Promyelocytic Leukemia Zinc Finger-Retinoic Acid Receptor α (PLZF-RARα), an Oncogenic Transcriptional Repressor of Cyclin-dependent Kinase Inhibitor 1A (p21WAF/CDKN1A) and Tumor Protein p53 (TP53) Genes*

Background: Promyelocytic leukemia zinc finger-retinoic acid receptor α (PLZF-RARα) is a transcriptional repressor generated by a chromosomal translocation between the PLZF and RARα genes in acute promyelocyteic leukemia patients. Results: The transcriptional regulation of CDKN1A by PLZF-RARα involves the competitive binding of p53, RARα, and Sp1, the modification of histones, and DNA methylation at the promoter. Conclusion: PLZF-RARα increases cell proliferation by repressing p21 expression. Significance: Oncoprotein PLZF-RARα represses transcription of the CDKN1A. Promyelocytic leukemia zinc finger-retinoic acid receptor α (PLZF-RARα) is an oncogene transcriptional repressor that is generated by a chromosomal translocation between the PLZF and RARα genes in acute promyelocytic leukemia (APL-type) patients. The molecular interaction between PLZF-RARα and the histone deacetylase corepressor was proposed to be important in leukemogenesis. We found that PLZF-RARα can repress transcription of the p21WAF/CDKN1A gene, which encodes the negative cell cycle regulator p21 by binding to its proximal promoter Sp1-binding GC-boxes 3, 4, 5/6, a retinoic acid response element (RARE), and distal p53-responsive elements (p53REs). PLZF-RARα also acts as a competitive transcriptional repressor of p53, RARα, and Sp1. PLZF-RARα interacts with co-repressors such as mSin3A, NCoR, and SMRT, thereby deacetylating histones Ac-H3 and Ac-H4 at the CDKN1A promoter. PLZF-RARα also interacts with the MBD3-NuRD complex, leading to epigenetic silencing of CDKN1A through DNA methylation. Furthermore, PLZF-RARα represses TP53 and increases p53 protein degradation by ubiquitination, further repressing p21 expression. Resultantly, PLZF-RARα promotes cell proliferation and significantly increases the number of cells in S-phase.

ture upon binding to RA ligand, releasing a corepressor and recruiting a coactivator instead (5)(6)(7), PLZF-RAR␣ does not release the corepressor⅐HDAC complex in the presence of RA, thus acting as a dominant-negative mutant form of RAR␣ in APL (8). Accordingly, ATRA resistance of cells containing the PLZF-RAR␣ fusion gene disrupts the RA signaling pathway that mediates myeloid differentiation, resulting in arrest at the immature promyelocytic stage (6, 9 -12). Although some developmentally important PLZF and RAR␣ target genes have been reported, the targets of PLZF-RAR␣ that are important in cell proliferation and oncogenesis remain largely unknown but are presumed to be genes that contain a RARE in their promoters (e.g. CDKN1A). PLZF-RAR␣ antagonizes RARE-containing genes normally up-regulated by RAR␣ in the presence of retinoic acid. Thus, transcription of a battery of RAR␣ target genes important in differentiation, development, and cell cycle arrest can be aberrantly repressed, leading to proliferation of the undifferentiated promyelocytes (2,3). p21, encoded by CDKN1A, inhibits the activity of the cyclin/ cdk2 complex and is a major regulator of mammalian cell cycle arrest (13,14). CDKN1A is primarily regulated at the transcriptional and translational levels (15). Whereas the induction of p21 predominantly leads to cell cycle arrest, the repression of CDKN1A expression may have a variety of outcomes, including cell proliferation, depending on the cellular context (15). The CDKN1A gene also is a transcriptional target of p53, which acts on the CDKN1A promoter distal p53 regulatory elements (14,16) and plays a crucial role in mediating G 1 , G 2 , and S phase growth arrests upon exposure to DNA-damaging agents (15). In addition, Sp1 family transcription factors are major regulators that affect CDKN1A gene expression by binding to the proximal promoter (17). Recently, Krüppel-like transcription factors were also characterized as key regulators of CDKN1A expression that affect p53-and proximal Sp1-mediated regulation of CDKN1A transcription (18 -24). p21 expression is activated by retinoic acid, and the CDKN1A promoter has a RARE with RAR␣ interacts to activate transcription.
