Inactivation of the P53–KLF4–CEBPA Axis in Acute Myeloid Leukemia Katja Seipel1,2, Miguel Teixera Marques1, Marie-Ange Bozzini1, Christina Meinken1, Beatrice U

Published OnlineFirst September 25, 2015; DOI: 10.1158/1078-0432.CCR-15-1054

Biology of Human Tumors Clinical Cancer Research Inactivation of the p53–KLF4–CEBPA Axis in Acute Myeloid Leukemia Katja Seipel1,2, Miguel Teixera Marques1, Marie-Ange Bozzini1, Christina Meinken1, Beatrice U. Mueller1, and Thomas Pabst2

Abstract

Purpose: In acute myeloid leukemia (AML), the transcription of p53 function and concomitant reduction of KLF4 and CEBPA factors CEBPA and KLF4 as well as the universal tumor suppressor protein levels. Assessment of cellular p53 modulator proteins p53 are frequently deregulated. Here, we investigated the extent of indicated that p53 inactivation in leukemic cells correlated with dysregulation, the molecular interactions, and the mechanisms elevated levels of the nuclear export protein XPO1/CRM1 and involved. increase of the p53 inhibitors MDM2 and CUL9/PARC in the Experimental Design: One hundred ten AML patient samples cytoplasm. Finally, restoring p53 function following treatment were analyzed for protein levels of CEBPA, KLF4, p53, and p53 with cytotoxic chemotherapy compounds and p53 restoring non- modulators. Regulation of CEBPA gene expression by KLF4 and genotoxic agents induced CEBPA gene expression, myeloid dif- p53 or by chemical p53 activators was characterized in AML cell ferentiation, and cell-cycle arrest in AML cells. lines. Conclusions: The p53–KLF4–CEBPA axis is deregulated in Results: We found that CEBPA gene transcription can be AML but can be functionally restored by conventional chemo- directly activated by p53 and KLF4, suggesting a p53–KLF4– therapy and novel p53 activating treatments. Clin Cancer Res; 22(3); CEBPA axis. In AML patient cells, we observed a prominent loss 746–56. 2015 AACR.

Introduction and/or translational expression, and altered protein degradation (3). Moreover, CEBPA is a DNA damage–inducible p53-regulated CCAAT/enhancer-binding protein alpha (CEBPA) is a lineage- mediator of the G checkpoint in keratinocytes, but not in specific transcription factor expressed in hematopoietic stem and 1 fibroblasts (4). myeloid progenitor cells (1). CEBPA functions as a differentiation Similarly, Kruppel-like€ factor 4 (KLF4/GKLF) is a lineage-spe- factor by directly targeting lineage-specific genes. CEBPA activates cific transcription factor expressed in a monocyte-restricted and granulocyte- and eosinophil-specific genes while inhibiting ery- stage-specific pattern during myelopoiesis, and it functions to throid- and lymphoid-specific factors. In addition, CEBPA acts as a promote monocyte differentiation (5, 6). Alike CEBPA, KLF4 can potent cell-cycle inhibitor. Cell-cycle arrest and terminal differ- arrest cell proliferation via induction of the CDKN1A gene, and it entiation are coupled processes, and the ability of CEBPA to represents a potent p53 mediator, thus acting as a tumor sup- mediate both may provide an explanation for its ability to act pressor (7). KLF4 expression is downregulated in NPM1-mutant as a switch between uncommitted proliferating cells and differ- AML by miR10a (8), and it can be repressed by CDX2 and HDAC1 entiated cell-cycle arrested cells. The mechanisms by which CEBPA (9–11). mediates cell-cycle arrest have been extensively investigated, and The tumor suppressor p53 has been dubbed the guardian of the they involve the ability of CEBPA to upregulate the CDKN1A gene genome (12) reflecting its crucial role in dealing with all kinds of encoding the cyclin-dependent kinase inhibitor p21/WAF1/Cip1, stressful conditions that can damage DNA. Physiologically, p53 its interaction with the cyclin-dependent kinases CDK2 and induces cell-cycle arrest to enable repair of damaged components CDK4, and the CEBPA-mediated repression of the E2F complex or, in the case of excessive damage, to induce cell death by (2). CEBPA function is commonly deregulated in patients apoptosis (13). p53 exerts its cell-cycle arrest and apoptotic func- with AML due to genomic mutations, suppressed transcriptional tions in the nucleus, but it is shuttled to the cytoplasm where it can be retained, reshuttled to the nucleus, or destroyed (14). The tumor suppressor p53 is inactivated in a wide variety of cancers, 1Department of Clinical Research, University and University Hospital of either by mutation or by dysregulation, in particular via induction 2 Berne, Berne, Switzerland. Department of Medical Oncology, Univer- of the p53 inhibitor MDM2 or by disturbing the nucleocytoplas- sity and University Hospital of Berne, Berne, Switzerland. mic shuttling of p53, either by dysregulation of the transporter Note: Supplementary data for this article are available at Clinical Cancer proteins importin and exportin or induction of the cytoplasmic Research Online (http://clincancerres.aacrjournals.org/). retention protein CUL9/PARC. In AML cells, the incidence of p53 Corresponding Author: Thomas Pabst, Department of Medical Oncology, mutations is low; however, dysregulation of p53 function in AML University Hospital, Berne 3010, Switzerland. Phone: 41-31-6328430; Fax: 41- appears to be a frequent event (15). 31-6323410; E-mail: [email protected] In AML as well as in other hematologic malignancies, CEBPA, doi: 10.1158/1078-0432.CCR-15-1054 KLF4, and p53 can be deregulated. The extent of dysregulation, 2015 American Association for Cancer Research. the molecular interactions, and mechanisms involved in this

