Reduced Intranuclear Mobility of APL Fusion Proteins Accompanies Their Mislocalization and Results in Sequestration and Decreased Mobility of Retinoid X Receptor α

Abstract

Acute promyelocytic leukemia (APL) cells contain one of five chimeric retinoic acid α-receptor (RARα) genes (X-RARα) created by chromosomal translocations or deletion; each generates a fusion protein thought to transcriptionally repress RARα target genes and block myeloid differentiation by an incompletely understood mechanism. To gain spatiotemporal insight into these oncogenic processes, we employed fluorescence microscopy and fluorescence recovery after photobleaching (FRAP). Fluorescence microscopy demonstrated that the intracellular localization of each of the X-RARα proteins was distinct from that of RARα and established which portion(s) of each X-RARα protein—X, RAR, or both—contributed to its altered localization. Using FRAP, we demonstrated that the intranuclear mobility of each X-RARα was reduced compared to that of RARα. In addition, the mobility of each X-RARα was reduced further by ligand addition, in contrast to RARα, which showed no change in mobility when ligand was added. Both the reduced baseline mobility of X-RARα and the ligand-induced slowing of X-RARα could be attributed to the protein interaction domain contained within X. RXRα aberrantly colocalized within each X-RARα; colocalization of RXRα with promyelocytic leukemia (PML)-RARα resulted in reduced mobility of RXRα. Thus, X-RARα may interfere with RARα through its aberrant nuclear dynamics, resulting in spatial and temporal sequestration of RXRα and perhaps other nuclear receptor coregulators critical for myeloid differentiation.

Acute promyelocytic leukemia (APL) accounts for 10% of cases of acute myeloid leukemia. In over 95% of APL cases, there is a chromosomal translocation within leukemic cells involving the promyelocytic leukemia (PML) gene at 15q22 and the retinoic acid α-receptor (RARα) gene at 17q21, resulting in the PML-RARα gene and protein product (7, 35). Four variant translocations or deletions occur in the remaining cases of APL, each involving the RARα locus on chromosome 17 including t(11;17)(q23;q21), t(11;17)(q13; q21), t(5;17)(q35;q21), and del(17) (21, 35). These chromosomal abnormalities produce fusions of RARα with PLZF, NuMA, NPM, and STAT5b respectively (1, 4, 9, 10, 21). It is important to identify APL patients with these alternative chromosomal abnormalities since they may not respond as well as PML-RARα-positive APL patients to treatment regiments with all-trans-retinoic acid (ATRA) and other conventional chemotherapies for APL (17).

Nuclear hormone receptors (NRs) comprise a large family of ligand-dependent transcription factors that bind to hormone response elements of target genes and regulate their transcription (3). Type I nuclear hormone receptors such as estrogen receptor (ER), androgen receptor (AR), and glucocorticoid receptor (GR) bind to their response elements as homodimers, whereas type II nuclear hormone receptors, including RAR, thyroid hormone receptor (TR) and vitamin D receptor (VDR), bind to their response elements as heterodimers with retinoid X receptors (RXRs). Both type I and II receptors can recruit coactivators in the presence of ligand. In addition, type II receptors, notably RAR and TR, can recruit corepressors such as SMRT and NCoR in the absence of ligand and repress the transcription of target genes (6).

RARs belong to the superfamily of NRs, which affect many physiological processes including differentiation and growth arrest of various cell types including hematopoietic cells. RARs can dimerize with RXRs and bind to retinoic acid response elements (RAREs) located within promoter regions of specific target genes. In the absence of ligand, RAR-RXR heterodimers associate with nuclear receptor transcriptional corepressors (CoR), SMRT-NCoR, resulting in repression of basal transcription. Ligands, such as retinoic acid (RA), release the CoR complex this is followed by recruitment of transcriptional coactivators to the transcriptional complex, resulting in the activation of gene expression. We and others demonstrated that X-RARα (where X represents PML, PLZF, NPM, NuMA, or STAT5b) could bind to RAREs as homodimers or heterodimers with RXR. These findings and others (reviewed in reference 17) support the concept that X-RARα proteins interfere with normal myeloid differentiation by inhibiting wild-type RARα transcriptional activity. However, our understanding of this process at the molecular level is incomplete.

