Altered Localization of Retinoid X Receptor {alpha} Coincides with Loss of Retinoid Responsiveness in Human Breast Cancer MDA-MB-231 Cells

To understand the mechanism of retinoid resistance, we studiedthe subcellular localization and function of retinoid receptorsin human breast cancer cell lines. Retinoid X receptor ${alpha}$ (RXR ${alpha}$ )localized throughout the nucleoplasm in retinoid-sensitive normalhuman mammary epithelial cells and in retinoid-responsive breastcancer cell line (MCF-7), whereas it was found in the splicingfactor compartment (SFC) of the retinoid-resistant MDA-MB-231breast cancer cell line and in human breast carcinoma tissue.In MDA-MB-231 cells, RXR ${alpha}$ was not associated with active transcriptionsite in the presence of ligand. Similarly, ligand-dependentRXR homo- or heterodimer-mediated transactivation on RXR responseelement or RARE showed minimal response to ligand in MDA-MB-231cells. Infecting MDA-MB-231 cells with adenoviral RXR ${alpha}$ inducednucleoplasmic overexpression of RXR ${alpha}$ and resulted in apoptosisupon treatment with an RXR ligand. This suggests that nucleoplasmicRXR ${alpha}$ restores retinoid sensitivity. Epitope-tagged RXR ${alpha}$ and aC-terminus deletion mutant failed to localize to the SFC. Moreover,RXR ${alpha}$ localization to the SFC was inhibited with RXR ${alpha}$ C-terminuspeptide. This peptide also induced ligand-dependent transactivationon RXRE. Therefore, the RXR ${alpha}$ C terminus may play a role in theintranuclear localization of RXR ${alpha}$ . Our results provide evidencethat altered localization of RXR ${alpha}$ to the SFC may be an importantfactor for the loss of retinoid responsiveness in MDA-MB-231breast cancer cells.

INTRODUCTION

Top

Introduction

Retinoids are natural and synthetic vitamin A derivatives whichregulate development (36), cell proliferation (24), and differentiation(7). Retinoids also act as cancer preventive agents and arepresently being used successfully to treat certain types ofcancer (6, 45). Although many studies have shown retinoid effectivenesson inhibition of cancer cell growth in vitro and in vivo (18),the clinical usage of vitamin A and its derivatives is currentlylimited by the requirement of a large dosage to reach therapeuticefficacy. The combination of synthetic retinoid and tamoxifeninhibited the growth of estrogen-positive breast cancers inpremenopausal patients; however, it failed to show any significanteffect on advanced breast cancer patients (2, 3, 30). It islikely that the responsiveness of cancer cells to retinoid diminishes,along with malignant progression. Indeed, growth inhibitoryeffects of retinoids have been observed in estrogen receptor(ER)-positive breast cancer cell lines such as MCF-7 and T-47D(19), whereas the effectiveness of retinoid diminishes in highlymalignant ER-negative breast cancer cell lines such as MDA-MB-231and BT-20 (5, 13, 14, 35, 53). The existing hormonal and chemotherapeutictherapies have provided significant improvement for the survivalof patients with localized breast cancer; however, treatmentfor metastatic breast cancer still remains palliative (31).The 5-year survival ratio for patients diagnosed with metastaticbreast cancer is only 15%. Thus, there is an urgent need tounderstand the mechanism of retinoid resistance in order todevelop therapeutic agents for metastatic breast cancer.

The physiological actions of retinoids are mediated throughtwo distinct nuclear receptor families (12, 26): the retinoicacid receptors (RAR ${alpha}$ , RARß, and RAR ${gamma}$ ), each of whichbinds both all-trans-retinoic acid or 9-cis-retinoic acid (9-cis-RA),and the retinoid X receptors (RXR ${alpha}$ , RXRß, and RXR ${gamma}$ ),which preferentially bind 9-cis-retinoic acid. RARs and RXRsbind to a specific DNA response element (RARE and RXRE, respectively)in the 5'-flanking region of target genes as homodimers or heterodimers,thereby promoting gene transcription (25). Since RAR and RXRplay a central role in mediating retinoid action in the physiologyof normal cells, it is likely that they are also involved inanticancer effects of retinoids.

Retinoid receptors are members of the steroid-thyroid-vitaminD receptor superfamily. This superfamily can be subdivided intotwo classes according to the partitioning of the unligandedreceptor in the cells. The first group of receptors, the glucocorticoidreceptor (GR), the mineral corticoid receptor, the ER, the progesteronereceptor, and the androgen receptor (AR) localize predominantlywithin the cytoplasm in the absence of ligand (34). Receptorsin this group bind to hsp90 upon ligand binding and translocateto the nucleus (34). In contrast to these receptors, the thyroidand retinoid receptors tightly associate with the nucleus independentof the presence of the ligand (16, 34). The liganded nuclearreceptor proteins, AR, GR, ER, and aryl-hydrocarbon receptorhave been shown to localize in 250 to 400 microfoci (11, 38,49). They recruit transcription cofactors such as SRC-1, TIF-II,and p300/CBP into microfoci, and active transcription occurs(38). With respect to spatial distribution, the mechanism bywhich a specific transcription factor recruits its dimerizationpartner and accesses the active transcription site is not clearlyunderstood yet.

In the present study, we focused on subcellular localizationof the retinoid receptors to understand the mechanism wherebybreast cancer cells become refractory to retinoids.

MATERIALS AND METHODS

Top

Materials and Methods

Abbreviations. In addition to the abbreviations introduced above, the followingabbreviations are used here: PPAR, peroxisome proliferator-activatedreceptor; VDR, vitamin D receptor; SFC, splicing factor compartment;PML, promyelocytic leukemia; NPC, nuclear pore complex; TSA,trichostatin A; TTNPB, (E)-4-[2-(5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl)-1-propenyl]benzoicacid.

Reagents. 9-cis-RA, actinomycin D, TSA, and Hoechst 33342 were obtainedfrom Sigma (St. Louis, Mo.). TTNPB was purchased from BIOMOL(Plymouth Meeting, Pa.).

Cell culture. Human mammary epithelial cells (HMEC) were purchased from Cambrex(Rockland, Minn.) and maintained in mammary epithelial growthmedium. MCF-7 and MDA-MB-231 breast cancer cell lines were purchasedfrom the American Type Culture Collection (Rockville, Md.) andmaintained in Dulbecco modified Eagle medium (DMEM) containing10% fetal bovine serum in a humidified atmosphere supplementedwith 5% CO₂ at 37°C. The subconfluent cells were treatedwith trypsin and seeded onto two-well glass chamber slides.On the following day, the cells were subjected to analysis byimmunofluorescence.

