Enforced Expression of NUP98-HOXA9 in Human CD34 Cells Enhances Stem Cell Proliferation

Research Article

Enforced Expression of NUP98-HOXA9 in Human CD34+ Cells Enhances Stem Cell Proliferation

Ki Y. Chung,1 Giovanni Morrone,4 Jan Jacob Schuringa,5 Magdalena Plasilova,2 Jae-Hung Shieh,2 Yue Zhang,3 Pengbo Zhou,3 and Malcolm A.S. Moore2

1Department of Medicine and 2Moore Laboratory, Cell Biology Program, Memorial Sloan-Kettering Cancer Center; 3Department of Pathology and Laboratory Medicine, Weill Medical College and Graduate School of Medical Sciences of Cornell University, New York, New York; 4Department of Experimental and Clinical Medicine ‘‘Gaetano Salvatore,’’ University of Catanzaro Magna Graecia, Catanzaro, Italy; and 5Department of Hematology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands

Abstract facilitate mRNA export from the nucleus (6–8). FG repeats of The t(7;11)(p15;p15) translocation, observed in acute myelog- NUP98 also bind and may facilitate interaction with the tran- enous leukemia and myelodysplastic syndrome, generates a scriptional coactivators cyclic AMP-responsive element binding chimeric gene where the 5¶ portion of the sequence encoding protein–binding protein (CBP) and p300 (9). This latter interaction the human nucleoporin NUP98 protein is fused to the 3¶ region may provide a link between NUP98 protein mobility and active of HOXA9. Here, we show that retroviral-mediated enforced transcription. expression of the NUP98-HOXA9 fusion protein in cord blood– Mounting evidence suggests that HOXA9 plays an important role derived CD34+ cells confers a proliferative advantage in both in normal hematopoiesis. HOXA9, HOXA7, and Meis1 are expressed + cytokine-stimulated suspension cultures and stromal cocul- in early self-renewing CD34 cells and down-regulate with differ- + ture. This advantage is reflected in the selective expansion of entiation (10, 11). In normal CD34 cells, HOXA9 is preferentially hematopoietic stem cells as measured in vitro by cobblestone expressed in a subfraction enriched for hematopoietic stem cells area–forming cell assays and in vivo by competitive repopu- (HSC). HOXA9 has been shown to bind DNA cooperatively with lation of nonobese diabetic/severe combined immunodefi- either PBX1 or Meis1, two other members of the homeodomain cient mice. NUP98-HOXA9 expression inhibited erythroid family (12). Meis1 was originally described as a HOX cofactor that progenitor differentiation and delayed neutrophil maturation alters HOX DNA-binding specificity and affinity and increases HOX in transduced progenitors but strongly enhanced their serial transcriptional activity (12). Targeted disruption of HOXA9 in mice replating efficiency. Analysis of the transcriptosome of leads to reduced numbers of progenitor cells and to a profound transduced cells revealed up-regulation of several homeobox defect in HSC (13). Conversely, enforced expression of HOXA9 genes of the A and B cluster as well as of Meis1 and Pim-1 and promoted proliferative expansion of HSC and progenitor cells and down-modulation of globin genes and of CAAT/enhancer subsequently inhibited their differentiation (14). These data binding protein A. The latter gene, when coexpressed with highlight the importance of precise control of HOXA9 protein NUP98-HOXA9, reversed the enhanced proliferation of trans- levels during hematopoiesis. We have shown that the Cullin-4A duced CD34+ cells. Unlike HOXA9, the NUP98-HOXA9 fusion (CUL-4A) ubiquitin ligase regulated HOXA9 protein levels by was protected from ubiquitination mediated by Cullin-4A and ubiquitination and degradation of the protein (15). Knockdown of subsequent proteasome-dependent degradation. The resulting CUL-4A by small interfering RNA in interleukin (IL)-3-dependent protein stabilization may contribute to the leukemogenic 32D myeloid cells enhanced their proliferation and blocked their activity of the fusion protein. (Cancer Res 2006; 66(24): 11781-91) differentiation in response to granulocyte colony-stimulating factor (G-CSF). Introduction HOXA9 is one of the top 20 genes distinguishing acute myelogenous leukemia (AML) from acute lymphocytic leukemia and NUP98 is a component of the nuclear pore complex (1). It is correlates with poor prognosis (16). HOXA9, HOXA7, and Meis1 mapped to 11p15.5 and fused to several distinct partners as a genes are coexpressed strongly in all but the acute promyelocytic consequence of leukemia-associated chromosomal translocations subset of AML (13, 17). HOXA9 behaves as an oncogene in leukemia that juxtapose the NH2 terminus of NUP98 to the COOH terminus of the partner gene (2, 3). Nine of the 21 reported partner genes are following mutations that induce its persistent expression or that of the homeobox (HOX) family (HOXA9, HOXA11, HOXA13, convert it into a persistent transcriptional activator. Leukemias HOXC11, HOXC13, HOXD11, and HOXD13) or HOX related (PMX1 associated with mixed lineage leukemia (MLL) gene translocations and PMX2; refs. 2, 3). NUP98 is present both at the nuclear pore show uniform activation of HOXA9, which may be the common pathway that unifies diverse initiating events in many myeloid complex and within the nuclear interior (4, 5). The NH2-terminal M9, GLEB, and FG domains of mammalian NUP98 bind to multiple leukemias (18). Although no individual HOX gene is essential, RNA export factors, including Kap, h2B, RaeI, and TAP, which Kumar et al. (19) proposed that the ‘‘HOX code,’’ minimally defined by the HOXA5-A9 cluster, is central to MLL leukemogenesis. HOXA9 overexpression can immortalize myeloid progenitors in vitro and inhibits some of their differentiation pathways (20, 21). Requests for reprints: Malcolm A.S. Moore, Moore Laboratory, Cell Biology When mice are transplanted with bone marrow cells overexpress- Program, Memorial Sloan-Kettering Cancer Center, New York, NY 10021. Phone: 212- ing HOXA9, they develop AML after 8 to 12 months, a period that 639-7090; Fax: 212-717-3618; E-mail: [email protected]. I2006 American Association for Cancer Research. is shortened to 50 to 60 days by Meis1 coexpression (20). Pineault doi:10.1158/0008-5472.CAN-06-0706 et al. (21) noted that Meis1 caused leukemic transformation of www.aacrjournals.org 11781 Cancer Res 2006; 66: (24). December 15, 2006