Recent investigations examining fluorescence-labeled ERα in live cells have demonstrated interdependence between their intranuclear dynamics and transcription function (27). Some ligand-dependent variation of AR subnuclear targeting has been shown by similar approaches (24, 32), but recent dynamic assessment of AR expressed at physiological levels indicate clear ligand-dependent differences in AR solubility and mobility (D. L. Stenoien, S. Simeoni, K. Patel, M. G. Mancini, I. Agoulnik, N. L. Weigel, and M. A. Mancini, submitted for publication). Moreover, we have demonstrated that an inactive AR point mutant occurring in androgen-unresponsive prostate cancer cells (19) responded in a ligand-dependent fashion by localizing abnormally within nuclear aggregates and sequestering the bulk of steroid receptor coactivator 1 (SRC-1).

To gain spatiotemporal insight into how X-RARα interferes with normal RARα function, we employed fluorescence microscopy and fluorescence recovery after photobleaching (FRAP) of fluorescent protein-tagged RARα, RARα, X-RARα, and each of the X components in fixed and live cells following transient transfection or cotransfection. Our findings demonstrated that intracellular localization of each of the X-RARα proteins was distinct from that of RARα and that this altered localization could be attributed, in each instance, to the truncated X component (ΔX) or truncated RARα component (ΔRAR) or the fusion of the two. The intranuclear mobility of each X-RARα was reduced compared to that of RARα and was reduced further by addition of ligand. In contrast, RARα showed no change in mobility after addition of ligand. The altered localization and reduced mobility of X-RARα could be attributed to the protein interaction domain of X and could result in the mislocalization and slowing of RXRα.

MATERIALS AND METHODS

Cell lines and reagents.

COS-7 cells were cultured in Dulbecco modified Eagle medium (Invitrogen) with 10% (fetal bovine serum) (FBS); HeLa cells were maintained in Opti-MEM I medium (Invitrogen) with 5% dialyzed FBS treated with charcoal-dextran (HyClone Labs, Logan, Utah). ATRA was obtained from Sigma (St. Louis, Mo.).

Plasmids.

The cyan fluorescent protein (CFP)-tagged RARα, NuMA, NuMA-RARα, and NuMA-RARα(ΔCC) expression constructs (in pECFP-C1 [Clontech]) have been described previously (9). The cDNA sequences for truncated RARα (amino acids [aa] 60 to 462), PML, truncated PML (aa 1 to 552), PML-RARα, PLZF, truncated PLZF (aa 1 to 455), PLZF-RARα, NPM, truncated NPM (aa 1 to 160), NPM-RARα, STAT5b, STAT5b-RARα, and RXRα were subcloned from their respective expression constructs (5, 9-11, 25) into pECFP-C1 vector or pEYFP-C1 vector (Clontech) to make CFP-tagged ΔRARα, PML, ΔPML, PML-RARα, PLZF, ΔPLZF, PLZF-RARα, NPM, ΔNPM, NPM-RARα, STAT5b, STAT5b-RARα, and yellow fluorescent protein (YFP)-tagged RXRα expression constructs, respectively. The cDNAs for CFP-RARα, CFP-PML-RARα, CFP-PLZF-RARα, CFP-NPM-RARα, CFP-NuMA-RARα, and CFP-STAT5b-RARα contained within the pECFP-C1 construct were subcloned into the pcDNA3.1 expression construct (Invitrogen) to facilitate the in vitro generation of proteins. The (RARE)3-tk-luciferase reporter construct has been reported previously (9). All plasmid constructs were confirmed by DNA sequencing.

Cell transfections and immunoblotting.

For transient transfections, COS-7 or HeLa cells were grown in six-well plates to 50 to 70% confluence. At 12 h later, the cells were transiently transfected with the indicated expression constructs and/or reporter genes using GeneJuice reagent as reported previously (10). After 24 to 48 h of transfection, the cells were lysed in lysis buffer as previously reported (10). Equivalent amounts of protein were electrophoresed on sodium dodecyl sulfate-7.5 or 10% polyacrylamide gels and transferred to a polyvinylidene difluoride membrane (Millipore). Antibodies used in this study were RARα rabbit polyclonal antibody (C-20; Santa Cruz BioTechnology, Santa Cruz, Calif.) and PLZF mouse monoclonal antibody (Oncogene Research Products, San Diego, Calif.).

In vitro translation and gel shift DNA-binding assays.

CFP-RARα, CFP-PML-RARα, CFP-PLZF-RARα, CFP-NPM-RARα, CFP-NuMA-RARα, and CFP-STAT5b-RARα proteins were generated in vitro using the TNT-coupled rabbit reticulocyte lysate system (Promega) as specified by the manufacturer. Gel shift assays were performed using the in vitro-translated proteins and DR5G RARE as described previously (11).

FRAP.