MBD3 (methyl-CpG-binding domain protein-3) is a component of the Mi-2/NuRD (Mi-2/nucleosome remodeling and deacetylase) chromatin remodeling complex that contains a nucleosome remodeling ATPase, HDAC1 and HDAC2 (histone deacetylases-1 and -2), and metastasis-associated protein 2 (MTA2) (25). MBD3, which has no intrinsic DNA binding activity, is targeted to methylated promoters through interactions with MBD2. At the promoter, MBD3 maintains transcriptionally repressed chromatin (26). Interestingly, the MBD3 protein was shown to be associated with the proximal promoter of CDKN1A in cancer cells and was released upon treatment of the cells with an HDAC inhibitor (27). However, the function and mechanism of MBD3 association with the CDKN1A promoter remains largely uncharacterized. By recruiting HDACs and DNA methyltransferases (DNMTs), MBD3 may act as an important transcriptional repressor of p21 during oncogenic transformation and cell proliferation (28).
Consequently, we investigated whether and how the CDKN1A gene encoding p21, a key regulator of cell cycle control and cell proliferation, is controlled by PLZF-RAR␣ at the transcriptional level. Here, we show how various molecular interactions between PLZF-RAR␣, p53, Sp1, and MBD3 are all involved in regulation of CDKN1A. We found that the transcriptional regulation of CDKN1A by PLZF-RAR␣ involves competitive binding of the transcription factors described above, modification of histones, and DNA methylation at the proximal CDKN1A promoter.
Promoter DNA Methylation Analysis by Bisulfite DNA Sequencing-Genomic DNA was purified using the Wizard genomic DNA purification kit (Promega). Methylation analyses were performed by bisulfite conversion of genomic DNA using the EpiXplore TM Methyl Detection kit (Clontech). The primer sequences used to amplify the CDKN1A promoter region were sense, 5Ј-AGGAGGGAAGTGTTTTTTTGTAGTA-3Ј, and antisense, 5Ј-ACAACTACTCACACCTCAACTAAC-3Ј. The PCR product was cloned using the pGEM-T Easy vector System I kit (Promega). Mini-scale plasmid DNA was prepared from more than 30 individually transformed Escherichia coli clones and sequenced.
Flow Cytometry for Cell Cycle and Apoptosis Analysis-HEK293, HCT116 p53 ϩ/ϩ , HCT116 p53 Ϫ/Ϫ , and HL-60 cells were transfected with either a PLZF-RAR␣ expression or control vector. Transfected cells were washed, fixed with methanol, and stained with a solution containing propidium iodide (50 g/ml) and ribonuclease A (100 g/ml) for 30 min at 37°C in the dark. The DNA content, cell cycle profiles, and forward scatter of the cells were analyzed using a FACSCalibur (BD Biosciences) flow cytometer set to 488 (excitation) and 575 nm (peak emission). The data were analyzed using ModFit LT 2.0 (Verity Software House) and WindMDI 2.8 (Joseph Trotter, Scripps Research Institute).
MTT Assays-Confluent HEK293, HCT116 p53 ϩ/ϩ , HCT116 p53 Ϫ/Ϫ , and HL-60 cells grown on 10-cm culture dishes were transfected with either a PLZF-RAR␣ expression vector or control vector, transferred to 6-well culture dishes and grown for 0 -4 days. At days 0, 1, 2, 3, and 4, the cells were incubated for 1 h at 37°C with 20 l of MTT/well (2 mg/ml). The precipitates were dissolved with 1 ml of dimethyl sulfoxide and the levels of cellular proliferation was determined by analyzing the conversion of MTT to formazan using a SpectraMAX 250 spectrophotometer (Molecular Devices) at 570 nm.
Statistical Analysis-Student's t test was used for all statistical analyses. p Ͻ 0.05 values were considered significant.

PLZF-RAR␣ Stimulates Cell Proliferation and Increases the Number of Cells in S-phase-
We tested whether the oncoprotein PLZF-RAR␣ can promote cell proliferation in HEK293, HCT116 cells, and eventually in HL-60 myeloid cells in the later part of this study. Flow cytometry analysis of HEK293 and HCT116 cells transfected with a PLZF-RAR␣ expression plasmid showed that PLZF-RAR␣ stimulated cell cycle progression and increased the number of cells in S-phase (from 31.8 to 43.3%) (Fig. 1A). In agreement, MTT assays showed that PLZF-RAR␣ significantly increased cell proliferation (Fig. 1B).
Because PLZF acts as a tumor suppressor with apoptotic activity in cells with a hematopoietic origin, we investigated whether PLZF-RAR␣ induces apoptosis by analyzing HEK293 cells stained with Annexin V and propidium iodide. The cell populations undergoing early and late apoptosis were either minimal or negligible (from 1.11 to 3.75%) (Fig. 1C).
