Leukemia (2006) 20, 417–425 & 2006 Nature Publishing Group All rights reserved 0887-6924/06 $30.00 www.nature.com/leu ORIGINAL ARTICLE

The GATA site-dependent hemogen promoter is transcriptionally regulated by GATA1 in hematopoietic and leukemia cells

LV Yang1,9, J Wan2,9,YGe3,4,,ZFu2, SY Kim5, Y Fujiwara6, JW Taub3,7, LH Matherly,3,4, J Eliason8 and L Li1,2

1Center for Molecular Medicine and Genetics, Wayne State University School of Medicine, Detroit, MI, USA; 2Department of Internal Medicine, Wayne State University School of Medicine, Detroit, MI, USA; 3Experimental and Clinical Therapeutics Program, Barbara Ann Karmanos Cancer Institute, Detroit, MI, USA; 4Department of Pharmacology, Wayne State University School of Medicine, Detroit, MI, USA; 5Cold Spring Harbor Laboratories, Cold Spring Harbor, NY, USA; 6Children’s Hospital, Harvard Medical School, Boston, MA, USA; 7Children’s Hospital of Michigan, Detroit, MI, USA and 8Asterand Inc., Detroit, MI, USA

Hemgn (a symbol for hemogen in mouse, EDAG in human pancreas, tonsil, colon and peripheral blood mononuclear and RP59 in rat) encodes a nuclear that is highly cells.2 EDAG is highly expressed in acute myeloid leukemia expressed in hematopoietic tissues and acute leukemia. To characterize its regulatory mechanisms, we examined the (AML) and acute lymphoblastic leukemia (ALL) and has been activities of a Hemgn promoter containing 2975 bp of 50 flanking demonstrated to be involved in erythroleukemia differentiation, sequence and 196 bp of 50 untranslated region (50 UTR) survival and apoptosis and cell transformation of NIH3T3 sequence both in vitro and in vivo: this promoter is preferen- cells.4,5 ALL these studies suggest that overexpression of EDAG tially activated in a hematopoietic cell line, not in nonhemato- in acute leukemia may play an important role in leukemogensis. poietic cell lines, and is sufficient to drive the transcription of a More recently, assocaition of overexpression of EDAG and no lacZ transgene in hematopoietic tissues in transgenic mice. Mutagenesis analyses showed that the 50 UTR including two remission in de novo AML has been reported: this result suggests highly conserved GATA boxes is critical for the promoter that EDAG may play a modulator role in AML and could be a 6 activity. GATA1, not GATA2, binds to the GATA binding sites new target in the treatment of AML. and transactivates the Hemgn promoter in a dose-dependent A number of hematopoietic-specific transcription factors are manner. Furthermore, the expression of human hemogen found to regulate the molecular pathways of hematopoiesis (EDAG) transcripts were closely correlated with levels of through direct binding to regulatory elements of key hemato- GATA1 transcripts in primary acute myeloid leukemia speci- 7 mens. This study suggests that the Hemgn promoter contains poietic . For instance, GATA bind to the critical regulatory elements for its transcription in hematopoie- consensus sequence WGATAR in the promoter regions of a 8 tic tissues and Hemgn is a direct target of GATA1 in leukemia variety of erythroid genes. Members of the GATA family are cells. generally categorized as hematopoietic GATA factors and Leukemia (2006) 20, 417–425. doi:10.1038/sj.leu.2404105; cardiovascular GATA factors.9 Among the former, GATA1 and published online 9 February 2006 GATA2 play critical roles in hematopoiesis, particularly Keywords: hemogen; EDAG; GATA1; promoter; AML; leukemia erythropoiesis, whereas GATA3 is primarily involved in T-cell development.10 To gain further insights into the regulatory mechanism(s) of Hemgn, we isolated and characterized a Hemgn promoter. In Introduction this report, we show that the Hemgn promoter exhibits hematopoietic-specific activities both in vitro and in vivo. Hemgn is highly expressed in the hematopoietic system.1–3 The Mutagenesis, gel shifting and chromatin immunoprecipitation developmental expression pattern of Hemgn marks its hemato- (ChIP) analyses were used to demonstrate the transcriptionally poietic ontogeny. In early embryogenesis, the expression of important roles of the GATA boxes and their binding protein, Hemgn is detected in blood islands of the yolk sac and in GATA1, in the Hemgn 50 untranslated region (50 UTR). Finally, primitive blood.1 In later development, the expression is we report that expression of GATA1 and EDAG appear to be sequentially localized in active hematopoietic sites, such as coregulated in primary AML cells. fetal liver and bone marrow.1,3 Among hematopoietic lineages, Hemgn is predominantly expressed in erythroid and megakaryo- cytic precursor cells; it is also expressed in hematopoietic Materials and methods stem cells and early progenitors.1,3 However, it is absent in 1,3 matured lymphocytes. Just like Hemgn, the human hemogen Hemgn promoter isolation and sequence analysis orthologue (EDAG) exhibits specific expression in human A BAC clones (# 567P14) containing the full-length Hemgn gene hemotopoietic tissues and cells, including adult bone marrow 1 was screened from a mouse BAC genomic library CITB-CJ7-B and fetal liver. However, no EDAG transcripts are detected in (Research Genetics, Catalogue # 96021) (for details, see adult liver, heart, brain, skeletal muscle, kidney, spleen, Supplementary Information). A Hemgn promoter spanning 3171 bp (positions À2975 to þ 196) was amplified with pfu Correspondence: Dr L Li, 421 E. Canfield Ave. #1107, Detroit, Turbo DNA polymerase (Stratagene, CA, USA) using the BAC as Michigan 48201, USA. a template. The Hemgn promoter was confirmed by DNA E-mail: [email protected] 9These authors contribute equally to the paper. sequencing. The comparison of human and Received 20 November 2005; accepted 29 November 2005; mouse Hemgn genes was performed using VISTA (http:// published online 9 February 2006 dcode.org) (Figure 1b).11 The sequences of mouse and human Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 418

