PML-RARalpha alleviates the transcriptional repression mediated by tumor suppressor Rb.

A fusion between the promyelocytic leukemia (PML) protein and the retinoic acid receptor-alpha (RARalpha) results in the transforming protein of acute promyelocytic leukemia, PML-RARalpha. PML has growth-suppressive properties and is localized within distinct nuclear structures referred to as nuclear bodies. PML participates in numerous cellular functions, including transcriptional activation, apoptosis, and transcriptional repression, whereas PML-RARalpha blocks these functions. However, the role played by PML-RARalpha in leukemogenesis remains unclear. Here we report that PML is required for transcriptional repression mediated by the tumor suppressor Rb. Rb interacts with the histone decaetylase (HDAC) complex containing co-repressors and represses the transcription of the E2F target genes. Overexpression of PML enhanced Rb-mediated repression. The degree of Rb-mediated repression was weakened by injecting anti-PML antibodies and was lower in Pml-deficient mouse embryonic fibroblasts. PML-RARalpha inhibited Rb-mediated repression, and two co-repressor-interacting sites on the PML-RARalpha molecule were required for this activity. Furthermore, PML-RARalpha blocked the interaction between Rb and HDAC. Thus, aberrant binding of PML-RARalpha to co-repressor-HDAC complexes may inhibit their association with Rb, resulting in the abrogation of Rb activity. Thus, the disruption of Rb-mediated repression may be a contributory factor in leukemogenesis.

The retinoblastoma tumor suppressor protein (Rb) 1 inhibits the G 1 /S transition in the cell cycle by repressing a subset of genes that are controlled by the E2F family (1,2). Rb binds to the activation domain of E2F and actively represses the promoter by recruiting the histone deacetylase (HDAC) complex (3)(4)(5). Rb associates not only with HDAC1 (3)(4)(5) but also with the co-repressors Ski and mSin3A (6). Co-repressors N-CoR/ SMRT, mSin3A, and Ski/Sno interact with each other, and all of these co-repressors are required for the transcriptional repression mediated by Mad and thyroid hormone receptor ␤ (TR␤) (7)(8)(9)(10)(11)(12). The N-CoR or SMRT complex contains the class II HDAC (13)(14)(15), while the mSin3 complex contains the class I HDAC (16). The mSin3 complex also contains two Rb-associated proteins, RbAp46 and RbAp48 (16), which directly interact with histone H4 (17). These results suggest that Rb represses the E2F target gene by recruiting multiple complexes containing both class I and II HDACs and co-repressors N-CoR/SMRT, mSin3, and Ski/Sno. In fact, we recently demonstrated that loss of one copy of sno leads to up-regulation of a target gene of Rb, cdc25A, and increases susceptibility to tumorigenesis in mice (18).
The promyelocytic leukemia (PML) protein was originally identified as a fusion partner of the retinoic acid receptor-␣ (RAR␣) in the transforming protein (PML-RAR␣) found in acute promyelocytic leukemias (19 -21). However, further analysis of PML indicated that PML has a capacity to suppress cellular proliferation (22,23). PML has the RBCC motif consisting of a C 3 HC 4 -type zinc finger (RING finger) motif and two Cys-rich regions (B-boxes), followed by a coiled-coil region. Wild-type PML is localized to nuclear dot-like structures known as PML nuclear bodies (NBs), which are normally comprised of 5-30 discrete domains (24 -26). PML-RAR␣ alters the dot-like NB structure into numerous small speckles, and normal NB structure can be regained by treating the cells with all-trans-retinoic acid (RA). Although PML appears to be involved in multiple functions, including apoptosis and transcriptional activation (reviewed in Ref. 27), we recently demonstrated that PML directly associates with multiple corepressors (N-CoR/SMRT, mSin3A, and Ski) and that it is required for Mad-mediated transcriptiona repression (28). Unlike PML, which directly interacts with co-repressors through its coiled-coil domain, PML-RAR␣ bears two sites that interact with the co-repressors: one is the coiled-coil region on PML and the other is the CoR box on RAR␣. Via these two sites, PML-RAR␣ may aberrantly bind to the co-repressor-HDAC complex, leading to the Mad-mediated repression. We were interested in determining whether PML/PML-RAR␣ is specific for Mad-mediated silencing or whether this is a common feature of other repressors that utilize co-repressors such as Ski.