Immunostaining and microscopy. Cells were rinsed twice with phosphate-buffered saline (PBS)and fixed in 4% paraformaldehyde for 15 min, followed by twowashes in PBS and permeabilization with 0.2% Triton X-100. Aftera brief wash with PBS, the slides were blocked with 10% goatserum for 1 h at room temperature and then incubated overnightat 4°C or for 1 h at room temperature with the followingprimary antibodies diluted in goat serum: rabbit anti-RXR ${alpha}$ (1:200;Santa Cruz Biotechnology, Santa Cruz, Calif.), mouse anti-PML(1:100; Santa Cruz Biotechnology), mouse anti-NPC (1:1,000;BabCO, Richmond, Calif.), mouse anti-SC-35 (1:2,000; Sigma),mouse anti-p150 (1:10; ICN, Cleveland, Ohio), goat anti-adenovirustype 5 (1:50; Chemicon, Temecula, Calif.), mouse anti-bromodeoxyuridine(anti-BrdU; 1:50; Roche, Indianapolis, Ind.), mouse anti-hemagglutinin(anti-HA; 1:25; Roche), and fluorescein isothiocyanate (FITC)-coupledmouse anti-Flag (1:500; Invitrogen, Carlsbad, Calif.). The sameamount of affinity-purified normal immunoglobulin G (IgG) fromthe corresponding species was used as a negative control. Slideswere washed with PBS extensively and incubated with goat anti-rabbitFITC and/or goat anti-mouse Texas red, donkey anti-goat TRITC(tetramethyl rhodamine isothiocyanate; 1:300; Jackson Laboratories,West Grove, Pa.) in 0.5% IgG-free bovine serum albumin for 1h at room temperature in the dark. The slides were then washedwith PBS and rinsed with deionized water. Nuclear counterstainingwas performed with Hoechst 33342 (Sigma) for 5 min at room temperature.Paraffin-embedded blocks of human breast tissue from the CooperativeHuman Tissue Network of the University of Pennsylvania MedicalCenter, Philadelphia, were sliced to obtain 6-µm-thicksections for immunohistochemistry of RXR ${alpha}$ according to the methodof Lawrence et al. (22). Antigen was retrieved in 10 mM citratebuffer (pH 6.0) in a microwave for 15 min past boiling. Rabbitanti-RXR ${alpha}$ and mouse anti-SC-35 antibodies were applied in a 1:50and 1:1,000 dilution, respectively. Each section was stainedwith hematoxylin and eosin for pathological diagnosis. Confocalimages were captured by using a Zeiss confocal microscope andprocessed by using Zeiss LMS 5 Image. Series of images wereobtained under the same conditions when fluorescence intensityneeded to be compared.

Western blotting. The cell pellets were lysed in Laemmli buffer and centrifuged,and the lysates were loaded onto 4 to 12% bis-Tris polyacrylamidegels and transferred to a polyvinylidene difluoride membrane.The membrane was blocked with 5% blocking milk, probed withrabbit anti-RXR ${alpha}$ , RAR ${alpha}$ , p21, and mouse anti-hnRNP K (Santa CruzBiotechnology), bcl-2, and Bax (Calbiochem, La Jolla, Calif.).The blots were washed with PBS-Tween, incubated with horseradishperoxidase-labeled anti-rabbit or anti-mouse antibody, and visualizedby chemiluminescence (Pierce, Rockford, Ill.).

Plasmid construct and transfection. RXR ${alpha}$ insert was excised from pCMX-hRXR ${alpha}$ and then subcloned intophrGFP-N1 (Stratagene) and pcDNA 3.1 (N-terminus Flag tag).C-terminus GFP-tagged RXR ${alpha}$ was generated by PCR with sense primer(5'-GGATCCATGGACACCAAACATTTC-3'), antisense primer (5'-CTCGAGAGTCATTTGGTGCGGCGCCTC-3')ligated into phrGFP-C (Stratagene, La Jolla, Calif.), and C-terminusHA tag pMH (Roche). The full-length DNA sequence of RXR ${alpha}$ wasconfirmed by using a dye terminator reaction (ABI Prism, FosterCity, Calif.). Then, 10⁶ of trypsin-treated MCF-7 and MDA-MB-231cells were mixed with 5 µg of supercoiled plasmid DNA,followed by electroporation at 180 V for ca. 70 ms in a 4-mmcuvette by using ECM 830 (BTX, Holliston, Mass.). The electroporatedcells were seeded in a two-well slide and cultured for 24 to48 h. Live cell image was taken 24 h after transfection. NontaggedRXR ${alpha}$ , Flag-tagged RXR ${alpha}$ , and HA-tagged RXR ${alpha}$ transfected cells werefixed and used for immunofluorescence with corresponding antibodyas described above.

Labeling of transcription site. The cells were grown in two-well slides for 2 days. The cellswere placed on ice and washed twice with ice-cold glycerol buffer(20 mM Tris-HCl [pH 7.4], 25% glycerol, 5 mM MgCl₂, 0.5 mM EGTA,0.5 mM phenylmethylsulfonyl fluoride [PMSF]), followed by permeabilizationwith 0.05% Triton X-100 for 10 min on ice. The cells were washedwith glycerol buffer to remove excess detergent and then incubatedfor 10 min at 37°C with transcription buffer (50 mM Tris-HCl[pH 7.4], 100 mM KCl, 25% glycerol, 5 mM MgCl₂, 1 mM PMSF, 25U of RNase inhibitor/ml) supplemented with 2 mM ATP, 0.5 mMCTP and GTP, and 0.2 mM BrUTP. After incorporation, the cellswere fixed with 4% paraformaldehyde in PBS for 20 min and evaluatedby immunofluorescence with mouse anti-BrdU and rabbit anti-RXR ${alpha}$ antibodies.

Nuclear digestion. Cells were digested, with a modification of the method describedby Nickerson et al. (32). The cells were washed three timesin cytoskeleton buffer {10 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonicacid); pH 6.8], 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mMEDTA, 0.05 mM PMSF, 10 µg of aprotinin/ml} and incubatedwith the cytoskeleton buffer containing 0.5% Triton X-100 for10 min at 4°C. The cells were rinsed twice with RSB buffer(42.5 mM Tris [pH 8.3], 8.5 mM NaCl, 2.6 mM MgCl₂, 0.05 mM PMSF,10 µg of aprotinin/ml) and then incubated with RBS buffercontaining 1% Tween 20 and 0.3% of deoxycholate for 10 min at4°C. Chromatin was removed by treatment with 100 U of RNase-freeDNase I/ml in digestion buffer (10 mM PIPES [pH 6.8], 50 mMNaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mM EDTA, 0.05 mM PMSF, 10µg of aprotinin/ml) for 1 h at 37°C. To complete digestion,ammonium sulfate was added to a final concentration of 0.25M, and the mixture was incubated for 5 min at 4°C, followedby 2 M NaCl for 5 min. For the removal of RNA, cells were incubatedin digestion buffer containing the indicated concentrationsof RNase A for 30 min at 37°C. The cells were washed extensivelywith ice-cold PBS, fixed with 4% paraformaldehyde for 15 min,and then used for immunofluorescence or collected and lysedin Laemmli buffer for Western blotting.