746 Clin Cancer Res; 22(3) February 1, 2016

p53–KLF4–CEBPA Axis in AML

CAC-30)andp53_mut_R (50-GTGGATAGATGCGGGAATTCA- Translational Relevance 0 GAGG-3 ) and to create pGL4.23-36-kbM1 using primer pairs The transcription factors CEBPA and KLF4 act as key reg- p53-mut F2 (50-CCTCTGCATGCCCGCATCTATCCAC-30)and ulators of normal differentiation and suppress malignant p53_mut_R2 (50- GTGGATAGATGCGGGCATGCAGAGG-30). transformation, whereas in acute myeloid leukemia (AML), Oligonucleotide primers were supplied by Microsynth. the functions of CEBPA, KLF4, and of the tumor suppressor To test the KLF4-binding sites in the 9-kb 30 HCNE, a partial p53 are frequently suppressed. Our study indicates that CEBPA construct pGL4.23-9kb3C (190 bp, 9-kb 30) was created by gene transcription can be directly activated by p53 and KLF4, internal SacI deletion in pGL4.23-9kb30 (730 bp, 9-kb 30). The suggesting a p53–KLF4–CEBPA axis. We observed in primary other 540-bp sequence was PCR-amplified using primers with AML samples a characteristic loss of p53 function and a restriction sites XhoI(50) and HindIII (30) and subcloned into concomitant reduction in KLF4 and CEBPA protein levels. pGL4.23 to create pGL4.23-9kb3N (540 bp, 9-kb 30). The 730-bp Importantly, restoring p53 function following treatment with 9-kb 30 region encoded four KLF4 sites, whereas the 190-bp 9-kb 30 cytotoxic chemotherapy together with p53 activating non- and the 540-bp 9-kb 30 constructs encoded two KLF4 sites. genotoxic agents induced CEBPA gene expression and initiated myeloid differentiation and cell-cycle arrest in AML cells. Our Luciferase reporter assays data pave the way toward therapeutic concepts combining CEBPA-firefly luciferase reporter and Renilla luciferase reference conventional chemotherapy and novel p53 activating treat- plasmids together with pcDNA3.3_KLF4 (16) and pcDNA_TP53 ment based on the observation that conventional chemother- expression plasmids were transfected into H1299 cells using apy together with novel p53 activating treatment can restore Lipofectamine2000 (Invitrogen). After 24 hours, cells were lysed CEBPA function. and luciferase activity was assessed using the Dual-Luciferase- Reporter Assay System (Promega) on a microplate reader Infinite 200 (Tecan). Assays were performed in at least three independent experiments. Statistical analysis was done with GraphPad Prism dysregulation are unclear. We investigated whether KLF4 and p53 software using two-tailed t tests. Data are depicted in column bar can activate CEBPA gene expression, and we assessed the extent of graphs plotting mean with SD values. The TP53 expression dysregulation in the protein levels of the three transcription plasmid was provided by Carol Prives and the KLF4 expression factors and of cellular p53 modulators in primary AML patient plasmid by Derrick Rossi (Addgene #26815). samples. CEBPA-luc-BAC stable cell lines Materials and Methods The CEBPA-BAC transgene P391 was created by insertion of a CEBPA reporter plasmids luciferase-puromycin cassette into RP11-270I13 (Gene Bridges). Highly conserved noncoding elements of the CEBPA gene P391 BAC DNA was prepared from the E. coli HS996 bacterial (HCNE) were defined using the genome browser genome.ucsc.edu. culture grown in LB medium with Cm (15 mg/mL), linearized by HCNEs were PCR-amplified from human gDNA using primers SgfI digestion, and subsequently used to generate stable cell lines. with restriction sites XhoI(50) and HindIII (30) for all HCNEs, with Leukemic HL60 cells were transfected with linearized P391 BAC 6 the exception of SacI(50) and BamHI (30) for the 33-kb 30 and 3- DNA (4 mg/10 cells) using the Nucleofector Kit V Program X-001 1kb 30 HCNEs, and BamHI (50 and 30) for the 6-kb 30 HCNE, and (Lonza). Cells were cultured for 24 hours before addition of they were cloned into pCR4-Topo (Invitrogen) before being puromycin (0.5 mg/mL). Colonies emerged after 10 to 14 days. transferred to the pGL4.23 firefly luciferase reporter plasmid Single-cell colonies were expanded, and P391-HL60 cells were (Promega). The Renilla luciferase (Rluc) plasmid was used for tested by transfection with pcDNA3.3_KLF4 (16), pcDNA_TP53, normalization. or pcDNA-AML1-ETO (17) expression plasmids (2.5 mg plasmid 5 HCNEs were located on the NCBI 36/hg18 assembly at chr19: per 2 10 cells) and Renilla reference plasmid using Transit2020 38443446–38443899 (454 bp, 38-kb 30), chr19:38446041– transfection reagent (Mirus Bio LLC). After 24 hours, cells were 38446632 (592 bp, 36-kb 30), chr19:38449792–38450687 lysed and luciferase activity was assessed using the Dual-Lucifer- (895 bp, 33-kb 30), chr19:38451402–38451655 (254 bp, 31-kb ase-Reporter Assay System (Promega) on a plate reader (Tecan). 30), chr19:38452815–38453385 (571 bp, 30-kb 30), chr19: Assays were performed at least in three independent experiments. 38463178–38463622 (360 bp, 19-kb 30), chr19:38464933– Statistical analysis was done with GraphPad Prism software using 38465373 (441 bp, 17-kb 30), chr19:38472626–38473335 two-tailed t tests. Data are depicted in column bar graphs plotting (730 bp, 9-kb 30), chr19: 38475809–38476523 (715 bp, 6-kb mean with SD values. 30), chr19:38478999–38479794 (796 bp, 3-kb 30), and 0 chr19:38485160–38486494 (1,334 bp, 1-kb 5 ). All HCNEs were Patient samples analyzed for transcription factor–binding sites using MatInspec- A cohort of 110 consecutive patients with AML diagnosed and tor software (Genomatix). treated at the University Hospital (Bern, Switzerland) between 0 0 A smaller construct pGL4.23-36kb3 (180 bp, 36-kb 3 ) 2002 and 2012 was included in this study. Informed consent from encoding the region flanking the putative p53-binding sites in all patients was obtained according to the Declaration of Helsinki, 0 the 36-kb 3 HCNE was created by internal SacI deletion in and the studies were approved by decisions of the local ethics 0 0 pGL4.23-36kb3 (592 bp, 36-kb 3 ). To test the p53-binding committee of Bern. Mutational screening for NMP1, CEBPA, and 0 sites in the 36-kb 3 HCNE, one or both sites were mutated by FLT3, as well as conventional karyotype analysis of at least 20 site-directed mutagenesis to create pGL4.23-36-kbM12 using metaphases were performed for each patient. Peripheral blood 0 primer pairs p53_mut_F (5 -CCTCTGAATTCCCGCATCTATC- mononuclear cells (PBMC) and bone marrow mononuclear cells

www.aacrjournals.org Clin Cancer Res; 22(3) February 1, 2016 747

Seipel et al.

(BMMC) were collected at the time of diagnosis before initiation branes were scanned and quantiﬁed on LI-COR Odyssey Infrared of treatment. Scanner (LI-COR Biotechnology).