Figure 1 Sequence analyses of mouse and human hemogen promoters. (a) Schematic diagram of exon and intron distribution of hemogen. Please note that 1 h and 1t refer to the first exons of hematopoietic and testicular isoforms, respectively. (b) VISTA plot of DNA sequence conservation between mouse and human hemogen. The homology analysis is set at 70% identity within a 20-bp window: the blue and yellow peaks indicate coding sequence and UTRs of the hemogen cDNA, respectively; the red and pink peaks indicate evolutionarily conserved sequence in the promoter and intronal regions, respectively. (c) The alignment of proximal promoters and the first exons of mouse (GenBank accession#: DQ204723) and human hemogen. The conserved nucleotides are marked with *. The TATA box is labeled in red color; GATA boxes in blue color; and the translation initiation codon ATG in green color. The GATA boxes are boxed, and the sites for AML-1, Evi-1 and Ets are indicated by lines (Please refer the html file on the web for the colored Figure 1).

proximal promoters and the first exons were aligned using CMS, an acute myeloid leukemia cell line (established from a ClustalW (http://www2.ebi.ac.uk/clustalw). The MatInspector 2-year-old girl with acute megakaryocytic leukemia (AMkL))13 program (http://www.gene-regulation.com/) was used to identify was a gift from Dr A Fuse (National Institute of Infectious putative hematopoietic-specific transcription factor binding sites Diseases, Tokyo, Japan). The AMkL cell line, Meg-01, was in the sequence (Figure 1c). obtained from the American Type Culture Collection (Manassas, VA, USA).14 Single colonies of CMS cells with stably transfected pcDNA3-GATA115 were screened for GATA1 expression by real Cell culture, transfections, and reporter assays time RT–PCR (for details see Supplementary Information). 10T1/2 (a mouse fibroblast cell line), PAC1 (a rat smooth muscle cell line), COS-7 (a monkey kidney cell line), C2C12 (a mouse skeletal muscle cell line) and K562 cells (a human erythroleu- Construction of Hemgn promoter deletion and mutant kemia cell line) were co-transfected with indicated plasmids and constructs 200 ng of pRL-SV40 Renilla Luciferase reporter (as an internal Deletion mutants of the Hemgn promoter were cloned into the control). The culture condition and transfection conditions are luciferase reporter vector pXP1.16 The constructs were verified detailed in Supplementary Information. Cells were harvested by sequencing. The primers for amplifying the corresponding 24 h after transfection and luciferase activities were measured Hemgn promoter and its deletion mutants were: 0 using Dual-Luciferaset reporter assay system (Promega, Madi- pHemgnÀ2975 þ 196:5-CGC GGA TCC CAC ATC AGA GAC son, WI, USA). ACC TTG CC and 50-CCG CTC GAG GGT ATT GGC TTT GAC 0 Drosophila Mel-2 cells (D. Mel-2, a Schneider S2 insect cell TTC AC; pHemgnÀ831 þ 196:5-CGC GGA TCC TTG AAC TAG line) from Invitrogen (Carlsbad, CA, USA) were co-transfected GGT GGC TCT GG and 50-CCG CTC GAG GGT ATT GGC TTT 0 with 1 mg of the Hemgn-luciferase reporter gene construct GAC TTC AC; pHemgnÀ404 þ 196:5-CGC GGA TCC AAC AGC (pHemgn-2975 þ 196luc) and GATA1 (50–400 ng pPacGA- CTA CCT AGG AAG AG and 50-CCG CTC GAG GGT ATT GGC 12 0 TA1) using Fugenet 6 reagent (Roche Diagnostics Corpora- TTT GAC TTC AC; pHemgnÀ2975 þ 12:5-CGC GGA TCC CAC tion, IN, USA). Cells were harvested after 24 h for luciferase ATC AGA GAC ACC TTG CC and 50-CCG CTC GAG ACA CTG assays using the Single Luciferase Assay System (Promega). CAC AGG TGT GAG GG. Luciferase activities were normalized to total cell protein, The core GATA binding sequences in the plasmid measured by the Bio-Rad protein assay system. pHemgnÀ2975 þ 196luc construct were mutated to cATA, a