Reporter assay. RXRE-luciferase, RARE-thymidine kinase (TK)-luciferase, or TK-luciferasereporter plasmid was transfected into MCF-7 and MDA-MB-231 cellsfor 5 h. Cells were treated with 10^–6 M RXR-selectiveagonist for 20 h, lysed, spun down briefly to remove debris,and analyzed for luciferase activity according to the manufacturer'sinstructions (Promega, Madison, Wis.). The light intensity wasmeasured by a luminometer Tropix TR717 (Perkin-Elmer, Boston,Mass.). The transfection efficiency was normalized to TK. Thedata shown are the means of at least six separate experiments.

Peptide delivery. Peptide delivery was performed according to the manufacturer'sinstructions (GTS, San Diego, Calif.). A total of 1 µgof (i) C-terminus peptide from RXR ${alpha}$ (441-IDTFLMEMLEAPHQMT-456),(ii) N-terminus peptide from RXR ${alpha}$ -N (1-MDTKHFLPLDFSTQVNSSLT-20),or (iii) the control peptide, the C-terminus peptide of RAR ${gamma}$ (441-VPGGQGKGGLKSPA-454), which has no homology to the C-terminalsequence of RXR ${alpha}$ , was delivered to the cells in the presenceof the delivery agent. The cells were further incubated for24 to 48 h after peptide delivery. For simultaneous plasmidtransfection and peptide delivery, the cells were cultured inantibiotic-free DMEM supplemented with 10% serum for 24 h. Onthe following day, the cells were preincubated with 1 µgof RXR ${alpha}$ -C-terminus, RXR ${alpha}$ -N, or RAR ${gamma}$ -C-terminus peptide for 5 min,followed by plasmid transfection for 3 h. The cells were recoveredovernight and incubated with ligand for 24 h, and the reporteractivity was measured.

TUNEL assay. Flow cytometry TUNEL (terminal deoxynucleotidyltransferase-mediateddUTP-biotin nick end labeling) assay was performed accordingto the manufacturer's instructions (Pharmingen, San Diego, Calif.).Stained cells were analyzed by flow cytometry using the FACSCalibur(BD Biosciences, San Jose, Calif.).

Recombinant adenovirus construction and propagation. The complete coding sequence of wild-type RXR ${alpha}$ cDNA was subclonedinto shuttle vector containing cytomegalovirus promoter andsimian virus 40 poly(A). To obtain recombinant plasmid, pAdEasy-1and shuttle vector (Stratagene) were mixed and transformed.The recombinant plasmid was transfected to HEK-293 cells togenerate adenovirus. Adenovirus was purified, and the titerwas measured. A day after seeding, MDA-MB-231 cells were infectedwith adenovirus at a multiplicity of infection (MOI) of 100for 24 h at 37°C in 10% fetal bovine serum containing DMEM.

RESULTS

Top

Results

Altered localization of RXR ${alpha}$ coincides with lack of responsiveness to retinoid. Retinoic acid inhibited growth of HMEC and MCF-7 human breastcancer cells (not shown). Immunofluorescence confocal microscopyrevealed that RAR ${alpha}$ localized throughout the nucleoplasm diffuselyand in microspeckles in HMEC, MCF-7 cells, and MDA-MB-231 cells(Fig. 1, top panels). In sharp contrast to RXR ${alpha}$ homogeneous nucleardistribution in retinoid-sensitive HMEC and MCF-7 cells, RXR ${alpha}$ showed a distinct punctate pattern in the retinoid-resistantMDA-MB-231 cells (Fig. 1, bottom panels). Control experimentswith the same amount of normal rabbit IgG or with blocking peptidesverified that the RXR ${alpha}$ and RAR ${alpha}$ signal was specific (data notshown).

View larger version (76K):
[in this window]
[in a new window]
FIG. 1. Altered localization of RXR ${alpha}$ coincides with lack of retinoid responsiveness. Cells were grown in two-well chamber slides. The slides were fixed, and then subcellular localization of RAR ${alpha}$ and RXR ${alpha}$ was detected by immunofluorescence. Bar, 10 µm. The fluorescence intensity does not correspond to the level of protein expression.

Identification of subnuclear localization of RXR ${alpha}$ in MDA-MB-231 cells. Confocal microscopy was used to pinpoint subnuclear localizationof RXR ${alpha}$ in the MDA-MB-231 cells (Fig. 2A). The subnuclear structureswere visualized by using Texas red-labeled secondary antibodyto different nuclear protein antibodies (Fig. 2A, panels a tod). RXR ${alpha}$ was visualized by using FITC-labeled secondary antibody(Fig. 2A, panels e to h). RXR ${alpha}$ did not colocalize with NPC (Fig.2A, panel i) or PML bodies (Fig. 2A, panel j). However, extensivecolocalization was observed between RXR ${alpha}$ and SC-35 (Fig. 2A,panel k), a splicing factor known to localize to SFC. RXR ${alpha}$ wasalso found to colocalize with p105 (Fig. 2A, panel l), a componentof interchromatin granule cluster, which is a substructure ofthe SFC.

View larger version (37K):
[in this window]
[in a new window]
FIG. 2. Colocalization of RXR ${alpha}$ with nuclear organelle proteins. (A) Fixed MDA-MB-231 cells were double immunostained with RXR ${alpha}$ (e to h) and the indicated nuclear protein antibody anti-NPC (a), anti-PML (b), anti-SC-35 (c), or anti-p105 (d). Confocal overlays of double-immunostained images are shown in merged panels (i to l). (B) HMEC, MCF-7, and MDA-MB-231 cells were double immunostained with anti-RXR ${alpha}$ and anti-SC-35. (C) RXR ${alpha}$ antisense adenovirus was infected to MDA-MB-231 and double immunostained with anti-RXR ${alpha}$ (green) and anti-adenovirus type 5 (red). (D) The dimerization partner (RXR ${gamma}$ , PPAR ${alpha}$ , PPAR ${gamma}$ , or RAR ${alpha}$ ) did not colocalize with SC-35. All antibodies were diluted at 1:200. (E) Paraffin-embedded human breast tissue sections were deparaffinized and hydrated. Antigen was retrieved in 10 mM citrate buffer in a microwave for 15 min past boiling.