PLZF-RAR␣ Is a Transcriptional Repressor of the CDKN1A Gene Encoding p21-To understand how PLZF-RAR␣ increases cell proliferation and exerts oncogenic properties, we investigated whether the oncoprotein PLZF-RAR␣ could stimulate cell proliferation by controlling genes of the p53 pathway, important for cell cycle regulation. Transient transcription assays of HEK293 cells showed that PLZF-RAR␣ repressed transcription of ARF, MDM2, TP53, and in particular, CDKN1A ( Fig. 2A). RT-qPCR and Western blot analyses revealed that ectopic PLZF-RAR␣ also repressed the expression of endogenous TP53, MDM2, and CDKN1A at the transcriptional level (Figs. 2B and 3C). Thus, PLZF-RAR␣ regulates the upstream regulatory genes that eventually affect CDKN1A expression. We also examined which region of the CDKN1A promoter is important for transcriptional repression by PLZF-RAR␣ in HEK293 cells. PLZF-RAR␣ repressed the transcription of the four different CDKN1A promoters in a similar fashion, suggesting that PLZF-RAR␣ may repress transcription by acting at the proximal promoter, which has six Sp1-binding GC-boxes (Fig. 2D).
PLZF-RAR␣ Represses Transcription of the CDKN1A Gene by Binding to the Distal p53 Binding Elements and Decreasing p53 Stability and TP53 Transcription-Treatment with the DNA damaging agent etoposide increased CDKN1A expression by inducing p53 in HCT116 cells, which was repressed by PLZF-RAR␣ (Fig. 3A). In HCT116 p53 Ϫ/Ϫ cells, ectopic p53 expression increased CDKN1A expression, which was also repressed by PLZF-RAR␣ (Fig. 3B). An additional transcriptional analysis using a pG5-6x(p53RE)-Luc construct with five copies of the distal p53 binding elements of the CDKN1A showed that PLZF-RAR␣ blocked transcriptional activation of CDKN1A by p53 in Saos-2 cells (Fig. 3C). We observed a similar PLZF-RAR␣-mediated transcriptional repression of pG13-Luc, which contains 13 copies of the putative p53 binding element (Fig. 3D). Overall, our data suggest that PLZF-RAR␣ can inhibit transcriptional activation of the CDKN1A gene by p53 at the p53 response element (p53RE) of the distal CDKN1A promoter.
We next analyzed whether ectopic PLZF-RAR␣ affected p53 binding induced by etoposide in HCT116 cells. Although etoposide treatment did not affect the expression of PLZF-RAR␣ significantly, transcriptional activation of TP53 and CDKN1A by etoposide was potently repressed by PLZF-RAR␣ at both the mRNA and protein levels (Fig. 3, E-H).
PLZF-RAR␣ also repressed transcriptional activation of CDKN1A by etoposide or ectopic p53 (Fig. 3). We also tested whether PLZF-RAR␣ repressed transcription of CDKN1A in the absence of p53. Transient transcription assays in HCT116 p53 Ϫ/Ϫ cells showed that PLZF-RAR␣ could repress transcription of CDKN1A (Fig. 4A), and MTT assays of the same cells showed that PLZF-RAR␣ significantly increased cell proliferation by 2.5-fold (Fig. 4B). Western blot and RT-qPCR analyses revealed that ectopic PLZF-RAR␣ also repressed the expression of endogenous CDKN1A at both the protein and mRNA levels in HCT116 p53 Ϫ/Ϫ cells (Fig. 4, C-F). These results suggest that transcriptional repression of CDKN1A by PLZF-RAR␣ can be independent of p53.
Accordingly, PLZF-RAR␣ may directly repress transcription of CDKN1A or indirectly, by repression of p53 activity or expression. Oligonucleotide pulldown assays showed that PLZF-RAR␣ binds to and decreases p53 binding to p53REs (Fig.  5B). ChIP assays showed similar results in vivo (Fig. 5, C-G). Together, these results suggest that PLZF-RAR␣ competes with p53 to bind to the two p53 binding elements and that this binding competition is important for transcriptional repression of CDKN1A. The transcription repression of TP53 by PLZF-RAR␣ may also contribute indirectly to the repression of CDKN1A (Figs. 3, F and H; 5B, input lane 2, and 5I, input lane 2).