HOXA9-immortalized progenitors, which did not have long-term free QBSF-60 medium (Quality Biological, Gaithersburg, MD) supplemented repopulating capacity, supporting a two-step model of leukemo- with c-Kit ligand (100 ng/mL; Kirin, Gunma, Japan), Flt3 ligand (100 ng/mL; genesis. Imclone Systems, New York, NY), and thrombopoietin (100 ng/mL; Kirin). Clinically, NUP98-HOXA9 is associated with AML and myelodys- Retrovirus-containing supernatants were harvested from PG13 clones incubated in QBSF for 8 to 12 hours, supplemented with 100 ng/mL of plastic syndrome (MDS), both de novo and therapy related, and c-Kit ligand, Flt3 ligand, and thrombopoietin and 4 Ag/mL polybrene, filtered with blastic crisis of chronic myelogenous leukemia (CML; refs. through a 0.45-Am filter (Costar, Acton, MA), and applied immediately to the 2, 3, 22, 23). Expression of NUP98-HOXA9 in murine bone marrow CD34+ cells on retronectin-coated six-well plates (Takara, Otsu, Japan). Three resulted in a myeloproliferative disease in transplanted mice, with consecutive transduction rounds, every 8 to 12 hours, were done. neutrophil leukocytosis and extramedullary hematopoiesis, pro- Cell culture and stromal cell lines. Human umbilical cord blood was gressing to AML by 7 to 8 months (20, 23). Retroviral insertional kindly provided by the Cord Blood Bank subdivision of the New York Blood mutagenesis identified several cofactors that collaborate with Bank from healthy full-term pregnancies. Human CD34+ cells were selected NUP98-HOXA9 in leukemia progression, the most frequent being from the Ficoll-separated mononuclear cord blood cells using the Meis1. Collaboration between Meis1 and NUP98-HOXA9 reduced MiniMACS CD34 isolation kit (Miltenyi Biotech, Auburn, CA). The murine the latency of AML development to 4 to 5 months (20). Coexpres- MS-5 stromal cell line was kindly provided by Dr. Itoh (Department of a sion of NUP98-HOXA9 with the BCR-ABL fusion oncoprotein Biology, Niigata University, Niigata, Japan) and grown in -MEM (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum reduced the latency period to AML development even further to (FBS; Hyclone, Logan, UT). The AGM-S2 stromal line was developed from 21 days (24). Thus, it is clear that NUP98-HOXA9 plays a causative the aorta-gonad-mesonephros region of a 10.5-day mouse embryo (26). H29 role in the development of AML, although additional genetic or cells were cultured in DMEM (Life Technologies) supplemented with 10% epigenetic changes are necessary for progression to overt AML. FBS, penicillin-streptomycin, and 200 mmol/L glutamine. MS-5 stromal We have evaluated the biological effects of NUP98-HOXA9 by cocultures and long-term culture–initiating cell (LTC-IC) experiments were retrovirally transducing the most physiologically relevant cells, done in Gartner’s medium [a-MEM containing 12.5% FBS, 12.5% horse human CD34+ HSC. Enforced expression of NUP98-HOXA9 con- serum (Hyclone), penicillin and streptomycin, 200 mmol/L glutamine, ferred a proliferative advantage and enhanced HSC self-renewal, 57.2 Amol/L h2-mercaptoethanol (Fisher Scientific, Fair Lawn, NJ), and A evident in stromal coculture and in the competitive repopulation of 1 mol/L hydrocortisone (Sigma)]. nonobese diabetic/severe combined immunodeficient (NOD/SCID) Cytokine-stimulated suspension cultures were done in Iscove’s modified Dulbecco’s medium (Life Technologies) containing 20% FBS and 100 ng/mL mice. It also alters differentiation, inhibiting erythropoiesis relative of c-Kit ligand, Flt3 ligand, and thrombopoietin. Population doubling curves to myelopoiesis, delaying neutrophil maturation, and inhibiting were calculated based on weekly total nucleated cell counts and replating of myeloid colony formation in responses to G-CSF or granulocyte 1 Â 105 cells. macrophage colony-stimulating factor (GM-CSF). Several of these Colony-forming cell, cobblestone area–forming cell, and LTC-IC and changes may be attributed to altered transcriptional activity, and serial cobblestone area–forming cell assays. Colony assays were done in we implicate down-modulation of CAAT/enhancer binding protein triplicate in 35-mm plates using 1.2% methylcellulose (Dow, Niagra Falls, NY) a (C/EBPa) as playing a significant role. We further show that the 30% FBS, 57.2 Amol/L h2-mercaptoethanol, 2 mmol/L glutamine, 0.5 mmol/L HOX protein component of the fusion is protected from CUL-4A– hemin (Sigma), 20 ng/mL of IL-3, IL-6, G-CSF, and c-Kit ligand, and 6 units/ mediated ubiquitination, resulting in increased stability of HOXA9 mL erythropoietin. IL-3 and IL-6 were from PeproTech (Rocky Hill, NJ), G-CSF protein that may contribute to HSC transformation. was from Amgen (Thousand Oaks, CA), and erythropoietin was from Ortho Biotech (Bridgewater, NJ). In some assays, single cytokines (G-CSF, GM-CSF, and erythropoietin) were used. Colonies were scored 14 days after plating. Cobblestone area–forming cell (CAFC) assays were done as described Materials and Methods (27) by plating 1 Â 104 cord blood CD34+ cells onto MS-5 monolayers in Retroviral vectors and constructs. A murine stem cell virus–based T12.5 tissue culture flasks (Becton Dickinson, Franklin Lanes, NJ) in retroviral expression vector (MIGR1) was used for all experiments. The triplicate. Weekly demi-depopulations were done, with phenotypic analysis MIGR1 plasmid (kindly provided by Dr. Warren Pear, University of of nonadherent cells. Cobblestone areas were defined as groups of at least Pennsylvania, Philadelphia, PA) contains a multicloning site upstream of 10 phase-contrast dark cells tightly associated beneath the MS-5 monolayer an IRES2 element and the coding sequence for the enhanced green and were scored over the course of 5 weeks. fluorescent protein (EGFP) downstream of the latter (Fig. 1A). The cDNA Limiting dilution week 5 CAFC assays were done by plating sorted EGFP+ encoding human NUP98-HOXA9 was kindly provided by Dr. Nabeel Yaseen MIGR1-transduced and NUP98-HOXA9–transduced CD34+ cells in 24-well (Department of Pathology, Northwestern University, Chicago, IL) and cloned tissue culture plates containing MS-5 monolayers at five progressive into the MIGR1 retrovirus. To generate stable, high-titer retroviral dilutions with 12 replicate wells per dilution. Additional experiments were À packaging cell lines, H29 cells were transiently transfected with 10 Agof done using sorted EGFP+ CD34+/CD38 and CD34+/CD38+ to determine the vector DNA by calcium phosphate precipitation. After 72 hours, the H29 CAFC frequency. supernatant, containing vesicular stomatitis virus (VSV)-pseudotyped Secondary CAFC assays were done by trypsin harvesting week 5 adherent retrovirus, was used for cross-transduction of PG13 packaging cells in the cells and replating all cultures, or sorted EGFP+ cells, on fresh MS-5 presence of polybrene (8 Ag/mL; Sigma, St. Louis, MO). High-titer monolayers. retrovirus-producing PG13 subclones were selected by EGFP+ cell sorting. Western blots and morphologic analysis. Sorted EGFP+ MIGR1- pRRL-C/EBPa-ER lentiviral vectors that encoded a 4-hydroxytamoxifen producing and NUP98-HOXA9–producing PG13 cells (3 Â 105) were (4-OHT)-inducible C/EBPa-ER fusion were cloned by swapping the EcoRI centrifuged, and immunoblotting was done as described previously (25). fragment from C/EBPa-ER from MiNR1-C/EBPa-ER as described previously Antibodies (Upstate Technologies, Lake Placid, NY) to HOXA9 were used at (25) blunt into the SnaB1 site of pRRL-YFP. Lentiviral particles were a dilution of 1:1,000 and the secondary horseradish peroxidase–conjugated produced in 2.5 Â 106 293T human embryonic kidney cells that were rabbit anti-mouse antibodies at 1:200. transduced with 3 Ag pCMV y8.91, 0.7 Ag VSV-G, and 3 Ag pRRL-C/EBPa-ER Cell cytospins were stained with the Max-Gru¨nwald-Giemsa method for (pRRL-YFP was a kind gift from Dr. C. Baum, Department of Experimental morphologic evaluation. Hematology, Hannover Medical School, Hannover, Germany). Reverse transcription-PCR and microarray analysis. Reverse tran- Retroviral transduction protocol. Human cord blood–derived CD34+ scription-PCR (RT-PCR) was done using total RNA isolated from 0.5 Â 106 cells isolated as detailed below were prestimulated for 48 hours in serum- to 1.0 Â 106 sorted cells using the RNeasy kit and One-Step RT-PCR kit

Cancer Res 2006; 66: (24). December 15, 2006 11782 www.aacrjournals.org

(Qiagen, Valencia, CA) following the manufacturer’s protocol. All RT-PCR included in each transfection to monitor the transfection efficiency. The primers were from Sigma and are available on request. cells were harvested 48 hours after transfection and lysed in radio- For microarray analysis, 4 Ag of total RNAs were isolated from sorted immunoprecipitation assay buffer, immunoprecipitated with the anti-HA EGFP+ MIGR1 and NUP98-HOXA9 cells using the RNeasy kit (Qiagen), (12CA5) monoclonal antibody, and analyzed by Western blotting using labeled, and hybridized to Affymetrix (Santa Clara, CA) Human Genome antibodies against HA (HA11, Covance, Berkeley, CA), FLAG (M2; Sigma), U133A chips. Comparative gene expression profiles were determined and h-actin (Santa Cruz Biotechnology, Santa Cruz, CA). h null null between MIGR1-transduced and NUP98-HOXA9–transduced cells. A signif- Mice. NOD/SCID, NOD/SCID 2M , and NOD/SCID g2 mice (The icant difference in gene expression was defined as a fold change of z1.87, Jackson Laboratory, Bar Harbor, ME) were housed and maintained in with a detection P value of <0.05 and a signal value of >200. laminar flow cabinets under specific pathogen-free conditions in facilities Cell culture, plasmids, and transfection. HeLa cells stably expressing under an institute-approved animal protocol in accordance with the tetracycline-repressible rtTA (HtTA) were cultured in DMEM containing Principles of Laboratory Animal Care. Female mice (10–12 weeks old) 10% FCS, 0.5 mg/mL G418, and antibiotics. The plasmids expressing were sublethally irradiated with 3 cGy from a cesium g-radiation source hemagglutinin (HA)-tagged HOXA9, HA-tagged NUP98-HOXA9 (gift of J. van and inoculated i.v. or into the right femur with 1 Â 105 CD34+ cells Deursen, St. Jude Children’s Research Hospital, Memphis, TN), and FLAG- transduced with either MIGR1-EGFP or MIGR1-NUP-HOXA9-EGFP. tagged CUL-4A were transfected using Fugene 6 transfection reagent Animals were sacrificed at 5 to 7 weeks, and bone marrow, spleen, and (Roche, Chicago, IL). pGreenLantern (1 Ag) that expressed EGFP was also liver were removed for fluorescence-activated cell sorting (FACS) analysis.