HeLa cells were cultured in Opti-MEM I medium (Invitrogen) with 4% FBS (30). At 24 h before transfection, the cells were plated onto poly-d-lysine-coated coverslips in 35-mm wells at 10⁵ cells per well for fixation studies or 40-mm coverslips in a 60-mm dish at 2 × 10⁵ cells per dish for FRAP analysis. Transient expression of plasmids (CFP tagged or YFP tagged) was accomplished using GeneJuice transfection reagent (Novegen) as described previously (8, 10). The cells were fed with fresh medium the day after transfection and allowed to recover for approximately 24 h prior to addition of vehicle (ethanol) or ATRA at a final concentration of 10⁻⁶ M for 2 h as indicated. The cells were fixed as described previously (26) and imaged with on LSM 510 confocal microscope (Carl Zeiss, Inc.). FRAP analysis has been described previously (28-30). The fluorescent molecules in FRAP experiments are essentially irreversibly photobleached in a small area of the cell by brief (1- to 2-s) exposure to a focused high-power laser beam. Subsequent diffusion of surrounding nonbleached fluorescent molecules into the bleached area leads to a recovery of fluorescence, which is recorded using time-lapse photography at low laser power and quantified using software available with the microscope; this was carried out with an LSM 510 confocal microscope. A single z-section was imaged before and at intervals after the 2-s bleach. The bleach was carried out at a wavelength of 458 nm (CFP) or 514 nm (YFP) and at maximum power for 100 iterations of a box representing ∼20% of the nuclear volume. For dual FRAP experiments, both were bleached with the same laser setting and simultaneous images corresponding to CFP and YFP fluorescence were obtained using the multitracking function of the microscope. Fluorescence intensities of regions of interest were obtained using LSM software, and data were analyzed using Microsoft Excel. Representative images from single focal planes were imported as TIFF files (29, 30). To allow pooling of data from each cell analyzed, the initial fluorescence at the end of bleaching was assigned a value of 0 and the final fluorescence recovery was assigned a value of 1.

Statistical analysis.

Unless indicated otherwise, data presented are the mean ± standard error of the mean; differences between means were assessed for significance by using Student's t test.

RESULTS

Biochemical characterization of CFP-tagged RAR, X-RARα, and X partners.

To study the intracellular localization and mobility of RARα, X-RARα, and X partners in living cells, we created constructs expressing CFP chimeras fused to the N-terminal ends of RARα and X-RARα, as well as its normal X partners (Fig. 1). Immunoblot analysis of whole-cell extracts from transfected cells and in vitro translation protein analysis showed that each CFP-tagged construct encoded proteins of the correct size (Figure 1a) (reference 9 and data not shown). To be certain that the addition of CFP to RARα or X-RARα did not significantly alter their functional characteristics compared to untagged RARα or X-RARα, we evaluated the DNA-binding activity of CFP-RARα or CFP-X-RARα by a gel shift assay. Similar to their untagged counterparts, each CFP-X-RARα bound to RARE as a homodimer (Fig. 1b) and each CFP-X-RARα, as well as CFP-RARα, bound to RARE as a heterodimer with RXRα (Fig. 1b). To further characterize the functional properties of CFP-RARα or CFP-X-RARα, we examined their transcriptional activity by using the luciferase reporter system as described previously (9-11). CFP-RARα or CFP-X-RARα expression plasmids were cotransfected into HeLa cells with the RARE-containing luciferase reporter construct, and each demonstrated ligand-dependent transactivation to levels equal to or greater than those of their non-CFP-tagged counterparts (Fig. 1c) (reference 9 and data not shown).

FIG. 1.

Biochemical characterization of CFP-RARα and CFP-X-RARα. (a) Immunoblot analysis of CFP-tagged X-RARα proteins expressed in COS-7 cells separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotted with the antibody indicated below each panel. The positions of the molecular mass markers are indicated on the left of each panel. (b) Gel shift assay analysis using the RARE duplex oligonucleotide DR5G and in vitro-translated CFP-tagged X-RARα proteins without or with RXRα as indicated. The black arrowheads indicate the positions of the CFP-X-RARα homodimer plus RARE complexes. (c) Luciferase reporter assay analysis of untagged (top panel) and CFP-tag (bottom panel) RARα and X-RARα proteins in HeLa cells in the absence or presence of ATRA at 10⁻⁶ M. The amounts of plasmid DNA used per well were 250 ng of RARE reporter construct, 500 ng of CFP-untagged or tagged expression construct, and 200 ng of β-galactosidase expression construct as transfection control. Luciferase activity was measured in a luminometer, expressed in arbitrary units, and normalized to the transfection control. Each point is the mean of three independent experiments.