Measurement of protein expression by ELISA Measurement of mRNA expression by qPCR fi Nuclear and cytoplasmic proteins were extracted from Ficoll- RNA was extracted from AML cells and quanti ed using separated mononucleated cells (Lymphoprep, Axis-Shield). Cell qPCR. The RNA extraction kit was supplied by Macherey-Nagel. pellets were lysed in 100 mL HEPES buffer containing 10 mmol/L Reverse transcription was done with MMLV-RT (Promega). Real-time PCR was performed on the ABI7500 Real-Time PCR KCl, 1.5 mmol/L MgCl2, 1% NP40, 1 mmol/L dithiothreitol (DTT), and 10 mmol/L PMSF. After centrifugation, supernatants Instrument using ABI universal master mix (Applied Biosys- fi containing cytoplasmic proteins were collected. Pelleted nuclei tems) and gene-speci c probes Hs00269972_s1 (CEBPA), were extracted in 100 mL HEPES buffer containing 420 mmol/L Hs00355782_m1 (CDKN1A), Hs00167918_m1 (GCSFR), and Hs02758991_g1 (GAPDH). Measurements for CEBPA were NaCl, 1.5 mmol/L MgCl2, 25% glycerol, 0.2 mmol/L EDTA, and 1 normalized with GAPDH values (ddCt relative quantitation). mmol/L PMSF. A280 nanodrop measurements were used to determine total protein concentration equivalents. Active p53 protein Assays were repeated in at least three independent experiments. Statistical analysis was done with GraphPad Prism software levels were assessed by colorimetric assay (OD450) using the TF- Detect Human p53 Activity Assay Kit (GeneCopoeia). CEBPA, using two-tailed t tests. Data are depicted in column bar graphs KLF4, MDM2, CUL9, XPO1, LMNA, and GAPDH protein levels plotting mean with SD values. were measured by colorimetric assay (OD450) using Sandwich ELISA kits supplied by USCN Life Science Inc. Measurements for Antibodies and flow cytometry nuclear proteins (p53, CEBPA, KLF4, MDM2, XPO1) were nor- Staining for cell surface markers was performed using CD11b- malized with nuclear LMNA, and measurements for cytoplasmic PE Cy7 (BioLegend) in FACS staining buffer for 30 minutes at 4 C proteins (MDM2, CUL9) were normalized with cytoplasmic and AnnexinV-FITC (ImmunoTools) in AnnexinV buffer for 15 GADPH levels. Statistical analysis was performed using the minutes followed by 7AAD cell viability staining (BD BioS- Mann–Whitney test on GraphPad Prism software. Data are ciences). For cell-cycle analysis, cells were fixed in 1% parafor- depicted as box and whiskers graphs plotting Tukey range. maldehyde (PFA)/PBS overnight at 4 C. Staining of intracellular antigens was achieved on fixed cells using Ki67-PE from eBioscience in permeabilization buffer for 30 minutes at 4C, AML cell lines followed by DNA staining using DAPI (Roche). Data were ana- OCI-AML3 (AML-M4, DNMT3Amut R882C, NPM1mut, lyzed using FlowJo and Kaluza software (Beckmann Coulter). TP53wt), MOLM-13 (AML-M5, t(9;11), FLT3-ITD, TP53wt), ML2 (AML-M4, t(6;11)(q27;q23), TP53wt), ME-1 (AML-M4eo, inv16, Chromatin immunoprecipitation TP53mut/null), and HL60 (AML-M2, MYCamp, TP53del/null) OCI-AML3 cells were treated with conventional induction cells were supplied by the Leibniz Institute DSMZ, German therapy, Nutlin-3A, or Leptomycin-B for 24 hours before fixation Collection of Microorganisms and Cell Cultures. AML cells were in paraformaldehyde. Chromatin was prepared using the CHIP-IT grown in RPMI-1640 (SIGMA-ALDRICH) supplemented with Express Enzymatic Kit (Active Motif Europe). Immunoprecipita- 20% FBS (Biochrom GmbH). ME-1 cells were transfected with tion was performed using rabbit-anti-human KLF4 (sc-2069) and pcDNA3.3_KLF4 (16) and pcDNA_TP53 expression plasmids mouse anti-human p53 (sc-126X, Santa-Cruz Biotechnology). using Nucleofector Kit V (Lonza). Isotype antibodies served as control. Input and Chromatin immunoprecipitation (ChIP) DNA samples were purified using Wizard Cytotoxicity assays PCR Cleanup System (Promega) and analyzed by qPCR using ABI AML cells were treated with the genotoxic compounds cytar- universal master mix (Applied Biosystems) and CEBPA gene– abine (SIGMA-ALDRICH) and idarubicin (Selleck Chemicals) specific probes. For the 36-kb 30 HCNE, the probes were 36-kb F in equimolar concentrations. To activate p53, cells were treated (50-GTACACCTCCTGGCCCTTGA-30), 36-kb R (50-GCTAAGC- with the MDM inhibitor Nutlin-3A (Tocris Bioscience) and the TTGGCTGGGGACGTGGGGCG-30) and 36-kb qPCR (50-FAM- exportin inhibitor Leptomycin-B (Biovision Inc.). Cell viability GCCTCT GCATGCCCGCACCTGT-TAMRA-30). For the 9-kb 30 was determined using the MTT-based in vitro toxicology assay HCNE, the probes were 9-kb F (50-CATCTCGAGACTAGCTGACT- (SIGMA-ALDRICH) with four repeat measurements per dosage. GATATGAATTGG-30), 9-kb R (50-GGGAGGTAGAGGA TGGTA- Data are depicted as scatter plots of mean and SD values. For GGA-30) and 9-kb qPCR (50-FAM-GGTCCAAAGGAGCTTTG- bone marrow samples, three samples each of normal and AML TCCCT-TAMRA-30). ChIP qPCR data were normalized applying bone marrow were analyzed. Statistical analysis was done on the percent input method. GraphPad Prism in grouped tables and significance calculated by two-way ANOVA. Results CEBPA activation by p53 and KLF4 in AML cells Measurement of protein levels by Western blotting In human AML cells, CEBPA, KLF4, and p53 are frequently Total protein extracts were prepared by RIPA lysis. One hundred deregulated. However, the extent of dysregulation, the molecular micrograms total protein extracts was PAGE-separated, trans- interactions, and mechanisms leading to dysregulation largely ferred to nitrocellulose membrane, and stained against mouse remain to be elucidated. We hypothesized that the CEBPA gene is anti-GAPDH (G8795; SIGMA-ALDRICH) and mouse anti-human a p53 and KLF4 downstream target in myeloid cells. Supporting p53 (sc-126; Santa-Cruz Biotechnology), followed by IRDye this hypothesis, we found that both p53 and KLF4 were able to 680RD goat anti-mouse IgG (LI-COR Biotechnology). Mem- induce cellular CEBPA mRNA expression in AML cells in a dose-

748 Clin Cancer Res; 22(3) February 1, 2016 Clinical Cancer Research

p53–KLF4–CEBPA Axis in AML

dependent manner (Fig. 1A). To study the activation of the CEBPA fected into HL60 cells. Similar to the endogenous CEBPA gene, we gene in a different genomic setting, we created a comprehensive found that the CEBPA-BAC transgene expression was induced by CEBPA-BAC transgene with a luciferase reporter which was trans- KLF4 and p53 (Fig. 1B) and repressed by AML1-ETO (17).

Figure 1. CEBPA gene transcriptional activation by KLF4 and p53. CEBPA mRNA quantitation in ME-1 AML cells induced by KLF4 and p53. A total of 10 mg plasmid DNA was nucleofected into 106 cells (A). CEBPA reporter activity in P391-HL60 cells induced by KLF4 and p53 and suppressed by AML1-ETO (B). The linearized CEBPA-BAC P391 is shown schematically. , signiﬁcant P values for t tests in subgroup versus pcDNA-transfected control. CEBPA gene HCNEs regulated luciferase reporter activity in H1299 cells induced by p53 and KLF4 (C). p53 induction of the CEBPA 36-kb 30HCNE-driven reporter is reduced in p53 single-site and lost in p53 double-site mutation (D). KLF4 induction of the CEBPA 9-kb 30 enhancer-driven reporter containing four putative KLF4-binding sites is reduced in the deletion constructs (D). p53 occupancy on the CEBPA 36-kb 30 HCNE (E) and KLF4 occupancy on the CEBPA 9-kb 30 HCNE (F) in OCI-AML3 cells treated with cytarabin/ idarubicin (CI; 100 nmol/L), with the MDM2 inhibitor Nutlin-3A (Nut; 2.5 mmol/L), or with the exportin inhibitor Leptomycin-B (LMB; 2 ng/mL) for 24 hours.

www.aacrjournals.org Clin Cancer Res; 22(3) February 1, 2016 749

Seipel et al.