Here, we report that PML is required for Rb-and TR␤mediated silencing and that PML-RAR␣ inhibits this silencing. Inhibition of the activity of the tumor suppressor Rb may be an important mechanism for PML-RAR␣-induced leukemogenesis.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ ‡ To whom correspondence should be addressed.

MATERIALS AND METHODS
Subcellular Localization of Rb and PML-Immunostaining of endogenous Rb and endogenous PML in NB4 cells were performed with anti-Rb mouse monoclonal antibody (G3-245, PharMingen) and anti-PML rabbit polyclonal antibodies (28) after treatment of the cells with RA (1 M) for 3 days. These procedures and confocal microscopy were performed as described previously (12).
Effect of PML on Rb-and TR␤-mediated Repression-CV-1 cells (1.5 ϫ 10 5 cells per six-well plate) were transfected using Lipo-fectAMINE (Invitrogen) with a mixture containing the Gal4 site-containing luciferase reporter (1 g) and either Gal4-Rb (0.01 g), Gal4-TR␤ (0.05 g), or the Gal4 expression plasmid (0.01 or 0.05 g) together with the PML expression plasmid pact-PML (0.25 or 0.5 g) and the internal control plasmid pRL-TK (Promega) (0.25 g). The total amount of plasmid DNA was adjusted to 3 g by addition of the control plasmid DNA lacking the cDNA. After transfection, cells were cultivated using charcoal-treated serum, and dual luciferase assays were then performed. Similar experiments were performed using the Gal4-␦EF1 (0.03 g) expression plasmid.
Single-cell Microinjection Assay-Rabbit polyclonal antibodies raised against GST-PML were purified using antigen columns. Microinjection assays were performed using Rat-1 cells as described previously (12).
Repressor Activity of Rb and TR␤ in Mouse Embryonic Fibroblasts (MEFs)-Wild-type and Pml Ϫ/Ϫ MEFs (1.5 ϫ 10 5 cells per six-well plate) were transfected using LipofectAMINE with a mixture containing the Gal4 site-containing luciferase reporter (1.5 g) and either 2.5 g each of Gal4-Rb, Gal4-TR␤, the control Gal4 expression plasmid or Gal4-␦EF1, and the internal control plasmid pRL-TK (0.5 g). Dual luciferase assays were performed as described above.
Effect of PML-RAR␣ on the Rb-and TR␤-mediated Repression-CV-1 cells (3 ϫ 10 5 cells per 60-mm dish) were transfected by the CaPO 4 method with a mixture containing the Gal4 site-containing luciferase reporter (1.5 g) and either Gal4-Rb, Gal4-TR␤, or the Gal4 expression plasmid (0.25 or 0.5 g) together with the plasmid pact-PML-RAR␣ (0.5 or 1.0 g), which expresses various forms of PML-RAR␣, and the internal control plasmid pRL-TK (0.5 g). Dual luciferase assays were then performed. The total amount of plasmid DNA was adjusted to 6 g by addition of the control plasmid DNA lacking the cDNA. Similar experiments were performed using Gal4-␦EF1 (0.03 g) expression plasmid. RA (1 M) treatment was performed for 30 h before lysate preparation.
Effect of PML-RAR␣ on E2F-dependent Transcription-A mixture containing 0.1 g of the E2F sites containing luciferase reporter, 0.1 g of the E2F1 expression plasmid or the control plasmid, and 1.5, 2.0, or 2.5 g of the PML-RAR␣ expression plasmid together with 0.1 g of the internal control plasmid pRL-TK was transfected into CV-1 cells (3 ϫ 10 5 cells per 60-mm dish) using LipofectAMINE, and luciferase assays were performed. The total amount of plasmid DNA was adjusted to 3 g by addition of the control plasmid DNA lacking the cDNA.
For co-immunoprecipitation of endogenous Rb and HDAC proteins of NB4 cells, cells were treated with RA (1 M) or control solvent for 70 h. Lysates were prepared by mild sonication in LDLD buffer. Anti-Rb rabbit antibodies (C-15, Santa Cruz) were used for immunoprecipitation. The immunocomplexes were washed with TBS (10 mM Tris-HCl (pH 7.5), 150 mM NaCl) and used for Western blotting with anti-HDAC1 (C-19, Santa Cruz) and anti-HDAC2 (C-19, Santa Cruz) goat polyclonal antibodies.