Cellular localization of CFP-tagged APL-related proteins.

We next examined the intracellular distribution of CFP-RARα, CFP-X-RARα, and each of their normal CFP-X components (Fig. 2; Table 1). Similar to previous immunohistological observations for wild-type RARα, X-RARα, and the normal X partners (Fig. 2) (12, 14, 15, 22, 33, 34), CFP-RARα and YFP-RXRα demonstrated a diffuse and predominantly intranuclear pattern, CFP-PML displayed a typical PML oncogenic domain (POD) structure, CFP-PLZF demonstrated an intranuclear microspeckled pattern, CFP-NPM localized to nucleoli, CFP-NuMA was localized almost exclusively to the nucleoplasm in a reticular pattern, and CFP-STAT5b was localized predominantly within the cytoplasm and demonstrated a diffuse pattern within the nucleus.

FIG. 2.

Fluorescence microscopy characterization of CFP-tagged RARα, normal X, X-RARα, and truncated X and YFP-tagged RXR following transient transfection and fixation of HeLa cells. The fluorescence photomicrographs shown were obtained at ×1,000 magnification and are representative of cells expressing the indicated constructs at low levels, as outlined in the text.

TABLE 1.

Summary of the intracellular localization of CFP-RARα, CFP-X-RARα, and CFP-X

Construct	Cytoplasm (C)			Nucleus (N)
	Localized	Diffuse	Aggregated	Localized	Diffuse	Structure
CFP-RARα	±	Yes	No	++	Yes	No
CFP-ΔRARα	±	Yes	No	++	Yes	No
CFP-PML	±	Yes	No	++	No	POD
CFP-ΔPML	±	Yes	No	++	No	POD
CFP-PML-RARα	±	Yes	No	++	No	Microspeckled
CFP-PLZF	+	Yes	No	++	Yes	Microspeckled
CFP-ΔPLZF	++	Yes	No	±	No	No
CFP-PLZF-RARα	+	Yes	No	++	Yes	Microspeckled
CFP-NPM	±	Yes	No	++	No	Nucleoli
CFP-ΔNPM	++	Yes	No	++	No	No
CFP-NPM-RARα	+	Yes	No	++	Yes	Microspeckled
CFP-NuMA	±	Yes	No	++	No	Reticular
CFP-NuMA-RARα	++	Yes	No	+	Yes	Microspeckled
CFP-NuMA-RARα(ΔCC)	+	Yes	No	++	Yes	No
CFP-STAT5b	++	Yes	No	+	Yes	No
CFP-STAT5b-RARα	+	Yes	No	++	Yes	Microspeckled

Each of the CFP-X-RARα proteins distributed within the nucleus in a pattern distinct from that of RARα, and four of the five CFP-X-RARα proteins (where X is PML, PLZF, NPM, and STAT5b) distributed within the nucleus in a pattern distinct from those of their corresponding CFP-X counterparts (Fig. 2; Table 1), confirming previous immunohistological reports (12, 14, 15, 22, 33, 34). CFP-PML-RARα localized predominantly to the nucleus in a microspeckled pattern distinct from that of PODs. CFP-PLZF-RARα demonstrated a microspeckled pattern similar to that of CFP-PLZF. CFP-NPM-RARα distributed predominantly in a diffuse pattern with occasional microspeckles within the nucleoplasm, distinct from that of CFP-NPM, which localized exclusively to nucleoli. In addition, a small portion of CFP-NPM-RARα colocalized with YFP-NPM within nucleoli (see Fig. 5f), distinct from RARα, as detailed below. CFP-STAT5b-RARα distributed predominantly within the nucleus in a microspeckled pattern along with the more diffuse pattern seen with CFP-STAT5b. As recently reported by us (9), while CFP-NuMA-RARα localized predominantly to the cytoplasm in the absence of ATRA and shifted to predominantly nuclear localization in the presence of ATRA, when present in the nucleus it distributed in a reticular pattern similar to that of CFP-NuMA.

FIG. 5.

Fluorescence microscopy of HeLa cells cotransfected with CFP- and YFP-tagged proteins. (a) Colocalization of CFP-ΔPML and YFP-PML within the POD structure, (b to f) Colocalization of CFP-PML-RARα and YFP-RXRα (b), CPF-PLZF-RARα and YFP-RXRα (c), CFP-STAT5b-RARα and YFP-RXRα (d), CFP-NuMA-RARα and YFP-RXRα (e), and CFP-NPM-RARα and YFP-NPM (f) within the nucleolus. (g) Cells were cotransfected with CFP-NPM, CFP-NPM-RARα, and YFP-RXRα. While YFP-RXRα colocalized with CFP-NPM-RARα in the nucleoplasm, colocalization of YFP-RXRα and CFP-NPM-RARα within the nucleolus could not be detected.