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 419 change reported to abolish GATA binding,17–19 using the Real time RT-PCR Quickchanget site-directed mutagenesis kit (Stratagene, La Total RNAs were extracted from mock-transfected, and GATA1 0 Jolla, CA, USA). The primer for GATAbox1 mutation was: 5 - stably transfected CMS cells or primary myeloblasts of newly CCT CAC ACC TGT GCA GTG TcA TAA AGA AAG TGG. The diagnosed acute myeloid leukemia patients using TriReagent 0 primer for GATAbox2 mutation was: 5 -GCT GTG GTC TAA CCA (Molecular Research Center Inc.). First strand cDNAs were cAT AAA ACT TTT AGG CGG G. Double mutations of both prepared from 1 mg RNA using random hexamer primers and a boxes were also generated. The primer for GATAbox3 mutation RT-PCR kit (Perkin Elmer), and purified with the QIAquick PCR was: 50-GTGTTGGTCTTaAgcTTTTCAA AAAGC. The mutant Purification Kit (Qiagen). GATA1, EDAG and 18S RNA mut mut mut constructs (named as GATAbox1luc, GATAbox2luc, GA- transcript levels were quantitated using a LightCycler real-time mut TAbox3luc and GATAbox1 þ 2luc) were recloned into pXP1 PCR machine (Roche). The primer sequences and the real-time vector (to eliminate possible mutation in the vector) and were PCR condition are detailed in Supplementary Information. verified by sequencing. EDAG and GATA1 transcript levels were expressed relative to 18S RNA. Real-time PCR results were expressed as mean values from two to three experiments using the same cDNA Generation and analysis of Hemgn promoter in preparation. transgenic mice The Hemgn promoter (pHemgnÀ2975 þ 196) linked to a lacZ reporter was used to generate transgenic mice (Cold Spring Clinical AML samples Harbor Transgenic Core Facility). The transgenic mice were Myeloblasts from children diagnosed with AML were obtained identified by PCR-genotyping using lacZ-specific primers. from the Children’s Hospital of Michigan leukemia cell bank. RNAs from embryos and tissues were isolated using Trizol Mononuclear cells were isolated on Ficoll-Hypaque gradients to (Invitrogen) and the expression of the lacZ transgene was obtain highly purified mononuclear cell fractions consisting examined by RT-PCR using lacZ-specific primers. The lacZ mostly of leukemic blasts. Total RNAs were extracted from the expression was also detected by X-gal staining as detailed in samples using TRIzol reagent (Life Technologies). The research Supplementary Information. protocol was approved by the Human Investigation Committee of Wayne State University School of Medicine.

Gel shift and ChIP assays Statistical analysis The nuclear extracts (3 mg) from a mouse erythroleukemia cell The nonparametric Spearman rank correlation coefficient was line (MEL) (Active Motif, Carlsbad, CA, USA) were used in gel used to analyze HEMGN transcript levels and their relationship shift assays. The experimental procedures were described 20 to GATA1 transcript levels. Statistical analyses were performed before. The oligonucleotides used are as follows: GATAsite1: with StatView (Version 4.5, for Windows). 50-CTGTGCAGTGTGATAAAGAAAGTGG; GACACGTCACAC TATTTCTTTCA-CC-50. mutGATAsite1: 50-CTGTGCAGTGTCTT 0 AAAG-AAAGTGG; ACACGTCACAGAATTTCTTTC-ACC-5 ; Results GATAsite2: 50-GGTCTAACCAGATAAAACTT-TTAGG; CCAGA 0 0 TTGGTCTATTTTGAAAAT-CC-5 ; mutGATAsite2: 5 -GGTCTAA Isolating mouse Hemgn promoters CCACTTAAAA-CTTTTAGG; CCTGATTGGTGAATTTTGAAA-A 0 Our previous studies characterized the temporospatial expres- TCC-5 . For the competition studies, 100-fold molar excess of sion patterns of Hemgn in hematopoiesis. To investigate the unlabeled oligonucleotide was included in the binding reaction. underlying regulatory mechanisms, we examined the Hemgn For supershift analysis, 2 mg anti-GATA1 or anti-GATA2 promoters both in vitro and in vivo. Sequence analyses antibodies (Santa Cruz Biotechnology) were incubated with confirmed a BAC clone (#567P14) containing the Hemgn locus nuclear extracts respectively at room temperature for 20 min (Figure 1a). As demonstrated by the Vista plot (Figure 1b), we prior to their use in the binding reactions. 14 compared sequences of mouse and human Hemgn genes. The ChIP assay was performed as described previously. Although mouse and human Hemgn were highly conserved in Purified chromatin fragments from Meg-01 (an AMkL cell line) the exons, there were few homologous regions in the introns. were incubated with anti-GATA1 antibodies (C-20, Santa Cruz However, several evolutionarily conserved regions (ECRs) are Inc.). Standard PCR for the HEMGN promoter region was located in the 50 UTR and 50 flanking sequence. These highly performed using forward (50-CCAGACACTTCCTGGCAGAT) 0 homologous regions contain potential regulatory elements that and reverse (5 -CACTTGACTTCCCGCCTAAA) primers spanning may control the Hemgn promoter activities. positions –141 to þ 89. A coding region (exon 3) of the human The sequences of the highly homologous proximal promoters GATA1 gene was also amplified using forward (50-TGGAGA 0 and the first exons of mouse and human Hemgn were aligned CTTTGAAGACAGAGCGGCTGAG) and reverse (5 -GAAGC (Figure 1c). The promoters contained a TATA box (TATAAA) that TTGGGAGAGGAATAGGCTGCTGA) primers to validate the was conserved in mouse and human. In the search for the specificity of the ChIP assays. binding sites of hematopoietic transcription factors, two conserved GATA boxes, designated GATAbox1 and GATAbox2, were identified in the 50 UTR that were exact matches to the 8 Western blot analyses consensus sequence WGATAR. The GATAbox3 has 1 bp The presence of GATA1 and GATA2 in MEL nuclear extracts mismatch with the consensus sequence. The presence of the was detected by Western blot assays using anti-GATA1 or anti- GATA boxes were in the Hemgn 50 UTR was confirmed by RT- GATA2 antibodies (Santa Cruz Biotechnology), as described PCR using mRNA from mouse spleen and human K562 cells. A 21 before. Primary antibodies were detected with POD con- forward primer containing the GATAbox1 and a reverse primer in jugated anti-goat-IgG and a BM Chemiluminescence Western exon 3 were used and the expected 897 and 607 bp amplicons Blotting kit (Roche). were amplified from spleen and K562 cDNAs, respectively (data

