendulin in preference to GST alone (Fig. 9B). Further, Western analysis of pendulin-bound material with an hsp90 antibody showed that hsp90 was associated with the unliganded GR bound to pendulin but that no hsp90 was associated with GST alone (Fig. 9C). Thus, as predicted by our in vivo results, the association of GR into its chaperone complex did not prevent GR-pendulin binding. However, phosphorimager analysis of multiple repetitions of this experiment indicated that the binding of liganded GR to GST-pendulin was 2 (± 0.6)-fold higher than the binding of unliganded, hsp-associated receptor, suggesting the possibility that hsp association may be partially effective in decreasing access of importin alpha  to NL1.

Taken together, our results suggest that NL1 mediated the nuclear transfer of unliganded and liganded GR through the pendulin/importin alpha  nuclear import pathway. By contrast, NL2 seems to mediate the nuclear transfer of GR through a pathway that functions independently of pendulin.


    DISCUSSION
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

Our results distinguish the nuclear import of GR by NL1 from that by NL2. Specifically, they suggest that NL1 mediates the nuclear localization of unliganded and liganded GRs through the importin alpha -importin beta  pathway, while NL2 is a hormone agonist-specific NLS that may direct the nuclear import of GR through an importin alpha -independent pathway. Moreover, our results provide the first evidence that the export of GR from the nucleus may be accomplished through the CRM1-dependent nuclear export pathway.

The basic NLSs of GR (NL1), ER, and PR are similar in position and sequence (35, 52, 69, 108), and the three unliganded receptors are complexed into highly similar hsp-immunophilin-containing 8S complexes (72). For PR and ER, these basic motifs are responsible for the localization of the unliganded, hsp-associated receptors to the nucleus (35, 108). However, it has long been stated that the cytoplasmic localization of GR in the absence of hormone results from a physical masking of NL1 by the hsps (69, 78, 101) in a manner similar to that by which Ikappa B masks the NLS of NFkappa B (3, 27). The primary evidence for this has been that a monoclonal antibody to the sequences around the GR NL1 exhibits a decreased affinity for in vitro-translated GR stabilized by the addition of 20 mM sodium molybdate. However, molybdate also stabilizes the hsp-complexed forms of other receptors, and the microinjection of molybdate into live cells leads to the rapid transfer of unliganded PR to the cytoplasm (106). Together, these results support the alternative possibility that molybdate artificially stabilizes steroid receptor-hsp complexes in a way that masks their basic NLS.

Our results offer strong support to the argument that the GR NL1 is functional in the context of the hsp-immunophilin-associated receptor. First, the partial transfer of overexpressed naive GRs to the nucleus was eliminated by site-directed mutagenesis of the NL1 core motif. Second, the persistence of hsp-immunophilin-associated GRs in the nucleus following the withdrawal of hormone treatment (36, 78) also was completely sensitive to mutation of NL1. Further, our previous heterokaryon experiments (36) and the present leptomycin B treatment of the cells indicated that the localization of hsp-associated GRs in the nucleus was not due to a gross defect in nuclear export.

The GR NL1 is a basic motif similar to those shown to mediate the nuclear import of proteins through the importin alpha -importin beta  pathway. Our results indicated that NL1 is very likely to mediate the nuclear import of GR through the same pathway. The NL1-dependent binding of GR to murine pendulin/importin alpha  was observed both in a yeast two-hybrid system and in vitro in a GST pulldown experiment. Further, in vitro binding occurred for both liganded free GR and hsp-associated receptors. Notably, the binding of hsp-associated GR to pendulin was strongly decreased by the addition of 20 mM molybdate to the binding assay (85). This is the first report of the interaction of a nuclear hormone receptor with importin alpha .

If the GR NL1 is constitutively accessible, what then may be responsible for the localization of naive GR to the cytoplasm and the slow return of hormone-withdrawn GR to the cytoplasm? At the present time, two possibilities seem most likely. First, it is possible that GR is maintained in the cytoplasm through a specific cytoplasmic retention signal and that the transfer of overexpressed GR reflects the saturation of cytoplasmic retention sites. The nature of such a signal is not obvious, but it is interesting that chimeric ER-GR receptors have been observed to be localized to the cytoplasm when the chimeras include the DBD and LBD of GR (68).

In this scenario it would also be necessary to account for the slow relocalization of rapidly shuttling GRs to the cytoplasm that follows hormone withdrawal. One possibility is that liganded GRs would become modified in a way, perhaps through phosphorylation, that would mask the cytoplasmic retention signal prior to hormone withdrawal but which would only slowly be reversed upon the reassociation of the unliganded receptor with the hsps. Alternatively, it has been proposed that following hormone withdrawal, hsp-associated GRs become localized to the nuclear matrix, which could provide nuclear retention sites for the shuttling receptor (93). Retention at the nuclear matrix could also be reversed by a slow reversal of ligand-induced posttranslational modification. We note that GR is a phosphoprotein and that the phosphorylation of GR following treatment with hormone agonist is only slowly reversed upon hormone withdrawal (4, 38, 44, 65). Further, GRs are differentially phosphorylated following treatment with RU486 (42), which appears to lead to the permanent localization of GR to the nucleus following hormone withdrawal (36, 78, 83).

