Cytoplasmic Function of Mutant Promyelocytic Leukemia (PML) and PML-Retinoic Acid Receptor-α*

The promyelocytic leukemia (PML) tumor suppressor of acute promyelocytic leukemia (APL) regulates major apoptotic and growth-suppressive pathways. In APL, PML is involved in a chromosomal translocation generating the PML-retinoic acid receptor-α (RARα) fusion protein. Two missense mutations in the remaining PML alleles have been identified, which give rise to a truncated cytoplasmic PML protein (Mut PML). APL patients carrying these mutations display resistance to retinoic acid (RA) and very poor prognosis. Here we show that Mut PML associates with the cytoplasmic regions we refer to as PML-cytoplasmic bodies (PML-CBs). Mut PML interacts with PML-RARα in PML-CB and potentiates PML-RARα-mediated inhibition of RA-dependent transcription. Remarkably, Mut PML stabilizes PML-RARα and inhibits differentiation induced by pharmacological doses of RA. A mutant form of PML-RARα that accumulates in the cytoplasm inhibits RA-dependent transcription and differentiation, thus suggesting that cytoplasmic localization of PML-RARα may contribute to transformation. Finally, we show that the bcr3 PML-RARα form is predominantly cytoplasmic and accumulates in PML-CBs. Taken together, these findings reveal novel insights into the molecular mechanisms contributing to APL.

The promyelocytic leukemia (Pml) gene encodes a tumor suppressor involved in the t(15;17) chromosomal translocation associated with acute promyelocytic leukemia (APL), 2 and which produces the fusion protein PML-RAR␣ (1). Two major isoforms are generated, depending on the location of the PML breakpoint, named bcr1 (long) and bcr3 (short) (2,3). PML-RAR␣ is able to act as a dominant-negative retinoic acid receptor, thus inhibiting the response to RA and blocking RAinduced differentiation (4). This function is mainly exerted through changes at the chromatin level through the recruitment of histone deacetylases and methyltransferases (5)(6)(7). However, it is presently unclear whether other mechanisms are involved. Treatment with pharmacological doses of all-trans-retinoic acid (ATRA) causes the degra-dation of the fusion protein and the maturation of the leukemic cells, thus resulting in clinical remission in patients (8). However, 20 -30% of patients are insensitive to such treatment and suffer relapse (8). Treatment regimens of such patients include the use of arsenic trioxide, which has a high efficacy (9,10). On the other hand, PML-RAR␣ inhibits PML tumor-suppressive functions thus lending the leukemic cells a growth and survival advantage (1). Because the effects of PML-RAR␣ on PML are dose-dependent (1), the product of the remaining PML allele may oppose PML-RAR␣-mediated leukemogenesis. Indeed, progressive reduction of the PML gene dosage in a murine APL model results in increased incidence and decreased latency of the disease (11). PML is a member of the Ring-B-box-coiled coil (alternatively named tripartite motif, TRIM) family and exists as multiple splicing variants (12). PML nuclear splice variants localize to a subnuclear structure known as the PML-nuclear body (PML-NB), which has been implicated in transcriptional regulation and is disrupted in APL (1). Nuclear PML is essential for the proper formation and stability of these subnuclear structures (1). Several nuclear proteins localize to the PML-NB, such as p53, the p53 acetyltransferase cAMP-response element-binding protein-binding protein (CBP), the cell death regulator Daxx, and Sp100 (1). By contrast, cytoplasmic isoforms are less characterized. One specific cytoplasmic isoform has been associated with transforming growth factor-␤ signaling (13). Remarkably, a recent study conducted on a cohort of 17 RAresistant APL cases identified two Pml mismatch mutations in the remaining allele, which are associated with very aggressive disease (14). Both mutations (1272delAG and IVS3-1G-A) generate a premature stop codon upstream of the nuclear localization signal. Thus, the resulting truncated proteins accumulate in the cytoplasm (14). Furthermore, another study identified a murine plasmacytoma cell line harboring a PML mutation (mut ex3), which also leads to a premature stop codon at the end of exon three, cytoplasmic localization and dominant-negative properties (15). PML-RAR␣ itself undergoes alternative splicing to produce truncated proteins that are predicted to localize in the cytoplasm (16,17). Localization of PML-RAR␣ in the cytoplasm has been reported to cause an endoplasmic reticulum stress response and the ubiquitination of N-Cor (18). Moreover, the proteolytic cleavage of PML-RAR␣ by elastase results in cytoplasmic accumulation of the PML portion of the fusion protein and is probably required for disease progression (19). Thus, accumulating evidence suggests that both PML and PML-RAR␣ might have a functional role in the cytoplasm.