Figure 1. Enforced expression of control MIGR1 and NUP98-HOXA9 in cord blood CD34+ cells. A, map of MIGR1 and NUP98-HOXA9 retroviruses and an immunoblot of whole-cell protein extracts from control (MIGR1) and NUP98-HOXA9 retroviral packaging cells withanti-HOXA9 antibody. B, transduction efficiencies determined by FACS analysis of EGFP expression using purified cord blood CD34+ cells prestimulated for 48 hours in QBSF-60 supplemented withc-Kit ligand, Flt3 ligand, and thrombopoietin (100 ng/mL) and then subjected to three transduction rounds as described in Materials and Methods. C, proliferative advantage of cord blood CD34+ cells transduced with NUP98-HOXA9 (NH) relative to MIGR1 control vector– transduced cells in nonsorted MS-5 stromal coculture. Left, advantage in total cell expansion; right, fold increase in percentage of EGFP+ cells over time in a comparison of MIGR1-EGFP+ control and NUP98-HOXA9-EGFP+ cells relative to nontransduced cells over time. D, EGFP+ cells were sorted from MIGR1 control or NUP98-HOXA9 CD34+ populations 24 hours after transduction and placed in MS-5 (left) or AGM-S2 (right) stromal coculture. Cultures were subjected to weekly demi-depopulation of suspension cells withFACS analysis for EGFP expression. Absolute expansion of EGFP+ cells in NUP98-HOXA9–transduced versus MIGR1 control cultures.

www.aacrjournals.org 11783 Cancer Res 2006; 66: (24). December 15, 2006

Flow cytometry. Phycoerythrin-conjugated antibodies were from definition of a colony) revealed >2-fold more clusters in NUP98- PharMingen (San Diego, CA). Cells were incubated with the relevant HOXA9 cultures relative to control. Replating of primary colonies j antibodies at 4 C for 1 hour, washed once with PBS/2% FCS, and analyzed from cultures stimulated with a combination of cytokines yielded on a FACSCalibur (Becton Dickinson). The data were acquired and analyzed secondary myeloid colonies but no BFU-E, with NUP98-HOXA9– with the CellQuest (Becton Dickinson) software. All cell sortings were done expressing primary colonies generating significantly more second- using MoFlo (DakoCytomation, Denver, CO). ary colonies than MIGR1 control colonies (P < 0.01; Fig. 3A). Tertiary passage was obtained only with NUP98-HOXA9 cells. This Results suggests that the expression of NUP98-HOXA9 resulted in Retroviral transduction of human cord blood CD34+ cells. acquisition of some measure of self-renewal by immature myeloid Gene transfer efficiencies into CD34+ cells were determined by progenitors. measuring the EGFP fluorescence 24 hours after the final round of Weekly evaluation of progenitor [colony-forming cell (CFC)] transduction. In multiple independent experiments, mean trans- production in stromal coculture showed that there was a duction efficiencies of 43 F 12% and 21 F 10% were obtained from consistently greater generation of CFC in the NUP98-HOXA9 control (MIGR1) and NUP98-HOXA9 viral supernatants, respective- cultures than in MIGR1 control, with a 4- to 5-fold greater colony- ly (Fig. 1B). Expression of the NUP98-HOXA9 fusion protein in the forming unit granulocyte, erythroid, monocyte, megakaryocyte NUP98-HOXA9 retroviral packaging cell line was verified by (CFU-GEMM), CFU-GM, and BFU-E output at week 5 (Fig. 3B). Western blotting (Fig. 1A). Enhanced CAFC formation by NUP98-HOXA9–expressing Selective in vitro proliferative advantage of NUP98-HOXA9– cells. NUP98-HOXA9–transduced CD34+ cells were evaluated for expressing cells. The proliferative potential of NUP98-HOXA9– HSC function using weeks 5 to 6 CAFC assay on MS-5 stromal cells transduced or MIGR1 control-transduced cells was evaluated (Fig. 3C). NUP98-HOXA9–expressing cells began to form cobble- in vitro in stromal coculture and cytokine-stimulated suspension stone areas by 1 week, and these progressively increased in number culture. In nonsorted cocultures, NUP98-HOXA9–expressing cells until approaching confluence by 3 to 4 weeks (Fig. 3C and D). showed a proliferative advantage over the course of 5 weeks as Limiting dilution assays (Fig. 3C and D) were used to evaluate determined by the increase of EGFP+ nonadherent progeny by weeks 5 to 6 CAFC frequency, and in three separate experiments, week 5 and a progressive increase in the percentage of total cells there were consistently greater numbers (4- to 8-fold more) of these expressing EGFP, whereas the MIGR1 EGFP+ population remained candidate HSC in NUP98-HOXA9 cultures compared with control. constant (Fig. 1C). We then evaluated EGFP+ sorted cells to remove CAFC comprised 2% to 4% of NUP98-HOXA9–transduced CD34+ any possible bystander effect of nontransduced cells. The cells compared with 0.3% to 0.5% for control cells. This reflects an proliferative advantage of NUP98-HOXA9 expression was still expansion of the primitive stem/progenitor cell pool. This was shown by total suspension cell production through week 5 of further confirmed by the replating capacity of week 5 NUP98- coculture on both MS-5 and AGM-S2 stroma (Fig. 1D). HOXA9 CAFCs in fresh MS-5 cocultures, showing generation of Cytokine-stimulated suspension cultures also confirmed the secondary (Fig. 3C) and tertiary (data not shown) week 5 CAFCs proliferative advantage of NUP98-HOXA9–expressing cells, with an (Fig. 3D). Week 5 cobblestone areas were harvested, and cells were 8- to 10-fold greater expansion of EGFP+ cells relative to control sorted by EGFP expression, characterized morphologically and (MIGR1) over 5 weeks (Fig. 2A). immunophenotypically, and plated in semisolid medium for Morphologically, nonadherent NUP98-HOXA9–expressing cells progenitor evaluation. The isolated EGFP+ cells seemed to be in culture were predominantly myeloblast/promyelocyte, lacking predominately myelomonocytic blasts (Fig. 3D), and their immu- the terminal band and segmented neutrophil differentiation noted nophenotype showed 10% CD34+ cells, 17% CD14+, and >90% in control cultures (Fig. 2B). However, by immunophenotype, these CD33+ (data not shown). Colony assays with sorted EGFP+ week 5 cells were indistinguishable from control MIGR1 cells in their cobblestone area–derived cells highlighted the following: (a) the expression of myeloid markers, CD33, and CD14 (data not shown). presence of a much higher number of progenitors in the NUP98- Defective myeloid and erythroid colony formation by HOXA9 cobblestone areas than in those derived by MIGR1- NUP98-HOXA9–expressing CD34+ cells. Clonogenic assays were transduced cells and (b) a significant proportion of CFCs were used to evaluate lineage commitment at the progenitor cell level BFU-Es and CFU-GEMMs in the NUP98-HOXA9 cobblestone areas following NUP98-HOXA9–transduction. Sorted NUP98-HOXA9– but not in the MIGR1 cobblestone areas. These data imply that expressing and MIGR1 control CD34+ cells cloned in the presence enforced expression of the fusion protein, particularly in the of IL-3, IL-6, G-CSF, c-Kit ligand, and erythropoietin (6 units) context of the hematopoietic microenvironment, maintains the showed comparable numbers of myeloid granulocyte-macrophage transduced cells in a more immature state. À colony-forming unit (CFU-GM), whereas blast-forming unit-ery- Experiments where sorted HSC-enriched CD34+/CD38 and throid (BFU-E) was significantly (P < 0.01) fewer in NUP98-HOXA9 progenitor-enriched CD34+/CD38+ NUP98-HOXA9–transduced cultures (Fig. 2C). A comparable reduction of BFU-E relative to populations were assayed in MS-5 cocultures revealed that only CFU-GM was noted when both NUP98-HOXA9–transduced CD34+, the former subset was able to generate weeks 5 to 6 CAFCs (data À CD38 lo and CD34+, CD38+ fractions were cultured separately not shown), suggesting that the proliferative advantage and (Fig. 2D). Significantly fewer BFU-E developed in NUP98-HOXA9 enhanced self-renewal of NUP98-HOXA9-expressing cells reflect cultures relative to control when erythropoietin was used at both an effect at the level of primitive HSCs. high and low concentrations, either alone or with c-Kit ligand Engraftment and expansion of NUP98-HOXA9–expressing null null (Fig. 2C). In contrast to results seen with a cocktail of cytokines, cells in NOD/SCID, NOD/SCID B2M , and NOD/SCID ;2 when cultures were stimulated by GM-CSF alone or by G-CSF at mice. To determine whether the in vitro proliferative advantage of both high and low concentrations, significantly fewer myeloid enforced NUP98-HOXA9 expression in CD34+ cells persisted in vivo, colonies developed (Fig. 2C). In G-CSF–stimulated cultures, analysis equivalent numbers of nonsorted NUP98-HOXA9–transduced and of the colony to cluster ratio (with >40 cells as the minimum MIGR1-transduced cells were transplanted by the i.v. route into