To determine the molecular basis for the distinct localizing features of each CFP-X-RARα compared to CFP-RARα or its corresponding CFP-X counterpart, we examined the intracellular distribution of the individual truncated protein components within each fusion protein. When CFP-ΔRARα (aa 60 to 462) and CFP-ΔPML (aa 1 to 552), which are the portions of RARα and PML, respectively, retained in the PML-RARα chimera, were expressed in HeLa cells, each displayed the same cellular localization as its full-length counterpart (Fig. 2; Table 1) i.e., CFP-ΔRARα had a diffuse intranuclear pattern, and CFP-ΔPML had a typical POD structure. When CFP-ΔPML and YFP-PML were cotransfected into HeLa cells, they colocalized within the POD structure (see Fig. 5a), indicating that the RBCC region of PML is sufficient for localization within PODs and that loss of the C-terminal region did not affect POD formation. However, fusion of ΔRARα (aa 60 to 462) to ΔPML (aa 1 to 552) does not result in the POD structure but, rather, leads to the formation of a distinct microspeckled pattern (Fig. 2), indicating that it is the addition of the ΔRARα portion that interferes with POD formation. CFP-ΔPLZF (aa 1 to 455), which is the portion of PLZF retained in PLZF-RARα, localized within the cytoplasm, presumably due to loss of its nuclear localization signal located in the C-terminal end of PLZF and deleted in ΔPLZF. With the fusion of ΔRARα, which contains a nuclear localization signal, to ΔPLZF to generate PLZF-RARα, the ability to localize within the nucleus was regained (Fig. 2). CFP-ΔNPM (aa 1 to 160), which is the portion of NPM retained in NPM-RARα, was localized within the cytoplasm due to deletion of its nuclear localization signal, similar to the findings with ΔPLZF. Also, similar to PLZF-RARα, fusion of ΔRARα to ΔNPM to form NPM-RARα restored its ability to localize to the nucleus. However, unlike NPM, the localization of NPM-RARα within the nucleus is predominantly diffuse and not limited to the nucleolus (Fig. 2). To determine whether NPM-RARα does localize to the nucleolus at all, we cotransfected HeLa cells with YFP-NPM and CFP-NPM-RARα. We found that the CFP-NPM-RARα protein also localized in the nucleolus with wild-type NPM (see Fig. 5f), presumably through heterodimerization mediated by the protein interaction domain within NPM. We previously determined that the coiled-coil domain within the NuMA component of NuMA-RARα was responsible for its colocalization with NuMA in the nucleus (9). C-terminal truncation mutants of STAT5b resembling the ΔSTAT5b component of STAT5b-RARα have been identified (2, 13, 16, 23), and while these truncations may function as dominant negatives of full-length STAT5, their intracellular distribution is similar to that of full-length STAT5b, indicating that the altered localization of STAT5b-RARα to microspeckles is a unique feature of the fusion of ΔSTAT5b and ΔRARα that cannot be attributed to either component alone.

Reduced mobility of X-RARα in comparison with RARα.

To study the intranuclear mobility of RARα, X-RARα, and X, we performed FRAP of live cells transfected with CFP- and YFP-tagged constructs (Fig. 3 and 4, Table 2). In all FRAP experiments, only cells expressing low levels of protein were examined, to avoid artifacts of overexpression as previously described (30). Specifically, we adhered strictly to the procedure of viewing cells at the lower end of detection and excluded overexpressers (>20-fold) from analysis. Furthermore, we demonstrated, using antibodies to endogenous receptors indirectly detected by Alexa 594, that expression in the CFP-positive cells was within a factor of 2 to 3 of the endogenous expression. Following photobleaching of HeLa cells transfected with CFP-RARα, fluorescence recovery occurred rapidly, with a half-maximal fluorescence recovery time (t_1/2) of 0.67 ± 0.15 s. In comparison with wild-type RARα, each of the CFP-X-RARα proteins had reduced mobility in the absence of ligand, ranging from t_1/2 of 0.84 ± 0.11 s for CFP-NPM-RARα to t_1/2 of >5 min for CFP-PML-RARα. The mobility of CFP-ΔRARα was similar to that of CFP-RARα, indicating that loss of the A domain in RARα did not contribute to the slowing of X-RARα. Rather, in each instance except for STAT5b, the reduction in the mobility of X-RARα compared to RARα could be attributed to reduced mobility of X (Table 2).

