Dimerization with Retinoid X Receptors Promotes NuclearLocalization and Subnuclear Targeting of Vitamin D Receptors*

Received for publication, May 4, 2000, and in revised form, September 21, 2000Published, JBC Papers in Press, September 22, 2000, DOI 10.1074/jbc.M003791200

Kirsten Prufer, Attila Racz, Grace C. Lin, and Julia Barsony‡

From the Laboratory of Cell Biochemistry and Biology, NIDDK, National Institutes of Health, Bethesda, Maryland 20892

The vitamin D receptor (VDR) acts as heterodimerwith the retinoid X receptor a (RXR) to control tran-scriptional activity of target genes. To explore the influ-ence of heterodimerization on the subcellular distribu-tion of these receptors in living cells, we developed aseries of fluorescent-protein chimeras. The steady-statedistribution of the yellow fluorescent protein-RXR wasmore nuclear than the unliganded green fluorescentprotein (GFP)-VDR. Coexpression of RXR-blue fluores-cent protein (BFP) promoted nuclear accumulation ofGFP-VDR by influencing both nuclear import and reten-tion. Fluorescence resonance energy transfer micros-copy (FRET) demonstrated that the unliganded GFP-VDR and RXR-BFP form heterodimers. The increase innuclear heterodimer content correlated with an in-crease in basal transcriptional activity. FRET also re-vealed that calcitriol induces formation of multiple nu-clear foci of heterodimers. Mutational analysis showed acorrelation between hormone-dependent nuclear VDRfoci formation and DNA binding. RXR-BFP also pro-moted hormone-dependent nuclear accumulation andintranuclear foci formation of a nuclear localization sig-nal mutant receptor (nlsGFP-VDR) and rescued its tran-scriptional activity. Heterodimerization mutant RXRfailed to alter GFP-VDR and nlsGFP-VDR distributionor activity. These experiments suggest that RXR has aprofound effect on VDR distribution. This effect of RXRto promote nuclear accumulation and intranuclear tar-geting contributes to the regulation of VDR activity andprobably the activity of other heterodimerizationpartners.

Proteins of the nuclear receptor superfamily mediate re-sponse to hormones or intracellular signals into transcriptionalresponses and regulate an array of important cellular func-tions. A member of the nuclear receptor superfamily, the vita-min D receptor (VDR)1, mediates effects of calcitriol on bonedevelopment and maintenance, calcium homeostasis, immunefunctions, endocrine functions, vitamin D metabolism, and cel-

lular proliferation and differentiation. Like other class II nu-clear receptors, such as the thyroid hormone receptor, theretinoic acid receptor, and many orphan receptors, VDR re-quires heterodimerization with the retinoid X receptor (RXR)for high affinity binding to target genes (1, 2). VDR and RXRcan heterodimerize in the absence of calcitriol, and these het-erodimers regulate basal transcriptional activity of targetgenes and exert transcriptional silencing functions (3). Theaddition of calcitriol stabilizes the heterodimers and promotestheir binding to the vitamin D response elements (4). Theimportance of heterodimerization in VDR functions led us toinvestigate the spatial and temporal relationships betweenthese receptors in living cells.

Recently we and others have used green fluorescent proteinchimeras of VDR to study the receptor distribution in livingcells (5–7). Unlike the glucocorticoid receptor (GR), which staysin the cytoplasm without the ligand, the unliganded VDR dis-tributes evenly between the cytoplasm and the nucleus. Thisindicates that the regulation of VDR distribution is more com-plex than the regulation of GR distribution. The fact that VDRdistribution is similar to thyroid hormone receptor distribution(8) raised the possibility that heterodimerization with RXRcould account for the partial nuclear localization of both VDRand thyroid hormone receptor.

Most of what we know about the subcellular distribution ofRXR comes from immunohistological and cell fractionationstudies. In these studies, RXR was found in the nuclei ofnormal human skin cells (9), the rat kidney (10), and thehuman pituitary (11). RXR was also detected in the cytoplasmby cell fractionation experiments (12). Although a nuclear lo-calization of RXR is to be expected, looking in live cells some-times gives surprises, as has been the partial cytoplasmic lo-calization of the thyroid hormone receptor (8). In a recentpaper, focusing on the localization of an orphan nuclear recep-tor, the transiently expressed mouse GFP-RXRb was found inthe nuclei of phaeochromocytoma cells (36). Because previousstudies indicated that VDR preferentially associates withRXRa to stimulate transcription (13), the subcellular localiza-tion of the RXRa had to be explored in detail. Even less isknown about the subcellular location of the VDR/RXR het-erodimers. Receptors that heterodimerize with RXR are notonly synthesized in the cytoplasm but reside in the cytoplasmto some degree before ligand binding. Thus, heterodimerizationmay influence subcellular localization of RXR or its partnerreceptors. Using transcriptionally active and stably expressedyellow, green, and blue fluorescent protein chimeras of RXRand VDR and their mutants, we visualized the differences inthe distribution of VDR and RXR in living cells by confocallaser-scanning microscopy.

Current developments of multicolor fluorescent protein vari-ants also allow visualization of protein interactions in livingcells by fluorescence resonance energy transfer microscopy

* The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked“advertisement” in accordance with 18 U.S.C. Section 1734 solely toindicate this fact.

‡ To whom correspondence should be addressed: Bldg. 8, Rm. 422,NIDDK, NIH, 8 Center Dr., Bethesda, MD 20892. Tel.: 301-402-2868;Fax: 301-496-9431; E-mail: jul@helix.nih.gov.

1 The abbreviations used are: VDR, vitamin D receptor; RXR, humanretinoid X receptor a; GFP, green fluorescent protein; YFP, yellowfluorescent protein; BFP, blue fluorescent protein; hd, heterodimeriza-tion; NLS, nuclear localization signal; 9c-RA, 9-cis-retinoic acid; ROS,17/2.8 rat osteosarcoma cells; CYR, CV-1-derived cell line stably ex-pressing YFP-RXR; GL48, 293-derived cell line stably expressing GFP-VDR; 24OH/Luc, luciferase gene that is under control of the 25-hy-droxyvitamin D3 24-hydroxylase promoter; FRET, fluorescenceresonance energy transfer; GR, glucocorticoid receptor.

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 275, No. 52, Issue of December 29, pp. 41114–41123, 2000Printed in U.S.A.

This paper is available on line at http://www.jbc.org41114

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

(FRET) (14). This method was used to demonstrate ho-modimerization of Pit-1 transcription factor and heterodimer-ization of Pit-1 with cEts-1 in the cell nucleus (15) as well as theassociation of retinoic acid receptors and estrogen receptorswith coactivators (16). Here, we adapted FRET to establishthat GFP-tagged VDR and RXR heterodimerize in living cellsand used this technique to explore the distribution ofheterodimers.

Recent results reveal that transcriptional activities of nu-clear receptors and other signaling proteins could be regulatedby the control of translocation across the nuclear membrane(17). We generated a cell line stably expressing GFP-VDR thatallowed us to correlate nuclear receptor content with basaltranscriptional activity. Our experiments revealed that RXRpromotes nuclear accumulation of VDR. To study the mecha-nisms of this RXR effect we explored both nuclear import andexport of the unliganded VDR by mutational analysis and bytemperature changes.

During hormone-induced receptor activation, a redistribu-tion from the cytoplasm into the nucleus has been shown inliving cells for GFP chimeras of the mineralocorticoid receptor(18), the GR (19), the thyroid hormone receptor (8), the proges-terone receptor (20), and the androgen receptor (21). We foundearlier that the cytoplasmic portion of VDR-GFP also translo-cates into the nucleus in a ligand-dependent fashion (5, 6). Themechanisms controlling VDR translocation, however, re-mained to be elucidated. Nuclear import of proteins involvesthe binding of their nuclear localization signal (NLS) sequencesto import receptors (22, 23). Recently, NLSs have been identi-fied in the VDR (7, 24). The effects of mutations in these NLSregions have not yet been evaluated in living cells, and the NLShas not yet been identified in the RXR sequence. We usedmutational analysis to clarify the impact of heterodimerizationon the rapid calcitriol-induced VDR translocation and correlateit with changes in transcriptional activity. Here we coexpresseda heterodimerization-competent RXR with an NLS mutant ofthe GFP-VDR to demonstrate that RXR also interacts with thenuclear import machinery.