Cancer Res 2006; 66: (24). December 15, 2006 11784 www.aacrjournals.org

Figure 2. Selective expansion of NUP98-HOXA9-EGFP–expressing cells in liquid cytokine-stimulated cultures in nonsorted competitive repopulation experiments. A, fold expansion of EGFP+ cells over 5 weeks in cultures of MIGR1 control-transduced or NUP98-HOXA9– transduced cells. B, morphology of suspension cells at 3 weeks in MIGR1 control-transduced (left) and NUP98- HOXA9–transduced (middle) cultures. Note the significant (P = 0.001) reduction in mature granulocytes (bands and segmented forms) and predominance of immature myeloid lineage cells (myeloblasts and promyelocytes) in the NUP98-HOXA9 suspension cultures relative to control cultures (right). C, comparison of myeloid colony formation (CFU-GM)byEGFP+ MIGR1 (Control) and NUP98-HOXA9–transduced (NHA9) CD34+ cells. Note no significant difference when stimulated by a combination of cytokines (IL-3, IL-6, G-CSF, c-Kit ligand, and erythropoietin) but significantly (P < 0.05–0.01) fewer colonies developed from NUP98-HOXA9–transduced cells stimulated by single myeloid growthfactors (G-CSF and GM-CSF). Erythroid burst formation was significantly less (P < 0.05–0.01) with NUP98-HOXA9– transduced cells than with MIGR1 control cells irrespective of the cytokine combinations used, including a combination of cytokines (IL-3, IL-6, G-CSF, c-Kit ligand, and erythropoietin), erythropoietin alone at a high and low concentration, and erythropoietin with 20 ng/mL of c-Kit ligand. D, CD34+ cells transduced with NUP98-HOXA9 (NH) or control vector (MIGR1) were separated into EGFP+ CD34+ populations and further divided into CD38+ and CD38À/lo fractions that were plated for colony formation in the presence of IL-3, IL-6, G-CSF, c-Kit ligand, and erythropoietin. Erythroid, myeloid, and multilineage progenitors were distributed in all fractions, but the ratio of myeloid to erythroid colonies was significantly different when either the CD38+ or CD38À/lo fractions of MIGR1 (erythroid predominant) were compared withNUP98-HOXA9 (erythroid depleted relative to myeloid).

null null sublethally irradiated NOD/SCID h2M and g2 mice (nine CFU-GM colonies (data not shown). The greatest degree of human NUP98-HOXA9 and six MIGR1). In addition, four NOD/SCID engraftment (31–34%) was observed in NOD/SCID g2null mouse mice were engrafted locoregionally by intrafemoral injection of marrow (Fig. 4D), with 20% to 28% of the human cells expressing transduced cells. Engraftment was measured 5 to 7 weeks after the myelomonocytic marker CD14, 3% to 5% expressing transplantation by FACS analysis of human CD45 and EGFP the erythroid marker glycophorin A, and 6% to 9% expressing expression in bone marrow mononuclear cells (Fig. 4). In all three lymphoid markers CD7, CD8, or CD4. The expression of EGFP in models, human hematopoietic engraftment was obtained in bone CD8 T cells (43% in control and 19–29% in the NUP98-HOXA9 marrow (2–42%). A selective in vivo proliferative advantage of engrafted mice) indicated that expression of the fusion protein was NUP98-HOXA9–expressing cells relative to the nontransduced cells not incompatible with T lymphocyte differentiation, although this coinjected was revealed by a mean 3.3-fold increase in the percen- pathway of differentiation may be quantitatively impaired (Fig. 4D). tage of human CD45 cells recovered at 5 to 6 weeks that expressed Validation of in vitro phenotype and discovery of potential EGFP relative to the EGFP percentage in the input population downstream targets by microarray analysis. Comparative (9–15%; Fig. 4A). This was significantly greater (P < 0.028) than the microarray analysis was undertaken on RNA isolated from freshly average expansion of EGFP cells seen in control mice (1.6-fold). sorted CD34+ NUP98-HOXA9–expressing cells and MIGR1 control The proliferative advantage of NUP98-HOXA9–expressing cells was cells (Table 1). Consistent with several other published studies also accompanied by extramedullary engraftment in the spleen and (3, 19, 28), expression of NUP98-HOXA9 up-regulated the expression liver. These mononuclear EGFP+ cells were predominantly myelo- of many HOX genes, including endogenous HOXA5, HOXA6, HOXA7, monocytic blasts (Fig. 4C). Additional colony assays from engrafted HOXA9, and HOXB5, important myeloid transcription factors, such CD34+-immunopurified mononuclear cells from bone marrow of as Meis1 and AML1, as well as the transcripts encoding the CD44 the engrafted mice revealed a predominance of CFU-GEMM and surface membrane protein, the Pim-1 serine/threonine kinase, and www.aacrjournals.org 11785 Cancer Res 2006; 66: (24). December 15, 2006

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2006 American Association for Cancer Research. Cancer Research hepatic leukemia factor (HLF). The level of mRNA for C/EBPa was nancies due to proviral integration (33). Pim-1 is a downstream decreased in NUP98-HOXA9+ cells. The up-regulation of endoge- target gene of signal transducer and activator of transcription 5 nous HOX genes and Pim-1 as well as the down-regulation of (Stat5) and of constitutively activated Flt3 mutants in AML, and its C/EBPa were validated by RT-PCR in transduced CD34+ cells in expression inhibits apoptosis and increases hematopoietic cell independent experiments (Fig. 5A). HOXA9, HOXA7, HOXA6, proliferation and survival independent of growth factors (34–36). HOXA5, HOXB5, and Meis1 have all been implicated in leukemo- Up-regulation of HLF is consistent with the observed enhanced genesis (19–23). Up-regulation of HOXA5 may also play a role in the HSC proliferation because it has been reported that ectopic suppression of erythropoiesis as reflected in the down-regulation of expression of HLF enhances HSC engraftment and inhibits several globin genes (Table 1). Enforced expression of HOXA5 in apoptosis (37). human CD34+ cells has been shown to preferentially support Validation of the role of C/EBPA down-modulation in myeloid differentiation, with a reduced frequency of erythroid NUP98-HOXA9 transformation. The down-regulation of C/EBPa progenitors (BFU-E; refs. 29, 30). HOXB5 up-regulation has also by NUP98-HOXA9–expressing CD34+ cells seems to be consistent been implicated in leukemia (31), and high expression levels of with the observed primitive myelomonocytic phenotype of their HOXB5 are associated with poor outcome (32). The Pim-1 serine/ progeny and their delayed neutrophil maturation. C/EBPa plays an À À threonine kinase was highly up-regulated in NUP98-HOXA9– important role in myeloid differentiation as shown in C/EBPa / expressing cells. This was first identified as a gene whose mice that display an accumulation of myeloblasts with myeloid expression was elevated in various murine hematopoietic malig- maturation arrest (38) and as recently reported in human CD34+