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 420 not shown). This establishes that GATA boxes are located in the during embryogenesis (Figure 3A and B). In the adults, we also 50 UTR of both mouse and human Hemgn mRNAs. detected the lacZ transgene expression in hematopoietic tissues including spleen and bone marrow, but not in nonhematopoie- tic tissues such as heart, liver, kidney and skeletal muscle The Hemgn promoter is specifically transactivated in a (Figure 3A). However, the expression of the lacZ transgene was hematopoietic cell line also unexpectedly detected in the brain, thymus, lung and testis To investigate the regulatory mechanisms of Hemgn expression (Figure 3A). during hematopoiesis, we isolated a putative Hemgn promoter Histological analyses demonstrated that the lacZ staining was that contained 2975 bp of 50 upstream sequence and 196 bp of present in adult bone marrow cells (indicated by yellow arrow, the 50 UTR. To determine whether this region was sufficient to Figure 3B-a,b). In adult spleen, the lacZ staining was localized activate Hemgn transcription, we generated a plasmid in the red pulp (RP), which is active in erythropoiesis (Figure 3B- pHemgnÀ2975 þ 196luc in which the luciferase reporter was c,d). The x-gal staining was also observed in the hematopoietic driven by the Hemgn promoter. When transfected into an islands (yellow arrow) of the fetal liver (E13.5) (Figure 3B-e,f). erythroleukemic cell line (K562), the pHemgnÀ2975 þ 196 pro- These results demonstrate that the pHemgnÀ2975 þ 196 promoter moter was sufficient to activate the transcription of the luciferase contains critical regulatory elements for Hemgn expression in reporter (Figure 2). This promoter construct showed much lower hematopoietic tissues in vivo. transactivation activities in nonhematopoietic cells including 10T1/2 (a fibroblast cell line), PAC1 (a smooth muscle cell line), COS-7 (a kidney cell line) and C2C12 (a skeletal muscle cell Functional analysis of GATA binding elements in the line) (Figure 2). Furthermore, a shorter pHemgnÀ831 þ 196 Hemgn promoter promoter construct that excluded the ECRs at À2kb and À3kb To identify the regulatory elements for Hemgn promoter, we (Figure 1b) did not significantly affect the transcriptional generated deletion mutants of the Hemgn promoter using activities and specificity (Figure 2). These results demonstrate pHemgnÀ2975 þ 196luc as a template (Figure 4). The wild-type that the Hemgn promoters are selectively activated in hemato- poietic cells in vitro.

The Hemgn promoter is activated in hematopoietic tissues in vivo To determine whether the Hemgn promoter is sufficient to activate transcription in vivo, we examined the expression of a lacZ transgene driven by the pHemgnÀ2975 þ 196 promoter in transgenic mice. Given that the endogenous Hemgn is specifically expressed in hematopoietic tissues during hemato- poiesis, we examined the transgene expression in hematopoietic tissues from fetal and adult tissues. We generated seven independent transgenic founders that show similar expression pattern of transgene. One of the representative transgenic mouse lines is described here. We demonstrated that the pHemgnÀ2975 þ 196 promoter was sufficient to activate the transcription of lacZ transgene in the yolk sac and fetal liver

Figure 3 The Hemgn promoter is transcriptionally active in hematopoietic tissues in vivo. The lacZ transgene driven by the pHemgnÀ2975 þ 196 promoter was used to examine the promoter Figure 2 The Hemgn promoter is specifically expressed in hemato- activities in transgenic mice. (A) The expression of lacZ transgene poietic cells in vitro. The luciferase reporter driven by indicated was detected in RNAs extracted from yolk sac at E9.0, fetal liver at Hemgn promoter or the promoter-less vector was transfected into both E13.5, and indicated adult tissues in transgenic mice by RT-PCR hematopoietic (K562) and nonhematopoietic (PAC1, COS7, C2C12 and assays. (B) X-gal staining shows that the lacZ transgene was expressed 10T1/2) cell lines. The data are expressed as the fold increased activity in adult bone marrow cells (indicated by yellow arrow, (a, b), in the of pHemgnÀ2975 þ 196luc and pHemgnÀ831 þ 196luc versus the promo- red pulp (RP), not white pulp (WP) of the spleen (c, d). The x-gal ter-less reporter vector pXP1. All transfection experiments were staining was also observed in the hematopoietic islands (yellow arrow) performed at least three times in triplicate. The relative luciferase of the fetal liver (E13.5) (e, f). Bars represent 25 mm(a, b), 10 mm(c, d), activity was the mean7s.e.m. of these experiments. 100 mm(e, f), respectively.