As protein-protein interactions between transcription factors can also repress transcription, we investigated whether PLZF-RAR␣ directly interacts with p53. Co-immunoprecipita- tion and Western blot assays of HEK293 cells transfected with a PLZF-RAR␣ expression vector revealed that PLZF-RAR␣ interacts with p53 in vivo (Fig. 5H). Interestingly, we noticed that PLZF-RAR␣ decreased the expression and acetylation of p53 (Fig. 5I). Ubiquitination assays and Western blot analyses further revealed that PLZF-RAR␣ considerably increased p53 ubiquitination, likely decreasing p53 protein stability. In cells expressing ectopic PLZF-RAR␣, p53 expression was most likely low because of the combination of decreased p53 stability and transcriptional repression of TP53 (Figs. 3, F and H; and 5, H-J).
We also tested which GC-box at the CDKN1A promoter is important for transcriptional repression by PLZF-RAR␣. Sitedirected mutagenesis of the proximal GC-boxes of reporter plasmids and transient transcription assays revealed that GCboxes 3, 4, and 5/6 are important for transcriptional repression by PLZF-RAR␣, as mutations in these elements inhibited CDKN1A transcriptional repression by PLZF-RAR␣ (Fig. 6D).
Oligonucleotide pulldown assays using HEK293 cells transfected with a control or PLZF-RAR␣ expression vector also indicated that PLZF-RAR␣ binds to these three elements and decreases Sp1 binding (Fig. 6E). PLZF-RAR␣ expression did not affect Sp1 expression (Fig. 6, E and I). ChIP assays further revealed that HA-tagged PLZF-RAR␣ or untagged PLZF-RAR␣ bound to the proximal promoter region in vivo, and that the increased PLZF-RAR␣ expression vector decreased endogenous Sp1 binding to the proximal promoter (Fig. 6, H and I).
Additionally, co-immunoprecipitation and Western blot assays of HEK293 cells transfected with a PLZF-RAR␣ expression vector revealed that PLZF-RAR␣ and Sp1 interact with each other in vivo (Fig. 6, G-I). Our data indicate that PLZF-RAR␣ is a GC-box-binding transcription factor that can compete with Sp1 at GC-boxes 3, 4, and 5/6 to repress the transcription of CDKN1A.
PLZF-RAR␣ Binds to a Distal RARE Promoter Element to Repress Transcription of CDKN1A-It has previously been shown that RAR␣-RXR complexes can activate transcription of CDKN1A by acting on the RARE in the CDKN1A distal promoter region (bp Ϫ1212 to Ϫ1194) (15,30). In addition, the PLZF-RAR␣ fusion protein retains a functional RAR␣ DNA binding domain, and dysregulation of RAR␣ target gene expression has been proposed to be an underlying cause of oncogenesis (5,31). Accordingly, we tested whether PLZF-RAR␣ could represses transcription of a CDKN1A-reporter fusion gene and the endogenous CDKN1A gene in HEK293 cells. Co-expression of RXR and RAR␣ increased transcription and, in the presence of ATRA, further activated transcription. Regardless of the presence of ATRA, PLZF-RAR␣ repressed both the CDKN1A reporter (Fig. 7A) and endogenous CDKN1A at their protein and mRNA levels (Fig. 7, B and C). Interestingly, we noticed that ATRA treatment increased p53 expression and PLZF-RAR␣ decreased p53 expression, but ATRA did not affect PLZF-RAR␣ expression (Fig. 7, B and C). Decrease in p53 expression by PLZF-RAR␣ could further decrease CDKN1A expression.
Site-directed mutagenesis of the RARE of the CDKN1A promoter in the reporter plasmid and transient transcription assays revealed that the RARE is important for PLZF-RAR␣mediated transcriptional repression because the mutation of any of the bipartite RARE elements caused a loss of transcriptional repression (Fig. 7, D and E). Oligonucleotide pulldown assays indicated that PLZF-RAR␣ binds the RARE (Fig. 7F). Moreover, ChIP assays of PLZF-RAR␣ binding with both anti-PLZF and anti-HA antibodies revealed that ectopic PLZF-RAR␣ binds to the RARE (Fig. 7G), resulting in CDKN1A transcriptional repression.