Ligand-induced intranuclear pattern changes, including fociformation, have been reported for the GR (19), the androgenreceptor (21), the mineralocorticoid receptor (18), the estrogenreceptor (26, 27), and the VDR (5). Here, we report similarhormone-induced intranuclear foci of YFP-RXR and GFP-VDR.In addition, we used FRET microscopy to study ligand-inducedchanges in the intranuclear distribution of the GFP-VDR/RXR-BFP heterodimers. We used mutational analysis to explore aconnection between DNA binding and nuclear foci formationand used FRET microscopy to visualize the effect of RXR topromote DNA binding of VDR in living cells.

EXPERIMENTAL PROCEDURES

Cell Culture—ROS17/2.8 rat osteosarcoma cells (ROS), CV-1 andCOS-7 African green monkey cells, and 293 human kidney epithelialcells were obtained and grown as described previously (5). Cell cultureswere maintained in Dulbecco’s modified Eagle’s medium supplementedwith 10% fetal bovine serum (HyClone), 2 mM glutamine, 0.1 mM insu-lin, and 0.1 mg/ml gentamicin. For microscopy, cells were subculturedinto glass coverslip chambers (Nalge Nunc Int.) and, for transactivationassays, into 12-well plates (Costar). During experiments, cells weremaintained in serum-free minimal essential medium (Life Technolo-gies, Inc.,) containing insulin, transferrin, selenium additive (Collabo-rative Biomedical Products) and 2 mM L-glutamine.

Construction of Expression Vectors and Generation of Stable CellLines—A transcriptionally active GFP-VDR was generated by cloningthe human VDR cDNA (a gift from Dr. J. W. Pike, University ofCincinnati College of Medicine, OH) C-terminal to the GFP of thepQBI25 plasmid (Quantum Biotechnologies). 293 cells were transfectedwith this GFP-VDR expression plasmid using LipofectAMINE Plusreagents (Life Technologies) according to manufacturer’s instructions.G418-resistant cell clones were then selected for 3–5 wk by culturing

cells in media with 1 mM G418 (Life Technologies). One of the high-expressing clones, GL48, was characterized and used for subsequentexperiments (49). The RXR-BFP was generated by cloning the humanRXRa cDNA (RXR) (a gift from P. Chambon, CNRS/INSERM, Stras-bourg, France) N-terminal to the BFP sequence of the blue fluorescentvector pQBI50 from Quantum Biotechnologies. First, the HindIII andKpnI restriction sites were generated, and then the RXRa cDNA wasinserted between these sites. The YFP-RXR was constructed by insert-ing the human RXRa cDNA to the C-terminal end of the YFP using theXhoI and XbaI restriction sites of the 10C plasmid (a gift from J.Ellenberg, NICHD, National Institutes of Health (28)). This YFP-RXRwas stably expressed in CV-1 cells and a high-expressing clone (CYR)was selected by culturing with G418 and used for experiments.

Point mutations were introduced into the coding sequences of theGFP-VDR, the YFP-RXR, and the RXR-BFP using the QuickChangemutagenesis kit (Stratagene). Up- and downstream oligonucleotideswere designed according to manufacturer’s instructions. To create aGFP-VDR mutant impaired in nuclear localization (nlsGFP-VDR), weintroduced point mutations in the NLS1 of VDR at amino acids 53–55(K53Q, lysine to glutamate; R54G, arginine to glycine; K55E, lysine toglutamic acid) (7). To create the DNA binding mutant of GFP-VDR(dnaGFP-VDR), we introduced a point mutation at amino acid 77(R77Q). This mutation was found in a patient with hereditary resist-ance to calcitriol, and in vitro studies demonstrated that the mutationcauses DNA binding defect without altering hormone binding (29) ornuclear import (30). To create heterodimerization mutants of RXR-BFP(hdRXR-BFP), we mutated the amino acids 419 and 420 of the RXRfrom leucine to arginine, because these amino acids are crucial for theheterodimerization function of RXR (31). In all constructs, including themutants, the coding sequences for the fusion proteins were confirmedusing the ABIPrism sequencing kit (PerkinElmer).

Determinations of Receptor Distribution—Cells were transientlytransfected with one or two plasmids using LipofectAMINE Plus re-agents with 0.125 mg of DNA/chamber for each plasmid. Transfectedcells were used for microscopy within 48 h. GL48 and CYR cells weresubcultured into coverglass chamber slides (Nunc) 24 h before micros-copy. An incubator was built around the microscope to control temper-ature during microscopy. Images were collected from a Zeiss Axiovert100 fluorescent microscope equipped with a LSM 410 laser-scanningunit (Carl Zeiss Inc.) using a 100 3 1.4 NA objective. The 488-nm lineof a krypton-argon laser with a band-pass 510–525-nm emission filterwas used for GFP and YFP detection, and the 364-nm line of an UVlaser with a 397-nm long-pass emission filter was used for BFP detec-tion. For morphometric analysis, mean brightness values were obtainedfrom areas of cytoplasm, nuclei, and background. The ratios of nuclearand cytoplasmic brightness values were calculated after backgroundsubtraction from 50 cells. The experiments were repeated at least fourtimes.

Detection of Heterodimer Distribution in Living Cells—Protein-pro-tein interactions of GFP-VDR and RXR-BFP were studied using FRETexperiments. CV-1 cells were grown on 24.5-mm diameter coverslipsand transfected as described above. Cells were transfected with BFP,GFP-VDR, RXR-BFP, and hdRXR-BFP alone or in combination; theGFP-VDR with either the RXR-BFP or the hdRXR-BFP (0.125/0.25mg/chamber DNA for GFP-VDR/RXR-BFP). Before microscopy, sampleswere placed into an open chamber containing insulin, transferrin, se-lenium additive media. Hormones were added to the media as indi-cated. Images were taken from a fluorescence microscope (Carl Zeiss,Photomicroscope III) using a digital CCD camera (TEA/CCD-512-FFT,Princeton Instruments), computer-controlled illumination, and a mo-torized stage (Ludl Electronic Products, Inc.). Filter sets were optimizedto discriminate between the spectra of GFP and BFP (Chroma Tech-nology). Blue fluorescence was detected using a 370–390-nm excitationfilter and a 435–485-nm emission filter; green fluorescence was de-tected using a 480–495-nm excitation filter and a 515–555-nm emissionfilter. FRET was detected with blue excitation and green emissionfilters; the ratios of the green and blue images were then calculated toobtain the FRET ratio image. Fluorescence intensities were obtainedfrom these ratio images over 5–10 areas of cell nuclei, cytoplasm, andbackground. Mean and S.D. of brightness values were calculated foreach group, and data were analyzed by Student’s t test. We used theMetaMorph software package from Universal Imaging Corporation forimage acquisition and morphometric analysis.