Double immunostaining demonstrated that RAR ${alpha}$ microspeckles colocalizedwith PML bodies but not with SC-35 and p105 (data not shown).Figure 2B shows that RXR ${alpha}$ localization to the SFC was found onlyin retinoid-resistant MDA-MB-231 cells and not in retinoid-sensitiveHMEC and MCF-7 cells. MDA-MB-231 cells infected with RXR ${alpha}$ antisenseadenovirus were immunostained for RXR ${alpha}$ (green) and adenovirus(red) (Fig. 2C). The RXR ${alpha}$ was found in the SFC in mock- and null-infectedcells but absent from RXR ${alpha}$ antisense-infected MDA-MB-231 cells(Fig. 2C), indicating that RXR ${alpha}$ localization to the SFC was specific.RXR dimerization partners, RAR ${alpha}$ , RXR ${gamma}$ , and PPARß andPPAR ${gamma}$ did not localize to the SFC, suggesting that they do notform dimers with RXR ${alpha}$ in the SFC (Fig. 2D). These staining patternswere commonly observed in all cells tested. To confirm our invitro results, colocalization between RXR ${alpha}$ and SC-35 was testedin human breast tissue. We tested tissue sections and adjacentnormal region from 12 patients with invasive breast cancer.RXR ${alpha}$ was found in the SFC of mesenchymal cells of 5 of 12 invasivebreast carcinomas (Fig. 2E). However, SFC localization of RXR ${alpha}$ was not detected in any of the normal tissue or in benign hyperplasiasof the three breast tissues tested (data not shown). Hematoxylinand eosin staining showed a representative area.

RXR ${alpha}$ was silenced due to SFC localization. To examine whether the localization of RXR ${alpha}$ to the SFC dependson the transcription status of MDA-MB-231 cells, the cells weretreated with actinomycin D or transcription activator. Someof the SFC became bigger when transcription was inhibited inthe presence of actinomycin D (Fig. 3A). In order to mimic activetranscription, cells were treated with TSA, 9-cis-RA, the naturalligand of RXR ${alpha}$ , or AGN194204, a pan-agonist of RXR. Against ourexpectation, RXR ${alpha}$ remained in the SFC in the presence of excessamounts of TSA, 9-cis-RA, and AGN104204 (Fig. 3A). These resultsdemonstrate that the ligand did not induce relocalization ofRXR ${alpha}$ from the SFC to the nucleoplasm, where the active transcriptionalmachinery is present. In order to determine whether RXR ${alpha}$ associateswith nucleic acid in the SFC, the nuclei were digested withDNase I or RNase. Interactions between chromatin, RNA, and theirbinding proteins are generally disrupted during this preparation.The speckled pattern of RXR ${alpha}$ remained intact after DNA and RNAremoval, suggesting that RXR ${alpha}$ does not interact with DNA or RNAin MDA-MB-231 cells (Fig. 3B, panel f). Western blotting confirmedthat RXR ${alpha}$ was resistant to DNase and RNase digestion. In contrast,RAR ${alpha}$ was sensitive to DNase digestion, suggesting that RAR ${alpha}$ isphysically associated with DNA (Fig. 3B). Concentrations ofRNase up to 1,000 µg had no effect or slightly reducedthe RXR ${alpha}$ speckled pattern (Fig. 3B, panels k to n). Correspondingsamples were analyzed by Western blotting and showed that RXR ${alpha}$ was resistant to RNase digestion, whereas hnRNP, which localizesto the SFC and tightly interacts with RNA, was removed efficientlyby RNase (Fig. 3B). In contrast to MDA-MB-231 cells, RXR ${alpha}$ wascompletely removed by DNase in MCF-7 cells, and no speckledpattern was revealed (data not shown), suggesting that RXR ${alpha}$ interactswith DNA in MCF-7 cells. To examine whether RXR ${alpha}$ participatesin transcription, nascent transcripts were labeled with BrUTPand double immunostained with RXR ${alpha}$ . RXR ${alpha}$ did not colocalize withactive transcription sites in MDA-MB-231 cells, but it colocalizedextensively in MCF-7 cells (Fig. 3C). These results demonstratethat RXR ${alpha}$ is actively participating in transcription in MCF-7cells but not in MDA-MB-231 cells. These staining patterns werecommonly observed in all cells tested.

View larger version (37K):
[in this window]
[in a new window]
FIG. 3. RXR ${alpha}$ is silenced due to SFC localization. (A) The cells were treated with vehicle alone (a), 0.5 µg of actinomycin D/ml (b), 100 nM TSA (c), 1 µM 9-cis-RA (d), or 1 µM RXR agonist (e) for 24 h. (B) MDA-MB-231 cells were permeabilized and then digested with DNase, followed by treatment with 2 M NaCl and 0.25 M (NH₄)₂SO₄ extraction. The cells were also digested with different concentrations of RNase A for 30 min at 37°C. Digested cells were fixed and immunostained with RXR ${alpha}$ (d to f and k to n), followed by counterstaining with Hoechst 33342 (a to c and g to j). The series of images were taken by confocal microscope under the same condition setting. For Western blotting, the cells were either untreated (Un) or treated with DNase (DN), RNase (RN), and different concentrations of RNase (0 to 1,000 µg/ml) as indicated. The cells were lysed in the Laemmli buffer after digestion. Then, 10 µl of whole-cell lysate was separated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis and probed with the indicated antibodies (1:1,000). (C) MDA-MB-231 and MCF-7 cells were treated with 0.2 µM 9-cis-RA for 24 h. Active transcription sites were labeled with BrUTP, followed by double immunostaining with mouse anti-BrdU and rabbit anti-RXR ${alpha}$ . (D) RXRE- and RARE-Luc was transfected into MCF-7 and MDA-MB-231 cells (a), and RXRE-Luc was cotransfected with RXR ${alpha}$ expression vector or empty vector (b). Cells were treated with vehicle alone or 1 µM RXR ligand for 20 h, and luciferase activity was assessed. The data are presented as means ± the standard deviations of six replicates from two independent experiments.

To test whether RXR ${alpha}$ is silent in MDA-MB-231 cells, RXRE (RXRhomodimer target) and RARE (the RAR/RXR heterodimer target)linked to the luciferase reporter gene were transfected intoMCF-7 and MDA-MB-231 cells. As expected, RXR selective ligandhad minimal effect on transcription via the RXRE in MDA-MB-231,whereas it activated transcription in MCF-7 cells (Fig. 3Da).MDA-MB-231 cells transfected with RARE also barely respondedto a RAR selective ligand (TTNPB). These results support thatRXR ${alpha}$ -mediated transcription is significantly reduced due to sequestrationof RXR ${alpha}$ in the SFC. Cotransfection of RXRE and RXR ${alpha}$ expressionvector restored the response to RXR ligand in the MDA-MB-231cells and caused a 10-fold increase in transactivation. Thesedata show that RXR ${alpha}$ is the limiting factor for MDA-MB-231 cellsin mediating transactivation via RXRE.