Although APL-associated PML mutations have been shown to cause PML relocation to the cytoplasm (14), the evidence that PML delocalization has an impact on the sensitivity to maturation signals such as RA was lacking. Here we show that mutant PML potentiates PML-RAR␣mediated block of differentiation and inhibition of retinoic acid-dependent transcription. Moreover, PML mutants form cytoplasmic bod-ies (PML-CB), where selective PML-NB components accumulate. PML-RAR␣ itself accumulates in PML-CB and colocalizes with Mut PML. Finally, we found that a PML-RAR␣ mutant that accumulates in the cytoplasm retains the ability to inhibit RA-dependent transcription and differentiation.

MATERIALS AND METHODS
Cell Culture-Primary fibroblasts and established cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 20 or 10% fetal bovine serum, 100 units/ml penicillin, 100 mg/ml streptomycin, 4.5 mg/ml glucose, and L-glutamine. Hematopoietic cell lines HL60, NB4, RA-resistant NB4 (336 and 006, a kind gift of Carlo Gambacorti, Istituto Tumori, Milan, Italy), and U937 were cultured in RPMI 1640 supplemented with 10% fetal bovine serum and 2 mM GlutaMAX TM . All reagents were purchased from Invitrogen.
Cloning and Plasmids Generation-PML mutants (14) as well as PML-RAR␣ bcr3 were generated by using PCR-based strategies, tagged at the amino terminus with HA or Myc epitope, and then subcloned in into pcDNA3.0 (Invitrogen) and pBABE PURO (20). PML-RAR␣ bcr1 (17) was cut from PINCO-PML-RAR␣ (a kind gift of Dr. Saverio Minucci) and subcloned into pSG5, pcDNA3.0, and pBABE PURO. When not specified, the PML-RAR␣ form used in this study is brc1. QuikChange site-directed mutagenesis kit was used according to the manufacturer's instructions to delete PML or both PML and RAR nuclear localization signals (nls) of PML-RAR␣ to generate pcDNA-PML-RAR␣-⌬NLS1 and -⌬NLS2, respectively (⌬1PR and ⌬2PR). PML-RAR␣ ⌬2 mutant that lacks the capability to bind RXR␣ has been described elsewhere (21) and was generated using standard PCR-based strategy. Finally, RXR␣ cDNA was amplified using RNA prepared from human primary fibroblasts (BJ) and subcloned into pcDNA 3.0.
Subcellular Fractionation-For each cell type, subcellular fractionation was performed as described previously (22). Cells were harvested, washed twice in ice-cold PBS, and homogenized using a loose-fitting Dounce homogenizer in ice-cold hypotonic buffer (10 mM Hepes, 10 mM MgCl 2 , 1 mM EDTA in the presence of a complete protease/inhibitor mixture). Alternatively, hematopoietic NB4 cells were homogenized using a tight-fitting Dounce homogenizer in ice-cold hypotonic buffer with added 0.25 M sucrose. After that, the homogenate was centrifuged at 1000 ϫ g for 10 min at 4°C to obtain pellet (P1) and supernatant (S1) consisting of nuclei and whole cytoplasmic extracts, respectively. Pellet P1 was lysed in ice-cold hypertonic buffer (150 mM NaCl, 10 mM Hepes, 10 mM MgCl 2 , 1 mM EDTA supplemented with a complete protease/inhibitor mixture), whereas S1 was further centrifuged at 100,000 ϫ g for 1 h at 4°C to obtain pellet (P100) and supernatant (S100) which represent, respectively, the insoluble (membranes enriched) and soluble cytosolic fractions. Aliquots of each fraction as well as whole cellular lysates were analyzed by SDS-PAGE, transferred onto a nitrocellulose membrane, and probed with relevant antibodies.