Induction was consistently higher in stably transfected BAC Table 1. Patient characteristics transgenic cells, as they most likely carry multiple copies of the n (%) reporter gene. Total 110 (100) fi To define the p53 and KLF4 responsive sites in the CEBPA gene, FAB classi cation M0 9 (8.2) 11 HCNEs including the promoter of the CEBPA locus were M1 24 (21.8) cloned into luciferase reporter plasmids and tested for activation M2 15 (13.6) by p53 and KLF4. The HCNEs located 36-kb and 9-kb downstream M3 5 (4.5) of the CEBPA encoding gene sequence selectively mediated a M4 22 (20.0) strong response to p53 and KLF4, respectively (Fig. 1C). Direct M5 13 (11.8) activation of CEBPA gene expression by p53 was demonstrated by M6 1 (0.9) 0 M7 0 (0.0) complete loss of activation in the 36-kb 3 HCNE constructs Pathogenesis mutated in both half-sites of the predicted p53-binding site, with De novo 91 (82.7) minor activation still residing in the constructs with only one Prior MDS 14 (12.7) mutated half-site (Fig. 1D). To define the KLF4 responsive sites, Treatment-related 5 (4.5) deletion constructs of the 9-kb 30 HCNE were created and tested Molecular diagnosis for activation by KLF4. Maximum transcriptional activation was FLT3/NPM1/CEBPA wt 53 (48.2) FLT3-ITD 29 (26.4) observed in the 9-kb HCNE-driven reporter containing four NPM1 mut 21 (19.1) predicted KLF4 sites indicating that KLF4 proteins bound to FLT3-ITD and NPM1 mut 9 (8.2) multiple sites may cooperate. In the reporter constructs with only CEBPA mut 3 (2.7) two predicted sites, KLF4 activation was retained in the N-termi- Cytogenetic diagnosis nal region of the 9-kb 30 HCNE indicating that the major KLF4 Normal karyotype 41 (37.3) sites are located in this region (Fig. 1E). To confirm the function- 5, 5q 3 (2.7) 7, 7q 9 (8.2) ality of the presumed binding sites, leukemic cells were treated þ8 9 (8.2) with conventional cytotoxic drugs and non-genotoxic p53 acti- inv16 5 (4.5) vating compounds, and transcription factor occupancy was tested t(6;9) 1 (0.9) by ChIP. These experiments consistently identified p53 occupancy t(6;11) 2 (1.8) on the CEBPA 36-kb 30 HCNE (Fig. 1F), and KLF4 occupancy on t(8;21) 6 (5.5) the 9-kb 30 HCNE (Fig. 1G). t(15;17) 5 (4.5) Complex karyotype 12 (10.9) Risk categorya CEBPA KLF4 Reduced protein levels of and in AML cells Favorable 25 (22.7) The nuclear protein levels of the transcription factors CEBPA Intermediate 34 (30.9) and KLF4 were determined in samples from patients with AML at Adverse 51 (46.4) diagnosis and from healthy volunteers. Clinical characteristics of aFavorable risk AML comprised t(8;21), t(15;17), inv16, and normal karyotype AML the AML study cohort are summarized in Table 1 and in further with NPM1mut or CEBPAmut and FLT3-ITD negative; intermediate risk com- details in a Supplementary Table S1. We found that the median prised NK AML without any mutations or with FLT3-ITD and NPM1 mut; adverse risk AML comprised 5, 5q, 7, 7q, t(6;9), t(6;11), 11q23, NK with FLT3-ITD, or levels of nuclear CEBPA protein were reduced by 20% (P ¼ 0.03) complex karyotype abnormalities. in patients with AML versus PBMCs from healthy volunteers (Fig. 2A). The lowest CEBPA protein levels were found in FLT3- ITD and NPM1-mutant AML samples (P ¼ 0.002) and in normal P < 0.0001). In summary, CEBPA and KLF4 protein levels were þ CD34 selected hematopoietic stem cells (HSC; P < 0.0001), with concomitantly reduced in AML cells, with significant reduction in median reductions of 50% compared with normal PBMCs. With the least differentiated M0 subtype and in FLT3-ITD and FLT3/ no apparent differences in CEBPA levels between AML blasts in NPM1 double mutant AML. this cohort and normal HSCs, we investigated whether differing stages of differentiation might account for reduced CEBPA levels. Inactivation of the p53 pathway in AML cells As a surrogate for the degree of differentiation in leukemic cells, To further assess the p53 pathway dysregulation in AML cells, CEBPA levels were assessed in subgroups according to the FAB functionally active p53 levels as well as the nuclear protein levels classification. Following this approach, significantly reduced of MDM2 and XPO1/CRM1 and the cytoplasmic protein levels CEBPA levels were found in the minimally differentiated M0 and of CUL9/PARC were determined in samples from patients with M1 FAB subtypes as well as in AML that progressed from mye- AML and healthy volunteers. We detected a strong inactivation lodysplastic syndrome (MDS; Supplementary Fig. S1A). Consis- of p53 function in leukemic cells. The median levels of functional tently higher CEBPA levels were present in the more differentiated nuclear p53 protein were reduced by 84% in patients with AML AML subtypes M3 and M4. (P < 0.0001) versus healthy PBMCs, with a 98% reduction in FLT3- Likewise, KLF4 protein was present at very low levels in HSCs ITD and NPM1 mutant samples (P < 0.0001; Fig. 2C). Normal with a median 67% reduction compared with normal PBMCs (P < HSCs had a 99% reduced p53 level compared with normal PBMCs 0.0001). Median levels of nuclear KLF4 protein were reduced by (P ¼ 0.004). With respect to FAB classification, minimum p53 21% (P ¼ 0.3) in patients with AML compared with normal levels were found in M0 cells, followed by M4, M5, and progressed PBMCs (Fig. 2B), with a 52% reduction in FLT3-ITD and NPM1- MDS (Supplementary Fig. S1C), with significant reductions also mutant samples (P ¼ 0.04). With respect to FAB subtypes, in M1, M2, and therapy-related AML, whereas AML-M3 samples significant reduction of KLF4 levels was found in the M0 and expressed distinctly higher levels. Finally, we observed a correla- M4 subtypes (Supplementary Fig. S1B), and there was a correla- tion of p53 and KLF4 levels (Spearman r ¼þ0.29; P ¼ 0.008) and tion of CEBPA and KLF4 levels (Spearman coefficient r ¼þ0.472; of p53 and CEBPA levels (r ¼þ0.22, P ¼ 0.018).

