Leukemia (2001) 15, 362–370  2001 Nature Publishing Group All rights reserved 0887-6924/01 $15.00 www.nature.com/leu The leukemogenic E2a-Pbx1 induces expression of the putative N- and target NDRG1 in Ba/F3 cells MN Rutherford1, GRL Bayly1, BP Matthews1, T Okuda2, WMN Dinjens3, H Kondoh2 and DP LeBrun1

1Department of Pathology, Queen’s University, Kingston, Ontario K7L 3N6, Canada; 2Institute for Molecular and Cellular Biology, Osaka University, Osaka, Japan; and 3Department of Pathology, Erasmus University Rotterdam, Rotterdam, The Netherlands

The chimeric transcription factor E2a-Pbx1 is expressed as a It encodes the Pbx1a (long form) and Pbx1b (short result of the 1;19 chromosomal translocation in some 5% of form), which are generated through alternative pre-mRNA cases of pediatric acute lymphoblastic leukemia. We investi- gated the biological and transcriptional consequences of for- splicing. Pbx1 contains a homeodomain, a DNA binding ced expression of E2a-Pbx1 in the interleukin-3 (IL-3) depen- module that, like the bHLH, defines a large family of transcrip- dent, bone marrow-derived cell line Ba/F3. We show that forced tion factor proteins.8 The affinity and sequence specificity of expression of E2a-Pbx1 induces apoptosis in Ba/F3 cells with- DNA binding by Pbx1 are markedly enhanced through coop- out apparent effects on cell cycle progression. This pro-apop- erative interactions with Hox proteins, a large subfamily of totic effect is enhanced on cytokine deprivation. Furthermore, using cDNA representational difference analysis (RDA), we homeodomain-containing proteins with critical roles in show that these cellular effects are associated with marked determining segment identity during embryonic develop- induction of the gene NDRG1, which was previously identified ment.9–11 Gene targeting of PBX1 in mice is associated with as a target of transcriptional repression by N-myc and induc- embryonic lethality and homeotic developmental abnormali- tion by the tumor suppressor p53. We identify a portion ties despite unperturbed expression levels of key Hox pro- of the NDRG1 promoter capable of mediating transcriptional 12 induction by E2a-Pbx1 and show that NDRG1 is also induced teins. Therefore, the currently available evidence supports a on simple IL-3 deprivation of BaF3 cells. Although we show that model in which Pbx1, by modulating DNA binding by Hox E2a-Pbx1 induction of NDRG1 is not impaired as a result of proteins, serves as a cofactor in the transcriptional regulation targeting p53 using HPV E6, and therefore does not appear to of developmentally important target . be p53-dependent, our results overall are consistent with the E2a-Pbx1 is capable of inducing neoplastic transformation notion that induction of NDRG1 by E2a-Pbx1 may represent 13–15 part of an apoptotic or cytostatic cellular response to oncogene in several experimental assays. It contains the N-terminal activation. Leukemia (2001) 15, 362–370. two thirds of E2a, including the two transcriptional activation Keywords: leukemia; E2a-Pbx1; Ndr1; RTP; NDRG1; represen- domains, and most of Pbx1, including the DNA binding tational difference analysis homeodomain.1 These structural features and some functional data support a model in which E2a-Pbx1 contributes to neo- plastic transformation through disrupting mechanisms that Introduction normally regulate the transcription of particular genes. Con- sistent with this notion, E2a-Pbx1 can induce the transcription In roughly 5% of cases of pediatric acute lymphoblastic leuke- of reporter constructs bearing Pbx1 binding elements, or of mia (ALL), a somatic chromosomal translocation results in the several endogenous genes identified with the use of differen- ′ juxtaposition of 5 elements from the gene E2A, on chromo- tial cloning strategies in fibroblasts or lymphoid cells.16–21 some 19, with a large portion of the gene PBX1, on chromo- Here, we have characterized the cellular and molecular some 1. In this manner, t(1;19) creates a fusion gene, E2A- consequences of enforced expression of E2a-Pbx1 in the PBX1, that encodes a chimeric transcription factor protein, interleukin-3 (IL-3) dependent, bone marrow-derived cell line 1 E2a-Pbx1. Ba/F3. E2a-Pbx1 enhances the tendency of Ba/F3 cells to The products of E2A belong to a large family of transcription undergo apoptotic death, particularly in the presence of sub- factors defined by the presence of a basic helix–loop–helix optimal concentrations of IL-3, without consistent effects on (bHLH) module. Whereas the two amphipathic helices within cell cycle progression. Using the differential cloning tech- the bHLH mediate homo- or heterodimerization with other nique of cDNA representational difference analysis (RDA), we HLH-containing proteins, the adjacent group of basic amino have demonstrated marked induction of the gene NDRG1 acids mediates sequence-specific binding to DNA.2 Alterna- downstream of E2a-Pbx1. We demonstrate that DNA regulat- tive splicing of the E2a transcript leads to the expression of ory elements derived from the NDRG1 promoter can mediate two proteins, E12 and E47, that differ minimally, exclusively within the bHLH domain. Mapping studies in which portions reporter gene induction by E2a-Pbx1, suggesting that induc- of E2a have been fused with heterologous DNA binding tion appears to occur, at least in part, at the transcriptional domains have defined two, separable transcriptional acti- level. NDRG1 is transcriptionally repressed downstream of N- vation domains called AD1 and AD2.3,4 Mice targeted for E2A Myc and is a known target of transcriptional induction by the are completely devoid of B-lymphocytes, manifest profoundly tumor suppressor protein p53. However, we show that disordered T-lymphopoiesis, and tend to develop T-lineage NDRG1 induction by E2a-Pbx1 appears to be p53-inde- lymphoid neoplasms.5–7 pendent. In aggregate, our results are most consistent with a PBX1 was first identified through its involvement in t(1;19).1 scenario in which NDGR1 induction by E2a-Pbx1 occurs through indirect mechanisms as part of a cellular response to oncogene activation. Correspondence: D LeBrun, Richardson Laboratory, Department of Pathology, Queen’s University, Kingston, Ontario, K7L 3N6, Canada; Fax: 613 533 2907 Received 9 October 2000; accepted 28 November 2000 NDRG1 induction