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 421

Figure 4 GATA boxes in the 50 UTR are indispensable for the transactivation activities of the Hemgn promoter in K562 cells. (a) The wild-type and deletion mutants of the Hemgn promoter were transfected into the K562 cells. The luciferase activity of the promoter-less vector (pXP1) was served as the basal level activities. (b) Site-direct mutagenesis analyses of GATA boxes in pHemgnÀ2975 þ 196. GATAbox1, GATAbox2 and GATAbox3 were mutated as indicated in pHemgnÀ2975 þ 196luc vector. The resulting mutants were transfected into K562 cells and the luciferase activities of each mutant were compared with the wild-type promoter. The relative luciferase activity was the mean7s.e.m. of at least three replicate experiments.

and mutated promoters were transfected into K562 cells and the Hemgn, via binding to the GATAbox1 and GATAbox2 sites in the transactivation activity of each promoter was compared with 50 UTR. To directly assess whether GATA1 and GATA2 can bind that of the promoterless reporter. The pHemgnÀ2975 þ 196 to the Hemgn promoter, we performed electrophoretic mobility promoter activated the reporter 80-fold over the basal activity shift assays using nuclear extracts from a mouse erythroleukemia of the promoterless vector (pXp1). The pHemgnÀ831 þ 196 cell line, MEL, and a DNA oligo containing either the GATAbox1 construct did not show significant reduction in promoter or the GATAbox2 as a probe. This analysis demonstrated that two activity, whereas the pHemgnÀ404 þ 196 promoter showed a protein complexes bound to both GATAbox1 and GATAbox2 40% decrease in the activity (Figure 4a). Remarkably, when the probes (Figure 5a). The faster migrating complex could be 184 bp sequence in the 50 UTR was deleted from the specifically competed away by excess (100 Â ) wild type (i.e., pHemgnÀ2975 þ 196 promoter, the activity of the resulting GATAbox1, GATAbox2) oligos and the consensus GATAbox oligo, pHemgnÀ2975 þ 12 promoter construct was reduced to the basal but not the GATAbox1 and GATAbox2 oligos including the level, indicating that the 184 bp DNA fragment contains critical identical mutations in the reporter gene assays (Figure 5a). regulatory elements (Figure 4a). Moreover, anti-GATA1 antibody disrupted the binding of this As shown in Figure 1c, the deleted 184 bp sequence contains complex to both GATAbox1 and GATAbox2. However, the anti- three GATA boxes. Given the established roles of GATA factors GATA2 antibody failed to disrupt the binding complex in hematopoietic gene regulation, it seemed likely that the (Figure 5a). Although the slower migrating complex could be GATA boxes in the 50 UTR are critical for Hemgn transactiva- specifically competed away unlabeled wild type oligos, the tion. Since previous studies have shown that the G in the GATA consensus GATAbox oligo and the anti-GATA antibodies had no core sequence is critical for the binding of GATA factors,22–24 effect on this DNA–protein complex, indicating its association the sequence was mutated to a/cATA to abolish GATA binding with DNA binding proteins other than GATA. Western blot 17–19 activity. Indeed, in our analysis mutation of the GATAbox1 analysis confirms that both GATA1 and GATA2 proteins are or GATAbox2 sites suppressed promoter activity in K562 cells by present in MEL nuclear extracts (Figure 5b). This study about 95 and 90%, respectively (Figure 4b). Mutation at both establishes that GATA1, but not GATA2, in MEL nuclear extracts GATA boxes completely eliminated the promoter activity binds to the GATAbox1 and GATAbox2 elements in the Hemgn (Figure 4b). However, mutation at GATAbox3 did not affect the promoter. promoter activities (Figure 4b). These results strongly suggest In vivo binding of GATA1 to the HEMGN promoter was that both GATAbox1 and GATAbox2 are essential for high-level confirmed by ChIP assays using primers flanking the proximal Hemgn promoter activity in hematopoietic cells. HEMGN promoter and the GATA1 antibodies-precipitated chromatin fragments from Meg-01 cells. The binding of GATA1 is only detected in anti-GATA1-precipitated chromatin frag- GATA1, not GATA2, binds and transactivates the ments, and not in nonspecific IgG precipitated chromatin Hemgn promoter fragments (Figure 5c). We also used the primers located in the Since GATA3 is predominantly expressed in lymphoid lineages, coding region (exon 3) of the GATA1 gene to confirm the GATA1 and GATA2 would seem to be the possible regulators for specificity of the precipitation: GATA1 fails to immunoprecipitate

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 422 of GATA1 co-transfection. GATA1 transactivated the wild-type Hemgn promoter (pHemgnÀ2975 þ 196) in a dose-dependent manner; however, it failed to transactivate the mutant Hemgn mut promoter p GATAbox1 þ 2HemgnÀ2975 þ 196 (Figure 5d). These results suggest that GATA1 is a key transcription factor that binds and transactivates the Hemgn promoter.