FIG. 3.

FRAP images of CFP-RARα, CFP-X-RARα, and its normal CFP-X counterpart in the absence of ATRA in live cells. The nuclei of HeLa cells transfected with the CFP-tagged proteins indicated were subjected to FRAP analysis. Images show single z-sections obtained before photobleaching (Pre-bleach), at the end of photobleaching (Bleach; 2 s), and at the indicated time points after photobleaching. The white rectangle represents the area photo-bleached within the nucleus.

FIG. 4.

Fluorescence recovery curves after photobleaching of HeLa cells transfected with CFP-RARα (a), CFP-RARα, CFP-ΔRARα, CFP-PML, CFP-ΔPML, or CFP-PML-RARα (b), CFP-RARα, CFP-PLZF, or CFP-PLZF-RARα (c), CFP-RARα, CFP-NPM, or CFP-NPM-RARα, (d), CFP-RARα, CFP-STAT5b, or CFP-STAT5b-RARα (e), and CFP-RARα, CFP-NuMA, CFP-NuMA-RARα, or CFP-NuMA-RARα(ΔCC) (f). Where indicated in panel a, cells were incubated with ATRA (10⁻⁶ M). Data presented are the mean fluorescence within the photobleached area of 10 cells.

TABLE 2.

t_1/2 of fluorescence recovery of CFP-RARα, CFP-X-RARα, and CFP-X constructs

No.a	Construct	t1/2 (−ATRA)	t1/2 (+ATRA, 10−6 M)
1	CFP-RARα	0.67 ± 0.15 s	0.57 ± 0.08 s
2	CFP-ΔRARα	0.50 ± 0.10 s	0.54 ± 0.11 s
3	CFP-PML	>5.0 min
4	CFP-ΔPML	>5.0 min
5	CFP-PML-RARα	>5.0 min	>5.0 min
6	CFP-PLZF	1.03 ± 0.27 s
7	CFP-PLZF-RARα	0.96 ± 0.12 s	1.35 ± 0.26 s (P < 0.001)
8	CFP-NPM	1.71 ± 0.20 s
9	CFP-NPM-RARα	0.84 ± 0.11 s	0.99 ± 0.18 s (P = 0.037)
10	CFP-NuMA	16.47 ± 2.3 s
11	CFP-NuMA-RARα	1.95 ± 0.41 s	2.54 ± 0.41 s (P = 0.005)
12	CFP-NuMA RARα(ΔCC)	0.80 ± 0.15 s	0.80 ± 0.11 s
13	CFP-STAT5b	0.47 ± 0.14 s
14	CFP-STAT5b-RARα	1.21 ± 0.14 s	1.44 ± 0.21 s (P = 0.01)

^aP ≤ 0.01 for 1 or 2 versus 5, 7, 9, 11, or 14; P ≤ 0.002 for 1 or 2 versus 6, 8, or 10.

ATRA slows X-RARα but not RARα.

We demonstrated previously that the addition of estrogen reduced the mobility of ER, a type I nuclear receptor (30). To determine if receptor slowing with ligand is a feature of NR of the type II class, we analyzed the behavior of CFP-RARα after addition of its ligand, ATRA (10⁻⁶ M), for 2 h. However, unlike ER, ligand addition did not slow CFP-RARα (Table 2). In contrast to CFP-RARα, however, ATRA did result in the slowing of CFP-PLZF-RARα, CFP-NPM-RARα, CFP-NuMA-RARα, and CFP-STAT5b-RARα, ranging from an 18% reduction of the t_1/2 for CFP-NPM-RARα to 30% for CFP-NuMA-RARα (Table 2). Slowing of CFP-PML-RARα with ligand addition could not be assessed because of its markedly reduced baseline mobility (t_1/2 >5 min).

The coiled-coil domain of NuMA-RARα is responsible for the reduced mobility of NuMA-RARα compared to RARα and its reduced mobility with ligand.