by E2a-Pbx1 MN Rutherford et al 363 Materials and methods Fluorescence signals from FITC and PI were then correlated using an EPICS Elite flow cytometer. Cell lines, tissue culture, expression constructs and To semi-quantify apoptosis using the DNA fragmentation transfections assay, 1 × 106 viable-appearing cells were pelleted and fixed in 70% ethanol. Cells were washed once with PBS then incu- ␮ m Ba/F3 cells (a kind gift of Dr Gerald Krystal, Terry Fox Labora- bated for 30 min in 40 l of a solution containing 0.1 m tory, University of British Columbia) were propagated in RPMI sodium citrate and 0.2 sodium phosphate to extract frag- mented DNA from cells. Cells were then pelleted and the containing 10% fetal bovine serum (FBS) and either 5 ng/ml ␮ of recombinant murine IL-3 (Sigma-Aldrich Canada, Oakville, DNA contained in 10 l of the supernatant was fractionated ON, Canada) or 5% WEHI-3B-conditioned medium. cDNAs on a 1.5% agarose gel. The TUNEL assay was performed using for expression in Ba/F3 cells were cloned using standard tech- a commercial kit (APO-BRDU, Pheonix Flow Technologies, niques into the plasmid pMT-CB6+ (a kind gift of Dr Frank San Diego, CA, USA) according to the manufacturer’s instruc- Rauscher III, The Wistar Institute), which confers expression tions. in mammalian cells regulated by a sheep metallothionein pro- moter and carries NEO as a selectable marker.22 For transfec- tions, 3 × 106 Ba/F3 cells in 400 ␮l of complete medium con- cDNA representational difference analysis taining 20 ␮g of DNA were placed in an electroporation RDA was performed as described previously using polyA+ cuvette (gap 0.4 cm) and subjected to an electrical pulse 7 23 (300 V, 1050 ␮F) using a Gene Pulser II apparatus (BioRad RNA purified from 10 cells. The second difference products Laboratories Canada, Mississauga, ON). Stably-transfected (DP-2) were separated on a 2% agarose gel, recovered from populations were selected by propagation for 10 days in agarose and subcloned into the BamHI site of pBluescript. medium containing 450 ␮g/ml of G418, which was added after 24 h of growth in non-selective medium. Stable transfec- tants were generally analyzed as uncloned populations. In the Western and Northern blots transient transfection of luciferase reporter constructs, elec- ␮ For the evaluation of protein expression by immunoblotting, troporation cocktails contained 4 g of expression plasmid, ␮ 10 ␮g of luciferase reporter, 1 ␮gof␤-galactosidase-express- cells were lysed in 70 l of protein sample buffer and boiled. ing construct and 5 ␮g of pBluescript plasmid DNA. Lucifer- Fifteen micrograms of protein from each sample was fraction- ase and ␤-galactosidase assays were carried out 24 h after ated on a discontinuous polyacrylamide gel under denaturing electroporation using chemiluminescence reagents (Tropix conditions and transferred electrophoretically to a nitrocellu- Inc, Bedford, MA, USA) according to instructions. For retrovi- lose membrane. The membrane was blocked overnight in PBS ral transduction of E6, ␾NX-Eco retroviral packaging cells containing 5% nonfat dried milk and 0.1% Tween 20 and (kindly provided by Dr Gary Nolan, Stanford University) were then subjected sequentially to an overnight incubation with transfected with pBabePuroE6 plasmid DNA (kindly provided the primary monoclonal antibody, three 5-min washes in by Dr Gordon Peters, Imperial Cancer Research Fund, Lon- PBS/Tween, and a 1 h incubation with horseradish peroxi- dase-conjugated, anti-mouse secondary antibody (Jackson don, UK) using a standard CaPO4 protocol. Roughly 16 h later, the transfection medium was replaced with 6 ml of RPMI ImmunoResearch Laboratories, West Grove, PA, USA). After containing 10% FBS and 105 exponentially growing Ba/F3 a final wash, secondary antibody on the membrane was cells that had been stably transfected with zinc-inducible E2a- detected by autoradiography using chemiluminescence Pbx1a. After 2 days of co-culture, the Ba/F3 cells were separ- reagents (NEN Life Science Products, Boston, MA, USA) ated from the adherent packaging cells by gentle pipetting and according to the manufacturer’s instructions. E2a-Pbx1 was subjected to selection in puromycin (1 ␮g/ml) for 10 days. detected using the Yae monclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at a 1:1000 dilution and p53 was detected using the Ab-3 monoclonal antibody (Oncogene Science, Cambridge, MA, USA) at a dilution of Cell cycle and apoptosis assays 1:200 from the stock solution. For Northern blots, total RNA was isolated using Trizol (Life For experiments involving IL-3 deprivation, cells were washed Technologies, Rockville, MD, USA) according to the manufac- three times in RPMI without IL-3 before resuspension in turer’s instructions. Fifteen micrograms of total RNA was frac- medium containing IL-3 at the specified concentrations. In tionated by electrophoresis on a 1% agarose-formaldehyde order to enumerate Ba/F3 cells in the various phases of the gel, transferred to a Hybond-N+ membrane (Amersham Phar- cell cycle, cells were labeled by adding 5-bromodeoxyuridine macia Biotech, Piscataway, NJ, USA), and hybridized with (BrdU, Roche Diagnostics, Laval, Canada) to 0.02 nm in the a 32P-labelled probe using standard methods. Signal intensity culture medium for 25 min, washed in phosphate-buffered on the blots was quantified using a PhosphorImager saline (PBS) and fixed in 70% ethanol. Cells were then washed (Molecular Dynamics, Sunnyvale, CA, USA). twice in PBS, resuspended in 1 ml of 1 N HCl containing 0.5% Triton X-100 and incubated for 30 min at 37°C. Cells were washed three times in PBS, resuspended in 50 ␮lof Results blocking solution (3% BSA and 0.5% Triton X-100 in PBS) containing 5 ␮g/ml of mouse monoclonal anti-BrdU (Roche E2a-Pbx1 promotes apoptosis in Ba/F3 cells Diagnostics) and incubated for 1 h at room temperature. Cells were then washed, incubated with a FITC-conjugated, goat Ba/F3 cells were transfected with a plasmid that confers zinc- anti-mouse secondary antibody, washed and resuspended in inducible expression of E2a-Pbx1a mediated by a sheep met- 1 ml of PBS containing 10 ␮g/ml propidium iodide (PI), allothionein promoter. Stably-transfected populations were 0.5 mg/ml RNAse A, 0.1% NP 40, and 0.1% trisodium citrate. selected in medium containing G418. Zinc-inducible