Increasing the endogenous HEMGN transcripts by GATA1 in CMS cells The effect of GATA1 on the endogenous HEMGN promoter was assessed by overexpressing GATA1 in CMS cells (an AML cell line) and determining changes in levels of endogenous EDAG transcripts. EDAG transcripts were measured by real time RT- PCR and normalized to 18S rRNA. Accordingly, expression of GATA1 increased levels of endogenous EDAG transcripts (Figure 6a and b). These results suggest that exogenous GATA1 is sufficient to transactivate the endogenous HEMGN promoter.

Correlation between EDAG and GATA1 transcripts in acute myeloid leukemia patient samples A recent study reported that EDAG is highly expressed in myeloblasts from acute myeloid leukemia patients.6 To further confirm that EDAG is a downstream target of GATA1, we performed the real time RT-PCR to quantify EDAG and GATA1 transcript levels in primary myeloblast specimens. As shown in Figure 6c, both EDAG and GATA1 transcripts varied over a substantial range (for EDAG: 0.06 to 19.85 with a mean of 0.6325; for GATA1: 0.016 to 3.988 with a mean of 0.1350). Importantly, EDAG transcripts significantly correlated with the GATA1 transcript levels (r ¼ 0.93, p ¼ 0.0003). This result strongly supports a regulatory role of GATA1 in EDAG transcription in AML.

Discussion

Our previous studies demonstrate that Hemgn is highly expressed in hematopoietic stem/progenitor cells during hema- Figure 5 GATA1, not GATA2, binds and transactivates the Hemgn topoiesis.1 In this report, we examined the regulation of Hemgn promoter. (a) EMSA assays were performed using MEL nuclear extract expression both in vitro and in vivo. We isolated a Hemgn with either the end labeled GATAbox1 or GATAbox2 oligonucleotides as probes. The molar ratio of the probe oligo versus the competing oligo promoter that contained 2975 bp of upstream flanking DNA 0 is 1:100. (b) Both GATA1 and GATA2 are expressed in MEL nuclear sequence and 196 bp of 5 UTR sequence. This promoter was extracts by western blot assays. (c) ChIP assays show that GATA1 preferentially expressed in a cultured hematopoietic cell line specifically binds to the HEMGN promoter, but not to the exon 3 of and hematopoietic tissues in transgenic mice. Mutagenesis GATA1 gene in the context of chromatin using the chromatin studies demonstrated that two GATA boxes in the 50 UTR were fragments from Meg-01 AML cell line and the primers described in required for promoter transactivation. GATA1, but not GATA2, ‘Supplementary Information’. (d) The indicated amount of GATA1 in nuclear extracts prepared from a mouse erythroleukemia cell expression vector and 1 mgofpHemgn2175 þ 196luc or its corresponding GATAboxes mutant reporter were co-transfected into a D-Mel2 cell line (MEL) bound the Hemgn promoter at these GATA binding line. The luciferase activities of the Hemgn promoters in the absence sites. of GATA1 expression vector were normalized to 100%. The relative The expression profiles of hemogen in mice, humans and rats 7 luciferase activity was the mean s.e.m. of three replicate experi- are highly enriched in hematopoietic tissues.1–3 Recently, we ments. reported a distinct Hemgn mRNA isoform in round spermatids of the testis.21 However, transcription of this testicular mRNA the exon 3 (the coding region) of the GATA1 gene (Figure 5c). isoform is driven by an alternative promoter located at 6 kb This result demonstrates that GATA1 binds to the proximal upstream of the hematopoietic promoter.21 Consistent with promoter of the HEMGN promoter in the context of these studies, the Hemgn promoter (pHemgnÀ2975 þ 196)is chromatin. predominantly expressed in hematopoietic tissues (Figure 3). To assess directly the role of GATA1 protein in Hemgn Therefore, it is likely that the Hemgn promoter tested in transactivation, GATA1 cDNA in pPac vector was co-trans- transgenic mice contains the critical regulatory elements fected with the wild-type or the double GATA-box mutant necessary for Hemgn expression in hematopoietic tissues. Hemgn promoter into D. Mel-2 cells in which GATA1 promoter Although the pHemgnÀ2975 þ 196 promoter is preferentially is not detected. Therefore, both the wild-type and the mutant expressed in hematopoietic tissues in vivo, promoter activities Hemgn promoters show basal luciferase activities in the absence were much lower than expected since the X-gal staining

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 423

Figure 6 The expression of EDAG correlates with that of GATA1 in human myeloblasts and human leukemia-derived cell lines. (a) The coding cDNA of GATA1 gene was amplified by PCR amplification and subcloned into a mammalian expression vector, pcDNA3. The GATA1 construct was stably transfected into CMS, an AML cell line. Overexpression of GATA1 in two of the stable clones was confirmed by real-time RT-PCR. (b) Increasing the endogenous EDAG transcripts by GATA1 in CMS cells. EDAG transcripts were measured by real time RT-PCR and normalized to 18S rRNA in the CMS GATA1 stable clones. The expression of GATA1 was in parallel with EDAG transcripts. (c) Correlation between EDAG and GATA1 transcripts was determined by real-time RT-PCR in 10 pediatric AML patient samples. The nonparametric Spearman rank correlation coefficient was used to analyze the relationship between EDAG and GATA1 transcripts and was found to be statistically significant.