These data imply the involvement of HDACs, DNMTs, and promoter DNA methylation in the transcriptional repression of CDKN1A by PLZF-RAR␣. Accordingly, we investigated whether the CDKN1A promoter region can be Extracts from HCT116 p53 ϩ/ϩ cells expressing ectopic PLZF-RAR␣ were incubated with biotinylated oligonucleotides, incubated with streptavidin-agarose beads, and precipitated. The precipitates were analyzed by Western blot with the antibodies indicated. C-F, qChIP assay showing HA-PLZF-RAR␣/PLZF-RAR␣ or p53 binding to the distal p53RE-1 and -2 of the endogenous CDKN1A gene in HCT116 p53 ϩ/ϩ cells. The cells were transfected with an increasing amount (0 -9 g) of pSG5-PLZF-RAR␣ expression vector. Antibodies against PLZF, RAR␣, and p53 were used for ChIP. IgG, control. G, Western blot (WB) analyses showing PLZF-RAR␣ and endogenous p53 expression in HCT116 p53 ϩ/ϩ cells transfected with an increasing amount (0 -9 g) of pSG5-PLZF-RAR␣ expression vector. GAPDH, control. H and I, co-immunoprecipitation (IP) of PLZF-RAR␣ and p53. HCT116 p53 ϩ/ϩ cell lysates were immunoprecipitated using an anti-RAR␣ antibody and the precipitates were analyzed by Western blot using an anti-p53 antibody. Alternatively, the anti-p53 antibody was used first in the co-IP, and the anti-RAR␣ antibody was used for Western blotting. J, ubiquitination assay for endogenous p53. H1299 cells were transfected with pcDNA3-His-ubiquitin or pSG5-PLZF-RAR␣ in the various combinations indicated. The cells were cultured and treated with MG132 for 3 h prior to harvest. The cell lysates were then incubated with MagneHis nickel particles and the precipitated pellets were washed, resolved by 10% SDS-PAGE, and analyzed by Western blot using a p53 antibody. *, p Ͻ 0.05; N.S., not significant; t test. methylated by Me-DIP (methylated DNA immunoprecipitation) assays. Ectopic PLZF-RAR␣ increased methylation of the CDKN1A promoter region, as in the positive control AlphaX1 promoter, indicating that PLZF-RAR␣ may repress transcription of CDKN1A through DNA methylation at the 15 CpGs of the proximal promoter region (bp, Ϫ139 to ϩ30) (Fig. 8E). In control cells transfected with the pcDNA3 control construct, methylated bisulfite DNA sequencing showed that although some of the CpGs were methylated, only 1 of 20 CDKN1A promoter DNA strands sequenced was strongly methylated (13 of 15 CpGs), and only 2-3 moderately methylated promoter DNA strands were detected. In particular, the core CpG of Sp1 binding site-3, which is critical for transcriptional activation of CDKN1A, was methylated in 50% of the promoter DNA strands sequenced (17). In contrast, PLZF-RAR␣ dramatically increased methylation at the CpG island of the CDKN1A proximal promoter, with virtually 70% (14 of 20) of the promoter DNA strands sequenced exhibiting extensively methylated CpGs (Fig. 8F). Interest-ingly, all of the core CpGs of the six Sp1 binding GC-boxes were heavily methylated, which may inhibit promoter DNA binding and transcriptional regulation by Sp1 family and other Krüppel-like transcription factors.
Co-immunoprecipitation and Western blot analysis of either HEK293 cells or HEK293 cells transfected with a PLZF-RAR␣ expression vector revealed that PLZF-RAR␣ interacts with MBD3, the Mi-2⅐NuRD-HDAC3 complex, and the NuRD complex-associated DNMTs and HP1 (Fig. 8G). ChIP analysis also showed that ectopic PLZF-RAR␣ significantly increased the binding of MBD3, Mi-2⅐NuRD-HDAC3 complex (as monitored by diagnostic subunit MTA2), DNMT1/3b, and HP1 to the CDKN1A proximal promoter (Fig. 8H). PLZF-RAR␣, by interacting with MBD3, recruits the Mi-2⅐NuRD-HDAC3 complex and the complex-associated DNMT1/3b and HP1, likely resulting in CDKN1A promoter DNA methylation. Together, these results suggest that the CDKN1A promoter may be epigenetically silenced by histone deacetylation and DNA methylation. construct and the mutants generated by site-directed mutagenesis of the GC-box. X, mutation introduced; the GC box core GGG was replaced with TTT (right). Transcription analysis in HEK293 cells. Cells were transiently co-transfected with pGL2-CDKN1A-Luc WT or a mutant construct (Ϫ131 bp), and PLZF-RAR␣ expression vector and luciferase activity were then measured. E, oligonucleotide pulldown assay showing PLZF-RAR␣ binding to the proximal GC-boxes. HEK293 cell extracts were incubated with biotinylated double-stranded oligonucleotides and precipitated, as described in the legend to Fig. 5. The precipitates were analyzed for PLZF-RAR␣ and Sp1 binding by Western blot (WB). F, co-immunoprecipitation (IP) of PLZF-RAR␣ and Sp1. HEK293 cell lysates were immunoprecipitated using an anti-RAR␣ antibody and analyzed by Western blot with an anti-Sp1 antibody. Conversely, the lysates were also immunoprecipitated with the anti-Sp1 antibody and analyzed by Western blot with an anti-RAR␣ antibody. G-I, qChIP assay showing competitive Sp1 and PLZF-RAR␣/HA-PLZF-RAR␣ binding at the endogenous CDKN1A proximal promoter in HEK293 cells. The cells were transfected with an increasing amount (0 -9 g) of the PLZF-RAR␣ expression vector. Antibodies against HA tag, PLZF, Sp1, and IgG (control) were used in the ChIP assays. *, p Ͻ 0.05; N.S., not significant; t test.