Immunoprecipitation-Cells were lysed in immunoprecipitation (IP) buffer (50 mM Tris-Cl, pH 7.6, 150 mM NaCl, 2 mM Na 2 VO 3 , 2.5 mM NaF, and 0.5% Triton X-100 in the presence of a complete protease/ inhibitor mixture). 500 g of lysates was subjected to immunoprecipitation using Sepharose beads coated with relevant antibodies for 3 h at 4°C. Immunocomplexes were washed five times with 500 l of IP buffer, resuspended in SDS sample buffer, and then subjected to Western blot analysis.
Western Blots-Western blot analysis was performed as described previously (23). In particular, the effect of Mut PML on PML-RAR␣ protein levels was assayed upon treatments with ATRA or arsenic tri-oxide (As 2 O 3 ) in COS1 cells. Briefly, 2.5 ϫ 10 5 cells were seeded in a 6-well plate for 12 h before transfection. Transfections were carried out using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. The following expression vectors were transfected: 1600 ng of pcDNA-HAPML-RAR␣ together with 400 to 200 ng of pcDNA-mycMut PML or empty vector. Eight hours later cells were washed twice with PBS and cultured in the presence or absence of 0.1 M ATRA (Sigma) or 1 M arsenic trioxide (Sigma). Finally, cells were harvested 12 and 24 h after treatments, and PML-RAR␣ protein levels were analyzed by Western blot.
Immunofluorescence-Immunofluorescence and confocal microscopy was performed as previously described (25).
Electron Microscopy and Immunogold Cytochemistry-Adherent cells were processed and embedded in epoxy resin. Duplicate pellets were fixed with 4% formaldehyde (freshly made up from paraformaldehyde) or a mixture of 4% formaldehyde and 0.1% glutaraldehyde in PBS, pH 7.4, for 1 h at room temperature and rinsed in PBS. A subset of these pellets was post-fixed in a mixture of osmium tetroxide (0.05%) and potassium ferrocyanide (0.05%). All were embedded in LR-White resin (Agar Scientific, Stansted, UK) and labeled with immunogold (British Biocell International, Cardiff, UK) (26). Ultra-thin sections were examined unstained or after staining with lead citrate and/or uranyl acetate. Control incubations involved the replacement of primary antibody with an equivalent concentration of the appropriate immunoglobulin.
Transcriptional Assays-Cells (COS1) were seeded 24 h prior transfection in 12-well plates and cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (Sigma), penicillin/ streptomycin, and GlutaMAX (Invitrogen). Transfection was carried out by using a calcium phosphate transfection kit (Invitrogen) according to the manufacturer's instructions. For RA-dependent transcription, the following expression vectors were used (100 ng unless otherwise stated): pCH110, 300 ng of pRep 4 -RARE-Luc and pcDNA-PML-RAR␣ alone or in combination with 100 ng of pcDNA-HAMut PML. Fresh medium containing 10 Ϫ7 M ligand (ATRA) was added 24 h post-transfection, and cells were incubated for 12 h. Finally, cells were harvested, and 5 l of extracts were assayed by using dual light system kit (Applied Biosystem) for luciferase activity according to the manufacturer's instructions. Transcriptional activity of PML-RAR␣ ⌬NLS-1 and -2 mutants was assayed as above. Cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and transfected with 1 g each of pcDNA-PML-RAR␣, pcDNA-PR ⌬NLS-1, and -2. Effects of PML-RAR ⌬2 on dihydroxyvitamin D 3 -dependent transcription were investigated in transiently transduced COS1 cells. Briefly, cells were transfected with 300 ng of vitamin D 3 luciferase reporter (consisting of four DR3-type vitamin D 3 -responsive elements inserted upstream of the tk-luciferase) (kind gift of Professor Carsten Carlberg), 600 ng of PML-RAR␣ (bcr-1), PML-RAR␣ ⌬2, or pcDNA3.0 vectors along with 100 ng of pCH110. Cells were cultured for 12 h in presence or absence of 10 Ϫ6 M of dihydroxyvitamin D 3 (Calbiochem) and subsequently assayed for luciferase and ␤-galactosidase activities.
Retroviral Infection-Retroviral infection was performed as described previously (23). For hematopoietic cells, spinoculation was performed by spinning viral supernatants on top of target cells at 2500 rpm for 1 h and 30 min.