detected in the transgenic mice become apparent only after Hemgn transcription. It will be of particular interest to establish overnight staining (Figure 3). These results suggest that the extent to which GATA protein binding within the 50 UTR enhancer-like regulatory elements may exist beyond the interacts with the upstream basal transcriptional machinery 3171 bp promoter region described herein. We also detected (e.g., TATA box) to modulate Hemgn transcription. We envisage ectopic expression of the Hemgn transgene in the brain, thymus, that a dynamic process of binding and dissociation of GATA lung and testis by RT-PCR. This implies that additional factors is involved in transcriptional initiation and elongation of regulatory elements are required to ensure the repression of the Hemgn gene. the endogenous Hemgn in thymus and nonhematopoietic Since both GATA1 and GATA2 bind to the consensus DNA tissues. sequence WGATAR that is present in promoters of a number of Although GATA elements appear to be essential in determin- hematopoietic genes,25 it is not fully understood how these ing the transactivation of the Hemgn promoter, it does not GATA family members distinguish different DNA sequences to exclude the possibility of an involvement of other regulatory regulate different genes. Previous studies have revealed some elements. According to current models, GATA factors are variations in DNA binding specificities among different GATA central mediators for hematopoietic gene regulation.25 There- factors. For instance, the N-terminal zinc fingers of GATA1, fore, it is very likely that GATA box binding factors regulate GATA2 and GATA3 recognize different DNA motifs and may Hemgn expression by recruiting other hematopoietic regulators contribute to their selectivity in binding target DNA sites.26 that bind directly or indirectly to the Hemgn promoter. We have Extensive studies using both in vitro and in vivo models found a cluster of evolutionarily conserved regulatory boxes for demonstrate that the functions and downstream targets of hematopoietic regulators in the Hemogen upstream region such GATA1 overlap with those of GATA2.25,27,28 Moreover, in as AML-1, Evi-I and Ets (Figure 1c). Studies are underway to GATA1 knockout mice, GATA2 is upregulated to partially further explore the possible molecular interactions between compensate for the loss of GATA1, and both GATA1 and GATA factors and other hematopoietic regulators in regulating GATA2 are critical for initiation of blood formation.28,29