Figure 3. Stem cell and progenitor cell expansion in NUP98-HOXA9–transduced cord blood CD34+ cell cultures. A, colony assays of week 0 sorted MIGR1 and NUP98-HOXA9–transduced (NHA9) CD34+ cells in the presence of IL-3, IL-6, G-CSF, c-Kit ligand, and erythropoietin. The primary day 14 colony contents of each35-mm Petri dish(140–180 CFC) were dissociated and passaged into secondary 1 ml cultures (in triplicate) and scored for secondary colonies after a further 14 days. Process repeated for a tertiary passage. Columns, mean of three experiments; bars, SE. B, left, total weekly progenitor cell production in MS-5 stromal cocultures of sorted EGFP+ cells from MIGR1 control-transduced or NUP98-HOXA9–transduced (NH) CD34+ populations. Note the 3- to 4-fold greater number of progenitors recovered at week 5 in the NUP98-HOXA9 cultures. Right, morphology of EGFP+ CFC generated at 5 weeks of stromal coculture. Note that, in addition to the 4.5-fold greater numbers of CFC in NUP98-HOXA9 cultures, multilineage (CFU-GEMM) and erythroid (BFU-E) progenitors were still produced, in contrast to the exclusive CFU-GM production in MIGR1 control cultures. C, cobblestone area formation (arrows)in MS-5 stromal cocultures at 1 and 2 weeks following addition of 2,000 sorted NUP98-HOXA9 (NHA9) or MIGR1 control CD34+ cells (left). Limiting dilution week 5 CAFC assay of NUP98-HOXA9 (NAH9) or MIGR1 control CD34+ cells on MS-5 stroma (0–800 input CD34+ cells per well) in control or transduced MS-5 cocultures (right). D, CAFC assay. Number of cobblestone areas formed on MS-5 stroma between 1 and 3 weeks following coculture with2,000 sorted EGFP + CD34+ cells transduced with NUP98-HOXA9 (NHA9) or MIGR1 control. Columns, mean; bars, SE. Data on week 5 cobblestone areas obtained by limiting dilution assay. Left, representative experiment. One third of week 5 cobblestone areas derived from NUP98-HOXA9–transduced CD34+ cells were able to generate secondary week 5 cobblestone areas following repassage onto freshMS-5, whereasno control cobblestone area had this capacity (right).

Cancer Res 2006; 66: (24). December 15, 2006 11786 www.aacrjournals.org

Figure 4. A, comparison of the hematopoietic differentiation and expansion of EGFP+ cells following injection of 105 unsorted NUP98/HOXA9-transduced or null null MIGR1-transduced cells i.v. into irradiated NOD/SCID h2M mice (D), i.v. into NOD/SCID g2 mice (o), or into the right femur of NOD/SCID mice (5). Transduction efficiency ranged from 9% to 20%. Data expressed as fold change in percentage of EGFP+ cells in engrafted bone marrow harvested at weeks 5 to 7 (2–100%) to the EGFP % in the input CD34+ cell population (9–20%). NUP98-HOXA9 mean increase = 3.31-fold; MIGR1 fold increase = 1.63-fold; P = 0.028. B, evaluation null + of bone marrow harvested at week 7 after i.v. inoculation of NOD/SCID h2M mice withnonsorted, transduced CD34 cells. Prominent engraftment of human CD45+ cells identified by FACS analysis, withEGFP + NUP98-HOXA9 cells representing f50% of the engrafted human CD45+ cells. C, cytospin of FACS-sorted NUP98-HOXA9–expressing cells showing a primitive myelomonocytic blast morphology similar to isolated week 5 CAFC cells. D, human hematopoietic engraftment in bone marrow of NOD/SCID g2null mice 5 weeks after i.v. injection of 105 NUP98-HOXA9–transduced (9.6% EGFP+) or MIGR1 control-transduced (14.1% EGFP+) CD34+ cells. Human CD45+ cells comprised 31% to 34% of the bone marrow with an additional 3% to 5% human glycophorin A+, CD45+ erythroid cells detected. EGFP+ cells comprised 30% of the MIGR1 CD45 population (a 2.1-fold increase in % EGFP+ cells). Left, EGFP+ NUP98-HOXA9 cells comprised 29% and 52% of the CD45 population (a 3- and 5.4-fold increase in % EGFP+ cells). Lymphoid differentiation was obtained, with 2.5% to 5.5% CD7+, CD45+ cells and 1.5% to 5% CD8+, CD45+ T cells detected in the marrow, with 43% of the control expressing MIGR1-EGFP and 19% and 29% of the experimental mice expressing NUP98-HOXA9-EGFP (right). cells expressing a dominant-negative C/EBPa (39). The clinical expressing NUP98-HOXA9 and C/EBPa-ER, or cells expressing C/ relevance is highlighted by reports of dominant-negative C/EBPa EBPa-ER alone, in the absence of 4-OHT, exhibited a relative mutations in AML patients leading to a block in myeloid expansion comparable with untransduced CD34+ cells. However, in differentiation (40). We have recently reported that down- the presence of 4-OHT, there was a very significant suppression of modulation of C/EBPa by STAT5 in CD34+ cells was a prerequisite proliferation relative to control or NUP98-HOXA9 cultures within 1 for STAT5-induced effects on enhancement of HSC self-renewal to 2 weeks (Fig. 5). and inhibition of myeloid differentiation (25). A 4-OHT-inducible Enhanced stability of the NUP98-HOXA9 fusion protein C/EBPa-ER protein was coexpressed with an activated STAT5 compared with wild-type HOXA9 may be mediated by mutant in CD34+ cells using a lentiviral/retroviral approach, and decreased sensitivity to CUL-4A–dependent ubiquitination. reexpression of C/EBPa restored myeloid differentiation and The CUL-4A ubiquitination machinery has recently been shown to inhibited HSC proliferation. We have used a comparable approach mediate ubiquitin-dependent proteolysis of HOXA9 (15). To to evaluate the relevance of C/EBPa down-modulation in the examine whether the stability of the NUP98-HOXA9 chimera was observed NUP98-HOXA9 phenotype. CD34+ cells were transduced also subjected to regulation by CUL-4A, HeLa cells were transiently with NUP98-HOXA9-EGFP,withC/EBPa-ER-YFP,orwithboth transfected with both HA-HOXA9 and HA-NUP98-HOXA9 together vectors, plated on MS-5 stroma, and cultured for 5 weeks in the with increasing doses of CUL-4A. As shown in Fig. 5C, HA-HOXA9 presence or absence of 4-OHT. The cultures were evaluated by protein was readily degraded in a CUL-4A dose-dependent manner weekly FACS analysis for cells expressing YFP, EGFP, or both. as described previously (15). In striking contrast, the steady-state Measurement of the relative expansion of the transduced cells levels of HA-NUP98-HOXA9 protein were not affected significantly reveals that NUP98-HOXA9–expressing cells expanded to a greater by the increased expression of CUL-4A. Pulse-chase analysis degree than control nontransduced cells by weeks 3 to 5 in the confirmed the difference in protein half-lives (Fig. 5D). These presence or absence of 4-OHT (Fig. 5). Double-transduced cells results indicate that the NUP98-HOXA9 chimera is resistant www.aacrjournals.org 11787 Cancer Res 2006; 66: (24). December 15, 2006

Table 1. Differential gene expression in cord blood CD34+ cells expressing NUP98-HOXA9 relative to CD34+ cells transduced withtheMIGR1 control vector