Subcellular localization of exogenous RXR ${alpha}$ . To further characterize RXR ${alpha}$ localization to the SFC, we comparedthe localization of exogenous RXR ${alpha}$ to that of endogenous RXR ${alpha}$ .GFP was fused to the C or N terminus of RXR ${alpha}$ and electroporatedto MCF-7 and MDA-MB-231 cells. Figure 4A shows the live cellimage of GFP-fused RXR ${alpha}$ . GFP fused either to the C or N terminusof RXR ${alpha}$ showed a diffuse nucleoplasmic localization, excludingnucleoli, and a number of bright microfoci in the nucleus ofboth MCF-7 and MDA-MB-231 cells (Fig. 4A). GFP-RXR ${alpha}$ accumulatedin these microfoci when the cells were treated with ligand for6 h (Fig. 4B, panel b). Next, we examined whether GFP-fusedRXR ${alpha}$ colocalizes with SC-35 in the SFC of MDA-MB-231 cells. Thecells were permeabilized and immunostained with anti-SC-35 antibody.Figure 4C shows that neither the C- nor N-terminus GFP-fusedRXR ${alpha}$ expression pattern overlaps with anti-SC-35 (Fig. 4C, panels cand i). Next, the cells expressing GFP-fused RXR ${alpha}$ were treatedwith actinomycin D to take advantage of its characteristicsthat facilitate SFC localization in the absence of transcription.Actinomycin D treatment did not cause relocalization of GFP-fusedRXR ${alpha}$ from the nucleoplasm to the SFC (Fig. 4C, panels f and l),whereas SC-35 accumulated in the SFC (Fig. 4B, panels e andk). Further, cell nuclei from GFP-RXR ${alpha}$ -overexpressing cells weredigested by DNase or RNase to remove nucleoplasmic GFP-RXR ${alpha}$ signal.GFP-RXR ${alpha}$ was not found in the SFC. Interestingly, GFP-RXR ${alpha}$ microfocidisappeared when transcription was inhibited by actinomycinD and diffused throughout the nucleus, including the nucleolus(Fig. 4C, panels d and j). These staining patterns were commonto all transfected cells tested.

View larger version (43K):
[in this window]
[in a new window]
FIG. 4. Subcellular localization of GFP-RXR ${alpha}$ . (A) GFP-RXR ${alpha}$ , RXR ${alpha}$ -GFP, or GFP vector alone was electroporated to MCF-7 (a and b) and MDA-MB-231 (c and d) cells. At 24 h after electroporation, live cell images were obtained by confocal microscopy. The images showed RXR ${alpha}$ -GFP (a and c) and GFP-RXR ${alpha}$ (b and d). The arrowhead points to nuclear microfoci. Bar, 5 µm. (B) GFP-RXR ${alpha}$ was electroporated to MDA-MB-231, incubated with dimethyl sulfoxide (a) or a mixture of 0.5 µM TTNPB and 0.5 µM RXR-selective ligand (b) for 6 h. The arrow points to nuclear microfoci. (C) GFP-RXR ${alpha}$ and RXR ${alpha}$ -GFP were introduced into the cell through electroporation, and then the cells were treated with vehicle alone (a to c and g to i) or actinomycin D (d to f and j to l) for 6 h. The cells were fixed and immunostained with anti-SC-35.

Possible role of RXR ${alpha}$ C terminus in SFC localization. To rule out the possibility that the GFP moiety might lead tochanges in secondary structure of RXR ${alpha}$ , smaller epitope tag,Flag (N-terminal 8 amino acid) or HA (C-terminal 9 amino acid)was fused to RXR ${alpha}$ . They were found to localize homogeneouslythroughout the nucleus but not in the SFC (Fig. 5A). Since exogenousepitope-tagged RXR ${alpha}$ did not show the same localization patternas endogenous RXR ${alpha}$ , the open reading frame of RXR ${alpha}$ of MCF-7 andMDA-MB-231 was sequenced and found to be identical to wild-typeRXR ${alpha}$ (data not shown). Thus, genetic alteration of RXR ${alpha}$ is notresponsible for this localization in MDA-MB-231. Next, MDA-MB-231cells were transfected with a nontagged RXR ${alpha}$ . The cell in thesquare (Fig. 5A, right panel) shows endogenous RXR ${alpha}$ , while thecell next to it, showing homogeneous nuclear staining patternin addition to distinct speckles, represents exogenous nontaggedRXR ${alpha}$ overexpression (Fig. 5A). Speckles found in cells overexpressingthe nontagged RXR ${alpha}$ were larger and markedly brighter than thoseof endogenous RXR ${alpha}$ in the SFC. These speckles colocalized withSC-35 (Fig. 5B) in the SFC. However, the staining signals reacheda plateau as we increased the amount of DNA used for transfection(data not shown), implying that SFC have limited capacity tointeract with RXR ${alpha}$ and that excess RXR ${alpha}$ distributes throughoutthe nucleoplasm (Fig. 5B). These results indirectly suggestthat the secondary structure of either N- or C-terminal RXR ${alpha}$ may be critical for SFC localization. To test this hypothesis,the C-terminus-truncated RXR ${alpha}$ ${Delta}$ 416 and full-length RXR ${alpha}$ , which havethe same plasmid backbone, were generated. Interestingly, RXR ${alpha}$ ${Delta}$ 416showed speckles of smaller size and reduced fluorescence comparedto the wild type (Fig. 5C). Approximately 100 transfected cellswere analyzed for quantitative evaluation, and the average specklefluorescence in those cells is summarized in the graph (Fig.5C). The average fluorescence of cells overexpressing full-lengthRXR ${alpha}$ was 119.7 ± 5.6, approximately twofold more thanthat of the endogenous RXR ${alpha}$ , whereas that of RXR ${alpha}$ ${Delta}$ 416 was 68.4± 7. Western blotting showed similar a expression levelin both RXR ${alpha}$ and RXR ${Delta}$ 416 (Fig. 5C). This excludes the possibilitythat the observed reduced fluorescence intensity of RXR ${Delta}$ 412 isdue to lower transfection efficiency.