PLZF-RAR␣ Stimulates Cell Proliferation and Represses CDKN1A Transcription in HL-60 Leukemia Cells through the Competitive Binding of p53, RAR␣, and Sp1, Histone
Modifications, and DNA Methylation-We showed the molecular mechanism underlying the oncogenic properties of PLZF-RAR␣ in HEK293 and HCT116 cells, and eventually tried to validate our findings in acute promyelocytic leukemia HL-60 cells. PLZF-RAR␣ promoted cell proliferation of human HL-60 cells and repressed CDKN1A and TP53 expression (Fig. 9, A-C). ChIP assays also showed that ectopic PLZF-RAR␣ decreased acetylation of histones H3 and H4 at the CDKN1A proximal promoter by 40 -65% and increased or decreased histone methylation markers of repression (H3K9-Me3) or activation (H3K4-Me3), respectively (Fig. 9, D-F). Furthermore, Me-DIP assays showed that ectopic PLZF-RAR␣ expression increased DNA methylation of the CDKN1A proximal promoter (Fig. 9G). These data imply that, as in HEK293 cells, PLZF-RAR␣ represses CDKN1A transcription by HDAC and promoter DNA methylation in HL-60 cells.
ChIP assays of HL-60 cells transfected with a PLZF-RAR␣ expression vector further revealed that ectopic PLZF-RAR␣ increased binding of MBD3, the Mi-2⅐NuRD-HDAC3 complex (as monitored by MTA2 binding), DNMT1/3b, and HP1 to the CDKN1A promoter (Fig. 9, H-K). These results suggest that PLZF-RAR␣ may repress CDKN1A expression epigenetically by histone deacetylation and/or methylation by recruiting the MBD3-Mi-2⅐NuRD-HDAC3 complex and its associated DNMT1/3b and HP1. These results show the molecular mechanisms we identified in HEK293 cells are also applicable to HL-60 human leukemia cells.
Epigenetic Repression of CDKN1A by PLZF-RAR␣ in HL-60 Cells Can Be Partially Reversed by the HDAC Inhibitor TSA, the DNMT Inhibitor 5-Aza-2Ј-deoxycytidine, ATRA or Any Combination of these Three Drugs-PLZF-RAR␣ can repress CDKN1A expression in HL-60 cells through epigenetic mechanisms that include histone deacetylation and promoter DNA methylation. Ectopic PLZF-RAR␣ repressed CDKN1A through the deacetylation of histones H3 and H4 (Fig. 10, A and B). The ChIP data on markers of transcriptional activation and repression indicated that treating the cells with epigenetic derepressive agents (TSA and 5-aza-2Јdeoxycytidine) combined with the RAR␣ ligand ATRA did not completely derepress CDKN1A transcription to the level found in control cells (Fig. 10, H and J). Treating the cells with any of these agents alone or in combination did not affect PLZF-RAR␣ (as judged by ChIP using an anti-PLZF antibody) binding or the control ChIP reactions (Fig. 9E). These results indicate that, of the several transcriptional repression mechanisms described above, CDKN1A transcriptional repression by the competitive binding of p53, PLZF-RAR␣, and Sp1 is quite significant. Because binding competition between these transcription factors is not likely affected by TSA and ATRA, the finding may explain why some APL patients with PLZF-RAR␣ translocation are resistant to TSA and ATRA combination therapy and relapse. Although ATRA supplemented with the HDAC inhibitor TSA is effective in leukemia treatment, the addition of a DNMT inhibitor such as 5-aza-2Ј-deoxycytidine appears to be more effective in inhibiting the proliferation of HL-60 cells transfected with the PLZF-RAR␣ expression vector (Fig. 10G). Because CDKN1A transcriptional repression caused by the competitive binding of p53, PLZF-RAR␣, RAR␣, and Sp1 and PLZF-RAR␣-mediated down-regulation of CDKN1A expression by PLZF-RAR␣ (Fig. 10, H-J) persists in leukemic cells expressing the PLZF-RAR␣ oncoprotein, a certain population of leukemic cells may still remain resistant to the three drug combination therapies. Accordingly, the fundamental goal for better treating RA-resistant APL leukemic patients may be to inactivate PLZF-RAR␣ activity or block the expression of the fusion protein.