Gene Unigene accession Gene symbol Fold change

Fibroblast growth factor 18 Hs.49585 FGF18 21.11 Urotensin 2 Hs.162200 UTS2 13.00 FBJ murine osteosarcoma viral oncogene homologue B Hs.75678 FOSB 12.13 Homeobox A9 Hs.127428 HOXA9 6.50 Homeobox A5 Hs.37034 HOXA5 4.92 CD44-Human CD44 antigen precursor Hs.285173 CD44 4.92 CD69 (p60, early T-cell activation antigen) Hs.82401 CD69 4.59 H2B histone family, member A Hs.352109 H2BFA 4.00 Piccolo (presynaptic cytomatrix protein) Hs.12376 PCLO 4.00 Phospholipase A2, group IVA Hs.211587 PLA2G4A 3.73 Cell division cycle 2-like 5 Hs.59498 CDC2L5 3.73 DEAD/H box polypeptide 17 (72 kDa) Hs.349121 DDX17 3.73 Homeobox B5 Hs.22554 HOXB5 3.48 KIAA1110 protein Hs.22479 KIAA1110 3.48 Serine/arginine repetitive matrix 1 Hs.18192 SRRM1 3.25 Rabconnectin-3 Hs.13264 RC3 3.25 Protease, serine 23 Hs.25338 SPUVE 3.03 Pre-B-cell leukemia transcription factor Hs.294101 PBX3 3.03 Pim-1 oncogene Hs.81170 PIM1 3.03 RNA-binding protein gene with multiple splicing Hs.80248 RBPMS 3.03 Testis specific protein 1 Hs.2042 TPX1 3.03 Heterogeneous nuclear ribonucleoprotein A1 Hs.249495 HNRPA1 3.03 Insulin-like growth factor binding protein 4 Hs.1516 IGFBP4 2.83 Spinocerebellar ataxia 7 Hs.108447 SCA7 2.83 Protein kinase, lysine deficient 1 Hs.184592 PRKWNK1 2.83 Bullous pemphigoid antigen 1 Hs.198689 BPAG1 2.83 Sex determining region Y (SRY)-box 4 Hs.83484 SOX4 2.83 Polymerase II polypeptide B Hs.296014 POLR2B 2.83 H2A histone family member O Hs.795 H2AFO 2.83 Nuclear receptor interacting protein 1 Hs.155017 NR1P1 2.64 Hepatic leukemia factor Hs.250692 HLF 2.64 Interleukin-1, a Hs.1722 IL1A 2.64 Homeobox A6 Hs.248073 HOXA6 2.64 Signal transducer and activator of transcription 1 Hs.21486 STAT1 2.64 Protein tyrosine phosphatase, nonreceptor type 13 Hs.211595 PTPN13 2.46 Hematopoietic PBX-interacting protein Hs.8068 HPIP 2.46 Eukaryotic translation initiation factor 4 g, 3 Hs.25732 EIF4G3 2.30 Rab6 GTPase activating protein Hs.55099 GAPCENA 2.30 Myeloid/lymphoid or mixed-lineage leukemia; t(3) Hs.404 MLLT3 2.30 Leptin receptor Hs.226627 LEPR 2.30 IFN-stimulated transcription factor 3, g Hs.1706 ISGF3G 2.14 3-Hydroxy-3-methylglutaryl-CoenzymeA synthase 1 Hs.77910 HMGCS1 2.14 Adducin 3 (g) Hs.324470 ADD3 2.14 Homeobox A7 Hs.355540 HOXA7 2.14 Runt-related transcription factor 1 Hs.129914 RUNX1 2.14 Aldehyde dehydrogenase 1 family, member A1 Hs.76392 ALDH1A1 2.14 Meis1 Hs.170177 MEIS1 2.00 Interleukin-7 Hs.72927 IL7 2.00 Dual specificity phosphatase 1 Hs.171695 DUSP1 1.87 BCL2-associated X protein Hs.159428 BAX 1.87 CCAAT/enhancer binding protein a Hs.76171 CEBPA À1.87 CD9 (p24) Hs.1244 CD9 À2.00 Platelet/endothelial cell adhesion molecule (CD31 antigen) Hs.78146 PECAM1 À2.00 Metallothionein 1G Hs.334409 MT1G À2.14 Metallothionein 1X Hs.278462 MT1X À2.14 Metallothionein 2A Hs.118786 MT2A À2.14 S100 calcium binding protein A8 Hs.100000 S100A8 À2.64

(Continued on the following page)

Cancer Res 2006; 66: (24). December 15, 2006 11788 www.aacrjournals.org

Table 1. Differential gene expression in cord blood CD34+ cells expressing NUP98-HOXA9 relative to CD34+ cells transduced withtheMIGR1 control vector (Cont’d)

Gene Unigene accession Gene symbol Fold change

Wingless-type MMTV integration site family, member 5B Hs.306051 WNT5B À2.64 Hemoglobin, a1 Hs.272572 HBA1 À2.83 Hemoglobin, a2 Hs.347939 HBA2 À3.03 Hemoglobin, e1 Hs.117848 HBE1 À3.25 Hemoglobin, gA Hs.266959 HBG1 À4.00 Hemoglobin, y Hs.36977 HBD À4.59 Hemoglobin, h Hs.155376 HBB À4.92 Platelet factor 4 Hs.81564 PF4 À8.00

NOTE: Cord blood CD34+ cells were prestimulated for 48 hours in QBSF-60 medium with c-Kit ligand, Flt3 ligand, and thrombopoietin (100 ng/mL of each) followed by three transduction rounds in the next 48 hours on retronectin. GFP+ cells were sorted and total RNA was isolated and used to hybridize gene arrays. Data shown are the comparison of GFP+ NUP98-HOXA9 and GFP+ MIGR1 cells. A change in gene expression was only considered significant when the fold change was >1.87, with a statistical P value of <0.05 and a signal value of >200.

to CUL-4A–mediated proteolysis. The increased stability of NUP98- transcription factor that binds DNA through the HOXA9 homeo- HOXA9 fusion protein may therefore further contribute to the domain and activates transcription through the NUP98 FG repeat generation of the phenotype observed. domain (9). Consistent with this notion, NUP98-HOXA9 is capable of binding to a HOX DNA recognition site (9). Because homeodomain Discussion proteins play important roles in both normal hematopoiesis and leukemogenesis (3), an aberrant homeodomain-containing The establishment of a retroviral transduction model to enforce + protein, such as NUP98-HOXA9, may cause leukemia by interfering expression of the NUP98-HOXA9 gene in human CD34 cells has permitted evaluation of its action in primary, nonleukemic human with the transcriptional programs of hematopoietic differentiation HSC and progenitor cells. We show that NUP98-HOXA9 expression and proliferation. In this study and an earlier one (42), we obtained confers a selective growth advantage on CD34+ cells as measured evidence that NUP98-HOXA9 acts as a strong transcriptional by in vitro expansion in stromal and cytokine-stimulated culture activator. When NUP98-HOXA9 was transduced into the blastic systems. Evidence that the fusion gene acts to enhance HSC CML cell line K562, we identified altered expression of 102 genes (42). proliferation was provided in two in vitro HSC assay systems. In the Of these, 92 were up-regulated whereas only 10 were down-regulated, LTC-IC assay, we showed an increased number of progenitors indicating that NUP98-HOXA9 acts primarily as a transcriptional activator. Our comparative microarray analysis of NUP98-HOXA9– generated by week 5 of culture. In the week 5 CAFC assay, we + showed a 4- to 8-fold expansion of CAFC and f33% of these could transduced CD34 cells revealed 157 genes increased and 79 genes be passaged for an additional 5 to 10 weeks whereas none of the decreased. There was little overlap in the gene targets in the two control CAFC could. The transduced CAFC resided in the CD34+, systems, indicating that NUP98-HOXA9 has a very different profile À CD38 /lo fraction, traditionally enriched for HSC, and not in the of target gene modification when expressed in normal HSC and HSC-depleted, progenitor-enriched CD34+, CD38+ fraction. In the progenitors than when expressed in BCR/ABL transformed leukemic + NOD/SCID assay for week 5+ human hematopoietic engraftment, cells. Of the genes up-regulated by NUP98-HOXA9 in normal CD34 traditionally considered as a HSC assay, NUP98-HOXA9–transduced cells, it was striking that the HOXA5-A9 cluster together with Meis1, cells were significantly more effective in competing against both implicated in leukemogenesis, were up-regulated. Defective coinjected nontransduced cells than were MIGR1 control cells. erythroid differentiation could be attributed to the down-modulation The data showing increased secondary and tertiary myeloid colony of several globin genes and to up-regulation of HOXA5 (29, 30). formation by NUP98-HOXA9-transduced primary colonies relative C/EBPa is a key regulator of granulopoiesis, and C/EBPa-binding to control colonies indicate that more committed progenitor cells sites have been identified in promoters of various myeloid restricted can acquire some measure of self-renewal. The differential genes, such as the G-CSF receptor and neutrophil elastase (25, 43). microarray analysis also supports our contention that NUP98- Conditional C/EBPa knockout in mice blocked the differentiation of HOXA9 enhances HSC proliferation because there is up-regulation the common myeloid progenitor with myeloblast accumulation in of genes, such as HOXA9 and HLF, implicated in HSC self-renewal marrow, absence of neutrophils, and enhancement of HSC (3, 37) and identified as up-regulated in leukemic stem cells (41). In competitive repopulating capacity and self-renewal (38). We have addition to proliferative defects, the fusion gene alters differenti- recently shown that down-modulation of C/EBPa is a prerequisite ation. Erythroid differentiation is suppressed, terminal neutrophil for STAT5-induced effects on HSC self-renewal and myelopoiesis maturation is inhibited, and clonogenic response to G-CSF or GM- (25). A comparable role for down-modulation of C/EBPa could CSF is blunted. In vivo, NUP98-HOXA9–transduced cells can explain the NUP98-HOXA9–induced myeloid maturation defects differentiate to CD8+ T cells but possibly with reduced efficiency. seen in long-term cultures, the impairment in G-CSF-induced and Two main theories have been developed to explain the role of GM-CSF-induced myeloid colony formation, and the proliferative NUP98-HOXA9 in leukemogenesis. In the first, the transcription advantage of transduced cells in vitro and in vivo. Support for activation model, NUP98-HOXA9 is thought to act as an aberrant this view was provided by our observation that overexpression of www.aacrjournals.org 11789 Cancer Res 2006; 66: (24). December 15, 2006