Differentiation Analysis-NB4 and U937 cells were infected as described previously (23) using empty and Mut PML or PML-RAR␣ ⌬2 retroviral constructs, respectively. Infection was carried out using spinoculation. Cells were selected with puromycin and used either as clones or mixed populations. Differentiation was induced as reported (27) culturing cells with 10 Ϫ7 or 10 Ϫ6 M ATRA. After 4 days, differentiation was assayed by flow cytometry analysis of differentiation surface marker CD11b (Caltag Laboratories). In addition, differentiation was evaluated by morphological analysis of U937 cells following hematoxylin-eosin staining.

Mutant PML Proteins Form Cytoplasmic Bodies-APL-associated
PML mutations (1272delAG and IVS3-1G-A) lead to the generation of truncated PML proteins (Mut PML 1 and 2) that lack the nuclear localization signal and accumulate in the cytoplasm (14). We analyzed whether Mut PML 1-and 2-transduced cells displayed alterations in cell Infected BJ cells were stained using an anti-HA antibody. Nuclei were visualized using DAPI. B, Mut PML accumulates in both the soluble and insoluble fraction of the cytoplasm. The pellet P100 and supernatant S100 fractions from Mut PML-transfected fibroblasts were stained with an anti-HA antibody. Tubulin was detected as loading control. C, Mut PML does not associate with endosomes. Mut PML-infected cells were fixed and subsequently stained with anti-HA and EEA1 antibodies, and nuclei were detected using DAPI. Slides were analyzed by confocal microscopy. D, Mut PML forms doughnut-shaped PML-CB. Immunogold death and proliferation at steady state and failed to detect any substantial changes, thus demonstrating that, unlike nuclear PML, these mutants are not growth-suppressive or pro-apoptotic per se (not shown). Then we studied the localization pattern of Mut PML 1 and 2 in transfected fibroblasts. Both proteins displayed an identical localization pattern and accumulated to discrete foci, which we refer to as PML cytoplasmic bodies (PML-CB; Fig. 1A and supplemental Fig. 1A). As both proteins displayed the same localization pattern, we decided to focus on Mut PML 1 (Mut PML) for all our subsequent analyses. In subcellular fractionation experiments, a portion of Mut PML accumulated in the P100 fraction, suggesting that the Mut PML-positive foci associate with the insoluble part of the cytoplasm, which contains intracellular membranes (Fig. 1B). As a recent study indicated that cytoplasmic PML isoforms are associated with early endosomes (13), we attempted to verify whether Mut PML colocalizes with the early endosomal marker EEA1 but failed to reveal a colocalization with these organelles (Fig. 1C). We also analyzed the distribution of Mut PML in relation to mitochondria, lysosomes, and endoplasmic reticulum using the appropriate markers but again failed to prove significant colocalization (not shown). To better assess the structural features of PML-CB, we performed electron microscopy studies using antibodies directed against PML or the HA tag. Mut PML formed doughnut-shaped structures as shown by EM, which are reminiscent of the PML-NB (Fig. 1D), and did not colocalize with intracellular organelles such as mitochondria and endoplasmic reticulum (m and r; Fig. 1D, bottom right). The PML-NB components Daxx and Sp100 did not associate with PML-CB upon coexpression of Mut PML in fibroblasts and hematopoietic cells (supplemental Fig. 1B and not shown). In contrast, although CBP was exclu-sively nuclear in the majority of control cells examined, coexpression of Mut PML caused CBP accumulation in PML-CB in both hematopoietic cells and fibroblasts ( Fig. 1E and supplemental Fig. 1, C and D). Furthermore, the number of CBP-containing PML-NBs was substantially reduced in Mut PML-expressing cells (supplemental Fig. 1C). Finally, CBP staining was competed out using a specific peptide (supplemental Fig. 1D).