750 Clin Cancer Res; 22(3) February 1, 2016 Clinical Cancer Research

p53–KLF4–CEBPA Axis in AML

Figure 2. Inactivation of CEBPA, KLF4, and p53 in AML blasts and normal HSCs. Relative quantitation of nuclear protein levels for CEBPA (A), KLF4 (B) p53 (C), and XPO1(D) normalized to Lamin-A (LMNA) and cytoplasmic protein levels for MDM2 (E) and CUL9 (F) normalized to GAPDH. AML and normal samples are depicted in white and grey boxplots, respectively. AML samples have been grouped into FLT-mutant (FLT3), NPM1-mutant (NPM), FLT3 and NPM1 double mutant (FLT3 þ NPM), þ and all AML (AML). Samples from healthy probands have been grouped into PBMC and normal HSCs (CD34 ). Whiskers indicate Tukey test range. , signiﬁcant P values using Mann–Whitney tests in AML cells versus normal PBMCs; #, signiﬁcant P values in AML cells versus normal HSCs.

We also assessed levels of the p53 inhibitor MDM2, the Restoring p53 and CEBPA function by conventional induction nuclear export protein XPO1/CRM1, and the cytoplasmic reten- therapy and by non-genotoxic p53 activators in AML cells. tion protein CUL9/PARC. The median levels of nuclear XPO1 On the basis of the results above indicating that p53 can directly protein were increased by 50% in AML samples (P ¼ 0.048) activate CEBPA gene expression, we investigated whether treating versus normal PBMCs, with a maximum (90%) increase in AML cells with p53 activators enhances CEBPA activity. Indeed, NPM1-mutant AML (P ¼ 0.002; Fig. 2D), which was 20% more we observed that CEBPA can be functionally induced by conven- than detected in normal HSCs (P ¼ 0.2). Again, higher XPO1 tional chemotherapy induction treatment as well as by non- levels were present in the least differentiated M0 subtype genotoxic p53 activators in OCI-AML3 cells (Fig. 3) but also in (Supplementary Fig. S1D). While nuclear MDM2 protein levels MOLM-13 and ML-2 cells (data not shown). While conventional were reduced by 20% in AML cells (P ¼ 0.4) versus normal genotoxic induction therapy induces p53 activation via the PBMCs (data not shown), the cytoplasmic MDM2 protein DNA damage response (18), Nutlin-3A works by blocking the levels were increased 6-fold in AML samples (P ¼ 0.0005) and function of the p53 inhibitor MDM2 (19), and Leptomycin-B 12-fold in FLT3 and NPM1 single mutant subtypes (P ¼ 0.003) by inhibiting the p53 nuclear export (20). Consequently, we which is twice above the levels in normal HSCs (P ¼ 0.4; Fig. tested conventional cytotoxic compounds, Nutlin-3A and Lep- 2E). Maximum cytoplasmic MDM2 levels were detected in tomycin-B, alone or in combination, in AML cells. Cells were samples from AML progressing from MDS, compared with collected after treatment for 24 hours. We observed an increase levels in normal HSCs (P ¼ 0.003; Supplementary Fig. S1E). of total cellular p53 protein (Fig. 3A), with highest levels in The median levels of cytoplasmic CUL9 protein were increased Nutlin-3A- and Leptomycin-B–treated cells. There was also an 5-fold in AML samples (P ¼ 0.004) versus normal PBMCs and induction of CEBPA gene expression (Fig. 3B) and transcrip- more than 7-fold in FLT3-ITD mutant samples (P ¼ 0.01) tional activation of the CEBPA target genes CDKN1A (Fig. 3C) whichis2-foldmorethanobservedinnormalHSCs(P ¼ and GCSFR (Fig. 3D), indicating induction of cell-cycle arrest 0.008; Fig. 2F). In summary, p53 appears to be broadly inacti- and granulocyte differentiation, respectively. In addition, these vated in AML cells, with lowest levels in the FLT3-ITD and treatments led to induction of the p53-inducible proapoptotic NPM1 mutant subsets and in the least differentiated M0 sub- gene NOXA (Fig. 3E), with highest levels following combina- type. At the same time, we observed induction of the cyto- tion treatment. Reﬂecting our measurement after 24 hours, it is plasmic protein levels of MDM2 and CUL9/PARC and a sub- possible that NOXA gene expression reached higher levels at stantial increase in the nuclear XPO1/CRM1 protein levels earlier time points (21), so the observed expression levels may across various AML subtypes. be at the tail-end of induction.

www.aacrjournals.org Clin Cancer Res; 22(3) February 1, 2016 751

Seipel et al.

Figure 3. Induction of CEBPA activity in OCI-AML3 cells by conventional induction treatment and small-molecule p53 activators. Induction of p53 protein expression (A), CEBPA mRNA (B), CDKN1A mRNA (C), GCSFR mRNA (D), NOXA mRNA (E), p53 DNA-binding activity (F), AnnexinV staining (G), CD11b staining (H), and cell-cycle arrest (J) in OCI-AML3 (NPM1 mut, p53 wt) cells treated with cytarabin/idarubicin (CI; 100 nmol/L), the MDM2 inhibitor Nutlin-3A (Nut; 2.5 mmol/L), and the exportin inhibitor Leptomycin-B (LMB; 2 ng/mL) for 24 hours. Cell-cycle stage were deﬁned as low in both Ki-67 and DAPI for G0, as high in Ki-67 and low in DAPI for G1, and as high in both Ki-67 and DAPI staining for G2–S–M.

752 Clin Cancer Res; 22(3) February 1, 2016 Clinical Cancer Research

p53–KLF4–CEBPA Axis in AML

A 110 The combination treatment was found to be more effective 100 than single-compound treatment, in particular with respect to 90 induction of CDKN1A and NOXA expression where maximum levels were detected following the combination treatment of 80 conventional therapy together with Nutlin-3A. The transcription- 70 al activation effects appeared to be additive for CEBPA and GSCFR 60 expression and synergistic for CDKN1A and NOXA gene expres- 50 DMSO sion, respectively. CI Along with the increase of total cellular p53 protein, we also