Transactivation Assays—CV-1 cells were subcultured into 12-wellplates (Costar), and after 24 h, cells were transfected using Lipo-fectAMINE Plus reagents. Transcriptional activities of wild type andmutant VDR fusion proteins were tested with a luciferase reporterassay. Expression plasmids were GFP-VDR, nlsGFP-VDR, or dnaGFP-

RXR Effect on VDR Distribution 41115

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

VDR and the same plasmids in combinations with either RXR-BFP orhdRXR-BFP (0.3 mg/well each). These plasmids were cotransfected withthe reporter plasmid p24OH/Luc-23 (0.07 mg/well; a gift from H. F.DeLuca, University of Wisconsin, Madison, WI) and the b-galactosidasestandardization plasmid pGL3 (0.07 mg/well; Promega). The p24OH/Luc-23 encodes a luciferase gene that is under the control of the 25-hydroxyvitamin D3 24-hydroxylase promoter (32). DNA contents wereequalized by the addition of herring sperm DNA (Roche MolecularBiochemicals). Transfected cells were incubated for 24 h with eithercalcitriol (Solvay Duphar) or vehicle in the media. Transcriptional ac-tivities of wild type and mutant RXR chimeras (0.3 mg/well) were alsotested by a luciferase reporter assay using the retinoid X responseelement-Luc (0.07 mg/well; a gift from Dr. Karathanasis, Wyeth-AyerstResearch, NY). This retinoid X response element-Luc encodes a lucif-erase gene that is under the control of a minimal ApoAI enhancer (33).Transfected cells were treated with 9c-RA (Sigma) or vehicle in themedia for 24 h. Then luciferase activities of cell lysates were determinedwith reagents from Promega, and b-galactosidase activities were deter-mined with reagents from Sigma as described previously (5). Lumines-cence data were normalized to b-galactosidase values and expressed asfold-induction relative to vehicle-treated controls. Experiments wereperformed with triplicate samples and repeated at least twice. Data arepresented as mean 6 1 S.E.

RESULTS

RXR Expression and Subcellular Localization—To deter-mine subcellular localization of RXR in living cells, we devel-oped expression constructs encoding functional GFP chimerasof RXR. Both N-terminal and C-terminal chimeras were gen-erated with the RXR and the yellow or blue fluorescing vari-ants of GFP. Also, point mutations were introduced into thewild-type RXR-BFP to generate a heterodimerization mutant(hdRXR-BFP). To characterize these fusion proteins, we eval-uated protein expression and transcriptional activities. We con-firmed the expression of intact fusion proteins by Western blotanalysis (not shown). In addition, fluorescence microscopyshowed similar expression levels for the wild-type RXR-BFPand the hdRXR-BFP. We evaluated the functionality of thefusion proteins by luciferase reporter assays. The ligand 9c-RA(100 nM) caused an 8-fold induction of retinoid X responseelement-luciferase reporter activity by YFP-RXR, a 7-fold in-duction by RXR-BFP, and a 5-fold induction by the untaggedRXR. As expected from a previous report (31), the mutanthdRXR-BFP displayed impaired transcriptional activity (2-foldinduction). These results show that the tagged RXR proteinsretained the ability of the untagged RXR to inducetransactivation.

To evaluate subcellular distribution of YFP-RXR and RXR-BFP, chimeric proteins were transiently expressed in CV-1,COS-7, 293, and ROS cells. Microscopy showed the unligandedYFP-RXR and RXR-BFP predominantly in the nucleus of eachcell type tested. The unliganded RXR distributed evenly withinthe nucleus excluding the nucleoli, as shown in ROS cells (Fig.1a). Treatment with 9c-RA induced development of prominentintranuclear foci (Fig. 1b). In addition to the nuclear YFP-RXR,a sub-population of cells contained variable amounts of cyto-plasmic YFP-RXR (Fig. 1c). Time-lapse recordings revealedthat this cytoplasmic YFP-RXR does not translocate from thecytoplasm into the nucleus after hormone treatment (notshown). During our previous studies with transiently ex-pressed GFP-GR and GFP-VDR, we noticed that overexpres-sion of receptors can cause predominantly nuclear mislocaliza-tion. To confirm that nuclear localization of RXR is not due toprotein overexpression, we generated a subclone of CV-1 cellsthat stably expresses the intact YFP-RXR (CYR cells). Fig. 1dshows that the YFP-RXR signal is detectable in every CYR cell,predominantly within the nucleus. The addition of 9c-RA toCYR cells also induced intranuclear YFP-RXR redistribution(not shown), suggesting that overexpression does not contrib-ute to foci formation. Taken together, our studies with Western

blot analysis, microscopy, and transactivation assays demon-strate that YFP-RXR and RXR-BFP encode intact, functional,and fluorescing receptors. These receptors appear to reside inthe cell nucleus and respond to hormone treatment by formingintranuclear foci.

VDR Expression and Subcellular Localization—We previ-ously reported that VDR partitions between the cytoplasm andthe nucleus without calcitriol and translocates into the nucleusafter calcitriol treatment in experiments using a transientlyexpressed N-terminal GFP chimera of VDR (5). Here we gen-erated a C-terminal fusion of VDR with GFP (GFP-VDR) andstably expressed it in 293 cells. One moderately expressingclone was selected for characterization (GL48). Western blotanalysis of GL48 cell extracts confirmed the expression of in-tact GFP-VDR. The expression level of GFP-VDR was approx-imately twice the expression level of the endogenous VDR inROS 17/2.8 cells (data not shown). Transactivation assay intransiently transfected COS-7 cells showed that the potency ofcalcitriol to induce transcription of the 24OH/Luc-23 reporter isthe same in GFP-VDR and in VDR-expressing cells. The half-maximal effect was detected at 0.5 nM calcitriol, and the max-imal induction was at 10 nM calcitriol (5-fold for GFP-VDR and4-fold for VDR over vehicle-treated controls).

Microscopy showed that without hormone, GFP-VDR parti-tions between the cytoplasm and the nucleus of GL48 cells (Fig.2a) and that after the addition of hormone (0.01–100 nM calcit-riol), GFP-VDR localizes predominantly in the nucleus (Fig.2b). The subcellular distribution of dnaGFP-VDR was indistin-guishable from the distribution of wild-type GFP-VDR (Fig. 2,c–d), and calcitriol induced dnaGFP-VDR translocation fromthe cytoplasm into the nucleus. This translocation is probablynot related to homodimerization and a “piggyback” mechanism(34), since CV-1 cells contain negligible amounts of endogenousVDR. We also generated a nuclear localization sequence mu-tant of the GFP-VDR (nlsGFP-VDR). The functional signifi-

FIG. 1. YFP-RXR is predominantly nuclear and responds tohormone by forming intranuclear foci. Confocal microscopyshowed YFP-RXR fluorescence predominantly in the nuclei of tran-siently expressing ROS cells (a–c) and in stably expressing CV-1 cells(d). Unliganded YFP-RXR distributed diffusely within the nuclei ex-cluding the nucleoli (a). The addition of 100 nM 9c-RA caused an in-tranuclear pattern change as YFP-RXR formed bright intranuclear foci(b). Cytoplasmic YFP-RXR fluorescence was also detectable; the inten-sity of this cytoplasmic signal was low in the majority of cells but higherin a number of cells (c). Bar, 10 mm.

RXR Effect on VDR Distribution41116

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

cance of this mutation in the NLS1 of VDR was recently char-acterized; the mutant receptors are competent to bind calcitrioland to form heterodimers with RXR, but their ability to accu-mulate in the nucleus, bind DNA, and induce transcription arecompromised (7). Consistent with previous findings on fixedcells, microscopy on living CV-1 cells showed significantly morenlsGFP-VDR in the cytoplasm than in the nucleus (Fig. 2e).Calcitriol induced only a slight translocation of nlsGFP-VDRfrom the cytoplasm into the nucleus (Fig. 2f). These resultssuggest that nuclear accumulation of VDR is unrelated to DNAbinding and is dependent on intact interaction with the nuclearimport receptors.