View larger version (49K):
[in this window]
[in a new window]
FIG. 5. Localization of epitope-tagged RXR ${alpha}$ and nontagged RXRa. (A) Flag-RXR ${alpha}$ (a), RXR ${alpha}$ -HA (b), or nontagged RXR ${alpha}$ (c) was electroporated into MDA-MB-231 cells. The cells were fixed and used for immunofluorescence with corresponding antibodies. The cell in the square represents a nontransfected cell. (B) Nontagged-RXR ${alpha}$ -overexpressing cells were labeled with anti-RXR ${alpha}$ and mouse anti-SC-35 antibody. (C) RXR ${alpha}$ deletion mutant was electroporated to MDA-MB-231 cells, followed by double immunostaining with anti-RXR ${alpha}$ and anti-SC-35. The double line represents the antibody recognition site. About 100 transfected cells were analyzed for quantitative evaluation, and average fluorescence intensities of the speckles in the cells overexpressing the deletion mutant are summarized. To determine how much RXR ${alpha}$ , including endogenous and exogenous, localizes to the speckles, the fluorescence from adjacent areas was subtracted from the fluorescence at the speckles. The average intensity of the speckles in RXR ${alpha}$ -transfected cells minus the average intensity from endogenous RXR ${alpha}$ represents the amount of exogenous RXR ${alpha}$ that localizes to the SFCs. The series of images was taken by confocal microscope under the same condition setting. The values represent the means ± the standard errors. The scheme shows the structure of the RXR ${alpha}$ deletion mutant. The double line represents the antibody recognition site. (D) A total of 1 µg of C-terminus peptide corresponding to RAR ${gamma}$ (a to c) or RXR ${alpha}$ (d to i) was incubated for 12 h with MDA-MB-231 cells. The cells were subjected to immunofluorescence after 24 and 48 h with RXR ${alpha}$ and SC-35 antibodies. The images were obtained at the same setting for the comparison. Western blotting of lysates prepared from MDA-MB-231 cells treated with 1 µg of C-terminus RAR ${gamma}$ , treated with RXR ${alpha}$ C or N peptide, or left untreated was carried out. Antibody to actin was used as a control. RXRE-Luc, RARE-Luc, or TK-Luc was transfected into MDA-MB-231 cells pretreated with C- or N-terminus RXR ${alpha}$ or RAR ${gamma}$ peptide or into MDA-MB-231 cells left untreated. The cells were treated with vehicle alone or 1 µM RXR ligand for 24 h, and the luciferase activity was assessed. The data are presented as means ± the standard deviations of four replicates from two independent experiments.

To establish a role of the C terminus of RXR ${alpha}$ in SFC localization,MDA-MB-231 cells were treated with RXR ${alpha}$ C-terminus-specific peptide.At 24 h, RXR ${alpha}$ was colocalized with SC-35 but was also detectedin the cytoplasm (Fig. 5D, panels d to f). At 48 h of incubationRXR ${alpha}$ distributed throughout the cell and no longer remained inthe SFC, and colocalization between SC-35 and RXR ${alpha}$ disappeared(Fig. 5D, panels g to i). The same amount of control peptidecorresponding to the C-terminus RAR ${gamma}$ did not affect the RXR ${alpha}$ punctatepattern (Fig. 5D, panels a to c). Furthermore, cells incubatedwith RXR ${alpha}$ C-terminus peptide showed a $~$ 2.5-fold increase in ligand-mediatedtransactivation on RXRE, whereas the RXR ${alpha}$ N-terminus peptidedid not show any effect, suggesting that the RXR ${alpha}$ C terminusmay interact directly with the SFC. RAR ${gamma}$ C-terminus peptide didnot affect ligand-dependent transactivation on RXRE (Fig. 5D).RXR ${alpha}$ immunoblotting showed that peptide delivery did not alterthe level of the RXR ${alpha}$ protein (Fig. 5D). Therefore, it appearsthat the F region of RXR ${alpha}$ is involved in intranuclear localization.

RXR ${alpha}$ restored the responsiveness to retinoid in MDA-MB-231 cells. MDA-MB-231 cells were infected with RXR ${alpha}$ adenovirus. With anadenovirus-RXR ${alpha}$ MOI of 100, RXR-selective ligand promoted transactivationvia RXRE (Fig. 6A). RXR ${alpha}$ was overexpressed throughout the nucleoplasmin addition to the SFC (Fig. 6B). MDA-MB-231 cells infectedwith RXR ${alpha}$ adenovirus for 24 h, followed by incubation with RXR-selectiveligand for 48 h, demonstrated a drastic decrease in the numberof viable cells by 40% of mock- and null-infected MDA-MB-231cells (Fig. 6C). Apoptosis was measured by TUNEL assay. AdenovirusRXR ${alpha}$ -infected MDA-MB-231 cells increased TUNEL-positive cellsby >60% in the presence of ligand within 96 h of adenovirusinfection, whereas the cells with mock and adenovirus null infectedin the presence of ligand showed only 5% TUNEL positivity (Fig.6D). These data indicate that nucleoplasmic overexpression ofRXR ${alpha}$ stimulates apoptosis in the MDA-MB-231 cells. Next, to testapoptosis-related protein expression, the cells were harvestedafter 96 h of incubation with ligand or vehicle alone. The inductionof ligand-induced apoptosis was correlated with the upregulationof p21 and downregulation of bcl-2, but Bax expression was unchanged(Fig. 6E).

View larger version (31K):
[in this window]
[in a new window]
FIG. 6. RXR ${alpha}$ restores responsiveness to retinoid in MDA-MB-231 cells. MDA-MB-231 cells were infected with RXR ${alpha}$ adenovirus. (A) RXRE was transfected into MDA-MB-231 cells, followed by adenovirus infection, and then the cells were treated with either vehicle or ligand for 24 h. (B) Immunofluorescence of MDA-MB-231 cells infected with adenovirus-null or adenovirus-RXR ${alpha}$ at an MOI of 100. (C) Adenovirus-infected MDA-MB-231 cells were harvested and fixed in 70% ice-cold ethanol and labeled for 2 h at 37°C for TUNEL assay. (E) Adenovirus-infected MDA-MB-231 cells were harvested, and 30 µg of total cell lysate was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The results for mock infection (Mock, lanes 1 and 2), adenovirus-null (Null, lanes 3 and 4), and adenovirus-RXR ${alpha}$ (RXR ${alpha}$ , lanes 5 and 6) infection in the presence of vehicle alone or ligand are presented.