Cell viability (%) 40 Nut found elevated p53 DNA-binding activity (Fig. 3F), with highest 30 CI+Nut levels following Leptomycin-B and combination treatments. Apo- 20 ptosis and differentiation were induced as demonstrated by an 10 increase of Annexin V–positive cells (Fig. 3G) and CD11b induc- 01 23 tion (Fig. 3H). Moreover, the proportion of cells in cell-cycle arrest and exit was induced (Fig. 3J), whereas conventional induction B 110 treatment alone or in combination induced cell-cycle arrest in G2 100 –S and cell-cycle exit, p53 activation by Nutlin-3A and Leptomy- cin-B induced cell–cycle arrest in the G phase. Flow cytometric 90 1 data have been summarized in the Supplementary Fig. S2 and in 80 Supplementary Table S2. 70 Sensitivity of AML cells to conventional induction treatment 60 DMSO and p53 activators 50 In vitro cytotoxicity assays were performed to determine the CI 40 sensitivity of AML cells to conventional induction therapy, Cell viability (%) LMB 30 Nutlin-3A, and Leptomycin-B alone or in combination. All CI+LMB treatment regimens led to a dose-dependent loss of cell via- 20 bility (Fig. 4). While the combination of conventional induc- 10 tion therapy and Nutlin-3A had synergistic effects on cell 01 23 viability in all tested doses (Fig. 4A), any combination with C 110 Leptomycin-B appeared to have antagonistic effects at low dosages but synergistic effects at higher dosages (Fig. 3G and 100 H). Finally, we performed cytotoxicity assays using AML and 90 normal bone marrow samples, and we observed that normal 80 bone marrow cells appeared to be signiﬁcantly less sensitive to 70 the treatment regimens described above than AML bone marrow cells (Fig. 4D). 60 DMSO In summary, our data indicate that CEBPA can be functionally 50 restored by p53 activators in AML cells and that combination Nut treatment may be more effective than single agents. We present

Cell viability (%) 40 LMB evidence for a p53–KLF4–CEBPA pathway which is frequently 30 Nut+LMB inactivated in AML cells. While p53 function is strongly repressed, 20 the KLF4 and CEPBA protein levels are also substantially and 10 concomitantly reduced in AML blast cells. CEBPA gene expression 01 23 can be directly activated by p53 and KLF4, and we have evidence that the p53–KLF4–CEPBA axis may be restored in AML cells D 100 following treatment with genotoxic chemotherapy and inhibitors of MDM2 and XPO1. Supposedly, the p53–KLF4–CEBPA axis may also be restored in AML cells by treatment with FLT3 or Cullin 80 inhibitors as summarized in Fig. 5.

60 Figure 4. Sensitivity of AML cells to conventional induction treatment and small- 40 molecule p53 activators. Cell viability in OCI-AML3 cells treated for 24 hours with cytarabin/idarubicin (CI) and Nultin-3A (Nut; A), CI and Leptomycin-B Cell viability (%) 20 (LMB; B), or Nut and LMB (D), respectively. CI was tested at 0, 100, 200, and 300 nmol/L CI; Nutlin-3A at 0, 2.5, 5, 7.5; and LMB at 0, 2, 8, 32 ng/mL in dosages 0, 1, 2, and 3, respectively. Cell viability in normal and AML bone 0 marrow cells (D) treated with CI (100 nmol/L), with the MDM2 inhibitor Nut CI Nut3A LMB CI+Nut CI+LMB Nut+LMB (2.5 mmol/L), and with the exportin inhibitor LMB (2 ng/mL) for 24 hours. , ﬁ P Normal BM AML-BM signi cant values using two-way ANOVA tests in normal versus AML bone marrow samples.

www.aacrjournals.org Clin Cancer Res; 22(3) February 1, 2016 753

Seipel et al.

Figure 5. Schematic representation of the TP53–KLF4–CEBPA axis (green) impaired by AML-speciﬁc mutations (white) and altered protein levels of key regulators (red). Therapeutic strategies are indicated to reactivate the p53 tumor suppressor (blue) leading to induction of cell-cycle arrest (black). Compounds used in this study include the XPO1 inhibitor Leptomycin-B and the MDM2 inhibitor Nutlin-3A.

The tumor suppressor p53 is rarely mutated in AML blasts, with Discussion the exception of secondary therapy–related AML (28). p53 func- CEBPA gene expression is regulated by transcription factors tion, however, is inactivated in the vast majority of AML samples, including the Wnt signal mediator LEF1 (22), NFKB (23), AML1/ with a loss of activity of 84% in our AML cohort, again with lowest RUNX1 (24), and SPI1/PU.1 (25). Here, we identified the tumor levels in the undifferentiated subtype M0, and a 98% loss in FLT3- suppressors p53 and KLF4 as novel transcriptional activators of the ITD and NPM1-mutant samples. In summary, there appeared to CEBPA gene, thereby suggesting a core activation pathway con- be a correlated reduction in the levels of the three transcription sisting of p53, KLF4, and CEBPA that is blocked in AML. Inacti- factors p53, KLF4, and CEBPA in AML cells compared with normal vation of the p53–KLF4–CEBPA axis in leukemic cell leads to loss PBMCs, with overall lowest levels in the M0 subtype and in FLT3/ of cell-cycle control, inadequate response to apoptotic signals and NPM1 double mutant cells; however, these levels were still higher loss of myeloid differentiation. Moreover, we showed that CEBPA than those observed in normal HSCs. This does not exclude the gene expression can be restored in AML cells by treatment with possibility that the reduced levels of these transcription factors conventional chemotherapy or novel p53 activating compounds. may reflect but not necessarily cause the increased proliferation CEBPA restricts HSC renewal and can trigger myeloid differ- and blocked differentiation of AML cells. entiation (1). CEBPA gene mutations are present in 7% to 12% of In contrast to CEBPA, KLF4, and p53, the nuclear XPO1 and the newly diagnosed AML (26). In addition, there is an apparent loss cytoplasmic protein levels of the p53 modulators MDM2 and of detectable CEBPA protein in our AML cohort, with lowest levels CUL9 in AML cells were not only higher than in normal PBMCs in the undifferentiated subtypes M0 and M1 and in AML that but also higher than in normal HSCs, indicating that MDM2 progressed from preceding MDS, with particularly low levels in assisted p53 export from the nucleus by XPO1/CRM1 (14) and FLT3-ITD and NPM1 mutant samples. p53 retention in the cytoplasm by CUL9 (29) may be essential for Similar to CEBPA, KLF4 is a transcription factor restricting the differentiation block in AML cells. Notably, XPO1 levels were HSC renewal and it induces myelomonocytic differentiation highest in NPM1-mutant AML, with a 90% increase of XPO1/ (5, 6). KLF4 mediates the trans-activating effect of p53 on the CRM1, and 6- and 12-fold induction of cytoplasmic CUL9/PARC CDKN1A promoter (7). In AML cells, the KLF4 gene is targeted and MDM2, respectively, in our cohort. NPM1 was reported to by miRNA10a which is induced in the presence of mutant interact directly with the tumor suppressor p53 in the nucleus, and NPM1 function (8). KLF4 expression has also been shown to it enhances its stability and transcriptional activation function be activated by AML1/RUNX1 (27), whereas the fusion protein (15). Mutant NPM1 (NPMc), however, is exported to the cyto- AML1-ETO/RUNX1-MTG8 encoded by the t(8;21) transloca- plasm (30). NPMc can no longer stabilize p53 in the nucleus but tion inhibits both KLF4 and CEBPA gene expression (17). KLF4 may further inhibit p53 by cytoplasmic retention. Increased levels gene mutations have not been reported so far in patients with of nuclear XPO1 protein lead to increased cytoplasmic levels of AML; however, we found an apparent loss of detectable KLF4 NPMc and MDM2 (31, 14). Low levels of Mdm2 mediate mono- protein in our AML study cohort, again with lowest levels in the ubiquitination of p53 and result in nuclear export of p53, whereas undifferentiated subtype M0 and in FLT3-ITD and NPM1- high levels of Mdm2 mediate p53 polyubiquitination and pro- mutant AML, possibly due to overexpression of miR10A, Cdx2, teasome-mediated degradation in the nucleus. CUL9/PARC was or HDAC1 as suggested previously (8–11). identified to function similar to NPMc to retain p53 in the