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 364 expression of E2a-Pbx1 by transfectants was confirmed by growth conditions (complete medium containing 5% WEHI- immunoblot using an anti-E2a monoclonal antibody 3B supernatant), expression of E2a-Pbx1 was associated with (Figure 1). Subsequent experiments were done using uncloned a slight increase in DNA fragmentation. This preferential DNA transfectant cell populations, except for the experiment fragmentation related to E2a-Pbx1 expression appeared involving p53 inactivation, in which cloned cells were used. greater when the concentration of IL-3 was reduced to 0.5 or We initially wished to evaluate potential effects of enforced 0.05%. Morphological evaluation indicated a three-fold E2a-Pbx1 expression on cell growth and viability. Growth greater proportion of apoptotic cells in E2a-Pbx1 transfectants curves indicated that expression of E2a-Pbx1 was associated relative to empty vector-transfected controls (64% vs 21%) with impaired cell accumulation relative to controls under after 24 h in medium containing 0.5% WEHI-3B-conditioned optimal or IL-3-deficient conditions (Figure 2a). These differ- medium (data not shown). We infer from these data that E2a- ences could be explained by either a reduced rate of cell Pbx1 increases the rate of apoptosis in Ba/F3 cells and division or accelerated cell death in E2a-Pbx1-expressing increases their sensitivity to IL-3 deprivation. cells. To evaluate possible effects on cell cycle progression in zinc-induced transfectants, S-phase cells were labelled with bromodeoxyuridine (BrdU) and the resulting fluorescence sig- E2a-Pbx1 induces expression of the gene NDRG1 nal was correlated with cellular DNA content using two- dimensional flow cytometry (Figure 2b). As expected, IL-3 In order to identify E2a-Pbx1-regulated genes whose deprivation was associated with delayed progression through expression might be altered in association with this pro-apop- G1 in both E2a-Pbx1 transfectants and controls. However, for- totic effect, we applied cDNA RDA to uncloned populations ced expression of E2a-Pbx1 was associated with only minor, of Ba/F3 cells. RNA was isolated from populations of cells inconsistent differences in cell cycle parameters of uncertain stably transfected with E2a-Pbx1a, or ‘empty’ vector, after significance. In order to evaluate possible effects of E2a-Pbx1 24 h of zinc induction in medium containing a concentration on apoptosis, stable transfectants were pre-incubated with of recombinant IL-3 sufficient to support proliferation and sur- zinc and then maintained for 24 h in the presence of decreas- vival (5 ng/ml). When cDNA from cells transfected with empty ing concentrations of IL-3, supplied as WEHI-3B-conditioned vector was used as ‘tester’ and cDNA from E2a-Pbx1-express- medium. Internucleosomal degradation of DNA, a character- ing cells as ‘driver’ (ie to identify transcripts that may be istic feature of apoptotic death in Ba/F3 cells, was then evalu- repressed by E2a-Pbx1), no discrete bands were visible on gel ated by agarose gel electrophoresis (Figure 2c). Under optimal electrophoresis of the second difference product (data not shown). However, when the subtraction was performed in the reverse direction, using cDNA from E2a-Pbx1-expressing cells as tester, five bands were resolved that ranged in size from roughly 200 to 600 bp (Figure 3). These polymerase chain reaction (PCR) products were cloned into pBluescript (clones 1M, 2C, 2F, 3A, and 4) and sequenced. Searches of NCBI databases were performed with the resultant sequence data. Clone 1M comprised a portion of the E2a-Pbx1a cDNA bracketed by DpnII sites at positions 1926 and 2549. This finding served to validate the differential cloning method. Clone 2C contained cDNA sequence of unknown function previously identified in a human BAC clone (GenBank accession No. AC004549). Clone 2F encoded a portion of elongation factor 2 (EF-2). Clone 3A comprised a portion of cDNA (positions 691 to 1036) from the gene Ndr1 (see below), and clone 4 represented a portion of 28S ribosomal RNA. To evaluate differential expression induced by E2a-Pbx1, these cDNA fragments were used to probe Northern blots of RNA purified from the populations of stably-transfected Ba/F3 cells (Figure 3). As expected, probing with the insert from clone 1M produced an intense signal in the roughly 3 kb range on zinc induction of E2a-Pbx1 transfectants (lane 4). We surmise that a weaker, higher molecular weight signal detected in uninduced cells (lane 2) was likely the result of low-level transcription from a constitutive promoter located more 5′ on the expression vector. Probing with clone 2F con- firmed constitutive expression of EF-2. Probing with clone 3A revealed a roughly 3 kb transcript of Ndr1 that was present exclusively in those Ba/F3 cells that had been transfected with E2a-Pbx1 and induced with zinc (lane 12). No signal was observed on probing with clone 2C and expression of clone Figure 1 Zinc-dependent expression of E2a-Pbx1a. A Western blot 4 (28S rRNA) was constitutive, as expected (data not shown). using an anti-E2a monoclonal antibody shows marked induction of Ndr1 is the murine homolog of human RTP, and the two E2a-Pbx1a expression on addition of zinc in stably transfected Ba/F3 cells. Expression is barely detectable in the uninduced state. The genes encode predicted proteins with 93% identity. RTP is monoclonal antibody interacts minimally with endogenous, murine widely expressed in tissues and encodes a protein of Mr E2a proteins. 43 000 with no recognizable functional motifs other than sev-