PLZF-RAR␣ was presumed to antagonize the function of RAR␣ by interfering with promyeloctye differentiation or/and disrupting myeloid-specific PLZF functions. PLZF-RAR␣ promotes proliferation of promyelocytes (immature granulocytes) through the aberrant regulation of cell cycleassociated genes such as MYC (1)(2)(3)32). However, the target genes and the mechanism by which the PLZF-RAR␣ oncoprotein stimulates cell proliferation and blocks myeloid differentiation has remained largely unknown.
PLZF-RAR␣ interacts with co-repressor⅐HDAC complexes such as NCoR/SMRT and Sin3A. PLZF-RAR␣ contains two co-repressor binding sites, the CoR box of RAR␣ and the POZ domain of PLZF. Retinoic acid releases HDAC complexes from the CoR box of PLZF-RAR␣, but not from the POZ domain, which accounts for the molecular basis of RA resistance in PLZF-RAR␣-type APL patients. Previously, an artificial minimal promoter system with a RARE was used to demonstrate that the histone deacetylase inhibitors TSA and ATRA can lift the transcriptional repression by PLZF-RAR␣ and synergistically activate reporter gene expression in CV-1, NB4, and U937 cells (5,33). That study provided a basis for the effective growth suppression of ATRA-resistant leukemic cells by HDAC inhibitors and ATRA (5). Although genome-wide PLZF-RAR␣ target genes were characterized by ChIP-on-ChIP in lymphoma U937 cells (34), the true targets of PLZF-RAR␣ that play important roles in cell proliferation, and the molecular mechanisms of PLZF-RAR␣ actions on these targets, remain largely unknown.
One of the key regulators of the cell cycle is p21. The CDKN1A gene encoding p21 was previously reported to be a direct target of RAR␣ (15). Our investigation revealed that PLZF-RAR␣ represses CDKN1A expression potently to stimulate cell proliferation. PLZF-RAR␣ also represses expression of another key regulator of the cell cycle, p53. Our investigation of the molecular mechanism of CDKN1A transcriptional regulation by PLZF-RAR␣ revealed several important novel features of the oncoprotein PLZF-RAR␣. These include CDKN1A repression through competitive binding of transcription factors (p53, Sp1, RAR␣, and PLZF-RAR␣), transcriptional repression of TP53, degradation of p53 by ubiquitination, and CDKN1A promoter histone deacetylation and DNA methylation. Moreover, the DNA binding specificity of PLZF-RAR␣ is rather promiscuous and is not limited to the RARE. PLZF-RAR␣ also binds to two distal p53 binding elements and prox-imal GC-boxes (thus competing with Sp1 for GC-box binding). These binding activities of PLZF-RAR␣ could potentially block both the constitutive and inducible expression of CDKN1A.
Consequently, we propose a hypothetical model for the transcriptional regulation of CDKN1A by PLZF-RAR␣ (Fig. 11). Under normal cellular conditions in which p53 levels are low and no PLZF-RAR␣ is present, p21 is expressed at a low level and cells proliferate normally. When cells are challenged with genotoxic stress, however, the tumor suppressor p53 is markedly induced and activates CDKN1A transcription by interacting with its p53REs and Sp1 bound at the Sp1-binding GC-box 3. The induced p21 protein then stops progression of the cell cycle and allows the cells to either repair DNA damage or undergo apoptosis. When cells express PLZF-RAR␣ following chromosomal translocation, PLZF-RAR␣ represses CDKN1A transcription by binding to proximal GC-boxes 3, 4, and 5/6, RARE, and the distal p53 binding elements of the CDKN1A promoter (Fig. 11B). Thus, transcriptional activation by p53, RAR␣, and Sp1 can be effectively blocked by PLZF-RAR␣ at their respective binding elements. Even when cells are under genotoxic stress, p53 and Sp1 cannot activate transcription due to PLZF-RAR␣ competitive binding to both the proximal GC boxes and distal p53REs that are critical for basal transcription and synergistic transcriptional activation by Sp1 and p53, leading to the accumulation of DNA damage and increased cell proliferation.

PLZF-RAR␣, a Transcriptional Repressor of CDKN1A and TP53
Furthermore, PLZF-RAR␣ potently represses transcription of TP53 and decreases p53 stability by inhibiting p53 acetylation and increasing p53 ubiquitination. p53 is an upstream transcriptional activator of CDKN1A. PLZF-RAR␣ blocks the induction of CDKN1A by repressing the expression of de novo p53 and promoting the degradation of p53 through decreased p53 acetylation and increased p53 ubiquitination (Figs. 5, B, G, I, and J; and 11, C and D). PLZF-RAR␣ significantly affects not only the transcription of CDKN1A, but may also affect other p53 and RAR␣ target genes important for apoptosis, differentiation, cell cycle regulation, etc.