754 Clin Cancer Res; 22(3) February 1, 2016 Clinical Cancer Research

p53–KLF4–CEBPA Axis in AML

cytoplasm (29). Notably, CUL9 levels were highest in FLT3- may also be restored by compounds inhibiting NPMc function in mutant and FLT3/ NPM1 double mutant AML, with a 7-fold myeloid leukemic cells (40), suggesting that multiple options are induction compared with normal PBMCs and twice the level available for therapeutic modulation of the p53–KLF4–CEBPA observed in normal HSCs. axis. In addition to MDM2, CUL9/PARC, and XPO1/CRM1, there may be other p53 inhibitors deregulated in AML. Indeed, MDM4, Disclosure of Potential Conflicts of Interest fi another RING nger ubiquitin ligase that interacts with MDM2 No potential conflicts of interest were disclosed. and inhibits p53, appears to be overexpressed in AML (32, 33). Moreover, it was shown that p53 is suppressed by overexpression Authors' Contributions of the transcriptional co-activator MN1 in some AML (34). Conception and design: K. Seipel, T. Pabst Finally, SIRT-1 also prevents p53 activation in AML cells (35). Development of methodology: T. Pabst Consequently, the list of p53 regulators and, thus, of potential Acquisition of data (provided animals, acquired and managed patients, drug targets steadily continues to grow (36). provided facilities, etc.): K. Seipel, M.T. Marques, M.-A. Bozzini, C. Meinken, Recently, potent and selective small-molecule inhibitors of T. Pabst MDM2 and XPO1/CRM1 have been identified (19, 37). These Analysis and interpretation of data (e.g., statistical analysis, biostatistics, studies have strengthened the concept that selective, non-gen- computational analysis): K. Seipel, B.U. Mueller, T. Pabst Writing, review, and/or revision of the manuscript: K. Seipel, M.T. Marques, otoxic p53 activation is a viable alternative to current cytotoxic fi M.-A. Bozzini, C. Meinken, B.U. Mueller, T. Pabst chemotherapy, and rst clinical studies with MDM2 inhibitors Administrative, technical, or material support (i.e., reporting or organizing have shown promising efficacy (36). Our data suggest that data, constructing databases): K. Seipel, T. Pabst treatment with p53 activators can reactivate the postulated Study supervision: K. Seipel, T. Pabst pathway leading to induction of CEBPA gene expression and concomitant inhibition of cell proliferation as well as induc- Acknowledgments tion of cellular differentiation toward the granulocytic lineages. The authors thank Marianne Messerli for compilation of CEBPA gene Indeed, we found that CEBPA as well as the CDKN1A and HCNEs, Sabine Hopner,€ and Lara Vanroth for technical assistance. They also GCSFR genes are upregulated in AML cells treated with Nutlin- thank Carol Prives for the p53 expression plasmid, and the pcDNA3.3_KLF4 3A and Leptomycin-B. plasmid was a gift of Derrick Rossi (Addgene plasmid #26815). CEBPA induction was also seen in leukemic cells treated with KPT-220, another XPO1/CRM1 inhibitor (37). This effect, how- Grant Support ever, may be forestalled in the presence of the AML1-ETO fusion This work was supported by a grant from the Swiss National Science protein in t(8;21) AML. Indeed, CEBPA gene expression was not Foundation (#310030-143584 to T. Pabst) and from the Swiss Cancer League induced by XPO1 inhibition in Kasumi-1 cells which carry the (KLS #2520-02-2010 to T. Pabst). The costs of publication of this article were defrayed in part by the payment of translocation t(8;21) (37). MDM2 and XPOI inhibitors are not the page charges. This article must therefore be hereby marked advertisement in only small-molecule compounds able to induce p53. Similar accordance with 18 U.S.C. Section 1734 solely to indicate this fact. effects may be observed in cells treated with FLT3 or Cullin inhibitors and with p53 activating compounds such as PRIMA- Received May 1, 2015; revised August 28, 2015; accepted September 20, 2015; 1 (38) and APR-246 (39). Moreover, KLF4 and CEBPA function published OnlineFirst September 25, 2015.

References 1. Rosenbauer F, Tenen DG. Transcription factors in myeloid development: 9. Faber K, Bullinger L, Ragu C, Garding A, Mertens D, Miller C, et al. CDX2- balancing differentiation with transformation. Nat Rev Immunol 2007; driven leukemogenesis involves KLF4 repression and deregulated PPARg 7:105–17. signaling. J Clin Invest 2013;123:299–314. 2. Nerlov C. C/EBPalpha mutations in acute myeloid leukaemias. Nat Rev 10. Rouhi A, Frohling€ S. Deregulation of the CDX2-KLF4 axis in acute myeloid Cancer 2004;4:394–400. leukemia and colon cancer. Oncotarget 2013;4:174–5. 3. Pabst T, Mueller BU. Complexity of CEBPA dysregulation in human acute 11. Huang Y, Chen J, Lu C, Han J, Wang G, Song C, et al. HDAC1 and Klf4 myeloid leukemia. Clin Cancer Res Off J Am Assoc Cancer Res 2009;15: interplay critically regulates human myeloid leukemia cell proliferation. 5303–7. Cell Death Dis 2014;5:e1491. 4. Yoon K, Smart RC. C/EBPalpha is a DNA damage-inducible p53-regulated 12. Lane DP. Cancer. p53, guardian of the genome. Nature 1992;358:15–6. mediator of the G1 checkpoint in keratinocytes. Mol Cell Biol 2004;24: 13. Attardi LD, DePinho RA. Conquering the complexity of p53. Nat Genet 10650–60. 2004;36:7–8. 5. Feinberg MW, Wara AK, Cao Z, Lebedeva MA, Rosenbauer F, Iwasaki H, 14. Xu L, Massague J. Nucleocytoplasmic shuttling of signal transducers. Nat et al. The Kruppel-like factor KLF4 is a critical regulator of monocyte Rev Mol Cell Biol 2004;5:209–19. differentiation. EMBO J 2007;26:4138–48. 15. Colombo E, Marine J-C, Danovi D, Falini B, Pelicci PG. Nucleophosmin 6. Cao Z, Sun X, Icli B, Wara AK, Feinberg MW. Role of Kruppel-like factors regulates the stability and transcriptional activity of p53. Nat Cell Biol in leukocyte development, function, and disease. Blood 2010;116: 2002;4:529–33. 4404–14. 16.WarrenL,ManosPD,AhfeldtT,LohY-H,LiH,LauF,etal.Highly 7. Zhang W, Geiman DE, Shields JM, Dang DT, Mahatan CS, Kaestner KH, efficient reprogramming to pluripotency and directed differentiation of et al. The gut-enriched Kruppel-like factor (Kruppel-like factor 4) mediates human cells with synthetic modified mRNA. Cell Stem Cell 2010; the transactivating effect of p53 on the p21WAF1/Cip1 promoter. J Biol 7:618–30. Chem 2000;275:18391–8. 17. Pabst T, Mueller BU, Harakawa N, Schoch C, Haferlach T, Behre G, et al. 8. Bryant A, Palma CA, Jayaswal V, Yang YW, Lutherborrow M, Ma DD. miR- AML1-ETO downregulates the granulocytic differentiation factor C/EBPal- 10a is aberrantly overexpressed in Nucleophosmin1 mutated acute mye- pha in t(8;21) myeloid leukemia. Nat Med 2001;7:444–51. loid leukaemia and its suppression induces cell death. Mol Cancer 18. yan AM, Anensen N, Bø TH, Stordrange L, Jonassen I, Bruserud ,etal. 2012;11:8. Genes of cell-cell interactions, chemotherapy detoxification and apoptosis