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 365

Figure 2 E2a-Pbx1 expression promotes apoptosis but not cell proliferation in Ba/F3 cells. (a) Stably transfected populations of Ba/F3 cells were pre-incubated for 24 h with or without zinc, washed three times and then plated in 96-well plates in medium containing optimal (5 ng/ml) or reduced concentrations of recombinant IL-3. Daily cell counts performed in triplicate indicate slower accumulation of cells specifically in the E2a-Pbx1a transfectant, zinc-induced population. Error bars represent one standard deviation. (b) Analysis of BrdU-labelled cells by two- parameter flow cytometry confirms G1 delay in association with reduced concentrations of IL-3 (supplied as WEHI-3B-conditioned medium), but shows no consistent cell cycle effects related to E2a-Pbx1a expression. (c) Agarose gel electrophoresis shows increased internucleosomal degradation of DNA in association with E2a-Pbx1a expression. This effect is enhanced by IL-3 deprivation.

eral predicted phosphorylation sites, a putative phosphopen- and may mediate some of p53’s cytostatic effects, including tetheine attachment site, and a previously undescribed 10 negative effects on cell proliferation, anchorage-independent amino acid motif that is repeated three times near the C-ter- growth and tumorigenicity in nude mice.27,28 Recently, two of minus. It was first identified as a transcript induced in human us (TO and HK) reported that Ndr1, the mouse homolog, is endothelial (HUVEC) cells on treatment with homocysteine.24 transcriptionally repressed downstream of the nuclear oncop- Subsequently, a number of groups have reported differential rotein N-myc.29 An agreement was reached recently to refer expression of this gene in diverse cell types in response to to both the murine and human genes as NDRG1. TDD5,a diverse stimuli, leading to the acquisition by a single gene of mouse gene highly identical to murine NDRG1 and possibly a confusing multitude of names, including Drg1, Cap43, and representing an alternative splice product, is transcriptionally RTP/rit42. Expression of RTP is induced on differentiation of repressed in response to testosterone.30 colon carcinoma cell lines, or on exposure of a variety of cell Considering the regulatory relationship that appears to exist types to nickel-containing compounds.25,26 Lower expression between murine NDRG1 and N-myc, the previously demon- has been observed in neoplastic cell lines of several lineages strated association of human NDRG1 expression with cellular relative to their non-neoplastic counterparts.27 Of perhaps gre- differentiation, cell cycle arrest and apoptosis, and the evi- atest relevance to the work described here, RTP is induced dence implicating human NDRG1 as a downstream effector downstream of the tumor suppressor p53 in several cell types of some p53 functions, we elected to investigate further the

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 366

Figure 3 Isolation of E2a-Pbx1a-induced cDNAs by RDA. The upper image illustrates the results of agarose gel electrophoresis of the PCR products representing the ‘second difference product’ (first lane) along with cloned counterparts after excision from a plasmid vector (other lanes). The lower three images show Northern blots probed with the indicated cloned inserts. Note that the transcript related to insert 3A (Ndr1/NDRG1) is detectable exclusively in E2a-Pbx1a-transfected, zinc-induced cells (lane 12). Not shown are the results of probing with insert 2C (no signal) or insert 4 (constitutive expression).

relationship between E2a-Pbx1 and NDRG1. ing DNA elements associated with the NDRG1 promoter. Ba/F3 cells were transiently co-transfected with a plasmid Cis-acting 5′ regulatory elements related to NDRG1 conferring constitutive expression of E2a-Pbx1a mediated by can mediate transcriptional induction by E2a-Pbx1 the CMV promoter along with reporter plasmids in which expression of firefly luciferase was regulated by various por- We investigated the possiblity that induction of NDRG1 tions of 5′ genomic DNA associated with the murine NDRG1 downstream of E2a-Pbx1 might be mediated through cis-act- promoter (Figure 4). The complete sequence of this promoter