DISCUSSION

Top

Discussion

Retinoids control cell proliferation and differentiation throughthe action of RARs and RXRs and have chemopreventive effectsin various tumor types, including the skin, prostate, ovary,leukemia, and breast (15, 23, 37, 42, 46). The anticancer effectof retinoids is lost during tumor progression; however, themechanism underlying the loss of retinoid responsiveness isstill poorly understood. In the present study, our data demonstratethat RXR ${alpha}$ localizes to the SFC in the highly malignant ER-negative,retinoid-resistant MDA-MB-231 human breast cancer cells. Furthermore,our histological data indicate that RXR ${alpha}$ localizes to the SFCin human breast carcinoma in vivo. Our data provide functionalevidence that most, if not all, endogenous RXR ${alpha}$ is sequesteredin the SFC and silenced and, consequently, retinoid signalingappears to be shut off in the MDA-MB-231 cells.

Splicing factors localize in 20 to 40 nuclear domains referredto as the SFC in the mammalian cell nucleus (28). SFC is composedof two distinct structures, the interchromatin granule clustersand perichromatin fibrils (9, 27, 43). Splicing factor localizationin the SFC is generally highly dynamic and is influenced bytranscriptional activity, cell cycle, and phosphorylation (44).Vitamin D receptor B1 (VDRB1), a heterodimerization partnerof RXR, is also found in the SFC (47), and it is redistributedthroughout the nucleoplasm upon exposure to its ligand, 1,25-dihydroxyvitaminD₃. We found that, in contrast to the ligand-induced dynamicintranuclear mobility of VDRB1, the ligand failed to redistributeRXR ${alpha}$ from the SFC to the nucleoplasm in MDA-MB-231 cells. Thisfinding allowed us to hypothesize that RXR ${alpha}$ might be sequesteredin the SFC, leading to retinoid unresponsiveness.

We demonstrated that RXR ${alpha}$ was not localized to active transcriptionsites in MDA-MB-231 cells but showed extensive colocalizationwith nascent transcripts in MCF-7 cells. This result was furtherconfirmed by reporter assays when ligand promoted RXRE (RXRhomodimer target) or RARE (RAR/RXR heterodimer target) transactivationin MCF-7 cells but failed to do so in MDA-MB-231 cells. Theabsence of ligand-dependent transcriptional activation in MDA-MB-231cells was not due to the reduction of RXR ${alpha}$ protein expressionlevel because RXR ${alpha}$ protein level in retinoid-sensitive HMEC wasthe same as in MDA-MB-231 cells (data not shown). Thus, ourinterest was then to investigate whether altered localizationof RXR ${alpha}$ could explain the loss of RXR ${alpha}$ activity and retinoid unresponsivenessof MDA-MB-231. When MDA-MB-231 cells were infected with adenoviralRXR ${alpha}$ , exogenous RXR ${alpha}$ localized throughout the nucleus in additionto the SFC. Nucleoplasmic overexpression of RXR ${alpha}$ induced apoptosisin accordance with p21 upregulation and bcl-2 downregulationin the presence of ligand. Taken together, RXR ${alpha}$ nucleoplasmiclocalization appears to be one of the major factors determiningthe retinoid sensitivity in the MDA-MB-231.

Several studies have reported that RXR-selective ligands aremore effective than RAR-selective ligands in particular celltypes (50, 53). In addition, these ligands may be also effectiveagainst cancer cells; for instance, the RXR pan-agonist LGD1069has been reported to cause complete regression of mammary carcinoma(1). These data implicate that there may be more than one malignantphenotype whose retinoid responsiveness is RXR dependent. RXRregulates multiple hormonal pathways through heterodimerizationwith several nuclear receptors, including RAR, PPAR, VDR, LXR,and nur77 (4, 8, 17, 33, 52). In MDA-MB-231, RXR ${alpha}$ was found inthe SFC, and this altered localization was associated with reducedRXR ${alpha}$ -mediated transcription. In this respect, other receptorfamily members, which require heterodimer formation with RXR ${alpha}$ for transcriptional activation, might be silenced in a similarfashion if RXR ${alpha}$ is the favorable isotype as an heterodimerizationpartner. RXR isotypes ${alpha}$ , ß, and ${gamma}$ are expressed in mostbreast cancer cells, and their expression is independent ofretinoid sensitivity (51). MDA-MB-231 cells express all threeRXR isotypes, and RXRß and RXR ${gamma}$ localize to the nucleoplasm(Fig. 2). Reporter assays showed that ligand-mediated transactivationon RXRE and RARE was not completely silenced (Fig. 3D), althoughRXR ${alpha}$ did not colocalize with active transcription sites in thepresence of ligand (Fig. 3C). This incomplete silencing suggeststhat functional RXRß and/or RXR ${gamma}$ participate in transcription.

RAR ${alpha}$ and RARß protein expression is suppressed in MDA-MB-231cells (51); thus, RXR ${alpha}$ sequestration does not appear to be theonly factor in the development of loss of retinoid sensitivity,but rather multiple factors may be involved. In support of thisidea, RAR ${alpha}$ and RARß overexpression restores retinoidsensitivity in MDA-MB-231 cells (40, 41). Wu et al. have shownthat an RXR pan-agonist activates RXR/nur77 heterodimer bindingto ßRARE and that RARß is activated by anRARß-selective ligand and induces apoptosis in MDA-MB-231(53). The difference between Wu et al.'s results and ours canbe explained on the basis that our data specifically refer toRXR ${alpha}$ as being sequestered in the SFC and leave open the possibilitythat RXRß and/or RXR ${gamma}$ may still be available for transcription.To date, differential roles or preferential heterodimerizationpartners for each RXR isotype have not been comprehensivelyunderstood. To further evaluate the role of RXR isotypes, developmentof each RXR-isotype-selective ligand and conditional gene knockoutapproaches are essential.

Altered localization of functional proteins occurs by variousmechanisms, including mutation (21), posttranslational modification,and selection of scaffold proteins (54), and results in themodification of cellular response. The reduction of retinoidresponsiveness is observed due to altered localization of RXRßwhen NGFI-B (Nur77) shuttles RXR to the cytoplasm from the nucleus(54). Kumar et al. (21) reported that mutated MTA1s, which isexpressed at a high level in ER-negative breast cancer cells,sequesters ER ${alpha}$ in the cytoplasm and causes loss of ER signalingin ER-negative breast cancer cells. Similarly, mutant E-cadherinis sequestered in the cell membrane instead of the nucleus,resulting in silencing Wnt signaling pathways (48). AlthoughRXR ${alpha}$ mutation is not responsible for SFC localization (data notshown), RXR ${alpha}$ may have a different posttranslational modificationin highly malignant cancer cells because it might have acquireda different set of interacting proteins that may shuttle RXR ${alpha}$ to the SFC. On the other hand, scaffold or chaperone proteinsthat do not interact with RXR ${alpha}$ in normal cells could be alteredin highly malignant cancer cells and misdirect RXR ${alpha}$ to the SFC.