RESULTS
As reported previously by Alcalay et al. (29), Rb was found to co-localize with PML. For this purpose, we used APL-derived NB4 cells, which express PML-RAR␣ and undergo differentiation after RA treatment (Fig. 1A). RA treatment of NB4 cells induces degradation of PML-RAR␣, resulting in the recovery of the normal dot-like structures of NBs containing PML. In RAtreated NB4 cells, endogenous Rb and PML co-localized to NBs.
We observed that some PML signals did not co-localize with Rb. This might be explained by the association of a proportion of PML molecules with co-activator such as CBP.
The physical association of PML with Rb raised the possibility that PML might be important for transcriptional repression mediated by Rb. To investigate this, we first examined the effect of overexpression of PML on Rb-mediated repression (Fig. 1B). The Gal4-Rb fusion, which consists of the Gal4 DNAbinding domain and the repressor domain of Rb, repressed transcription from a Gal4 site-containing reporter. This Gal4-Rb-induced repression was further enhanced by PML. Similar results were obtained with the Gal4-TR␤. This effect of PML was specific to Rb and TR␤, since PML did not enhance the repression by another repressor ␦EF1, whose repressor activity was previously shown not to be mediated by the co-repressor c-Ski (12).
Micro-injection experiments were then performed using a Gal4-lacZ reporter containing Gal4-binding sites ( Fig. 2A). Injection of the reporter into Rat-1 cells gave rise to many lacZpositive cells, whereas co-injection of the lacZ reporter with the plasmid encoding Gal4-Rb or Gal4-TR␤ resulted in a decrease in the number of lacZ-positive cells. This decrease was partially relieved by co-injection with anti-PML antibody. The effect of the anti-PML antibody could be blocked by co-injection with the PML expression plasmid. In contrast, anti-PML antibody did not affect repression by the Gal4-␦EF1 fusion, whose repressor activity does not require the c-Ski complex. The incomplete abrogation of the Gal-Rb and Gal4-TR␤ function by anti-PML antibody may be due to the presence of other PML-related protein(s) containing the RBCC motif.
To further confirm that PML is involved in Rb-and TR␤mediated repression, we used MEFs from a Pml-deficient mouse (Fig. 2B). The degree of repression by Gal4-Rb or Gal4- TR␤ in Pml Ϫ/Ϫ cells was lower than that in wild-type cells, whereas the repressor activity of Gal4-␦EF1 was similar in both types of cells. The incomplete abrogation of repression mediated by Gal4-Rb or Gal4-TR␤ in Pml Ϫ/Ϫ cells may be due to the presence of PML-like proteins containing the RBCC motif.
As we recently reported (28), PML-RAR␣ inhibits Mad-mediated repression. This activity of PML-RAR␣ required the presence of two sites that interact with the co-repressors: one is the coiled-coil region on PML, and the other is the CoR box on RAR␣ (Fig. 3A). This suggests that PML-RAR␣ may change the conformation of the co-repressor-HDAC complex by binding aberrantly to it, thereby abrogating Mad-mediated repression. Since PML is also required for Rb-mediated repression, we speculated that PML-RAR␣ also inhibits Rb-mediated repression. To investigate this possibility, co-transfection assays were performed (Fig. 3A). Normal PML-RAR␣ abrogated Rb-induced transcriptional repression, while the PML-RAR␣ mutant lacking the PML coiled-coil region and/or the functional RAR␣ CoR box did not. Furthermore, RA treatment of transfected cells blocked the effect of PML-RAR␣. Similar results were also obtained using Gal4-TR␤. In contrast, normal PML-RAR␣ did not affect the repression mediated by Gal4-␦EF1. These results indicate that the presence of two sites in PMLRAR␣, which interact with the co-repressors-HDAC complex, mediates the abrogation of repression by Rb and TR␤.