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 367

Figure 4 Induction of the NDRG1 promoter by E2a-Pbx1a. Transi- ent co-transfection of Ba/F3 cells with E2a-Pbx1a and luciferase reporter constructs containing 5′ genomic DNA sequence related to murine NDRG1 indicate that induction of NDRG1 by E2a-Pbx1 can be mediated by a small promoter segment that includes an Sp1 consensus recognition sequence and the TATA box. region, from nucleotide positions −637 to +194 relative to the transcriptional start site, is provided elsewhere.29 In these experiments, a relatively small portion of 5′ DNA beginning at nucleotide position −52 was associated with maximal luciferase activity, suggesting that silencing elements may exist further 5′. This small DNA fragment also conferred near- maximal, 21-fold responsiveness to E2a-Pbx1. Deletion from this fragment of an additional 19 nucleotides resulting in the removal of a predicted Sp1 binding site immediately 5′ to the TATA box led to a marked reduction in both absolute lucifer- ase expression levels and responsiveness to E2a-Pbx1. Deletion of the TATA box was associated with a further Figure 5 Time course of NDRG1 induction downstream of E2a- Pbx1a. A population of stably-transfected Ba/F3 cells was induced marked decrease in both luciferase activity and E2a-Pbx1 with zinc and subsequent expression levels of E2a-Pbx1a protein and responsiveness (data not shown). Inspection of the nucleotide NDRG1 transcript were followed using Western (upper image) or sequence from position −52 failed to uncover a predicted Northern blots (middle image), respectively. The graph compares fold- binding site for Pbx1 or a Pbx1-Hox heterodimer. This induction of NDRG1 transcript in E2a-Pbx1a-transfected (solid circles) absence of apparent Pbx1 binding sites within a restricted and empty vector-transfected (open circles) populations of Ba/F3 cells. region of DNA capable of mediating induction by E2a-Pbx1 Quantification was by Phosphorimager and all values were nor- malized to those obtained on reprobing of the blots for G3PDH. suggests that the regulatory relationship that exists between E2a-Pbx1 and NDRG1 may be indirect, perhaps involving additional, intermediary proteins. apoptotic effect, E2a-Pbx1 must function as a DNA-binding A time course experiment indicated that while expression transcription factor in the induction of NDRG1. Also observed of recombinant E2a-Pbx1a protein was near-maximal by 6 h consistently in several experiments was a relative induction of after induction with zinc, the abundance of the NDRG1 tran- NDRG1 on IL-3 deprivation of empty vector-transfected Ba/F3 script underwent a sharp increase between 12 and 15 h post- cells, albeit to lower levels than those seen in association with induction (Figure 5). This suggests a latency period of roughly E2a-Pbx1 (Figure 6, lanes 1–4). This observation lends support 9 h between the onset of elevated levels of E2a-Pbx1 protein to the existence of a functional relationship between NDRG1 and the consequent accumulation of NDRG1 transcript. and apoptosis. Quantitative evaluation of signal intensity by Phosphorimager indicated a 27-fold induction of NDRG1 by 24 h (Figure 5). As in all quantitative determinations in this study using North- Induction of NDRG1 downstream of E2a-Pbx1 does ern blots, values for NDRG1 signal were normalized by strip- not require p53 ping the membrane and re-probing for the housekeeping tran- script G3PDH. The relatively long latency between the Human NDRG1 is a target of transcriptional induction by p53 upsurge in E2a-Pbx1 levels and induction of NDRG1 is con- and has been said to play some role in mediating p53’s anti- sistent with the existence of an indirect regulatory relationship proliferative effects.27 The p53 protein can be functionally between these two events. induced by diverse stimuli, including oncogene activation and The ability of E2a-Pbx1 to induce apoptosis in a B-ALL cell growth factor deprivation. Therefore, we considered the possi- line has been shown previously to require the Pbx1 homeodo- bility that p53 might function as an intermediary in the induc- main.31 To investigate the role of the Pbx1 homeodomain in tion of NDRG1 downstream of E2a-Pbx1. A clone of stably induction of NDRG1, Ba/F3 cells were stably transfected with E2a-Pbx1-transfected Ba/F3 cells was infected with a retroviral a plasmid conferring zinc-inducible expression of the protein vector conferring expression of the human papilloma virus E2a-Pbx1⌬HD, which contains a targeted deletion of the Pbx1 (HPV) protein E6, which mediates inactivation and destruction homeodomain (amino acids 627–689 in E2a-Pbx1a). In two of p53.32 An anti-p53 immunoblot confirmed that retroviral experiments, this protein failed consistently to induce the transduction of E6 markedly attenuated the enhanced expression of NDRG1 (Figure 6, compare lanes 9–12 with accumulation of p53 protein that normally occurs on lanes 5–8). This observation suggests that, as with its pro- irradiation of Ba/F3 cells (Figure 7a).33 A Northern blot of RNA