FRET Shows VDR/RXR Heterodimers in the Nucleus—FRET experiments were performed to evaluate the abilities ofthe tagged receptors to form heterodimers and to elucidate thesubcellular distribution of heterodimers. Interactions betweentwo proteins tagged with appropriate fluorescent markers canbe detected by measuring the emission intensities of the accep-tor fluorophore after exciting the donor fluorophore. We usedRXR-BFP for the donor and GFP-VDR for the acceptor, whichwere cotransfected into CV-1 cells. We excited the donor RXR-BFP at 380 nm and recorded the emission of RXR-BFP with theblue filter set and the emission of the acceptor GFP-VDR withthe green filter set as described under “Experimental Proce-dures.” When the distance between the two proteins is lessthan 100 Å, an increase in the acceptor emission can take place(14). Since heterodimerization is a tight molecular interaction,we postulated that the acceptor emission intensities would behigher in the presence of RXR-BFP than in the presence of the

mutant hdRXR-BFP. To determine which levels of acceptoremission signify heterodimers, we conducted preliminary ex-periments. First, we generated ratio images of green emissionand blue emission from the same field to compensate for dif-ferences caused by different RXR-BFP expression and differentbackground fluorescence. We then applied a look-up color tableto the ratio image to show intensity gradients (Fig. 3). Second,we determined the contribution of nonspecific signal intensi-ties. The fluorescence intensities over cells expressing GFP-VDR alone were only approximately 20% of the intensitiesgenerated from FRET. Thus, we did not calculate ratios unlessthe cells expressed a blue protein. In cells expressing BFP orRXR-BFP, the blue signal intensities were high, whereas thegreen intensities remained low; the green intensities afterbackground subtraction were only 40% that of the intensitiesfrom FRET (Fig. 3c). As a result, ratio intensities from thebleed-through of BFP were 0.9 6 0.02, and intensities from thebleed-through of RXR-BFP were 0.9 6 0.1. To detect FRETsignals from heterodimers, we generated ratio images fromcells coexpressing GFP-VDR and RXR-BFP (Fig. 3a). Controlratio images were generated from cells coexpressing the GFP-VDR and the mutant hdRXR-BFP (Fig. 3b). The ratio values ofthe GFP-VDR/RXR-BFP-expressing cells were 1.7 6 0.5 overthe nuclei and 1.2 6 0.2 over the cytoplasm. However, controlratio values of the GFP-VDR/hdRXR-BFP-expressing cellswere significantly lower: 1.0 6 0.2 over the nuclei (p , 0.025)and 0.9 6 0.1 over the cytoplasm (p , 0.05). After a 30-minincubation with 50 nM calcitriol, heterodimers were no longerdetectable in the cytoplasm. This finding suggests that theheterodimers translocate from the cytoplasm into the nucleus,which is similar to calcitriol-induced GFP-VDR translocation(Fig. 3d). These FRET experiments demonstrated that thetagged receptors heterodimerize in intact living cells and indi-cated that heterodimers are present in both the cytoplasm andnucleus with significantly more heterodimers present in thenucleus than in the cytoplasm. Most importantly, FRETrevealed that calcitriol promotes nuclear redistribution ofheterodimers.

Heterodimerization between VDR and RXR Shifts VDR intothe Nucleus—The results obtained in living cells indicated thatRXR and VDR/RXR heterodimers localize predominantly in thenucleus. There was no apparent change in RXR-BFP distribu-tion when it was coexpressed with GFP-VDR in CV-1 cells.However, GFP-VDR was more nuclear when coexpressed withRXR-BFP. To explore the possibility that heterodimerizationwith RXR modifies VDR localization in more detail, we studiedthe effect of RXR-BFP expression on GFP-VDR localization inGL48 cells. Multi-channel confocal microscopy of living GL48cells showed steady-state GFP-VDR distribution in the greenchannel and RXR-BFP distribution in the blue channel. With-out RXR, the unliganded GFP-VDR is evenly distributed be-tween cytoplasm and nucleus (Fig. 2). When these cells weretransiently transfected with RXR-BFP, 30% of cells expressedthe blue-tagged RXR (Fig. 4b). In those cells expressing RXR-BFP (arrows), 69 6 7% GFP-VDR accumulated in the nucleus,whereas in the nontransfected cells 46 6 7% of total GFP-VDRsignal was in the nucleus (p , 0.0001) (Fig. 4a). This effect ofRXR-BFP to induce a nuclear shift of GFP-VDR could havebeen the result of heterodimerization, but several other non-specific effects had to be considered. To test the possibility thatnonspecific protein overexpression caused mislocalization ofVDR, we transfected GL48 cells with large amounts of BFP.This experiment showed that overexpression of the BFP pro-tein does not influence VDR distribution (Fig. 4, c-d) and clar-ified that the more intense nuclear GFP-VDR signal does notderive from a bleed-through of the BFP signal. Another possi-

FIG. 2. Mutation in the NLS of GFP-VDR perturbs subcellulardistribution and hormone-dependent translocation. GFP-VDRfluorescence distributed evenly between the cytoplasm and the nucleusin GL48 cells (a 293 cell-derived clone that stably expresses GFP-VDR)(a). Calcitriol treatment (10 nM for 30 min) induced GFP-VDR translo-cation into the nucleus (b). A point mutation in the DNA binding region(R77Q) of VDR (c–d) did not perturb dnaGFP-VDR-partitioning andcalcitriol-induced translocation into the nucleus in CV-1 cells (d). Apoint mutation in the nuclear localization signal sequence within theDNA binding region of VDR (nlsGFP-VDR) reduced both hormone-independent and hormone-dependent nuclear import of receptors inCV-1 cells (e–f). Bar, 10 mm.

RXR Effect on VDR Distribution 41117

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

bility was an effect due to changes in the structure and functionof RXR as a consequence of the addition of BFP tag. Thus, weexpressed the untagged RXR in GL48 cells and found that italso causes a shift of GFP-VDR into the nucleus (not shown).This effect of RXR to shift GFP-VDR into the nucleus wasreproducible in COS-7, CV-1, and 293 cells (not shown).

To test the role of heterodimerization, hdRXR-BFP wastransfected into GL48 cells. Unlike the wild-type RXR-BFP,expression of hdRXR-BFP did not intensify the intranuclearGFP-VDR signal. GFP-VDR remained evenly distributed be-tween the cytoplasm and the nucleus (Fig. 4e, arrows), al-though the hdRXR-BFP remained predominantly nuclear (Fig.4f). Taken together, these results demonstrate the specificityand reproducibility of the effect of RXR on VDR distributionand indicate a new function for heterodimerizing RXR in thecontrol of VDR localization. RXR could exert this effect byfacilitating retention of VDR in the nucleus and by facilitatingVDR import into the nucleus.

RXR Inhibits VDR Export—We noticed during microscopythat after keeping GL48 cells at room temperature, GFP-VDRmoves out of the nucleus into the cytoplasm (Fig. 5a–b). Thistemperature sensitivity was present in other cell types but to alesser extent. We took advantage of this temperature-depend-ent export to evaluate the ability of RXR to promote nuclearretention of VDR. We incubated GL48 cells for 2 h at either 37or at 22 °C before microscopy and maintained these conditionsduring microscopy. As expected, GFP-VDR moved out from thenucleus and accumulated in the cytoplasm when GL48 cellswere incubated at 22 °C for 2 h (Fig. 5, a–b). This redistributionwas specific for VDR, as YFP-RXR was not exported at 22 °C(Fig. 5, c-d). GFP-VDR remained functional at this tempera-ture, as the addition of 10 nM calcitriol induced a completetranslocation of cytoplasmic GFP-VDR within 2 h at 22 °C (not

shown). It was also reversible; increasing the temperature to37 °C restored nuclear localization. RXR-BFP and GFP distri-bution did not change when the temperature was lowered (notshown). Then we transfected GL48 cells with RXR-BFP andcompared temperature sensitivity of GFP-VDR distribution inRXR-BFP-expressing and nontransfected cells. GFP-VDR re-mained in the nucleus after lowering the temperature in cellsexpressing both proteins, but GFP-VDR moved into the cyto-plasm in the nontransfected cells (Fig. 5, e–f). Morphometricanalysis confirmed that RXR-BFP inhibits temperature-dependent GFP-VDR export. Without RXR-BFP, the nuclear/cytoplasmic ratios of GFP-VDR signal intensities in GL48 cellswere 1.28 6 0.03% at 37 °C and 0.86 6 0.01 at 22 °C (p , 0.01).Nuclear/cytoplasmic ratios of the YFP-RXR in the CYR cellsdid not decrease after lowering the temperature (6.3 6 1.1 at37 °C and 8.1 6 0.36 at 22 °C). When GL48 cells express RXR-BFP, VDR ratios were 5.1 6 0.49 at 37 °C and 3.8 6 0.49 at22 °C (p 5 0.167). This change is smaller than the changewithout RXR-BFP and suggests that nuclear retention of VDRby RXR could contribute to the nuclear accumulation.