Mut PML Potentiates PML-RAR␣ Function-PML-RAR␣ acts as a strong inhibitor of the RA receptor complex, thus causing a block of RA-dependent transcription and differentiation (4). We reasoned that Mut PML could affect PML-RAR␣-mediated inhibition of RA-dependent processes in leukemic cells. As both Mut PML and PML-RAR␣ contain the Ring-B-box-coiled coil motif, they can potentially interact and form heterodimers. To test this, we coexpressed Mut PML and PML-RAR␣ and found that they partially colocalized in cytoplasmic bodies ( Fig. 2A). Furthermore, we were able to coimmunoprecipitate PML-RAR␣ and Mut PML in transfected 293T cells and fibroblasts (Fig.  2B). Remarkably, we found that expression of Mut PML potentiated the inhibition of ATRA-dependent transcription exerted by PML-RAR␣ (Fig. 2C). To uncover the potential mechanisms underlying the Mut PML-mediated effect on PML-RAR␣, we analyzed the effect of Mut PML expression on PML-RAR␣ down-regulation upon ATRA treatment. Expression of Mut PML resulted in impaired PML-RAR␣ downmodulation in the presence of different concentrations of ATRA (0.1 and 1 M; Fig. 2D and not shown). Instead, Mut PML did not affect arsenic trioxide (As 2 O 3 )-mediated down-regulation of PML-RAR␣ (Fig. 2D). However, As 2 O 3 caused reduction of Mut PML protein levels as well, thus explaining the lack of protection (Fig. 2D). To determine whether Mut PML could affect differentiation in the context of APL cells, we transduced the APL cell line NB4 with Mut PML and analyzed the sensitivity to ATRA (Fig. 3A). Mut PML and PML-RAR␣ partially colocalized in cytoplasmic bodies in these cells (Fig. 3B). Vector-and Mut PML-transduced cells were cultured in the presence of ATRA and analyzed for the expression of the differentiation marker CD11b. Remarkably, in Mut PML-transduced NB4 cells differentiation induced by pharmacological concentrations of ATRA was reduced (Fig. 3C). This indicates that Mut PML is able to potentiate PML-RAR␣ activity in the context of APL.
Cytosolic PML-RAR␣ Inhibits the Response to RA-Previous reports have shown that overexpressed PML-RAR␣ localizes to the cytoplasm (18, 28 -30). We confirmed that in a fraction of transfected fibroblasts (ϳ40%) PML-RAR␣ accumulates in cytoplasmic structures reminiscent of PML-CB ( Fig. 4D and not shown). We then analyzed the distribution of overexpressed full-length PML-RAR␣ by subcellular fractionation and found that it accumulated in both nuclear and cytoplasmic fractions (Fig. 4A). Moreover, endogenous PML-RAR␣ accumulated in the cytosolic fraction of an APL cell line, NB4 (Fig. 4B).
To analyze the consequences of cytoplasmic localization of PML-RAR␣, we constructed bcr1 PML-RAR␣ mutants lacking PML nls only (⌬1 PR) or both PML and RAR␣ nls (⌬2 PR) (Fig. 4C). ⌬1 and ⌬2 PR mutants were expressed in both COS1 and U937 cells (Fig. 4D). Immunofluorescence analyses demonstrated that although wild type PML-RAR␣ was found either in the nucleus only (60% cells; Fig. 4D, upper panel) or in both the nucleus and the cytoplasm (40% cells; Fig. 4D, lower panel), ⌬2 PR was instead exclusively cytoplasmic (Fig. 4D). Interestingly, ⌬1 PR predominantly accumulated in the cytoplasm, thus suggesting that PML nls governs the nuclear localization of PML-RAR␣ (Fig. 4D). We then studied the transcriptional properties of the PR mutants in COS1 cells. Surprisingly, we found that both ⌬2 and ⌬1 PR are fully capable of inhibiting ATRA-dependent transcription (Fig. 4E and not shown). Furthermore, we observed that ⌬2 PR is not down-regulated upon RA treatment, thus suggesting that cytoplasmic localization of PML-RAR␣ results in its impaired degradation (Fig. 4F).