C/EBPa alone, or coexpressed with NUP98-HOXA9, profoundly A second hypothesis for the role of NUP98-HOXA9 in leukemo- suppressed CD34+ cell proliferation, possibly reflecting an impair- genesis is based on disruption of NUP98 function in the nuclear ment in HSC proliferation and enhancement of myeloid differenti- import of transcription factors or in the export of mRNAs. This ation comparable with our observations in the STAT5 system (25). in turn could disrupt pathways critical to both HSC proliferation The transforming potential of NUP98-HOXA9 may in part be and differentiation. We have obtained preliminary data showing related to its enhanced stability, with 3-fold longer half-life of the impaired assembly of the nuclear pore complex with formation fusion HOXA9 protein relative to wild-type HOXA9, due to the of distinct NUP98 colocalization bodies in NUP98-HOXA9– resistance of the fusion HOXA9 protein to CUL-4A-mediated transduced CD34+ cells (47). NUP98 maps to 11p15.5 and in AML ubiquitination and degradation. Enhanced NUP98-HOXA9 stability and MDS, among all regions studied, this region showed the would allow for more robust expression of endogenous HOX genes. highest frequency of loss of heterozygosity (LOH; 47%; ref. 48). We The ubiquitination site on the first helix of HOXA9 is present in the have also noted a high frequency of LOH specifically at the NUP98 fusion protein, but it is possible that some form of steric hindrance locus in AML (47). mediated by the NUP98 component, dimerization of NUP-HOXA9 It has been proposed that one class of leukemogenic mutant or its binding to wild-type NUP98, blocks interaction with CUL-4A. confers a proliferative or survival advantage, whereas a second An alternative mechanism may involve a CBP/p300-mediated class primarily interferes with differentiation and subsequent acetylation of lysine residues required for ubiquitination. p53 and apoptosis (49). Our studies point to multiple roles for the its structural and functional homologue p73 are protected from NUP98-HOXA9 fusion protein in the dysregulation of several ubiquitination by such an acetylation process (44, 45). Competition critical pathways that influence HSC and progenitor self-renewal, between ubiquitination and acetylation of overlapping lysine residues proliferation, differentiation, and interaction with bone marrow constitutes a novel mechanism regulating protein stability (46). stroma. However, additional mutations would seem to be needed

Figure 5. A, RT-PCR validation of selected genes differentially expressed in the comparative microarray analysis of NUP98-HOXA9 versus MIGR1. Total RNA was isolated and used as described in Materials and Methods. B, demonstration that overexpression of C/EBPa counteracts the proliferative advantage of CD34+ cells expressing NUP98-HOXA9. CD34+ cells were transduced withretroviral/lentiviral vectors expressing YFP and a 4-OHT–inducible C/EBPa-ER protein, EGFP and NUP98-HOXA9, or both vectors. Cells were cultured on MS-5 stroma in the presence or absence of 0.5 Amol/L 4-OHT and evaluated weekly for the number of cells expressing EGFP, YFP, or bothlabels. C, NUP98-HOXA9 is resistant to degradation by CUL-4A. Ectopic expression of CUL-4A reduced the steady-state levels of HOXA9 (HA9) but not NUP98-HOXA9 (NHA9). HA-tagged NUP98-HOXA9 (HA-NHA9), HOXA9, and FLAG-tagged CUL-4A (in Ag as indicated) were cotransfected in HtTA cells, and their steady-state levels were measured by Western blotting. D, pulse-chase analysis of the half-lives of coexpressed NUP98-HOXA9 (NHA9) and native HOXA9 (HA9) in response to ectopic CUL-4A expression in HtTA cells.

Cancer Res 2006; 66: (24). December 15, 2006 11790 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2006 American Association for Cancer Research. NUP98-HOXA9 Transformation for progression to acute leukemia. Transduction of normal CD34 Grant support: Leukemia and Lymphoma Society of America Specialized Center of Research and Gar Reichman Fund of the Cancer Research Institute cells with NUP98-HOXA9 and additional leukemogenic genes (e.g., (M.A.S. Moore), Italian Ministry for University and Research Interlink and Italian Flt3 internal tandem duplication; ref. 35) provides a model for Association for Cancer Research (G. Morrone), European Molecular Biology testing this concept in a human system. Organization grant ALTF-412-2001 (J.J. Schuringa), and National Cancer Institute grants CA118085 and CA092210 (P. Zhou). P. Zhou is a Scholar of the Leukemia and Lymphoma Society. Acknowledgments 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 Received 3/29/2006; revised 9/5/2006; accepted 10/20/2006. with 18 U.S.C. Section 1734 solely to indicate this fact.