RXR Promotes VDR Import—To test the ability of RXR toinfluence nuclear import, we coexpressed the nlsGFP-VDR andthe RXR-BFP in CV-1 cells. Multi-channel confocal microscopyshowed that the nlsGFP-VDR remained mostly in the cyto-plasm of cells, which did not express RXR-BFP but becamenuclear in cells that expressed the RXR-BFP (Fig. 6, a–b). Theability of RXR-BFP to shift the nlsGFP-VDR into the nucleuswas also reproducible in COS-7 and 293 cells (not shown). Totest the role of heterodimerization, we expressed the nlsGFP-VDR together with the heterodimerization mutant hdRXR-BFP. Microscopy showed that the hdRXR-BFP has no effect onthe nlsGFP-VDR distribution, as the nlsGFP-VDR remained inthe cytoplasm in both the hdRXR-BFP-expressing and nonex-

FIG. 3. FRET shows VDR and RXRheterodimers in the nucleus and inthe cytoplasm. CV-1 cells were cotrans-fected with GFP-VDR and RXR-BFP (aand d), GFP-VDR and hdRXR-BFP (b), orwith BFP alone (c). Representative ratioimages are shown from FRET experi-ments, with a color table that was appliedto show intensity differences. Ratio im-ages were generated from the imagestaken with a 370–390-nm excitation filterand with either a 515–555-nm or a 435–485-nm emission filter as indicated. Ratioimage of cells expressing both wild typereceptors show energy transfer betweenthe RXR-BFP and the GFP-VDR. The ra-tios in the green range indicate het-erodimers; most of them are in the nuclei,and a few of them are in the cytoplasm(a). The ratio image from cells expressingthe GFP-VDR and the hdRXR-BFP showsratio values in the blue range because theBFP emission is stronger than the GFPemission (b). The ratio image from cellsexpressing only the BFP shows similarratio values (c). After treating the GFP-VDR/RXR-BFP-expressing cells with 10nM calcitriol, ratio images indicate thestrongest green signal for heterodimerswithin intranuclear foci (d). Bar, 10 mm.

RXR Effect on VDR Distribution41118

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

pressing cells (Fig. 6, c–d). Thus, the effect of RXR on VDRlocalization requires an intact heterodimerization function ofRXR.

Next, we studied the effect of RXR on the calcitriol-inducedtranslocation of GFP-VDR. Because in RXR-BFP-expressingcells most of the GFP-VDR is already in the nucleus withoutthe hormone, the effect of calcitriol was not easily noticeable.However, RXR-BFP caused only a partial translocation of thenlsGFP-VDR without the hormone (Fig. 2, e–f), thus allowingfor the detection of calcitriol-induced translocation. Microscopyshowed that RXR-BFP restored the ability of calcitriol to in-duce translocation of the nlsGFP-VDR into the nucleus (Fig. 6).The mutant hdRXR-BFP was ineffective in rescuing the hor-mone-dependent translocation of the nlsGFP-VDR. Morpho-metric analysis of these images showed that without the pres-ence of RXR, calcitriol caused a minimal nuclear shift ofnlsGFP-VDR (p 5 0.06) (Fig. 7). In RXR-coexpressing cells,calcitriol caused a significant shift of the nlsGFP-VDR into thenucleus (p , 0.0001) (Fig. 7). Taken together, these data sug-

gest that dimerization of RXR with VDR facilitates VDR nu-clear import.

RXR Influences Calcitriol-dependent Intranuclear Redistri-bution of VDR—The addition of calcitriol to GL48 cells inducedformation of multiple small intranuclear foci (Fig. 8a). We havedetected similar intranuclear VDR-GFP foci formation afterhormone treatment (5). However, the addition of calcitriol didnot cause such focal nuclear accumulation of the dnaGFP-VDRand the nlsGFP-VDR (Fig. 8, b and c). Coexpression of RXR-BFP restored the ability of calcitriol to induce foci formation ofnlsGFP-VDR (Fig. 8d). In contrast, coexpression of RXR-BFPwith the dnaGFP-VDR did not allow calcitriol-dependent in-tranuclear redistribution (not shown). These results indicatean association between the lack of foci formation and the lackof DNA binding. The effect of RXR on foci formation could beinterpreted as an increase of DNA binding.

FRET experiments showed that calcitriol also induces in-tranuclear redistribution of VDR/RXR heterodimers. The ratioimage (Fig. 3d) showed that FRET is strongest in the foci.There were no focal intranuclear ratio increases in the hor-mone-treated cells expressing GFP-VDR/hdRXR-BFP (data notshown). Because previous studies showed that RXR and VDRbind to DNA as heterodimers, the hormone-induced het-erodimer accumulation within discrete nuclear foci could sig-nify receptor binding to DNA.

RXR Effect on VDR Distribution Is Functionally Signifi-cant—To explore the physiological importance of RXR-BFPeffect on GFP-VDR localization, we performed transactivation

FIG. 4. Heterodimerization with RXR-BFP induces nuclear ac-cumulation of GFP-VDR. Representative images are shown fromdual channel confocal microscopy experiments, with the GFP channelshown in green and the BFP channel shown in blue. When GL48 cellswere transiently transfected with RXR-BFP (a–b), BFP (c–d), orhdRXR-BFP (e–f), a subpopulation of cells expressed these blue fluo-rescing proteins (arrows). GFP-VDR distributed evenly between thecytoplasm and the nucleus in the nontransfected GL48 cells (cells notmarked with arrows on panel a). In contrast, GFP-VDR distributionwas predominantly nuclear in the cells that expressed RXR-BFP (cellsmarked with arrows on panel a). Expression of BFP or hdRXR-BFP didnot alter GFP-VDR distribution (compare cells that are marked witharrows and the unmarked cells on panels c and e). Bar, 10 mm.

FIG. 5. Temperature-dependent export of GFP-VDR is inhib-ited by heterodimerizing RXR-BFP. GFP-VDR shifted from thenucleus into the cytoplasm after a 2-h incubation at 22 °C in GL48 cells(a–b). In contrast, YFP-RXR remained nuclear after lowering the tem-perature from 37 to 22 °C in CYR cells (c–d). When GL48 cells weretransiently transfected with RXR-BFP, GFP-VDR fluorescence was pre-dominantly nuclear in every cell that expressed the blue fluorescingprotein (arrows) both at 37 °C (e) and at 22 °C (f). In the same experi-ment, all the cells that expressed the GFP-VDR without RXR-BFPshowed even distribution of fluorescence at 37 °C and predominantlycytoplasmic fluorescence at 22 °C (cells without arrows). Bar, 10 mm.

RXR Effect on VDR Distribution 41119

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

assays using a suitable vitamin D response element-luciferasereporter system (24OH/Luc). This 24-hydroxylase promotercontains vitamin D response element sequences separated bythree nucleotides (DR3), which typically bind VDR/RXR het-erodimers. Moreover, previous studies indicated that coexpres-sion of RXR stimulates basal transcription of this reporter byVDR (13). First, we compared basal transcriptional activities inGL48 cells transfected with increasing concentrations of RXR-BFP. Transfection with 0.003 mg/well RXR-BFP plasmid re-sulted in a 2.7-fold increase and, with 0.02 mg/well RXR-BFP,in a 3.6-fold increase in basal transcriptional activity comparedwith activities from GL48 cells transfected with the reporter

plasmid and herring sperm DNA. We then compared the lucif-erase activities of extracts from CV-1 cells that were trans-fected with 24OH/Luc, GFP-VDR, and RXR-BFP with activities

FIG. 6. RXR-BFP brings nlsGFP-VDR into the nucleus. Representativeimages are shown from dual channel con-focal microscopy experiments, with theGFP channel shown in green and the BFPchannel shown in blue. CV-1 cells weretransfected with nlsGFP-VDR togetherwith either RXR-BFP (a–b) or hdRXR-BFP (c–d). The nlsGFP-VDR accumu-lated in the nucleus in every cell that alsoexpressed the RXR-BFP (arrow) (a). Inthe nontransfected cells the nlsGFP-VDRremained in the cytoplasm (such as thecell that is not marked by an arrow in thelower left corner of panel a). The hdRXR-BFP did not influence the nlsGFP-VDRdistribution (c). Bar, 10 mm.