However, the transcriptional repression of CDKN1A mediated by PLZF-RAR␣ appears to be more complex than the repression by competitive binding among transcription factors and HDAC activity, and additionally involves epigenetic silencing by histone deacetylation and DNA methylation. Interestingly, the methylated DNA-binding protein MBD3, which is one of the subunits of the Mi-2⅐NuRD-HDAC3 complex, was found to be associated with the proximal promoter of CDNK1A (35). PLZF-RAR␣ interacts with MBD3 and recruits the Mi-2⅐NuRD-HDAC3 complex and the NuRD-associated DNMT1/ 3b and HP1, which eventually leads to CDKN1A promoter DNA methylation.
Using HEK293 and HCT116 cells, we were able to consistently show that PLZF-RAR␣ has proto-oncoprotein characteristics with the capacity to transform cells and stimulate cell proliferation by repressing CDKN1A expression. However, one might argue that the molecular mechanism of transcriptional regulation of CDKN1A by PLZF-RAR␣ revealed in HEK293 and HCT116 cells might not be true in leukemic cells, which is often true depending on the transcription factors and cellular contexts. We were able to demonstrate that PLZF-RAR␣ also represses transcription of CDKN1A in human promyelocytic leukemia HL-60 cells.
The PML-RAR␣ fusion protein has an abnormally high affinity for corepressors, and the protein switches to an activator by In the absence of PLZF-RAR␣, the CDKN1A gene is transcribed at a basal level, primarily by Sp1 family transcription factors. In PLZF-RAR␣ type leukemic cells, however, PLZF-RAR␣ represses CDKN1A transcription by binding to its two distal promoter p53-binding elements, a RARE, and the proximal GC-boxes 3, 4, and 5/6. Binding to these elements involves competition with p53, RAR␣/RXR, and Sp1 at their respective binding sites. PLZF-RAR␣ recruits co-repressor⅐HDAC complexes, deacetylates histones, and increases the level of H3K9-Me3. PLZF-RAR␣ also interacts with the MBD3⅐NuRD/ HDAC3 complex, which also modifies histone acetylation and methylation to reflect a repressed state. The MBD3⅐NuRD complex also contains DNMT1/3 and can methylate the CDKN1A promoter DNA. Because the MBD3⅐NuRD complex contains HDAC3 and DNMTs, it is not certain whether histone deacetylation and DNA methylation occurs simultaneously or sequentially. Presence of HP1 at the promoter region indicates heterochromatin formation. Tsp(ϩ1), transcription start site. ZF, zinc finger DNA binding domain. X, transcriptional repression. C, transcriptional repression of the TP53 gene by PLZF-RAR␣ and the corepressor⅐HDAC complex. D, post-translational ubiquitination of p53 by PLZF-RAR␣. Although acetylation of p53 by p300 increases p53 activity or stability, ubiquitination of p53 decreases p53 protein stability.
releasing corepressors at pharmacological doses of ATRA (5,36). However, the PLZF-RAR␣ fusion protein is ATRA-resistant and does not release corepressors. Previously, HDAC inhibitors such as TSA and butyrate were shown to block PLZF-RAR␣-mediated repression of reporter genes (5), and the combination of ATRA and the HDAC inhibitor suberoyl anilide hydroxamic acid (SAHA) was reported to be sufficient for clearing leukemic blasts from the peripheral blood of mice harboring PLZF-RAR␣ (12). Our study suggests that DNMT inhibitors such as 5-aza-2Ј-deoxycytidine in combination with ATRA and HDAC inhibitors may be more effective in derepression of the PLZF-RAR␣ target genes involved in cell differentiation, cell proliferation, and oncogenesis. Thus, PLZF-RAR␣type APL patients may be more effectively treated by the addition of DNMT inhibitors to the ATRA plus HDAC inhibitor regimen. However, because the repression of CDKN1A is, in part, due to competitive binding between p53, PLZF-RAR␣, RAR␣, and Sp1, and the fact that transcription repression of TP53 by PLZF-RAR␣ persists in leukemic cells with PLZF-RAR␣, a certain population of leukemic cells may still remain resistant to the above described three drug combination therapy. Thus, an improved treatment of ATRA-resistant APL patients could be to inactivate PLZF-RAR␣ activity or block the expression of the fusion protein.