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 368

Figure 6 The Pbx1 homeodomain is required for induction of NDRG1 by E2a-Pbx1. Incubation of Ba/F3 cells transfected with vec- tor alone for 24 h in decreasing concentrations of IL-3 is associated with induction of NDRG1, as demonstrated by Northern blot (lanes Figure 7 The role of p53 in NDRG1 induction downstream of E2a- 1–4). Forced expression of E2a-Pbx1⌬HD, a mutant form of E2a-Pbx1 Pbx1a. (a) Infection of a clone of E2a-Pbx1a-transfected Ba/F3 cells that lacks the Pbx1 homeodomain, fails to induce NDRG1 (compare with a retrovirus conferring expression of the E6 oncoprotein from lanes 9–12 with lanes 5–8). HPV markedly attenuates the accumulation of p53 protein that occurs on ␥-irradiation (4 Gy) of control cells infected with the retroviral vec- purified 26 h after zinc induction of E2a-Pbx1 indicated per- tor alone. (b) A Northern blot of RNA purified from the same cells sistence of a modest increase in NDRG1 expression irrespec- 26 h after zinc induction of E2a-Pbx1a expression shows increased NDRG1 expression in association with E2a-Pbx1 expression (lanes 2 tive of the presence of E6 (Figure 7b). The magnitude of and 4) irrespective of the presence of E6. NDRG1 induction in this clone was considerably less than that observed earlier in the uncloned BaF3 cell population. We speculate that this may be related to clonal variation. These include the relatively long latency of induction and the Nonetheless, these results indicate that activation of p53 is not absence of apparent Pbx1 binding elements in the murine required in the induction of NDRG1 downstream of E2a-Pbx1. NDRG1 promoter. Therefore, we considered the possibility In order to investigate the potential relevance to human leu- that induction could require the intercession of other factors. kemia of NDRG1 induction by E2a-Pbx1, we next evaluated A tumor suppressor protein mutated in greater that 50% of expression of human NDRG1 by Northern blot in human ALL human cancers, wild-type p53 can induce cell cycle arrest or cell lines (Figure 8). The Reh cell line is derived from a case apoptosis in response to diverse stimuli, including DNA dam- of B-lineage ALL. Raji is derived from Burkitt’s lymphoma, a age or oncogene activation.34 Perhaps most compelling in neoplasm with phenotypic features of fully differentiated B considering the possibility of a relationship between NDRG1 cells. UOCB1 is derived from B-lineage ALL and is associated function and neoplastic transformation are data implicating with expression of endogenous E2a-Hlf, a chimeric, leukem- NDRG1 as a target of induction by p53 and a downstream ogenic transcription factor expressed in a small proportion of mediator of p53-dependent, anti-oncogenic functions.27 These pediatric ALL cases. The B-lineage ALL lines 697 and RCH considerations led us to ask whether p53 could function ACV are associated with t(1;19) and express endogenous E2a- downstream of E2a-Pbx1 in the induction of NDRG1, and Pbx1. Finally, the lines LB-1, -2, -3, -4 and -5 are lymphoblas- whether NDRG1, in turn, could contribute to the pro-apop- toid lines established by immortalizing B cells from normal totic effect associated with E2a-Pbx1 expression in Ba/F3 cells. donors using Epstein–Barr virus. The Northern blot indicated In our study, antagonizing p53 with HPV E6 failed to impair elevated expression of NDRG1 in the three lines derived from NDRG1 induction by E2a-Pbx1. In a recent study using serial immature B-lineage progenitor cells that express endogenous analysis of (SAGE) to survey transcriptional E2a-fusion proteins (UOCB1, 697 and RCH ACV). target genes of p53, NDRG1 emerged as one of six p53- inducible genes that could also be induced by , a p53 homolog not targeted by E6.28,35 Furthermore, although Discussion NDRG1 expression can be induced by DNA damaging agents in several cell lines, this induction does not invariably require Several observations in our study suggest that the regulatory p53.28 These results lend further support to the highly plaus- relationship between E2a-Pbx1 and NDRG1 may be indirect. ible notion that multiple mechanisms exist for the induction

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 369 ily members. Alternatively, since enforced expression of N- myc can induce apoptosis under certain conditions, we con- sidered the possibility that the pro-apoptotic effects of E2a- Pbx1 might correlate with myc induction. However, evalu- ation by Northern blot failed to uncover any effect of enforced E2a-Pbx1 expression on the abundance of N-myc or C-myc transcripts (data not shown). Therefore, it appears that the regulation of NDRG1 transcription may be mediated through either myc-dependent or myc-independent pathways, and that regulation downstream of E2a-Pbx1 involves one of the latter. The short portion of 5′ genomic DNA sequence from the murine NDRG1 locus capable of mediating transcriptional induction by E2a-Pbx1 contains a predicted Sp1 binding element, raising the possibility that NDRG1 induction could be mediated by Sp1 or related proteins. Sp1 belongs to a fam- ily of closely related transcription