Dynamic Shuttling and Intranuclear Mobility of Nuclear Hormone Receptors*

We expressed green fluorescent protein (GFP) chimeras of estrogen, retinoic acid, and thyroid hormone receptors (ERs, RARs, and TRs, respectively) in HeLa cells to examine nucleocytoplasmic shuttling and intranuclear mobility of nuclear hormone receptors (NRs) by confocal microscopy. These receptors were predominantly in the nucleus and, interestingly, underwent intranuclear reorganization after ligand treatment. Nucleocytoplasmic shuttling was demonstrated by heterokaryon experiments and energy-dependent blockade of nuclear import and leptomycin-dependent blockade of nuclear export. Ligand addition decreased shuttling by GFP-ER, whereas heterodimerization with retinoid X receptor helped maintain TR and RAR within the nucleus. Intranuclear mobility of the GFP-NRs was studied by fluorescence recovery after photo-bleaching ± cognate ligands. Both GFP-TR and GFP-RAR moved rapidly in the nucleus, and ligand binding did not significantly affect their mobility. In contrast, estrogen binding decreased the mobility of GFP-ER and also increased the fraction of GFP-ER that was unable to diffuse. These effects were even more pronounced with tamoxifen. Co-transfection of the co-activator, SRC-1, further slowed the mobility of liganded GFP-ER. Our findings suggest estradiol and tamoxifen exert differential effects on the intranuclear mobility of GFP-ER. They also show that ligand-binding and protein-protein interactions can affect the intracellular mobility of some NRs and thereby may contribute to their biological activity.

Nuclear hormone receptors (NRs) 1 comprise a large family of ligand-activated transcription factors that bind to hormone response elements of target genes and regulate their transcription (1,2). Members of the NR superfamily include the steroid receptors, thyroid hormone, retinoic acid, and vitamin D receptors (TRs, RARs, and VDRs, respectively) and orphan receptors whose cognate ligands are not known. NRs are structurally characterized by three distinct domains, an N-terminal domain, a central DNA-binding domain, and a C-terminal ligandbinding domain.
Steroid receptors bind to their response elements as homodimers and are referred to as Type I receptors, whereas RARs, TRs, and VDRs bind to their response elements and are called Type II receptors as heterodimers with retinoid X receptors (RXRs). Both Type I and Type II receptors can recruit co-activators in the presence of ligand, whereas Type II receptors, notably TRs and RARs, can recruit co-repressors such as silencing mediator for retinoid and thyroid hormone receptors and nuclear co-repressor (NCoR) in the absence of ligand and repress the transcription of target genes (1,3). The formation of NR/co-activator complexes is associated with histone acetylation (1,4) of the promoter region whereas NR/co-repressor complexes recruit histone deacetylases. Although various protein-protein interactions and enzymatic activities seem to play important roles in mediating transcriptional activity by NRs, it is possible that other mechanisms, such as the subcellular distribution and intranuclear dynamics of NRs, may contribute to hormonal responses mediated by these receptors.
Recently, several groups have shown that progesterone and glucocorticoid receptors (PRs and GRs), VDRs, and TRs continuously shuttle between the cytoplasm and nucleus (5)(6)(7)(8). We previously used green fluorescent protein (GFP) fusions of TR␤ to show that entry into the nucleus is an energy-mediated process whereas export out of the nucleus is a passive one (9). Furthermore, our results showed that ablation of the DNAbinding capacity of TR did not prevent its nuclear accumulation. Using TR mutants that were defective in their interaction with various co-regulators, we showed that heterodimerization with RXR and interaction with NCoR help maintain unliganded TR within the nucleus.