Protein interaction domains within the X component of X-RARα fusion proteins have been demonstrated by us and others to be responsible for many of the oncogenic properties of X-RARα proteins (9, 10, 17). We previously demonstrated that the coiled-coil domain of NuMA was responsible for NuMA-RARα homodimerization, inhibition of ligand-dependent RARα-mediated gene transcription, and enhancement of Stat3-mediated transcriptional activation (9). To begin to examine the contribution of the protein interaction domains to altered X-RARα mobility, we compared the mobility of CFP-NuMA-RARα with that of CFP-NuMA-RARα(ΔCC), in which the coiled-coil domain of CFP-NuMA-RARα is deleted (Fig. 3; Table 2). While the fluorescence recovery t_1/2 of CFP-NuMA-RARα was 1.95 ± 0.41 s, the t_1/2 of CFP-NuMA-RARα(ΔCC) was reduced by 59% to 0.80 ± 0.15 s (P < 0.001), which was indistinguishable from the t_1/2 of CFP-RARα. In addition to increasing its mobility, deletion of the coiled-coil domain eliminated ligand-induced slowing of CFP-NuMA-RARα (Table 2). Thus, in addition to being essential for its other oncogenic functions, the coiled-coil domain of NuMA is responsible for the reduced mobility of CFP-NuMA-RARα compared to CFP-RARα and its slowing in response to ligand.

Effect of X-RARα on nuclear localization and mobility of RXRα.

To determine whether the altered localization and reduced mobility of X-RARα altered the localization and mobility of NR coregulators, we performed fluorescence microscopy and FRAP analysis of HeLa cells transfected with YFP-RXRα, without or with CFP-RARα or CFP-X-RARα (Fig. 2, 5, and 6). YFP-RXRα alone distributed predominantly within the nucleus in a diffuse fashion similar to that of RARα (Fig. 2). Cotransfection of cells with YFP-RXRα and CFP-RARα did not change the nuclear distribution of RXRα (Fig. 6). However, when coexpressed with CPF-X-RARα, YFP-RXRα changed its nuclear distribution and colocalized with each X-RARα (Fig. 5), presumably due to its ability to heterodimerize with the RARα portion of X-RARα. While YFP-RXRα colocalized with CFP-NPM-RARα within the nucleoplasm, we could not detect YFP-RXRα within the nucleoli of cells when coexpressed with both CFP-NPM-RARα and CFP-NPM (Fig. 5g).

FIG. 6.

Dual-FRAP analysis of RARα and RXRα as well as PML-RARα and RXRα. (a to c) HeLa cells transfected with YFP-RXRα alone (a), YFP-RXRα and CFP-RARα (b), or YFP-RXRα and CFP-PML-RARα (c) were incubated in the absence of ATRA and subjected to FRAP analysis. Images shown are single z sections obtained before photobleaching (Pre-bleach), at the end of photobleaching (Bleach; 2 s), and at the indicated time points after photobleaching. Images shown in panel a were collected at the emission wavelength of YFP; images shown in panels b and c were simultaneously collected at the emission wavelengths of CFP (top row) and YFP (bottom row). (d) Fluorescence recovery curves from HeLa cells transfected with RXRα alone, RXRα and RARα, or RXRα and PML-RARα. Data presented are the mean fluorescence within the photobleached area of 10 cells.

Similar to CFP-RARα, YFP-RXRα was rapidly mobile when expressed alone, with a fluorescence recovery t_1/2 = 0.97 ± 0.29 s (Fig. 6a). Cotransfection of cells with YFP-RXRα and CFP-RARα did not change the nuclear mobility of either YFP-RXRα (t_1/2 = 0.98 ± 0.23 s) or CFP-RARα (t_1/2 = 0.63 ± 0.19 s) (Fig. 6b). In contrast, cotransfection of cells with YFP-RXRα and CFP-PML-RARα strikingly reduced the mobility of YFP-RXRα by eightfold (t_1/2 = 7.81 ± 1.20 s, P < 0.001 [Fig. 6c and d]), indicating that colocalization of RXRα with PML-RARα, which is relatively immobile, resulted in its reduced mobility.

DISCUSSION

There is increasing evidence from studies using real-time, live-cell approaches that the majority of nuclear proteins, including the linker histone H1 and nuclear receptors, are highly dynamic and exhibit both rapid movement in the nucleoplasmic space and fast exchange with a variety of targets, including chromatin-binding sites (18). Growing evidence also suggests that the dynamic exchange of proteins on chromatin is essential for transcriptional activators to gain access to chromatin and that controlling the exchange rate of a protein on chromatin may contribute to the regulation of gene expression (31). To gain new spatial and temporal insights into how X-RARα interferes with normal RARα function, we employed fluorescence microscopy and FRAP of fluorescent protein-tagged RARα, RXRα, X-RARα, and each of the X components in fixed and live cells following transient transfection or cotransfection. Our findings demonstrate that (i) the intracellular localization of each X-RARα protein was distinct from RARα and that this altered localization could be attributed, in each instance, to ΔX, ΔRAR, or the fusion of the two; (ii) the intranuclear mobility of each X-RARα was reduced compared to that of RARα; and (iii) in contrast to RARα, which showed no change in mobility with addition of ligand, the mobility of each X-RARα was reduced with ligand. Similar to other oncogenic features of NuMA-RARα, the altered localization and reduced mobility of NuMA-RARα mapped to the coiled-coil domain of NuMA. The altered localization of X-RARα resulted in the mislocalization of RXRα, presumably due to the interaction of RXRα with the RARα portion of X-RARα. In the case of PML-RARα, which has a fluorescence recovery t_1/2 of >5 min, the mislocalization of RXRα was accompanied by a dramatic reduction in its mobility.