factors with at least 16 members in mammals.42 Although relative binding affinities vary, these proteins recognize very similar sites on DNA. Fur- thermore, their transcriptional functions appear to be regu- lated, to a considerable degree, by such factors as relative Figure 8 Expression of NDRG1 in lymphoid leukemia cell lines. expression levels of several family member proteins and post- A Northern blot prepared from 10 human lymphoid cell lines was translational modifications, including phosphorylation. We probed with a fragment of human NDRG1 cDNA. Reh, UOCB1, 697 and RCH ACV were established from cases of B-progenitor ALL, Raji chose not to pursue experimentally the possibility that Sp1- is derived from Burkitt lymphoma and LB-1, -2, -3, -4, and -5 are related transcription factors could mediate NDRG1 induction. lymphoblastoid lines established from B cells from normal donors. Nonetheless, the idea remains plausible. Note elevated NDRG1 expression in the lines associated with E2A In summary, we have shown that forced expression of E2a- translocations (UOCB1, 697 and RCH ACV). Pbx1 induces apoptosis in Ba/F3 cells associated with induc- tion of the gene NDRG1. NDRG1 is a target of transcriptional induction by p53 and a putative mediator of some anti-pro- of NDRG1 expression. Therefore, although our results with E6 liferative effects. Although induction of NDRG1 consequent argue against a requirement for p53 in NDRG1 induction by to E2a-Pbx1 expression does not require the intercession of E2a-Pbx1, they do not exclude the possibility that NDRG1 p53 in BaF3 cells, it remains likely that NDRG1 induction could be induced by other means as part of an apoptotic or represents part of a cytostatic or apoptotic cellular response cytostatic cellular response to oncogene activation. This idea to oncogene activation. is consistent with the previous observation that E2a-Pbx1- induced apoptosis is not dependent on functional p53.31 Exposure of prostatic adenocarcinoma cells to androgen, Acknowledgements colonic adenocarcinoma cells to low glucose concentration, myelomonocytic leukemia cells to retinoic acid or vitamin D3, We thank Drs Gerald Krystal, Frank Rauscher, Gary Nolan or trophoblastic cells to forskolin induces NDRG1 expression and Gordon Peters for providing reagents, as indicated in the in association with cellular differentiation and cell cycle text, and Dr Stephen Hunger for providing WEHI-3B cells and arrest.25,36–39 Furthermore, endogenous expression of murine helpful comments on the manuscript. This work was sup- NDRG1 correlates with tissue differentiation in embryonic ported by operating grants from The Hospital for Sick Children tissues and is repressed downstream of the nuclear oncoprot- Foundation and The National Cancer Institute of Canada. ein N-myc.29 Since terminal differentiation is associated with apoptosis in many cell types, including epithelial and myeloid lineages, the association of NDRG1 expression with cellular References differentiation is also consistent with a role in apoptosis. Also supportive of such a role has been the demonstration in pri- 1 Nourse J, Mellentin JD, Galili N, Wilkinson J, Stanbridge E, Smith mary human samples of preferential NDRG1 expression in SD, Cleary ML. Chromosomal translocation t(1;19) results in syn- thesis of a fusion mRNA that codes for a potential tissues in which apoptosis appears to play a particularly chimeric transcription factor. Cell 1990; 60: 535–545. important regulatory role, including periluminal, surface 2 Murre C, Bain G, van Dijk MA, Engel I, Furnari BA, Massari ME, colonic epithelium and placental trophoblast.25,40,41 We did Matthews JR, Quong MW, Rivera RR, Stuiver MH. Structure and not observe cytostatic effects on forced expression of recombi- function of helix–loop–helix proteins. Biochim Biophys Acta nant murine NDRG1 in Ba/F3 cells (data not shown). How- 1994; 1218: 129–135. ever, evaluation by Northern blot showed that the expression 3 Quong MW, Massari ME, Zwart R, Murre C. A new transcriptional- activation motif restricted to a class of helix–loop–helix proteins level of recombinant NDRG1 was considerably lower than is functionally conserved in both yeast and mammalian cells. Mol that of the endogenous gene in response to E2a-Pbx1 (data Cell Biol 1993; 13: 792–800. not shown), making our failure to observe phenotypic changes 4 Aronheim A, Shiran R, Rosen A, Walker MD. The E2A gene pro- difficult to interpret. duct contains two separable and functionally distinct transcription Two of us (TO and HK) have shown previously that activation domains. Proc Nat Acad Sci USA 1993; 90: 8063–8067. expression of NDRG1 is repressed downstream of N-myc.29 5 Zhuang Y, Soriano P, Weintraub H. The helix–loop–helix gene E2A is required for B cell formation. Cell 1994; 79: 875–884. Therefore, in considering possible intermediary factors, we 6 Bain G, Robanus Maandag EC, Izon DJ, Amsen D, Kruisbeek AM, wondered whether the effects of E2a-Pbx1 on NDRG1 Weintraub BC, Krop I, Schlissel MS, Feeney AJ, van Roon M, van expression might be mediated through repression of myc fam- der Valk M, te Riele HPJ, Berns A, Murre C. E2a proteins are