FIG. 7. Morphometric analysis demonstrates that RXR-BFPpromotes calcitriol-dependent translocation of nlsGFP-VDR.CV-1 cells were transiently transfected with expression plasmids andexposed to calcitriol for 1 h before microscopy. Images were taken fromvehicle-treated (open bars) and calcitriol-treated (shaded bars) livingcells. The ratios of nuclear and cytoplasmic brightness values werecalculated as described under “Experimental Procedures.” RXR-BFPincreased the nuclear GFP-VDR levels both in the presence and absenceof calcitriol and restored the ability of calcitriol to shift the nlsGFP-VDRinto the nucleus. Data represent mean 6 1 S.E.

FIG. 8. RXR-BFP promotes calcitriol-dependent intranuclearfoci formation of nlsGFP-VDR. CV-1 cells were transfected withGFP-VDR (a), dnaGFP-VDR (b), nlsGFP-VDR (c) or cotransfected withnlsGFP-VDR and RXR-BFP (d). Cells were treated with calcitriol for 1 hbefore microscopy. Confocal images were taken from living cells. Hor-mone treatment resulted in a nuclear accumulation of the wild-typeGFP-VDR and the formation of small bright foci within the nucleus (a).After calcitriol, the dnaGFP-VDR and the nlsGFP-VDR did not formintranuclear foci (b–c). Cotransfection of RXR-BFP restored the abilityof calcitriol to induce intranuclear nlsGFP-VDR foci formation (d). Bar,10 mm.

RXR Effect on VDR Distribution41120

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

from cells that were transfected with 24OH/Luc, GFP-VDR,and hdRXR-BFP. The coexpression of RXR-BFP with GFP-VDR increased basal transcription by 2.5-fold (Fig. 9, A–B,GFP-VDR, open bars). In contrast, the coexpression of thehdRXR-BFP had a moderate inhibitory effect (not shown), pos-sibly due to a dominant negative effect on the endogenous RXR

(35). Similar experiments were carried out with the nlsGFP-VDR. RXR-BFP coexpression increased reporter activity 4.2-fold (Fig. 9, A–B nlsGFP-VDR, open bars), whereas the coex-pression of the hdRXR-BFP had no effect (not shown). Thiseffect of RXR to increase basal transcription correlated withour observation that RXR increases nuclear import of VDR.The connection is further supported by the correlation betweenthe cytoplasmic localization and the 3-fold lower basal tran-scriptional activity of the nlsGFP-VDR compared with GFP-VDR. These results all suggest that heterodimerization withRXR increases the hormone-independent transcriptional activ-ity of VDR by facilitating its hormone-independent nuclearimport.

We used similar cotransfection assays to test if this shift inVDR localization by RXR influences hormone-dependent func-tions of VDR. Calcitriol (10 nM) addition for 24 h induced a15-fold increase in reporter activity compared with vehicle-treated controls in GFP-VDR-expressing cells (Fig. 9A, GFP-VDR, shaded bar). Coexpression of RXR did not influence cal-citriol-induced maximal luciferase activity by GFP-VDR (Fig.9B, GFP-VDR, shaded bar). Calcitriol did not increase lucifer-ase activity by the nlsGFP-VDR (Fig. 9A, nlsGFP-VDR, shadedbar). Coexpression of RXR-BFP, however, completely restoredthe calcitriol-dependent transcriptional activity of the nlsGFP-VDR (Fig. 9B, nlsGFP-VDR, shaded bar). Coexpression of thehdRXR-BFP with the nlsGFP-VDR failed to increase calcitriol-dependent transcriptional activity (not shown). Because muta-tions of the nlsGFP-VDR also impair DNA binding, we testedthe effect of RXR-BFP coexpression on the transcriptional ac-tivity of another DNA binding mutant of GFP-VDR (dnaGFP-VDR). Calcitriol had no effect on transcriptional activity ofdnaGFP-VDR (Fig. 9A, dnaGFP-VDR), and RXR-BFP coex-pression caused only a small increase in the calcitriol-depend-ent transcriptional activity by the dnaGFP-VDR (Fig. 9B,dnaGFP-VDR).

Collectively, these assays reveal that the effects of RXR onnuclear VDR accumulation and intranuclear reorganizationcorrelate with the effects of RXR on VDR transcriptional activ-ity. By bringing the unliganded GFP-VDR into the nucleus,RXR increases basal activity, and by bringing the nlsGFP-VDRinto the nucleus, RXR rescues its hormone-dependent tran-scriptional activity.

DISCUSSION

Our experiments with green and blue fluorescent chimeras ofthe receptors and FRET microscopy provided the initial evi-dence for the distribution of VDR/RXR heterodimers in livingcells. Confocal microscopy also allowed us to recognize thatdimerizing RXR facilitates nuclear accumulation of VDR. Inaddition, we observed that VDR, RXR, and their respectiveheterodimers all responded to hormone by forming intranu-clear bright foci. Mutational analysis showed a correlation be-tween receptor distribution and transcriptional activity andbetween nuclear foci formation and DNA binding.

At first, we generated N-terminal-tagged GFP-VDR andYFP-RXR and a C-terminal-tagged RXR-BFP expression plas-mids. The transcriptional activities of the tagged receptorswere confirmed by luciferase reporter assays, and the expres-sion of the correct size proteins were confirmed by Western blotanalysis. We also cloned cell lines to stably express the GFP-VDR (GL48) and the YFP-RXR (CYR) to establish reproduciblesystems for the investigation of hormone-independent and hor-mone-dependent receptor trafficking. Previously, we and oth-ers used alternate GFP chimeras of VDR to show hormone-de-pendent translocation from the cytoplasm into the nucleus (5,6). Confocal microscopy on GL48 cells confirmed our previousfindings on the steady-state subcellular distribution of VDR by

FIG. 9. RXR-BFP changes basal and calcitriol-induced tran-scriptional activities of wild type, nlsGFP-VDR, and dnaGFP-VDR. CV-1 cells were cotransfected with the 24OH/Luc-23 reporter andthe expression plasmids as described under “Experimental Procedures.”Luciferase activities of cells vehicle treated are shown as open bars, andthose treated with calcitriol are shown as shaded bars. Activities of thewild type (first group), nlsGFP-VDR (second group) and dnaGFP-VDR(third group) are shown without coexpression of RXR (A) and withcoexpression of RXR (B). Luminescence values were normalized withb-galactosidase activities, and data are expressed as mean 6 S.E. Basaltranscriptional activity of the GFP-VDR increased 2.5-fold by RXR-BFPcoexpression. RXR-BFP expression caused a similar increase in thebasal activity for the nlsGFP-VDR and the dnaGFP-VDR. RXR-BFPrescued ligand-dependent transcriptional activity of the nlsGFP-VDR.The calcitriol-induced transcriptional activity of the dnaGFP-VDR wasonly slightly increased.

RXR Effect on VDR Distribution 41121

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

showing that unliganded GFP-VDR remained evenly distrib-uted between the cytoplasm and nucleus (Fig. 2).