References nisms of cooperativity with Hoxa9 in myeloid leukemia. 33. Selten G, Cuypers HT, Boelens W, et al. The primary Proc Natl Acad Sci U S A 2001;98:13120–5. structure of the putative oncogene pim-1 shows 1. Cronshaw JM, Krutchinsky AN, Zhang W, Chait BT, 18. Ayton PM, Cleary ML. Transformation of myeloid extensive homology with protein kinases. Cell 1986;46: Matunis MJ. Proteomic analysis of the mammalian progenitors by MLL oncoproteins is dependent on 603–11. nuclear pore complex. J Cell Biol 2002;158:915–27. Hoxa7 and Hoxa9. Genes Dev 2003;17:2298–307. 34. Kim KT, Baird K, Ahn JY, et al. Pim-1 is up-regulated 2. Slape C, Aplan PD. The role of NUP98 gene fusions in 19. Kumar AR, Hudson WA, Chen W, Nishiuchi R, Yao Q, by constitutively activated FLT3 and plays a role in hematologic malignancy. Leuk Lymphoma 2004;45: Kersey JH. Hoxa9 influences the phenotype but not the FLT3-mediated cell survival. Blood 2005;105:1759–67. 1341–50. incidence of Mll-AF9 fusion gene leukemia. Blood 2004; 35. Chung KY, Morrone G, Schuringa JJ, Wong B, Dorn 3. Moore M. Converging pathways in leukemogenesis 103:1823–8. DC, Moore MA. Enforced expression of an Flt3 internal and stem cell self-renewal. Exp Hematol 2005;33:719–37. 20. Kroon E, Thorsteinsdottir U, Mayotte N, Nakamura T, tandem duplication in human CD34+ cells confers 4. Enninga J, Levy DE, Blobel G, Fontoura BM. Role of Sauvageau G. NUP98-HOXA9 expression in hemopoietic properties of self-renewal and enhanced erythropoiesis. nucleoporin induction in releasing an mRNA nuclear stem cells induces chronic and acute myeloid leukemias Blood 2005;105:77–84. export block. Science 2002;295:1523–5. in mice. EMBO J 2001;20:350–61. 36. Schuringa JJ, Chung KY, Morrone G, Moore MA. 5. Fontoura BM, Blobel G, Matunis MJ. A conserved 21. Pineault N, Abramovich C, Humphries RK. Trans- Constitutive activation of STAT5A promotes human biogenesis pathway for nucleoporins: proteolytic pro- plantable cell lines generated with NUP98-Hox fusion hematopoietic stem cell self-renewal and erythroid cessing of a 186-kilodalton precursor generates Nup98 genes undergo leukemic progression by Meis1 inde- differentiation. J Exp Med 2004;200:623–35. and the novel nucleoporin, Nup96. J Cell Biol 1999;144: pendent of its binding to DNA. Leukemia 2005;19: 37. Shojaei F, Trowbridge J, Gallacher L, et al. Hierarchi- 1097–112. 636–43. cal and ontogenic positions serve to define the 6. Shamsher MK, Ploski J, Radu A. Karyopherin h2B 22. Borrow J, Shearman AM, Stanton VP, Jr., et al. The molecular basis of human hematopoietic stem cell participates in mRNA export from the nucleus. Proc t(7;11)(p15;p15) translocation in acute myeloid leukae- behavior. Dev Cell 2005;8:651–63. Natl Acad Sci U S A 2002;99:14195–9. mia fuses the genes for nucleoporin NUP98 and class I 38. Zhang P, Iwasaki-Arai J, Iwasaki H, et al. Enhance- 7. Pritchard CE, Fornerod M, Kasper LH, van Deursen homeoprotein HOXA9. Nat Genet 1996;12:159–67. ment of hematopoietic stem cell repopulating capacity JM. RAE1 is a shuttling mRNA export factor that binds 23. Iwasaki M, Kuwata T, Yamazaki Y, et al. Identification and self-renewal in the absence of the transcription to a GLEBS-like NUP98 motif at the nuclear pore of cooperative genes for NUP98-HOXA9 in myeloid factor C/EBPa. Immunity 2004;21:853–63. complex through multiple domains. J Cell Biol 1999;145: leukemogenesis using a mouse model. Blood 2005;105: 237–54. 39. Schwieger M, Lohler J, Fischer M, Herwig U, Tenen 784–93. 8. Blevins MB, Smith AM, Phillips EM, Powers MA. DG, Stocking C. A dominant-negative mutant of C/ 24. Mayotte N, Roy DC, Yao J, Kroon E, Sauvageau G. a Complex formation among the RNA export proteins EBP , associated with acute myeloid leukemias, inhibits Oncogenic interaction between BCR-ABL and NUP98- Nup98, Rae1/Gle2, and TAP. J Biol Chem 2003;278: differentiation of myeloid and erythroid progenitors of HOXA9 demonstrated by the use of an in vitro purging 20979–88. man but not mouse. Blood 2004;103:2744–52. culture system. Blood 2002;100:4177–84. 9. Kasper LH, Brindle PK, Schnabel CA, Pritchard CE, 40. Frohling S, Dohner H. Disruption of C/EBPa function Cleary ML, van Deursen JM. CREB binding protein 25. Wierenga AT, Schepers H, Moore MA, Vellenga E, in acute myeloid leukemia. N Engl J Med 2004;351:2370–2. interacts with nucleoporin-specific FG repeats that Schuringa JJ. STAT5-induced self-renewal and impaired 41. Moore MAS, Dorn DC, Schuringa JJ, Chung KY, activate transcription and mediate NUP98-HOXA9 myelopoiesis of human hematopoietic stem/progenitor Morrone G. Constitutive activation of Flt3 and STAT5A a oncogenicity. Mol Cell Biol 1999;19:764–76. cells involves downmodulation of C/EBP . Blood 2006; enhance self-renewal and alters differentiation in 10. Lawrence HJ, Rozenfeld S, Cruz C, et al. Frequent co- 107:4326–33. normal and leukemic hematopoietic stem cells. Exp expression of the HOXA9 and MEIS1 homeobox genes in 26. Weisel KC, Moore MA. Genetic and functional Hematol. In press 2006. human myeloid leukemias. Leukemia 1999;13:1993–9. characterization of isolated stromal cell lines from the 42. Ghannam G, Takeda A, Camarata T, Moore MA, Viale 11. Kawagoe H, Humphries RK, Blair A, Sutherland HJ, aorta-gonado-mesonephros region. Ann N Y Acad Sci A, Yaseen NR. The oncogene Nup98-HOXA9 induces Hogge DE. Expression of HOX genes, HOX cofactors, 2005;1044:51–9. gene transcription in myeloid cells. J Biol Chem 2004; and MLL in phenotypically and functionally defined 27. Jo DY, Rafii S, Hamada T, Moore MA. Chemotaxis of 279:866–75. subpopulations of leukemic and normal human hema- primitive hematopoietic cells in response to stromal 43. Radomska HS, Huettner CS, Zhang P, et al. CCAAT/ topoietic cells. Leukemia 1999;13:687–98. cell-derived factor-1. J Clin Invest 2000;105:101–11. enhancer binding protein a is a regulatory switch 12. Shen WF, Rozenfeld S, Kwong A, Kom ves LG, 28. Calvo KR, Sykes DB, Pasillas MP, Kamps MP. Nup98- sufficient for induction of granulocyte development Lawrence HJ, Largman C. HOXA9 forms triple com- HoxA9 immortalizes myeloid progenitors, enforces from bipotential myeloid progenitors. Mol Cell Biol plexes with PBX2 and MEIS1 in myeloid cells. Mol Cell expression of Hoxa9, Hoxa7, and Meis1, and alters 1998;18:4301–14. Biol 1999;19:3051–61. cytokine-specific responses in a manner similar to that 44. Pearson M, Carbone R, Sebastiani C, et al. PML 13. Lawrence HJ, Helgason CD, Sauvageau G, et al. Mice induced by retroviral co-expression of Hoxa9 and Meis1. regulates p53 acetylation and premature senescence bearing a targeted interruption of the homeobox gene Oncogene 2002;21:4247–56. induced by oncogenic Ras. Nature 2000;406:207–10. HOXA9 have defects in myeloid, erythroid, and lym- 29. Crooks GM, Fuller J, Petersen D, et al. Constitutive 45. Bernassola F, Salomoni P, Oberst A, et al. Ubiquitin- phoid hematopoiesis. Blood 1997;89:1922–30. HOXA5 expression inhibits erythropoiesis and increases dependent degradation of p73 is inhibited by PML. J Exp 14. Thorsteinsdottir U, Sauvageau G, Humphries RK. Hox myelopoiesis from human hematopoietic progenitors. Med 2004;199:1545–57. homeobox genes as regulators of normal and leukemic Blood 1999;94:519–28. 46. Gronroos E, Hellman U, Heldin CH, Ericsson J. hematopoiesis. Hematol Oncol Clin North Am 1997;11: 30. Fuller JF, McAdara J, Yaron Y, Sakaguchi M, Fraser JK, Control of Smad7 stability by competition between 1221–37. Gasson JC. Characterization of HOX gene expression acetylation and ubiquitination. Mol Cell 2002;10:483–93. 15. Zhang Y, Morrone G, Zhang J, et al. CUL-4A during myelopoiesis: role of HOX A5 in lineage 47. Plasilova M, Chung K, Shieh JH, et al. Depletion of stimulates ubiquitylation and degradation of the HOXA9 commitment and maturation. Blood 1999;93:3391–400. NUP98: a common leukemogenic mechanism in AML? homeodomain protein. EMBO J 2003;22:6057–67. 31. Neben K, Schnittger S, Brors B, et al. Distinct gene Blood 2005;106:197a. 16. Golub TR, Slonim DK, Tamayo P, et al. Molecular expression patterns associated with FLT3- and NRAS- 48. Krskova-Honzatkova L, Cermak J, Sajdova J, Stary J, classification of cancer: class discovery and class activating mutations in acute myeloid leukemia with Sedlacek P, Sieglova Z. Loss of heterozygosity and prediction by gene expression monitoring. Science normal karyotype. Oncogene 2005;24:1580–8. heterogeneity of its appearance and persisting in the 1999;286:531–7. 32. Bullinger L, Dohner K, Bair E, et al. Use of gene- course of acute myeloid leukemia and myelodysplastic 17. Calvo KR, Knoepfler PS, Sykes DB, Pasillas MP, expression profiling to identify prognostic subclasses in syndromes. Leuk Res 2001;25:45–53. Kamps MP. Meis1a suppresses differentiation by G-CSF adult acute myeloid leukemia. N Engl J Med 2004;350: 49. Gilliland DG, Griffin JD. The roles of FLT3 in and promotes proliferation by SCF: potential mecha- 1605–16. hematopoiesis and leukemia. Blood 2002;100:1532–42.

www.aacrjournals.org 11791 Cancer Res 2006; 66: (24). December 15, 2006

Downloaded from cancerres.aacrjournals.org on September 28, 2021. © 2006 American Association for Cancer Research. Enforced Expression of NUP98-HOXA9 in Human CD34+ Cells Enhances Stem Cell Proliferation

Ki Y. Chung, Giovanni Morrone, Jan Jacob Schuringa, et al.

Cancer Res 2006;66:11781-11791.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/24/11781

Cited articles This article cites 48 articles, 28 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/24/11781.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/24/11781.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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/66/24/11781. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.