We and others (10 -12) have shown that GFP chimeras of ER, GR, and TR form a punctate pattern in the nucleus after ligand addition, suggesting that intranuclear diffusion and reorganization of NRs may be important regulatory processes. Recently, Mancini and co-workers (13) have shown that unliganded ER exhibits high mobility whereas estrogen-bound ER has reduced mobility in the nucleus. Additionally, we have shown that GR rapidly exchanges with regulatory DNA sequences and is not statically bound to hormone response elements as thought previously (14). In this paper, we have used GFP chimeras of ER, RAR, and TR as examples of Type I and II receptors to examine and compare nucleocytoplasmic shuttling and intranuclear diffusion of NRs in more detail. In particular, we have used energy depletion studies, heterokaryon experiments, and fluorescence after photo-bleaching to examine the movement of these receptors in the absence or presence of ligand and co-regulators. Our studies suggest that ligandbinding and protein-protein interactions can significantly in-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
fluence the mobility of these NRs and thereby contribute to their biological activity.

MATERIALS AND METHODS
Plasmids-Expression vectors for human ER␣, RAR␤, and TR␤ were described previously (15). GFP fusions were constructed into the multiple cloning site of pEGFP-C1 (Clontech, Palo Alto, CA) vector by PCR amplification at XhoI and EcoRI sites at the C-terminal end of the coding sequence of NRs. Molecular weight of the receptors was confirmed by Western blotting using GFP monoclonal antibodies (Clontech, Palo Alto, CA), and the integrity of all the constructs was confirmed by DNA sequencing (Veritas, Rockville, MD). Functional characterization was done by co-transfecting 0.1 g of the cDNA with 1.7 g of reporter thyroid hormone response element (F 2 ) and 1 g of ␤-galactosidase into HeLa cells in 35-mm dishes, and luciferase activity was determined in the lysates using a Berthold lumat-LB 9507 luminometer.
Microscopic Studies-The cells expressing various GFP-NRs were viewed under a Leica TCS SP laser scanning confocal microscope mounted on a DMIRBE inverted epifluorescent microscope equipped with a ϫ63 1.4-numerical aperture oil immersion lens (Heidelberg, Germany). The GFP fluorescence was excited with a 488-nm laser line from an air-cooled fiber-coupled argon laser (Coherent Inc., Santa Clara, CA). Typical laser output was less than 10% of its maximal power. DAPI fluorescence was excited by a 385-nm laser line from a water-cooled argon laser at 25% power (Coherent Inc., Santa Clara, CA). GFP emission was monitored between 505 and 590 nm, and DAPI emission was followed between 405 and 490 nm. Both GFP and DAPI were visualized with a pinhole of 1.0 (Airy units) and detected. Quantitative analysis of receptor distribution was done as described previously (9).
ATP Depletion Experiments-ATP depletion experiments were performed as described previously (9). Cells were incubated for 2 h with 10 mM sodium azide (Merck, Darmstadt, Germany) in the presence of 6 mM 2-deoxyglucose prior to imaging.
Heterokaryon Assays-The shuttling nature of NRs was studied by interspecies heterokaryon assay as described previously (17). Briefly, 2 ϫ 10 5 mouse (NIH3T3) cells were plated onto glass coverslips in 6-well multidishes on day one. Subsequently the cells were transfected with 0.1 g of GFP-NR along with carrier DNA by the calcium phosphate method. After 48 h 1 ϫ 10 6 human (HeLa) cells were plated onto the coverslips. After 3 h cells were washed thoroughly with Dulbecco's phosphate-buffered saline, 100 l of pre-warmed polyethylene gly-col1500 were added, and the cells were incubated at 37°C for 2 min. The cells were washed again with Dulbecco's phosphate-buffered saline and incubated with Dulbecco's modified Eagle's medium with 50 g/ml of cycloheximide incubated for another 3 h and fixed with 4% paraformaldehyde and mounted on the glass slides after counter-staining with DAPI for microscopy.
Fluorescence Recovery after Photo-bleaching (FRAP)-These studies were carried out as described previously with minor modifications (14). Briefly, HeLa cells were grown on coverslips in live cell chambers and maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum at 37°C. They were transfected with GFP-NR vectors and treated with ligands. FRAP experiments were carried out using a Leica TCS SP laser scanning confocal microscope. A full power beam of 488and 514-nm laser lines was focused on a defined region for 0.5 s to bleach the region and then followed the recovery of fluorescence that region was monitored over time. Fluorescence intensities of regions of interest were obtained using Leica TCS NT software and analyzed by Microsoft Excel. The images were imported as TIFF files, and final images were generated using Corel software, version 9.0.