By using the E2F1 site-containing luciferase reporter, we further confirmed the effect of PML-RAR␣ on Rb activity (Fig.  3B). Co-transfection of this reporter into CV-1 cells together with small amount of the E2F1 expression plasmid slightly enhanced the luciferase expression. This suggested that E2F binds to specific sites on the promoter to enhance transcription from the reporter. Co-expression of PML-RAR␣ with this re-porter and the E2F1 expression plasmid enhanced the luciferase expression in a dose-dependent manner, indicating that PML-RAR␣ inhibits the activity of endogenous Rb. These results indicate that PML-RAR␣ inhibits Rb-dependent transcriptional repression.
The results described above suggest that an interaction between Rb and two sites on PML-RAR␣ mediates inhibition of Rb activity. One possibility is that PML-RAR␣ blocks the interaction between Rb and the co-repressor-HDAC complex. We examined this by co-immunoprecipitation (Fig. 4A). The plasmids expressing FLAG-HDAC1 and Gal4-Rb were transfected into 293T cells together with the PML-RAR␣ expression plasmid or the control plasmid. Lysates from the transfected cells were used for immunoprecipitation with anti-Gal4 antibody. A significant amount of HDAC1 was co-precipitated with Gal4-Rb in the absence of PML-RAR␣, whereas PML-RAR␣ strongly reduced the amount of co-precipitated HDAC1.
To directly investigate the effect of PML-RAR␣ on the interaction between endogenous Rb and HDAC, we performed coimmunoprecipitation experiments using RA-treated and untreated NB4 cells (Fig. 4B). NB4 cells express PML-RAR␣, whereas RA treatment of NB4 cells induces degradation of PML-RAR␣. Anti-Rb antibody co-precipitated the endogenous HDAC1 and HDAC2 proteins in the lysates from RA-treated NB4 cells, A mixture of the E2F1 sitecontaining luciferase reporter, the E2F1 or control expression plasmid, and increasing amounts of PML-RAR␣ expression plasmid was transfected into CV-1 cells, and luciferase activity was measured. The amount of PML-RAR␣ expression plasmid was 1.5 (ϩ), 2.0 (ϩϩ), or 2.5 g (ϩϩϩ). An average of three experiments Ϯ S.E. is shown.

Role of PML in Rb-dependent Repression 43493
but not from lysates of untreated cells. Similar amounts of endogenous Rb were precipitated by anti-Rb antibody from RAtreated and untreated NB4 cells. These results suggest that PML-RAR␣ inhibits the association between Rb and HDAC. DISCUSSION Our results indicate that PML and PML-RAR␣ have opposite affects on Rb-mediated transcriptional repression; PML stimulates Rb-mediated transcriptional repression, whereas PML-RAR␣ inhibits Rb activity. The presence of two sites on the PML-RARa molecule, which interact with the co-repressors, is required for the inhibition of Rb activity. This suggests that the binding of PML-RAR␣ to the co-repressors complexes induces a conformational change in the complexes, which blocks the interaction between Rb and HDAC. Since we recently demonstrated that PML-RAR␣ also blocks Mad-mediated repression, the activity of two tumor suppressors, Mad and Rb, are probably inhibited through this mechanism. Thus, our results suggest an important role for Rb in acute promyelocytic leukemia. Consistent with this, it was reported that Rb protein is defective in binding to a viral oncoprotein in the M3 promyelocytic subtype of acute myeloid leukemia patients (30).
Rb function was only partly abrogated by injection of anti-PML antibody andonly partly diminished in Pml Ϫ/Ϫ cells. This may be due to the presence of PML-like proteins containing the RBCC motif. In fact, we found that another RBCC motif-containing protein, RFP (Ret finger protein), also directly interacts with multiple co-repressors including Ski. 2 It is interesting to note that the N terminus of RFP is fused to the ret protooncoprotein in transformed NIH3T3 cells (31). Since RFP was reported to associate with PML NBs (32), RFP-Ret fusion oncoprotein may also affect Mad and Rb activities like PML-RAR␣. At present, relatively little is unknown about how PML acts in transcriptional repression. However, one hypothesis is that PML and possibly other RBCC motif-containing proteins may act as key factors, such as scaffold proteins, to maintain the correct architecture or subcellular localization of co-repressors-HDAC complexes. Thus, PML and PML-RAR␣ may have the capacity to affect the activity of many other repressors.