PML-RARα had previously been demonstrated to sequester RXR within the cytoplasm when overexpressed transiently in COS cells (20) and within microspeckles tightly bound to chromatin within the nucleus of APL cells (33). In transfected COS cells, PML-RARα overexpression prevented the binding of the VDR to a target sequence in vitro and inhibited the vitamin D₃-dependent activation of a VDR-responsive reporter gene; in APL cells, ATRA treatment reversed RXRα sequestration. These findings raised the possibility that RXRα sequestration may be an important mechanism of PML-RARα oncogenesis. Our findings provide additional support for the sequestration hypothesis by demonstrating that RXRα and X-RARα, in cells expressing X-RARα at low levels, colocalize within the nucleus within microspeckles and further demonstrate that when PML-RARα and RXRα are coexpressed, the mobility of RXRα is reduced eightfold compared to its mobility when expressed alone or when coexpressed with RARα. Thus, in addition to sequestering RXRα, PML-RARα expression may interfere with normal type II NR functions, including RARα, by reducing the intranuclear mobility of RXRα and other critical NR coregulators.

Each of the X-RARα proteins had a nuclear predominance. The ΔRARα component was responsible for this feature in PLZF-RARα, NPM-RARα, and STAT5b-RARα, while the ΔX component mediated nuclear localization of PML-RARα and NuMA-RARα. Each of the X-RARα proteins has a pattern of intranuclear distribution distinct from that of ΔRARα. In all instances except NuMA-RAR, this pattern is microspeckled and distinct from that of ΔX. It is clear from our previously published results (9) that in the case of NuMA-RARα, the NuMA component is responsible for the reticular intranuclear pattern of NuMA-RARα. The microspeckled pattern observed in our studies for each of the other four X-RARα proteins could not be attributed to either ΔX or ΔRARα; rather, it appears to be a unique feature of the chimeric protein.

Studies such as those described in this report, using derivatives of GFP to obtain spatiotemporal information about X-RARα proteins and their individual components singly and in combination, cannot be performed on fixed cells by immunohistological techniques. However, the results are valid only if the fluorescent tag does not interfere with the biological function (if assessable) and the localization of the untagged protein. From previous studies by us and others (reviewed in reference 31), it is clear that attachment of GFP and its derivatives rarely affects the function and localization of the fusion protein. In this work, two lines of evidence provide strong support to the contention that our tagged proteins were not appreciably altered: (i) each tagged X-RARα exhibited biochemical and transcriptional properties similar to those of their untagged counterparts, and (ii) each localized within cells as reported previously in experiments using immunohistological techniques.

Previous studies by us demonstrated that ligand addition can differentially reduce the mobility of type I NR, ER and AR (29, 30; Stenoien et al., submitted), in contrast to the findings of others (24, 32). In our present study, while ligand addition had a modest effect on the mobility of X-RARα, it did not change the mobility of wild-type RARα. These findings raise two possibilities: (i) that the regulation of transcription and/or degradation by type I and type II NRs is distinct and (ii) that by being able to bind to RARE as homodimers, X-RARα may share some additional features with type I NRs involving their regulation and/or degradation. Alternatively, since at least ER and AR undergo posttranslational modification after ligand binding, e.g., ubiquitination, it is possible that the type I NRs are more influenced by ligand than are type II NRs, which are less clearly understood in terms of these modifications.

The reduced mobility of STAT5b-RARα compared to RARα, unlike the other X-RARα fusion proteins, could not be attributed to the reduced mobility of STAT5b, which had a high mobility, similar to that of RARα. However, STAT5b-RARα, like each of the other X-RARα proteins, formed microspeckles. Microspeckles formed by PML-RARα were demonstrated previously to correspond to PML-RARα protein bound tightly to chromatin (33), suggesting that rather than X within X-RARα, it is the ability of each X-RARα to form microspeckles that results in the reduced mobility of X-RARα including STAT5b-RARα.

.