Immunofluorescence Studies-Immunofluorescence was performed on GFP-tagged ER-transfected HeLa cells on coverslips using anti-CRM-1 antibodies (1:1000) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and visualized under a Leica-Fish microscope, and images were collected using Metamorph software. The images were processed using Corel software, version 9.0.

RESULTS
To study the intracellular distribution of NRs in living cells, we created vectors expressing GFP chimeras fused to the Nterminal ends of TR␤, ER␣, and RAR␤ (Fig. 1A). Western blot analysis of extracts obtained from transfected cells showed that all the fusion constructs encoded full-length proteins of correct size (data not shown). The transcriptional activities of the We examined the intracellular distribution of GFP-ER and GFP-RAR in live cells and found that these receptors localized mostly in the nucleus, similar to previous observations for TR␤ and ER␣ (9, 18) (see Fig. 2A and Fig. 3A). To quantitate the percentage of NRs present within the nucleus and cytoplasm, we measured the area-corrected intensity of GFP-NR fluorescence in both the nucleus and cytoplasm. From these analyses, we observed that only about 10 -15% of unliganded GFP-ER ( Fig. 2A) was present in the cytoplasm, and ER effectively translocated into the nucleus after estradiol (E 2 ; 10 Ϫ6 M) or tamoxifen (Txn; 10 Ϫ8 M) treatment for 1 h (Fig. 2, B and C). Similar to GFP-ER, ϳ20% of unliganded GFP-RAR (Fig. 3B) was present in the cytoplasm, and GFP-RAR almost entirely translocated into the nucleus after retinoic acid (RA; 10 Ϫ6 M) treatment (Fig. 3B). These changes in the nucleocytoplasmic distribution of ER and RAR after ligand addition contrast with our previous observations for TR␤, which maintained ϳ10% of receptor in the cytoplasm regardless of whether ligand was present (9). The observed nuclear/cytoplasmic distribution was independent of the amount of expression vector transfected, as it remained constant when varying concentrations of GFP-ER and GFP-RAR vectors were transfected into HeLa cells (data not shown). In these experiments, we also observed that ligand induced intranuclear reorganization of GFP-ER and GFP-RAR (see Fig. 2, A-C and Fig. 3, A and B) similar to previous observations for GR, ER, and, TR (9 -11). In the absence of ligand, GFP-ER and GFP-RAR displayed a diffuse, reticular pattern whereas in the presence of their cognate ligands, these receptors exhibited a discrete, punctate pattern. These effects were rapid with equilibrium patterns occurring within 10 min after ligand addition (data not shown).
To further delineate the nucleocytoplasmic shuttling of these receptors, the intracellular distribution of these receptors was studied after treatment with sodium azide, which depletes intracellular stores of ATP and blocks the active transport of proteins across the nuclear membrane (see Figs. 2 and 3). We showed previously (9) that nuclear import of GFP-TR␤ is an energy-mediated process and that its export is a passive one. We now observed similar findings for both GFP-ER and GFP-RAR. After a 2-h treatment with sodium azide, more than 20% of GFP-ER redistributed to the cytoplasm indicating that a subpopulation of GFP-ER shuttles between the nucleus and cytoplasm (Fig. 2D). Interestingly, treatment with either E 2 or tamoxifen prior to sodium azide treatment prevented cytoplasmic accumulation (Fig. 2, E and F) indicating that ligand bind- ing to GFP-ER decreases shuttling out of the nucleus. Similarly, a significant amount of GFP-RAR was found to undergo nucleocytoplasmic shuttling (Fig. 3D). This shuttling out of the nucleus was prevented by either ligand treatment or co-expression of unfused RXR (Fig. 3, E and F) and suggests that ligand binding and heterodimerization help retain GFP-RAR in the nucleus.