Second, it also is possible that the subcellular localization of unliganded GRs under various conditions reflects subtle changes or limitations in the rates of the nuclear import and export of the liganded and unliganded receptor. Although the rates of protein transfer through nuclear pores in the two directions appear to be comparable (25, 100), import and export are mediated through distinct signals on the proteins being transported, which may be differentially regulated. Several examples of the regulation of nuclear import through protein posttranslational modification have been reported (50, 59, 79). In this context it is interesting that serine 527 of GR, which occurs just C terminal to the core NL1 sequence, may be a phosphorylation site for DNA-dependent protein kinase (29). It is expected that determination of the events that lead to relocalization of hsp-immunophilin-complexed GR will require a careful reconstruction of the nuclear import and export of the receptor in a reconstituted system.

More simply, however, our GST pulldown assays indicated that pendulin bound approximately twofold more efficiently to liganded GR than to unliganded receptor, suggesting that hsp association decreased the binding of GR to pendulin. Thus, the hsp association of GR, while still allowing receptor import, may decrease the affinity of GR for pendulin/importin alpha  sufficiently to decrease the rate of GR import. Such a decrease in import rate could explain the predominant cytoplasmic localization of shuttling, unliganded GR.

Finally, one early paper suggested that a PR peptide including the NLS could function to mediate the slow, energy-independent export of a beta -galactosidase fusion protein from the nucleus (33). However, it appears clear from our results that the 3-amino-acid substitution in the GR NL1 element is unlikely to have significantly affected the export of GR, since GRNL1- rapidly relocalized to the cytoplasm following the withdrawal of hormone.

By contrast to the case for NL1, the nuclear transfer of GR mediated through NL2 appeared to be strictly hormone dependent. GRNL1- remained almost completely cytoplasmic in response to treatment with RU486 and was rapidly redistributed to the cytoplasm upon the withdrawal of steroid agonist. NL2-mediated nuclear import of GR occurred much more slowly than import of the WT receptor. This difference in kinetics, together with the inability of liganded GRNL1- to interact with pendulin in our experiments, suggests that the nuclear import of GR by NL2 occurred through a pathway distinct from that involving importin alpha -like proteins. These results also are consistent with the lack of an obvious basic motif in the GR LBD.

The C-terminal alpha -helix of GR contains determinants critical to the AF-2 transcriptional activation function of GR (15). One possibility suggested by the agonist dependence of NL2 is that nuclear uptake of GR through this motif may occur as the result of the association of GR with transcriptional coactivators that would cotransport GR to the nucleus. The AF-2 of GR is closely related to the AF-2s of ER and PR and interacts with a similar array of transcriptional coactivators (15). However, as neither PR nor ER has NL2 activity, it seems unlikely that the GR NL2 would function in this way.

The nature of the initial 30-min delay or lag in the transfer of GRNL1- to the nucleus in response to treatment with hormone agonist is unclear. Our control experiments indicate that the GRs appear to transform or dissociate from the hsp-immunophilin complex appropriately and that the substitutions in GRNL1- do not appear to affect the affinity of GR for DNA. Therefore, our data suggest that an additional event following receptor transformation may be required for GRNL1- to become competent to be transferred to the nucleus.

Finally, from our results it appears to be clear that, in fibroblasts synchronized to G0, NL2 is at best of modest importance for the nuclear transfer of GR. This contrasts with an earlier report that the GR LBD could mediate the efficient transfer of fusion proteins to the nucleus in response to steroid in asynchronously growing cells (69). Whether this difference reflects a contribution from a cytoplasmic retention signal in the full-length GR or is an indicator of the effects of the cell cycle on the nuclear localization of GR (41, 42) is not known at this time.

While the difference in the kinetics of NL1- and NL2-mediated GR import during G0 has allowed us to convincingly distinguish between the nuclear import of GR by each signal, it also presents the challenge of identifying conditions under which NL2 is important for the localization of GR to the nucleus. Two possibilities seem most likely. First, NL2 function may be particularly important in specific tissues or at a particular time in development. However, the subcellular trafficking properties of GR also are known to fluctuate during different stages of the cell cycle (41). Thus, it is also possible that the relative importance of NL1 and NL2 for the nuclear import of GR varies according to the phase of the cell cycle. Thus, in addition to identifying the NL2 receptor, it will be important to compare the NL1- and NL2-dependent nucleocytoplasmic trafficking of GR as the cell progresses through the cell cycle and to evaluate the relative importance of the two signals in the localization of GR in different tissues.


    ACKNOWLEDGMENTS

We thank K. Yamamoto, M. L. Waterman, and I. W. Mattaj for plasmids used in this work and B. Wolff at Novartis for providing a sample of leptomycin B. We also thank G. Bélanger for his critical commentary on the manuscript and S. Ginsburg for assistance in preparing the figures.