Our finding on the steady-state nuclear localization of YFP-RXRa in living cells was expected from previous immunolocal-ization studies, and it is similar to the nuclear localization ofthe transiently expressed GFP-RXRb in PC12 cells (36). Incontrast, the presence of YFP-RXR in the cytoplasm is new andsuggests that heterodimers of VDR and RXR could also form inthe cytoplasm (Fig. 1). Furthermore, the difference between thesteady-state localization of RXR and VDR raised the possibilitythat heterodimerization could influence the distribution of ei-ther VDR or RXR.

FRET microscopy has not been used extensively in the nu-clear receptor field. Here, utilizing a heterodimerization mu-tant of RXR as a control, we located dimers of RXR-BFP andGFP-VDR for the first time in intact living cells. These FRETstudies showed that the tagged receptors are competent to formheterodimers; therefore, they were suitable for studies on theimpact of dimerization on subcellular trafficking. FRET resultsdemonstrated that most of the VDR formed dimers with theRXR without the ligand and that these heterodimers werepredominantly in the nucleus in a diffuse pattern excluding thenucleoli.

Our experiments showed that the dimerization-competentRXR can shift the VDR from the cytoplasm into the nucleus,but the dimerization-deficient RXR is ineffective. This effect ofRXR on VDR distribution could involve changes in either VDRexport or import across the nuclear envelope. We tested theeffect of RXR on VDR export by taking advantage of the differ-ences in their temperature sensitivity. Lowering the tempera-ture of GL48 cells induced export of GFP-VDR into the cyto-plasm, but GFP-VDR/RXR-BFP heterodimers were lesssensitive to temperature change and remained in the nucleus(Fig. 5). This finding suggests that the export mechanisms forthe heterodimer differ from that of VDR. It is possible thatexport receptors are involved because both VDR and RXR haveleucine-rich sequences, which could serve as nuclear exportsignal (37). Mutational analysis of these GFP-tagged receptorsopens several potential avenues for the investigations of theseexport mechanisms.

We also found that RXR influences nuclear import of VDR.This is most apparent in our experiments with the nlsGFP-VDR, which harbors a mutation in one of the nuclear localiza-tion signals of the VDR (7). This mutation severely compro-mised nuclear import of VDR but did not completely abolish it.The remaining import is likely to be mediated by other domainsof the VDR, such as the one suggested within the hinge region(24). However, we did not find any defect in nuclear importwhen we deleted this region of GFP-VDR.2 The ability of RXRto bring the NLS mutant GFP-VDR into the nucleus can beexplained by a piggyback mechanism. Such a mechanism wasreported to bring the NLS mutant of progesterone receptor intothe nucleus by homodimerizing wild-type receptors (38). Theeffect of RXR could also be indirect; RXR could weaken theVDR interactions with cytoplasmic docking sites, as suggestedby a recent report showing that RXR increases the solubilityand perhaps assists the folding of the retinoic acid receptor inEscherichia coli (39). Finally, RXR could also act by complexingwith Hsp70 and Hsp90 chaperones (40). Interestingly, RXR notonly influenced the steady-state distribution of the nlsGFP-VDR but also restored the effect of calcitriol to induce a rapidnuclear accumulation of the mutant receptors (Figs. 6 and 7).This effect is unexpected and implies a role for RXR in theregulation of VDR nuclear import.

We found a strong correlation between the ability of RXR toshift VDR into the nucleus and to regulate its transcriptionalactivity. First, the increase in the nuclear VDR content corre-lated with an increase in basal transcriptional activity by VDRboth in GL48 and CV-1 cells. This hormone-like effect of RXR isconsistent with the phantom ligand effect of RXR describedearlier (41). This effect on basal activity may be physiologicallyimportant, since previous studies indicate that the unligandedVDR can activate several target genes (42, 43). Second, RXRfacilitated hormone-independent nuclear localization of thenlsGFP-VDR and increased basal transcriptional activity bythe nlsGFP-VDR. Third, RXR restored the ability of the nls-GFP-VDR to accumulate in the nucleus after calcitriol additionand to activate the reporter gene after calcitriol addition. Thisindicates that the regulation of VDR nuclear import by RXRmay play a role in hormone-dependent transcriptional regula-tion. Such regulation could explain how, in some target tissues,changes in the ratio between RXR and VDR can influencecalcitriol-induced transcription (44, 45). Potentially, the regu-lation of protein trafficking across the nuclear envelope by RXRis a general mechanism that controls the activity of manytranscriptional regulators.

Regulation of subnuclear trafficking can be one of the waysRXR influences the speed and specificity of transcriptionalactivities. Here we found that hormone induces YFP-RXR andGFP-VDR to accumulate in foci (Figs. 1 and 2). Our earlierfinding that a DNA binding mutant of GFP-GR failed to formfoci (46) and our present finding that the DNA binding mutantof GFP-VDR also fails to form nuclear foci suggest that thesefoci could signify receptor binding to DNA target sites. Thisnotion is further supported by the FRET experiments, whichshow that calcitriol treatment causes focal accumulation ofheterodimers in the nucleus (Fig. 3). The use of the het-erodimerization mutant of RXR-BFP for control makes thisargument even more powerful. Moreover, coexpression of RXR-BFP restored the ability of the nlsGFP-VDR to form foci andactivate transcription after calcitriol treatment (Figs. 7 and 9).The dimerization-deficient form of RXR-BFP had no effect.This observation is in agreement with earlier findings on theeffect of calcitriol to promote binding of VDR/RXR heterodimersto vitamin D response elements (47, 48) and suggests that RXRcould promote DNA binding by influencing intranuclear target-ing of VDR. A local change in receptor dynamics could accountfor the calcitriol-induced accumulation of heterodimers in dis-crete subnuclear regions and warrant further investigations byphotobleaching techniques.

Multicolor fluorescent protein tagging and advanced micros-copy methods allowed us to visualize subcellular distribution ofVDR, RXR, and their respective heterodimers in living cells.These studies revealed that RXR heterodimerization plays akey role in the translocation and nuclear targeting of VDR,which could explain tissue-specific differences in VDR distri-bution observed previously by immunocytology (25). We alsovisualized the ligand-dependent interaction of heterodimerswith possible DNA binding sites by FRET imaging and found acorrelation between intranuclear foci formation and transcrip-tional activity by mutational analysis. Mutational analysis alsoindicated that nuclear targeting is independent of DNA bind-ing. As the RXR forms heterodimers with other nuclear recep-tors and associates with coactivators and corepressors, RXRcould also regulate the activity of other partners by regulatingsubcellular trafficking. The techniques described here open thepossibility of exploring the location and kinetics for a multitudeof regulatory protein interactions and exploring the mecha-nisms of transcriptional regulation in living cells.

Acknowledgments—We are grateful to Dr. J. W. Pike (University of2 J. Barsony, unpublished data.

RXR Effect on VDR Distribution41122

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Cincinnati College of Medicine, Cincinnati, OH) for human VDR cDNA,Dr. P. Chambon (CNRS/INSERM, Strasbourg, France) for humanRXRa cDNA, Dr. J. Ellenberg (NICHD, National Institutes of Health)for the 10C plasmid, Dr. H. F. DeLuca (University of Wisconsin, Mad-ison, WI) for the p24OH/Luc23 plasmid, and Dr. S. K. Karathanasis(Wyeth-Ayerst Research, Radnor, PA) for the retinoid X response ele-ment-Luc plasmid. We thank Natasha Sefcovic for generating the DNAbinding mutant of VDR, Claudia Schroder for performing the Westernblot analyses, and Colin Segovis for image processing and editorialassistance.

REFERENCES

1. Sone, T., Kerner, S., and Pike, J. W. (1991) J. Biol. Chem. 266, 23296–233052. Blumberg, B., and Evans, R. M. (1998) Genes Dev. 12, 3149–31553. Tagami, T., Lutz, W. H., Kumar, R., and Jameson, J. L. (1998) Biochem.