Based on these findings, we conjectured that cytoplasmic PML-RAR␣ could induce the delocalization of essential transcriptional regulators to the cytoplasm. Previous reports have shown that PML-RAR␣ induces the delocalization of the RA receptor component RXR␣ to aberrant nuclear and cytoplasmic foci (29). Remarkably, we also found that ⌬2 PR causes the relocation of both endogenous and exogenous RXR␣ to PML-CB ( Fig. 5A and data not shown). To determine whether delocalization of RXR␣ is required for transcriptional inhibition, we constructed a ⌬2 PR mutated at residues essential for RXR␣ binding (⌬2 M883R/T886R) (21). ⌬2 M883R/T886R was less potent in repressing ATRA-dependent transcription, indicating that RXR␣ recruitment is required at least in part for transcriptional repression (Fig. 5B). We next tested whether vitamin D 3 (VD3)-dependent transcription, which relies on RXR␣ and is repressed by WT PML-RAR␣, was inhibited by ⌬2 PR (Fig. 5C). Indeed, endogenous VD3 receptor activity is repressed by cytoplasmic PR to an extent similar to WT PR (Fig. 5C).

Cytosolic PML-RAR␣ Inhibits Differentiation of Hematopoietic Cells-
To determine whether ⌬2 PR-mediated impairment of RA-dependent transcription has functional consequences on the maturation of hematopoietic cells, we transduced U937 cells with ⌬2 PR retroviruses and analyzed the sensitivity of the resulting clones to ATRA-induced differentiation (Fig. 6A). Strikingly, the morphology of ATRA-treated ⌬2 PRexpressing cells was almost indistinguishable from control cells, and many mitotic figures were readily detectable in ATRA-treated ⌬2 PR cells (Fig. 6B). Accordingly, the expression of the maturation marker CD11b upon ATRA treatment was inhibited in ⌬2 PR-transduced U937 clones (Fig. 6C).
bcr3 PML-RAR␣ Largely Localizes to the Cytoplasm and Colocalizes with RXR␣-We next analyzed the localization pattern of the bcr3 PML-RAR␣ form (Fig. 7A), which lacks the PML nls (2, 3), and found, remarkably, that it was predominantly cytoplasmic (Fig. 7, A and B). Moreover, as for ⌬2 PR, bcr3 PML-RAR␣ colocalized with both overexpressed and endogenous RXR␣ (Fig. 7, A and B). Finally, we analyzed the subcellular distribution of PML-RAR␣ in primary APL cells carrying the bcr3 breakpoint and found that a substantial amount of PML-RAR␣ accumulates in the cytoplasm (Fig. 7C). Furthermore, RXR␣ predomi-nantly partitioned to the cytoplasmic fractions, thus indicating that cytoplasmic PML-RAR␣ also induces RXR␣ delocalization in primary APL cells (Fig. 7C).

DISCUSSION
PML-RAR␣-mediated leukemogenesis is believed to occur mainly through chromatin remodeling, which leads to dramatic changes in the transcriptome of hematopoietic cells. However, limited information is available about the mechanisms that could promote transformation without a direct effect on chromatin. Our work indicates that cytosolic localization of PML and PML-RAR␣ can affect transcription and promote a differentiation block in hematopoietic cells.
Cytosolic PML Mutants Accumulate in Doughnut-shaped Cytoplasmic Bodies-We have demonstrated that Mut PML forms novel cytoplasmic structures, which we named PML-CB, and closely resembles the PML-NB in EM analyses. This suggests that essential components of the two structures could be shared. Alternatively, PML could be the core and essential component required for the assembly of the doughnut-shaped body. The fact that Mut PML associates with the insoluble part of the cytoplasm originally prompted us to investigate whether it could colocalize with cytoplasmic organelles. However, we were unable to demonstrate any colocalization with known organelles. In particular, Mut PML does not associate with endosomes, as it is known for a cytoplasmic PML isoform (13). It is therefore possible that the association with endosomes may be lost in APL cells expressing truncated PML mutants. While analyzing the distribution of various PML-NB components, we found that CBP relocates to PML-CB in Mut PML-expressing cells, whereas Daxx and Sp100 do not. It is reasonable to hypothesize that CBP redistribution could have functional consequences on the activation of selected transcription factors.