This work was supported by an operating grant from the Medical Research Council of Canada to Y.A.L. R.J.G.H. is a Scholar of the Medical Research Council of Canada and the Cancer Research Society Inc.

J.G.A.S., B.H., and I.R.L. contributed equally to this work.


    FOOTNOTES

* Corresponding author. Mailing address: The Loeb Health Research Institute, 725 Parkdale Ave., Ottawa, Ontario, Canada K1Y 4E9. Phone: (613) 761-5142. Fax: (613) 761-5036. E-mail for Robert J. G. Haché: rhache{at}lri.ca. E-mail for Yvonne A. Lefebvre: lefebvre{at}civich.ottawa.on.ca.


    REFERENCES
Top
Abstract
Introduction
Materials and methods
Results
Discussion
References

1. Allan, G. F., X. Leng, S. Y. Tsai, N. L. Weigel, D. P. Edwards, M. J. Tsai, and B. W. O'Malley. 1992. Hormone and antihormone induce distinct conformational changes which are central to steroid receptor activation. J. Biol. Chem. 267:19513-19520[Abstract/Free Full Text].
2. Bakke, O., and K. B. Eik-Nes. 1981. Cell cycle-specific glucocorticoid growth regulation of a human cell line (NHIK 3025). J. Cell Physiol. 109:489-496[Medline].
3. Beg, A. A., S. M. Ruben, R. I. Scheinman, S. Haskill, C. A. Rosen, and A. S. Baldwin, Jr. 1992. I kappa B interacts with the nuclear localization sequences of the subunits of NF-kappa B: a mechanism for cytoplasmic retention. Genes Dev. 6:1899-1913[Abstract/Free Full Text].
4. Bodwell, J. E., J. M. Hu, E. Orti, and A. Munck. 1995. Hormone-induced hyperphosphorylation of specific phosphorylated sites in the mouse glucocorticoid receptor. J. Steroid Biochem. Mol. Biol. 52:135-140[Medline].
5. Bresnick, E. H., F. C. Dalman, E. R. Sanchez, and W. B. Pratt. 1989. Evidence that the 90-kDa heat shock protein is necessary for the steroid binding conformation of the L cell glucocorticoid receptor. J. Biol. Chem. 264:4992-4997[Abstract/Free Full Text].
6. Brzozowski, A. M., A. C. Pike, Z. Dauter, R. E. Hubbard, T. Bonn, O. Engstrom, L. Ohman, G. L. Greene, J. A. Gustafsson, and M. Carlquist. 1997. Molecular basis of agonism and antagonism in the oestrogen receptor. Nature 389:753-758[Medline].
7. Burnstein, K. L., C. M. Jewell, M. Sar, and J. A. Cidlowski. 1994. Intragenic sequences of the human glucocorticoid receptor complementary DNA mediate hormone-inducible receptor messenger RNA down-regulation through multiple mechanisms. Mol. Endocrinol. 8:1764-1773[Abstract].
8. Cadepond, F., J.-M. Gasc, F. Delahaye, N. Jibard, G. Schweizer-Groyer, I. Segard-Masurel, R. Evans, and E.-E. Baulieu. 1992. Hormonal regulation of the nuclear localization signals of the human glucocorticosteroid receptor. Exp. Cell Res. 201:99-108[Medline].
9. Cadepond, F., G. Schweizergroyer, I. Segardmaurel, N. Jibard, S. M. Hollenberg, V. Giguere, R. M. Evans, and E. E. Baulieu. 1991. Heat shock protein-90 as a critical factor in maintaining glucocorticosteroid receptor in a nonfunctional state. J. Biol. Chem. 266:5834-5841[Abstract/Free Full Text].
10. Cato, A. C. B., R. Miksicek, G. Schütz, J. Arnemann, and M. Beato. 1986. The hormone regulatory element of mouse mammary tumor virus mediates progesterone induction. EMBO J. 5:2237-2240[Medline].
11. Chang, H. C., and S. Lindquist. 1994. Conservation of Hsp90 macromolecular complexes in Saccharomyces cerevisiae. J. Biol. Chem. 269:24983-24988[Abstract/Free Full Text].
12. Cidlowski, J. A., and N. B. Cidlowski. 1981. Regulation of glucocorticoid receptors by glucocorticoids in cultured HeLa S3 cells. Endocrinology 109:1975-1982[Abstract].
13. Cole, T. J., J. A. Blendy, W. Schmid, U. Strahle, and G. Schutz. 1993. Expression of the mouse glucocorticoid receptor and its role during development. J. Steroid Biochem. Mol. Biol. 47:49-53[Medline].
14. Dalman, F. C., R. J. Koenig, G. H. Perdew, E. Massa, and W. B. Pratt. 1990. In contrast to the glucocorticoid receptor, the thyroid hormone receptor is translated in the DNA binding state and is not associated with hsp90. J. Biol. Chem. 265:3615-3618[Abstract/Free Full Text].
15. Danielian, P. S., R. White, J. A. Lees, and M. G. Parker. 1992. Identification of