Biophys. Res. Commun. 253, 358–3634. Carlberg, C., and Saurat, J. H. (1996) J. Investig. Dermatol. Symp. Proc. 1,

82–865. Racz, A., and Barsony, J. (1999) J. Biol. Chem. 274, 19352–193606. Michigami, T., Suga, A., Yamazaki, M., Shimizu, C., Cai, G., Okada, S., and

Ozono, K. (1999) J. Biol. Chem. 274, 33531–335387. Hsieh, J. C., Shimizu, Y., Minoshima, S., Shimizu, N., Haussler, C. A.,

Jurutka, P. W., and Haussler, M. R. (1998) J. Cell. Biochem. 70, 94–1098. Zhu, X. G., Hanover, J. A., Hager, G. L., and Cheng, S. Y. (1998) J. Biol. Chem.

273, 27058–270639. Reichrath, J., Mittmann, M., Kamradt, J., and Muller, S. M. (1997) Histochem.

J. 29, 127–13310. Sugawara, A., Sanno, N., Takahashi, N., Osamura, R. Y., and Abe, K. (1997)

Endocrinology 138, 3175–318011. Sanno, N., Sugawara, A., Teramoto, A., Abe, Y., Yen, P. M., Chin, W. W., and

Osamura, R. Y. (1997) Neuroendocrinology 65, 299–30612. Janssen, J. J., Kuhlmann, E. D., van Vugt, A. H., Winkens, H. J., Janssen,

B. P., Deutman, A. F., and Driessen, C. A. (1999) Curr. Eye Res. 19, 338–34713. Kephart, D. D., Walfish, P. G., DeLuca, H., and Butt, T. R. (1996) Mol.

Endocrinol. 10, 408–41914. Periasamy, A., and Day, R. N. (1999) Methods Cell Biol. 58, 293–31415. Day, R. N. (1998) Mol. Endocrinol. 12, 1410–141916. Chauchereau, A., Georgiakaki, M., Perrin-Wolff, M., Milgrom, E., and Loosfelt,

H. (2000) J. Biol. Chem. 275, 8540–854817. Johnson, C., Van Antwerp, D., and Hope, T. J. (1999) EMBO J. 18, 6682–669318. Fejes-Toth, G., Pearce, D., and Naray, F. T. (1998) Proc. Natl. Acad. Sci.

U. S. A. 95, 2973–297819. Htun, H., Barsony, J., Renyi, I., Gould, D. L., and Hager, G. L. (1996) Proc.

Natl. Acad. Sci. U. S. A. 93, 4845–485020. Lim, C. S., Baumann, C. T., Htun, H., Xian, W., Irie, M., Smith, C. L., and

Hager, G. L. (1999) Mol. Endocrinol. 13, 366–37521. Georget, V., Lobaccaro, J. M., Terouanne, B., Mangeat, P., Nicolas, J. C., and

Sultan, C. (1997) Mol. Cell. Endocrinol. 129, 17–2622. Powers, M. A., and Forbes, D. J. (1994) Cell 79, 931–93423. Kohler, M., Haller, H., and Hartmann, E. (1999) Exp. Nephrol. 7, 290–294

24. Luo, Z., Rouvinen, J., and Maenpaa, P. H. (1994) Eur. J. Biochem. 223,381–387

25. Prufer, K., Veenstra, T. D., Jirikowski, G. F., and Kumar, R. (1999) J. Chem.Neuroanat. 16, 135–145

26. Htun, H., Holth, L. T., Walker, D., Davie, J. R., and Hager, G. L. (1999) Mol.Biol. Cell 10, 471–486

27. Stenoien, D. L., Mancini, M. G., Patel, K., Allegretto, E. A., Smith, C. L., andMancini, M. A. (2000) Mol. Endocrinol. 14, 518–534

28. Ellenberg, J., Lippincott-Schwartz, J., and Presley, J. F. (1998) Biotechniques25, 838–836

29. Sone, T., Marx, S. J., Liberman, U. A., and Pike, J. W. (1990) Mol. Endocrinol.4, 623–631

30. Barsony, J., Renyi, I., and McKoy, W. (1997) J. Biol. Chem. 272, 5774–578231. Chen, S., Costa, C. H., Nakamura, K., Ribeiro, R. C., and Gardner, D. G. (1999)

J. Biol. Chem. 274, 11260–1126632. Kimmel-Jehan, C., Jehan, F., and DeLuca, H. F. (1997) Arch. Biochem. Bio-

phys. 341, 75–8033. Harnish, D. C., Malik, S., Kilbourne, E., Costa, R., and Karathanasis, S. K.

(1996) J. Biol. Chem. 271, 13621–1362834. Guiochon-Mantel, A., Delabre, K., Lescop, P., Perrot-Applanat, M., and

Milgrom, E. (1994) Biochem. Pharmacol. 47, 21–2435. Feng, X., Peng, Z. H., Di, W., Li, X. Y., Rochette-Egly, C., Chambon, P.,

Voorhees, J. J., and Xiao, J. H. (1997) Genes Dev. 11, 59–7136. Katagiri, Y., Takeda, K., Yu, Z. X., Ferrans, J., Ozato, K., and Guroff, G. (2000)

Nat. Cell Biol. 2, 435–44037. Hopper, A. K. (1999) Curr. Biol. 9, 803–80638. Guiochon-Mantel, A., Loosfelt, H., Lescop, P., Christin-Maitre, S., Perrot-

Applanat, M., and Milgrom, E. (1992) J. Steroid Biochem. Mol. Biol. 41,209–215

39. Li, C., Schwabe, J. W., Banayo, E., and Evans, R. M. (1997) Proc. Natl. Acad.Sci. U. S. A. 94, 2278–2283

40. Holley, S. J., and Yamamoto, K. R. (1995) Mol. Biol. Cell 6, 1833–184241. Schulman, I. G., Li, C., Schwabe, J. W., and Evans, R. M. (1997) Genes Dev. 11,

299–30842. Castillo, A. I., Jimenez-Lara, A. M., Tolon, R. M., and Aranda, A. (1999) Mol.

Endocrinol. 13, 1141–115443. Dabrowski, M., Robinson, E., Hughes, S. V., Bland, R., and Hewison, M. (1998)

Mol. Cell. Endocrinol. 142, 131–13944. Whitfield, G. K., Hsieh, J. C., Nakajima, S., MacDonald, P. N., Thompson,

P. D., Jurutka, P. W., Haussler, C. A., and Haussler, M. R. (1995) Mol.Endocrinol. 9, 1166–1179

45. Kahlen, J. P., and Carlberg, C. (1997) Nucleic Acids Res. 25, 4307–431346. Barsony, J., Carroll, J., McKoy, W., Renyi, I., Gould, D. L., Htun, H., and

Hager, G. L. (1997) Microsc. Microanal. 3, Suppl. 2, 131–13247. Thompson, P. D., Jurutka, P. W., Haussler, C. A., Whitfield, G. K., and

Haussler, M. R. (1998) J. Biol. Chem. 273, 8483–849148. Lemon, B. D., and Freedman, L. P. (1996) Mol. Cell. Biol. 16, 1006–101649. Barsony, J., Lin, G. C., Kim, D. Y., Pacheco, J. M., and Racz, A. (1999) 81st

Annual Meeting of the Endocrine Society, June 12–15, San Diego, CA,p. 415, Endocrine Society Press, Bethesda, MD

RXR Effect on VDR Distribution 41123

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from

Kirsten Prüfer, Attila Racz, Grace C. Lin and Julia BarsonySubnuclear Targeting of Vitamin D Receptors

Dimerization with Retinoid X Receptors Promotes Nuclear Localization and

doi: 10.1074/jbc.M003791200 originally published online September 22, 20002000, 275:41114-41123.J. Biol. Chem. 

  10.1074/jbc.M003791200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/275/52/41114.full.html#ref-list-1

This article cites 48 references, 19 of which can be accessed free at

by guest on March 20, 2019

http://ww

w.jbc.org/

Dow

nloaded from