Mut PML Potentiates PML-RAR␣-We found that Mut PML colocalizes and interacts with PML-RAR␣ in both adherent cells and leukemic blasts. Surprisingly, Mut PML potentiates PML-RAR␣-mediated repression of RA-dependent transcription, thus suggesting that cytoplasmic localization of PML-RAR␣ may result in transcriptional inhibition in the absence of direct effects on chromatin. Remarkably, expres-  sion of Mut PML in APL cells counteracts differentiation induced by pharmacological doses of RA. Furthermore, we demonstrated that Mut PML hampers PML-RAR␣ degradation induced by RA, thus providing a mechanistic explanation for the observed resistance to RA. It is possible that Mut PML can interfere with the proteasome-mediated degradation of PML-RAR␣, which is as yet poorly characterized. Another possibility is that Mut PML modulates PML-RAR␣ SUMOylation, which has been shown to be essential for PML-RAR␣-transforming capacity (21).
Cytosolic PML-RAR␣ Inhibits Transcription and Differentiation-Consistent with previous reports (18, 28 -30), we found that overexpressed PML-RAR␣ can locate to the cytoplasm. In addition, we demonstrated that endogenous PML-RAR␣ accumulates in the cytoplasmic fraction of APL cells, thus suggesting it can bear cytoplasmic functions. To test this, we analyzed whether a PML-RAR␣ mutant (⌬2 PR) that localizes to the cytoplasm retains the ability of inhibiting RA-dependent transcription. Indeed, ⌬2 PR is almost as potent as WT PML-RAR␣ in repressing RA-triggered transcription. ⌬2 PR is not down-regulated upon RA, thus again supporting the hypothesis that cytoplasmic localization of PML-RAR␣ inhibits its RA-dependent degradation. Interestingly, ⌬2 PR colocalizes with RXR␣ in PML-CB. It is of note that the cytoplasmic redistribution of RXR␣ has been reported to cause an impairment in its transcriptional function (31). We have found that mutations abolishing RXR␣ binding substantially reduce ⌬2 PR repressing activity, thus suggesting that RXR␣ relocation is one of the potential mechanisms involved. As the rescue is not complete, it is conceivable that other mechanisms may contribute to the observed phenomena. ⌬2 PR appears to be a potent inhibitor of differentiation as it represses differentiation in U937 cells at pharmacological doses of RA. Interestingly, we found that although ⌬2 PR causes a 50% reduction of CD11bϩ cells compared with ATRA treatment, the morphology of RA-treated ⌬2 PR-infected cells is nearly undistinguishable from untreated cells. Moreover, proliferation is not substantially blocked in ⌬2 PR-expressing cells, which continue to proliferate even at high concentrations of ATRA (1 M). 3 This aspect warrants further investigation in the near future. Finally, we found that the bcr3 form of PML-RAR␣, which has been associated with poor prognosis in a number of studies (2, 3), localizes to the cytoplasm at a greater extent than the bcr1 form and causes the redistribution of RXR␣ to PML-CB. In cells from an APL patient 3 C. Bellodi and P. Salomoni, unpublished observations. The upper panel shows a schematic representation of Bcr1 and Bcr3 proteins. Lower panels, HA-bcr3 PML-RAR␣-transfected COS1 cells were probed with an anti-HA antibody, counterstained with DAPI, and analyzed by confocal microscopy. B, overexpressed RXR␣ colocalizes with bcr3 PML-RAR␣ in the cytoplasm. HA-bcr3 PML-RAR␣/RXR␣-and RXR␣transfected COS1 cells were probed with anti-HA and anti-RXR␣ antibodies, counterstained with DAPI, and analyzed by confocal microscopy. C, bcr3 PML-RAR␣ and RXR␣ accumulate in the cytosolic fraction of primary APL cells. APL cells from a patient carrying the bcr3 translocation were fractionated into nuclear extracts (N.E.) and cytosolic fractions (total cytoplasmic extracts (C.E.), pellet, P100; supernatant, S100). Filters were probed with anti-RAR␣, RXR␣, lamin A/C, and tubulin antibodies. D, working model. Cytoplasmic localization of PML-RAR␣, alone or through binding to Mut PML, does not affect its ability to repress differentiation. Delocalization of components of the RA receptor complex (ϫ), such as RXR␣, is one of the potential mechanisms involved.
carrying the bcr3 break point, a large fraction of PML-RAR␣ is found in the cytosol, where also RXR␣ accumulates. Overall, our findings suggest that PML-RAR␣ cytoplasmic pool may affect transcription and differentiation in the absence of direct effects on chromatin-dependent phenomena (Fig. 7D).