Leukemia NDRG1 induction by E2a-Pbx1 MN Rutherford et al 370 required for proper B cell development and initiation of immuno- 24 Kokame K, Kato H, Miyata T. Homocysteine-respondent genes in globulin gene rearrangements. Cell 1994; 79: 885–892. vascular endothelial cells identified by differential display analy- 7 Bain G, Engel I, Robanus Maandag EC, te Riele HP, Voland JR, sis. GRP78/BiP and novel genes. J Biol Chem 1996; 271: Sharp LL, Chun J, Huey B, Pinkel D, Murre C. E2A deficiency leads 29659–29665. to abnormalities in ␣/␤ T-cell development and to rapid develop- 25 van Belzen N, Dinjens WN, Diesveld MP, Groen NA, van der ment of T-cell lymphomas. Mol Cell Biol 1997; 17: 4782–4791. Made AC, Nozawa Y, Vlietstra R, Trapman J, Bosman FT. A novel 8 Gehring WJ, Affolter M, Burglin T. Homeodomain proteins. Ann gene which is up-regulated during colon epithelial cell differen- Rev Biochem 1994; 63: 487–526. tiation and down-regulated in colorectal neoplasms. Lab Invest 9 Lu Q, Knoepfler PS, Scheele J, Wright DD, Kamps MP. Both Pbx1 1997; 77: 85–92. and E2A-Pbx1 bind the DNA motif ATCAATCAA cooperatively 26 Zhou D, Salnikow K, Costa M. Cap43, a novel gene specifically with the products of multiple murine Hox genes, some of which induced by Ni2+ compounds. Cancer Res 1998; 58: 2182–2189. are themselves oncogenes. Mol Cell Biol 1995; 15: 3786–3795. 27 Kurdistani SK, Arizti P, Reimer CL, Sugrue MM, Aaronson SA, Lee 10 Chang CP, Shen WF, Rozenfeld S, Lawrence HJ, Largman C, SW. Inhibition of tumor cell growth by RTP/rit42 and its respon- Cleary ML. Pbx proteins display hexapeptide-dependent cooperat- siveness to p53 and DNA damage. Cancer Res 1998; 58: 4439– ive DNA binding with a subset of Hox proteins. Genes Dev 1995; 4444. 9: 663–674. 28 Yu J, Zhang L, Hwang PM, Rago C, Kinzler KW, Vogelstein B. 11 Phelan ML, Rambaldi I, Featherstone MS. Cooperative interactions Identification and classification of p53-regulated genes. Proc Natl between HOX and PBX proteins mediated by a conserved peptide Acad Sci USA 1999; 96: 14517–14522. motif. Mol Cell Biol 1995; 15: 3989–3997. 29 Shimono A, Okuda T, Kondoh H. N-myc-dependent repression of 12 Selleri L, Jacobs Y, Choe S, Chanda S, O’Gorman S, Cleary ML. Ndr1, a gene identified by direct subtraction of whole mouse The Hox-cofactor Pbx1 is required for normal development of embryo cDNAs between wild type and N-myc mutant. Mech Dev proximal limb and axial skeleton and for spleen morphogenesis. 1999; 83: 39–52. Blood 1999; 92: A1265-A1265. 30 Lin TM, Chang C. Cloning and characterization of TDD5, an 13 Monica K, LeBrun DP, Dedera DA, Brown RB, Cleary ML. Trans- androgen target gene that is differentially repressed by testosterone formation properties of the E2a-Pbx1 chimeric oncoprotein: fusion and dihydrotestosterone. Proc Natl Acad Sci USA 1997; 94: with E2a is essential, but the Pbx1 homeodomain is dispensable. 4988–4993. Mol Cell Biol 1994; 14: 8304–8314. 31 Smith KS, Jacobs Y, Chang CP, Cleary ML. Chimeric oncoprotein 14 Kamps MP, Baltimore D. E2A-Pbx1, the t(1;19) translocation pro- E2a-Pbx1 induces apoptosis of hematopoietic cells by a p53-inde- tein of human pre-B-cell acute lymphocytic leukemia, causes pendent mechanism that is suppressed by Bcl-2. Oncogene 1997; acute myeloid leukemia in mice. Mol Cell Biol 1993; 13: 351– 14: 2917–2926. 357. 32 Tommasino M, Crawford L. Human papillomavirus E6 and E7: 15 Dedera DA, Waller EK, LeBrun DP, Sen-Majumdar A, Stevens ME, proteins which deregulate the cell cycle. Bioessays 1995; 17: Barsh GS, Cleary ML. Chimeric homeobox gene E2A-PBX1 509–518. induces proliferation, apoptosis, and malignant lymphomas in 33 Canman CE, Gilmer TM, Coutts SB, Kastan MB. Growth factor transgenic mice. Cell 1993; 74: 833–843. modulation of p53-mediated growth arrest versus apoptosis. 16 Fu X, Kamps MP. E2a-Pbx1 induces aberrant expression of tissue- Genes Dev 1995; 9: 600–611. specific and developmentally regulated genes when expressed in 34 Levine AJ. p53, the cellular gatekeeper for growth and division. NIH 3T3 fibroblasts. Mol Cell Biol 1997; 17: 1503–1512. Cell 1997; 88: 323–331. 17 Fu X, McGrath S, Pasillas M, Nakazawa S, Kamps MP. EB-1, a 35 Marin MC, Jost CA, Irwin MS, DeCaprio JA, Caput D, Kaelin WG Jr. Viral oncoproteins discriminate between p53 and the p53 tyrosine kinase signal transduction gene, is transcriptionally acti- homolog p73. Mol Cell Biol 1998; 18: 6316–6324. vated in the t(1;19) subset of pre-B ALL, which express oncoprot- 36 Ulrix W, Swinnen JV, Heyns W, Verhoeven G. The differentiation- ein E2a-Pbx1. Oncogene 1999; 18: 4920–4929. related gene 1, Drg1, is markedly upregulated by androgens in 18 Fu X, Roberts WG, Nobile V, Shapiro R, Kamps MP. mAngiogenin- LNCaP prostatic adenocarcinoma cells. FEBS Lett 1999; 455: 3, a target gene of oncoprotein E2a-Pbx1, encodes a new angiog- 23–26. enic member of the angiogenin family. Growth Factors 1999; 17: 37 van Belzen N, Dinjens WN, Eussen BH, Bosman FT. Expression 125–137. of differentiation-related genes in colorectal cancer: possible 19 McWhirter JR, Goulding M, Weiner JA, Chun J, Murre C. A novel implications for prognosis. Histol Histopathol 1998; 13: 1233– fibroblast growth factor gene expressed in the developing nervous 1242. system is a downstream target of the chimeric homeodomain 38 Piquemal D, Joulia D, Balaguer P, Basset A, Marti J, Commes T. oncoprotein E2A-Pbx1. Development 1997; 124: 3221–3232. Differential expression of the RTP/Drg1/Ndr1 gene product in pro- 20 de Lau WB, Hurenkamp J, Berendes P, Touw IP, Clevers HC, van liferating and growth arrested cells. Biochim Biophys Acta 1999; Dijk MA. The gene encoding the granulocyte colony-stimulating 1450: 364–373. factor is a target for deregulation in pre-B ALL by the 39 Xu B, Lin L, Rote NS. Identification of a stress-induced protein t(1;19)-specific oncoprotein E2A-Pbx1. Oncogene 1998; 17: during human trophoblast differentiation by differential display 503–510. analysis. Biol Rep 1999; 61: 681–686. 21 McWhirter JR, Neuteboom ST, Wancewicz EV, Monia BP, Down- 40 Kokawa K, Shikone T, Nakano R. Apoptosis in human chorionic ing JR, Murre C. Oncogenic homeodomain transcription factor villi and decidua during normal embryonic development and E2A-Pbx1 activates a novel WNT gene in pre-B acute lymphoblas- spontaneous abortion in the first trimester. Placenta 1998; 19: toid leukemia. Proc Natl Acad Sci USA 1999; 96: 11464–11469. 21–26. 22 Cook DM, Hinkes MT, Bernfield M, Rauscher FJ III. Transcrip- 41 Runic R, Lockwood CJ, LaChapelle L, Dipasquale B, Demopoulos tional activation of the syndecan-1 promoter by the Wilms’ tumor RI, Kumar A, Guller S. Apoptosis and Fas expression in human protein WT1. Oncogene 1996; 13: 1789–1799. fetal membranes. J Clin Endocrinol Metab 1998; 83: 660–666. 23 Hubank M, Schatz DG. Identifying differences in mRNA 42 Philipsen S, Suske G. A tale of three fingers: the family of mam- expression by representational difference analysis of cDNA. malian Sp/XKLF transcription factors. Nucleic Acids Res 1999; 27: Nucleic Acids Res 1994; 22: 5640–5648. 2991–3000.

Leukemia