We then used the heterokaryon system to further study the nucleocytoplasmic shuttling by GFP-ER. In these studies, NIH3T3 cells expressing GFP-ER were fused with nonexpressing HeLa cells, in both the presence and absence of ligands. We observed that GFP-ER translocated from the NIH3T3 nucleus to the HeLa nucleus, indicating that GFP-ER rapidly shuttles between the nucleus and cytoplasm (Fig. 4, A and B). However, in the presence of either E 2 or tamoxifen, translocation of GFP-ER into the HeLa nucleus was reduced (Fig. 4, C-F) indicating decreased nucleocytoplasmic shuttling.
To characterize the mechanism of shuttling, we studied ER diffusion in the heterokaryon system, in the presence of leptomycin B, an inhibitor of CRM-1 (involved in nucleocytoplasmic export) (Fig. 5). GFP-ER was retained in the HeLa cell nuclei in the absence of ligand, E 2 , and tamoxifen suggesting that ER export occurs by CRM-1 regardless of its ligand-binding state. Similar results also were observed for GFP-TR␤ and GFP-RAR (data not shown).
These foregoing findings suggest that GFP-ER and GFP-RAR shuttle between the nuclear and cytoplasmic compartments. Additionally, our findings in Figs. 2 and 3 suggest that these receptors reorganize within the nucleus upon ligand binding. To better understand this latter phenomenon, we used FRAP to study the intranuclear mobility of GFP-ER, GFP-RAR, and GFP-TR in transfected cells in the presence and absence of ligand. In these studies, a full power laser beam was focused for 0.5 s on the nuclei of GFP-NR-expressing cells, causing fluorescence loss in a defined nuclear zone. The mobility of GFP-NR molecules was then measured as a function of fluorescence recovery over time in the bleached region.
Under these conditions, we determined the recovery rate of a mobile fraction of GFP-NRs in which fluorescence recovery could be measured (t1 ⁄2 ) and observed an immobile fraction in which no recovery occurred even after 3 min (original fluorescence (100%) Ϫ % of recovery after 3 min). The calculated data of ten individual cells from six different experiments are shown in Table I. The effect of ligands on the mobility of GFP-ER was studied using the FRAP technique (see Fig. 6 and Fig. 7A). In cells transfected with GFP-ER vector, bleaching for 0.5 s caused only a 55% reduction in fluorescence within the nuclear zone, indicating that unliganded GFP-ER was extremely mobile and rapidly moved back to the bleached zone, even during the time between the bleaching process and first measurement (see Fig. 6A and Fig. 7A). Rapid recovery was observed (t1 ⁄2 ϭ 1.6 s) as more than 90% of fluorescence recovery occurred within 3 s, and maximal recovery occurred in less than 12 s. The immobile fraction was 14%. To observe the effect of ligands on the mobility of GFP-ER, we incubated transfected cells with either E 2 or tamoxifen for 1 h and then subjected the cells to FRAP (see Fig. 6, B and C and Fig. 7A). In these studies, we found that cells treated with E 2 had more than 70% loss in fluorescence initially, and t1 ⁄2 ϭ 5.8 s, with an immobile fraction of 44% (see Fig. 7A and Table I). These findings demonstrated that E 2 reduced the rate at which GFP-ER returns to the bleached zone and increased the amount of ER that was unable to diffuse readily. Interestingly, treatment with the ER antagonist, tamoxifen, resulted in almost total loss in fluorescence, with little fluorescence recovery even after 3 min because of a large immobile fraction (see Fig. 6C, Fig. 7A, and Table I) (data not shown). In all these experiments, the recovered fluorescence was normalized to unbleached control cells in the same experiments. All the experiments were repeated more than six times with similar findings.