www.aacrjournals.org Clin Cancer Res; 22(3) February 1, 2016 755

Seipel et al.

are induced during chemotherapy of acute myeloid leukemia. BMC Cancer 29. Nikolaev AY, Li M, Puskas N, Qin J, Gu W. Parc: a cytoplasmic anchor for 2009;9:77–81. p53. Cell 2003;112:29–40. 19. Kojima K, Konopleva M, Samudio IJ, Shikami M, Cabreira-Hansen M, 30. Grisendi S, Mecucci C, Falini B, Pandolfi PP. Nucleophosmin and cancer. McQueen T, et al. MDM2 antagonists induce p53-dependent apoptosis in Nat Rev Cancer 2006;6:493–505. AML: implications for leukemia therapy. Blood 2005;106:3150–9. 31. Freedman DA, Levine AJ. Nuclear export is required for degradation of 20. Wolff B, Sanglier JJ, Wang Y. Leptomycin B is an inhibitor of nuclear export: endogenous p53 by MDM2 and human papillomavirus E6. Mol Cell Biol inhibition of nucleo-cytoplasmic translocation of the human immunode- 1998;18:7288–93. ficiency virus type 1 (HIV-1) Rev protein and Rev-dependent mRNA. Chem 32. Tan BX, Khoo KH, Lim TM, Lane DP. High Mdm4 levels suppress p53 Biol 1997;4:139–47. activity and enhance its half-life in acute myeloid leukaemia. Oncotarget 21. Oda E, Ohki R, Murasawa H, Nemoto J, Shibue T, Yamashita T, et al. Noxa, 2014;5:933–43. a BH3-Only Member of the Bcl-2 Family and Candidate Mediator of p53- 33. Li L, Tan Y, Chen X, Xu Z, Yang S, Ren F, et al. MDM4 Overexpressed in Acute Induced Apoptosis. Science 2000;288:1053–8. Myeloid Leukemia Patients with Complex Karyotype and Wild-Type TP53. 22.SkokowaJ,CarioG,UenalanM,SchambachA,GermeshausenM, PloS One 2014;9:e113088. Battmer K, et al. LEF-1 is crucial for neutrophil granulocytopoiesis and 34. Pardee TS. Overexpression of MN1 confers resistance to chemotherapy, its expression is severely reduced in congenital neutropenia. Nat Med accelerates leukemia onset, and suppresses p53 and Bim induction. PLoS 2006;12:1191–7. One 2012;7:e43185. 23. Wang D, Paz-Priel I, Friedman AD. NF-kappa B p50 regulates C/EBP alpha 35. Sasca D, H€ahnel PS, Szybinski J, Khawaja K, Kriege O, Pante SV, et al. SIRT1 expression and inflammatory cytokine-induced neutrophil production. prevents genotoxic stress-induced p53 activation in acute myeloid leuke- J Immunol 2009;182:5757–62. mia. Blood 2014;124:121–33. 24. Guo H, Ma O, Speck NA, Friedman AD. Runx1 deletion or dominant 36. Khoo KH, Hoe KK, Verma CS, Lane DP. Drugging the p53 pathway: inhibition reduces Cebpa transcription via conserved promoter and distal understanding the route to clinical efficacy. Nat Rev Drug Discov 2014; enhancer sites to favor monopoiesis over granulopoiesis. Blood 2012;119: 13:217–36. 4408–18. 37. Ranganathan P, Yu X, Na C, Santhanam R, Shacham S, Kauffman M, et al. 25. Burda P, Curik N, Kokavec J, Basova P, Mikulenkova D, Skoultchi AI, et al. Preclinical activity of a novel CRM1 inhibitor in acute myeloid leukemia. PU.1 activation relieves GATA-1-mediated repression of Cebpa and Cbfb Blood 2012;120:1765–73. during leukemia differentiation. Mol Cancer Res MCR 2009;7:1693–703. 38. Lambert JMR, Gorzov P, Veprintsev DB, Soderqvist€ M, Segerb€ack D, Berg- 26. Patel JP, Gonen€ M, Figueroa ME, Fernandez H, Sun Z, Racevskis J, et al. man J, et al. PRIMA-1 reactivates mutant p53 by covalent binding to the Prognostic relevance of integrated genetic profiling in acute myeloid core domain. Cancer Cell 2009;15:376–88. leukemia. N Engl J Med 2012;366:1079–89. 39. Ali D, Jonsson-Vides€ €ater K, Deneberg S, Bengtzen S, Nahi H, Paul C, et al. 27. Buchert M, Darido C, Lagerqvist E, Sedello A, Cazevieille C, Buchholz F, APR-246 exhibits anti-leukemic activity and synergism with conventional et al. The symplekin/ZONAB complex inhibits intestinal cell differentia- chemotherapeutic drugs in acute myeloid leukemia cells. Eur J Haematol tion by the repression of AML1/Runx1. Gastroenterology 2009;137:156– 2011;86:206–15. 64, 164.e1–3. 40. Balusu R, Fiskus W, Rao R, Chong DG, Nalluri S, Mudunuru U, et al. 28. Pedersen-Bjergaard J, Andersen MT, Andersen MK. Genetic Pathways in the Targeting levels or oligomerization of nucleophosmin 1 induces differen- Pathogenesis of Therapy-Related Myelodysplasia and Acute Myeloid Leu- tiation and loss of survival of human AML cells with mutant NPM1. Blood kemia. ASH Educ Program Book 2007;2007:392–7. 2011;118:3096–106.

756 Clin Cancer Res; 22(3) February 1, 2016 Clinical Cancer Research

Inactivation of the p53−KLF4−CEBPA Axis in Acute Myeloid Leukemia

Katja Seipel, Miguel Teixera Marques, Marie-Ange Bozzini, et al.

Clin Cancer Res 2016;22:746-756. Published OnlineFirst September 25, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-15-1054

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2015/09/25/1078-0432.CCR-15-1054.DC1

Cited articles This article cites 40 articles, 12 of which you can access for free at: http://clincancerres.aacrjournals.org/content/22/3/746.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/22/3/746.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/22/3/746. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.