RAR ${alpha}$ was found in both nucleoplasm and PML bodies, and this localizationpattern was common to all of the cells tested. PML bodies donot share the same intranuclear spatial partitioning with SFC.PML bodies are a cluster of proteins, including PML itself,p53, CBP, and pRb, but do not contain DNA in the structure andare thought to be involved in transcriptional regulation, aswell as posttranslational modification, or compartmentalization(57). In the nucleoplasm, PML acts as coactivator in the RAR/RXRheterodimer complex (56). We found a part of RAR ${alpha}$ localized inthe PML bodies, implying that RAR ${alpha}$ may be temporarily storedin the PML bodies to recruit essential coactivators such asPML into a complex prior to active transcription.

Liganded nuclear receptors fused to GFP (AR, GR, ER, VDR, mineralcorticoid receptor, and aryl hydrocarbon receptor) and theircoactivators have been shown to localize to intranuclear microfociwhere active transcription occurs (11, 38, 49). Here we showeda ligand-stimulated microfocal localization of GFP-RXR ${alpha}$ and thatthis localization was prevented by the transcription inhibitoractinomycin D. This suggests that the microfoci are a commonsite for intranuclear compartmentalization and active transcriptionfor RXR ${alpha}$ , as well as other nuclear receptors and their coactivators.Even though GFP-RXR ${alpha}$ localizes to the microfoci, it was not foundin the SFC in MDA-MB-231 cells. It is possible that the largeGFP moiety causes hindrance of the structural proteins of thescaffold, which interact with RXR ${alpha}$ , or the lack of specific posttranslationalmodification(s) that may be required for SFC localization. Sincethe microfoci and SFC are separate nuclear compartments (49,55), sequestered RXR ${alpha}$ in the SFC in MDA-MB-231 cells is unableto take part in active transcription.

In order to test the commonality of the RXR ${alpha}$ SFC localizationpattern, we tested additional cell lines, including the MDA-MB-453,MDA-MB-435, BT-20, and BT-549 breast cancer cell lines; theSK-OV-3 ovarian cancer cell line; and the I-7, HT-144, S1 cells,Sp1, A2058 keratinocyte, HeLa, and ras NIH 3T3 cell lines. Wecould not find the same RXR ${alpha}$ localization pattern in the SFC,suggesting that RXR ${alpha}$ localization in the SFC in MDA-MB-231 cellsmay represent a phenomenon occurring in a small subset of tissue-specificcell types. Interestingly, however, immunohistochemistry studiesrevealed that RXR ${alpha}$ SFC localization was found in mesenchymalcells of human invasive breast tumor tissue. Generally, thegeneration of a reactive stroma environment occurs in many humancancers and is likely to promote tumorigenesis (39), and molecularalterations in breast stroma during malignant progression havebeen reported (39). The link between alteration in RXR ${alpha}$ localizationin the SFC and breast microenvironment during malignant progressionneeds further investigation.

Signal peptide for targeting the SFC was recently describedand characterized by a region rich in arginine/serine dipeptides(RS domain) or multiple threonine-proline (TP) repeats (10,20). However, many non-SR proteins that are reported to localizeto the SFC do not carry a TR repeat or a RS domain, suggestingthat more complex mechanisms might be involved in this localization(29). Even though RXR ${alpha}$ does not contain the RS domain or theTP repeats, it remained in the SFC after DNase or RNase digestion.In addition to nuclease resistance, the RXR ${alpha}$ C-terminus peptidecaused the release of RXR ${alpha}$ from the SFC to the nucleoplasm, andthus this nucleoplasmic localization increased RXR homodimer-mediatedligand-dependent transactivation (Fig. 5D). This suggests thatthe RXR ${alpha}$ association with SFC may be mediated by a protein-proteininteraction between the RXR ${alpha}$ C terminus and a component of theSFC. The secondary structure or posttranslational modificationin the RXR ${alpha}$ C terminus may be important for the interaction betweenRXR ${alpha}$ and an SFC component and the consequent SFC localization.We also could not rule out that a specific set of molecularchaperones may be involved to guide RXR ${alpha}$ to SFC in malignantcancer cells. We initially hypothesized that both the C terminusand the N terminus are equally important because epitope tagsfused to each terminus interfered with exogenous RXR ${alpha}$ to localizeto the SFC (Fig. 5A). However, RXR ${alpha}$ N-terminus peptide did notrestore ligand-dependent transactivation (Fig. 5D), suggestingthat RXR ${alpha}$ N terminus presumably is not interacting directly withthe SFC. Nevertheless, RXR ${alpha}$ N terminus may be required to coordinatethe correct folding of the whole protein for SFC localization.Retinoid receptors are categorized into class II receptors whichdo not require ligand for their nuclear import (16, 34). Infact RXR ${alpha}$ was found in the nucleus without ligand stimuli. Controversially,RXR ${alpha}$ was found in the nucleoplasm and cytoplasm upon incubationwith C-terminus peptide. C-terminus peptide not only rescuedRXR ${alpha}$ from the SFC but also may block nuclear import of newlysynthesized RXR ${alpha}$ from the cytoplasm to the nucleus by maskingchaperones that might be involved in nuclear import of RXR ${alpha}$ .The mechanistic details underlying the RXR ${alpha}$ localization to variouscellular organelles need further investigation.

In conclusion, RXR ${alpha}$ was found in the splicing factor compartmentof the highly malignant breast cancer MDA-MB-231 cell line andin mesenchymal cells of tissue sections of several invasivecarcinomas. These findings suggest that localization of RXR ${alpha}$ changes during malignant progression since RXR ${alpha}$ is normally homogeneouslydistributed in the nucleoplasm. Our studies define a possiblemechanism, i.e., sequestration of RXR ${alpha}$ , for retinoid unresponsivenessas frequently encountered in breast cancer cells. These observationsoffer new insights into the silencing of molecular mechanismspossibly involved in retinoid signaling pathways.

ACKNOWLEDGMENTS

We thank Ronald M. Evans for generously providing RXR ${alpha}$ expressionvector. We also thank David J. Mangelsdorf for RXRE and RARE-lucreporter plasmid and Keiko Ozato for help with various RXR expressionvectors. We are grateful to Rosh Chandraratna of Allergan forthe RXR pan-agonist.

FOOTNOTES

* Corresponding author. Mailing address: National Institutes of Health, Bldg. 37, Rm. 4054C, 37 Convent Dr., Bethesda, MD 20892-4255. Phone: (301) 496-2698. Fax: (301) 496-8709. E-mail: delucal{at}mail.nih.gov.

REFERENCES

Top