Previous studies have shown that p160 co-activators such as SRC-1 interact with ER in a ligand-dependent manner and enhance its transcriptional activity (1). To study the effect of SRC-1 on the intranuclear mobility of GFP-ER, we performed FRAP on cells co-expressing both GFP-ER and unfused SRC-1, in both the absence and presence of E 2 (Fig. 6B). Interestingly, there was delayed recovery of fluorescence of unliganded GFP-ER in the presence of SRC-1 (t1 ⁄2 ϭ 3.6 s versus 1.6 s, respectively) with a maximal recovery occurring by ϳ20 s (Fig.  7B). These findings suggest a possible association between SRC-1 and unliganded GFP-ER that may decrease the mobility of the latter. In the presence of E 2 , more than 80% of fluorescence was bleached in cells transfected with GFP-ER and SRC-1, compared with 70% bleaching in cells transfected with only GFP-ER. Additionally, recovery time was slower in cells transfected with GFP-ER and SRC-1 than GFP-ER alone (t1 ⁄2 ϭ 6.9 s versus 5.8 s, respectively). These findings suggest that SRC-1 may reduce the intranuclear mobility of E 2 -bound ER.
To further investigate the effect of ligands on the mobility of nuclear hormone receptors, we performed FRAP on cells expressing GFP-RAR in the presence and absence of RA (Fig. 8A). GFP-RAR was highly mobile in the absence of ligand as there was only a 40% loss in fluorescence after photo-bleaching, and maximal recovery occurred by 10 s (Fig. 8A). In the presence of RA, there was little change in the intranuclear mobility of GFP-RAR (t1 ⁄2 ϭ 2.3 s versus 2.0 s). FRAP studies on unliganded GFP-TR␤ showed very similar recovery kinetics as GFP-RAR as there was only a 40% loss in fluorescence after photo-bleaching and almost total fluorescence recovery after 10 s (Fig. 8B).
In the presence of T 3 , there was little change in mobility of GFP-TR as maximal fluorescence recovery occurred by 5 s, and t1 ⁄2 remained unchanged (1.8 s). In contrast to the effect of SRC-1 on ER mobility, co-transfection of SRC-1, as well as NCoR and HDAC-1, did not affect the intranuclear mobility of GFP-TR (data not shown). We observed previously that nuclear export of ER, RAR, and TR was mediated by CRM-1; thus, we examined whether CRM-1 may be associated with GFP-ER and modulate its intranuclear mobility (Fig. 9). However, we did not observe co-localization of CRM-1 and GFP-ER in the absence or presence of ligand, suggesting that exportin does not modulate intranuclear mobility of ER. DISCUSSION We have used GFP chimeras of ER, RAR, and TR to examine the cellular localization of several prototypical Type I and Type II NRs in living cells. In the absence of ligand, all the NRs studied displayed predominantly nuclear distribution. In contrast, previous studies of some steroid hormone receptors, particularly GR and PR, showed mostly cytoplasmic distribution in the absence of ligand (12,19). This distribution is likely maintained by the formation of cytoplasmic complexes that include chaperone proteins such as hsp 70 and hsp 90 (20). The location of unliganded ER is still controversial as earlier biochemical studies suggested ER was mostly cytoplasmic, but later cell fractionation and immunohistochemical studies suggested it was predominantly nuclear (21)(22)(23)(24). Recent studies with GFP-ER have not addressed directly the issue of unliganded ER localization (10,11). Here, we show that a small pool of unliganded ER exists in the cytoplasm, and it constantly shuttles between the cytoplasm and nucleus in live cells. Moreover, the cellular distribution and shuttling of ER is markedly affected by estrogen treatment. Interestingly, the estrogen antagonist, tamoxifen, also has similar effects as estrogen on receptor cycling and retention, suggesting it does not interfere with these particular receptor roles. We also showed that the nuclear import of ER is an energy-dependent process similar to our present findings for RAR and previous studies for TR and PR (5,9). We also demonstrated that nuclear export of ER, RAR, and TR is mediated via the exportin, CRM-1 (25). Previous studies suggested that VDR, but not PR and GR, were blocked by leptomycin B (26 -28). Thus, it is possible there may  be more than one mechanism for nuclear export of nuclear hormone receptors.
The mechanism for the nuclear retention of ER is not known; however, it is possible that ligand binding may facilitate interactions with nuclear proteins or components. For instance, it is known that liganded ERs can form multimeric transcription complexes containing co-activators (1). Recent studies also have shown that E 2 -bound ER and T 3 -bound TR can bind to the nuclear matrix component (9,11), which in turn may affect intranuclear mobility and nuclear retention of these receptors.
Our studies also show that unliganded GFP-RAR and GFP-TR shuttle between the cytoplasmic and nuclear compartments, suggesting that such shuttling might be a general phenomenon among NRs. In contrast to ER, ligand binding did not promote major changes in RAR or TR cellular distribution. Instead, GFP-RAR exhibited reduced shuttling in the presence of RXR suggesting that heterodimerization plays a major role in maintaining the normal cellular distribution of RAR. We previously used GFP chimeras of TR mutants that were defective in homodimerization or nuclear localization to demon-strate that heterodimerization with RXR helped maintain TR in the nucleus. Similar results also have been observed for TR␣ in an oocyte system and GFP-VDR (7,8). Our previous studies also showed that unliganded TR␤ interaction with the co-repressor, NCoR, helped maintain nuclear localization (9). Furthermore, we showed previously that DNA binding by TR did not significantly contribute to nuclear retention. Taken together, these findings with the other GFP-NRs suggest that ligand-binding or protein-protein interactions are key processes that may be differentially employed by NRs for nuclear localization.
Previous studies have indicated that some NRs undergo intranuclear reorganization in response to ligand treatment (7, 9 -11, 18). Similarly, all the NRs studied here underwent intranuclear reorganization after hormone treatment. To better understand this phenomenon, we used FRAP to study the intranuclear mobility of the NRs and the effect of various co-regulators on their mobility. Our studies demonstrate that NRs move rapidly within the nucleus. Additionally, the mobility of NRs can be modulated by ligand binding as both E 2 and tamoxifen reduce the mobility of GFP-ER. However, E 2 treatment results in partial fluorescence recovery whereas tamoxifen treatment causes minimal or no fluorescence recovery as it generated a large immobile fraction of ER. These findings suggest that anti-estrogens can have profound effects on the intranuclear mobility of ER, and this process may be involved in antagonist activity. Of note, our data differ from a study published recently (11) in which ICI decreased ER intranuclear mobility whereas tamoxifen had minimal effect. Additionally, these studies are consistent with recent findings that showed steroid hormone receptors are dynamic within the nucleus and rapidly exchange between various nuclear components including chromatin (14).
In contrast to ER, ligand did not significantly affect the mobility of Type II receptors such as GFP-RAR and GFP-TR. Thus, ligands can exert differential effects on the intranuclear mobility of NRs. We also studied the effects of various transcriptional co-factors on the intranuclear mobility of NRs. GFP-ER had decreased mobility when co-expressed with SRC-1 perhaps by forming of an NR/co-activator complex or by stabilizing NR within larger multiprotein complexes. In this connection, nuclear hormone receptors have been shown to co-localize with co-activators and interact with proteins in the proteasome complex (29 -31). This may potentially be a mechanism for down-regulating transcriptional activity by NRs. It also is in-teresting to note that SRC-1 decreased the mobility of both unliganded and E 2 -bound ER, suggesting that SRC-1 itself may be in a dynamic equilibrium with ER, even when the latter is in an unliganded state.
In summary, we have shown that several representative NRs undergo nucleocytoplasmic shuttling in an energy-dependent manner. These NRs also have rapid intranuclear mobilities, which, in the case of ER, can be modulated by ligand binding. Estrogen and tamoxifen have differential effects on ER mobility, and the marked effect of tamoxifen on ER mobility may contribute to its anti-estrogenic properties. When our results of GFP-ER, GFP-RAR, and GFP-TR are taken together, they suggest that ligand binding and interaction of NRs with transcriptional co-factors may exert differential effects on the nucleocytoplasmic shuttling and intranuclear mobility of NRs. These, in turn, may represent novel regulatory processes, which affect the biological activity of NRs.