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 424 In this report, we investigated the potential roles of GATA1 tic development and its human homologue EDAG maps to and GATA2 in regulating Hemgn transcription. We found that 9q22, a region containing breakpoints of hematolo- only GATA1 can bind and transactivate the Hemgn promoter. gical neoplasms. Mech Dev 2001; 104: 105–111. Consistent with this observation, the expression level of EDAG 2 Lu J, Xu WX, Wang SY, Zhan YQ, Jiang Y, Cai WM et al. Isolation and Characterization of EDAG-1, A Novel Gene Related to closely correlated with that of GATA1 in human leukemia- Regulation in Hematopoietic System. Sheng Wu Hua Xue Yu derived cell lines and primary AML specimens (Figure 6). Sheng Wu Wu Li Xue Bao 2001; 33: 641–646. Although GATA1 and GATA2 bind to the same consensus 3 Kruger A, Ellerstrom C, Lundmark C, Christersson C, Wurtz T. nucleotides, flanking sequences may further influence their RP59, a marker for osteoblast recruitment, is also detected in binding preferences. The differential binding observed in our primitive mesenchymal cells, erythroid cells, and megakaryocytes. EMSA assay could be due to the binding affinity difference of Dev Dyn 2002; 223: 414–418. 4 Li CY, Zhan YQ, Xu CW, Xu WX, Wang SY, Lv J et al. EDAG GATA1 versus GATA2 to the Hemgn promoter. It remains to be regulates the proliferation and differentiation of hematopoietic investigated whether and how GATA2 is involved in the cells and resists cell apoptosis through the activation of nuclear expression of Hemgn in vivo. This question will be addressed factor-kappa B. Cell Death Differ 2004; 11: 1299–1308. using GATA1 and GATA2 knockout mice. 5 Lu J, Xu WX, Wang SY, Jiang Y, Li CY, Cai WM et al. In conclusion, we provide evidence to show that Hemgn is a Overexpression of EDAG-1 in NIH3T3 cells leads to malignant direct transcriptional target of GATA1, a key regulator of transformation]. Sheng Wu Hua Xue Yu Sheng Wu Wu Li Xue Bao erythropoiesis. GATA1 deficient mice die during early embry- (Shanghai) 2002; 34: 95–98. 6 An LL, Li G, Wu KF, Ma XT, Zheng GG, Qiu LG et al. High ogenesis from severe anemia with erythroid differentiation expression of EDAG and its significance in AML. Leukemia 2005; arrested at the proerythroblast stage and rapid apoptosis of 19: 1499–1502. these cells ,29 suggesting a role of GATA1 in erythroid cell 7 Perry C, Soreq H. Transcriptional regulation of erythropoiesis. Fine survival and maturation. Hemgn also appears to play a key role tuning of combinatorial multi-domain elements. Eur J Biochem in erythropoiesis. Overexpression of EDAG (the Hemgn human 2002; 269: 3607–3618. ortholog) results in resistance of cell apoptosis, and expression 8 Evans T, Reitman M, Felsenfeld G. An erythrocyte-specific DNA-binding factor recognizes a regulatory sequence common of EDAG in K562 cells promotes the cell growth and inhibits 4 to all chicken globin genes. Proc Natl Acad Sci USA 1988; 85: erythroid differentiation, which is consistent with the expres- 5976–5980. 1 sion of Hemgn in erythroblasts and the downregulation during 9 Patient RK, McGhee JD. The GATA family (vertebrates and erythroid differentiation.2 Hemgn may thus play a positive role invertebrates). Curr Opin Genet Dev 2002; 12: 416–422. in cell survival and a negative role in erythroid terminal 10 Ohneda K, Yamamoto M. Roles of hematopoietic transcription differentiation. The molecular mechanism by which Hemgn factors GATA-1 and GATA-2 in the development of red blood cell lineage. Acta Haematol 2002; 108: 237–245. exerts its biological function remains unknown, however, a 11 Mayor C, Brudno M, Schwartz JR, Poliakov A, Rubin EM, Frazer putative coiled-coil motif in its N-terminus suggests this protein KA et al. VISTA: visualizing global DNA sequence alignments of 1 may dimerize or bind with other proteins. Further experiments arbitrary length. Bioinformatics 2000; 16: 1046–1047. are warranted to investigate how hemgn interacts with other 12 Ge Y, Jensen TL, Stout ML, Flatley RM, Grohar PJ, Ravindranath Y nuclear factors such as EKLF and GATA1 in the regulation of et al. The role of cytidine deaminase and GATA1 mutations in the erythroid development. increased cytosine arabinoside sensitivity of Down syndrome A recent study has reported an inverse relationship between myeloblasts and leukemia cell lines. Cancer Res 2004; 64: 728–735. HEMGN/EDAG expression and the chemotherapy response for 6 13 Drexler HG. The leukemia-lymphoma cell lines factsbook. AML patients and our own results strongly suggest a causal Academic Press: San Diego California USA, 2001, pp 646–647. relationship between GATA1 and EDAG in primary AMLs. 14 Ge Y, Jensen TL, Matherly LH, Taub JW. Transcriptional regulation Since mutation of GATA1 has been shown to contribute to of the cystathionine-beta-synthase gene in Down syndrome and the chemotherapy sensitivity of AML in Down’s syndrome non-Down syndrome megakaryocytic leukemia cell lines. Blood patients,12 it is conceivable that the downregulation of 2003; 101: 1551–1557. 15 Ge Y, Stout ML, Tatman DA, Jensen TL, Buck S, Thomas RL et al. HEMGN may be involved in molecular mechanisms of the GATA1, cytidine deaminase, and the high cure rate of Down remarkable chemotherapy sensitivity of this unique group of syndrome children with acute megakaryocytic leukemia. J Natl AML patients. Cancer Inst 2005; 97: 226–231. 16 Nordeen SK. Luciferase reporter gene vectors for analysis of promoters and enhancers. Biotechniques 1988; 6: 454–458. Acknowledgements 17 Crossley M, Tsang AP, Bieker JJ, Orkin SH. Regulation of the erythroid Kruppel-like factor (EKLF) gene promoter by the erythroid We appreciate the access to microscopes in Dr Richard Vander transcription factor GATA-1. J Biol Chem 1994; 269: Heide’s laboratory and the Imaging Core Facilities of Karmanos 15440–15444. 18 Anderson KP, Crable SC, Lingrel JB. Multiple proteins binding to a Cancer Institute at Wayne State University. We thank Dr Giuseppe GATA-E box-GATA motif regulate the erythroid Kruppel-like factor Rossi and Scott Goustin for helpful discussions. This work was (EKLF) gene. J Biol Chem 1998; 273: 14347–14354. supported by a grant from the Children’s Research Center of 19 Anderson KP, Crable SC, Lingrel JB. The GATA-E box-GATA motif Michigan (to L Li), the seed grant from the department of Internal in the EKLF promoter is required for in vivo expression. Blood Medicine at Wayne State University (to L Li), NHLBI/NIH 2000; 95: 1652–1655. HL58916 (to L Li), NCI Grant RO1 CA92308 (to JW Taub), NCI 20 Qiu P, Feng XH, Li L. Interaction of Smad3 and SRF-associated complex mediates TGF-beta1 signals to regulate SM22 transcrip- Grant RO1 CA76641 (to L Matherly), the Leukemia and tion during myofibroblast differentiation. J Mol Cell Cardiol 2003; Lymphoma Society (to JW Taub), and the Children’s Research 35: 1407–1420. Center of Michigan (to Y Ge). 21 Yang LV, Heng HH, Wan J, Southwood CM, Gow A, Li L. Alternative promoters and polyadenylation regulate tissue-specific expression of Hemogen isoforms during hematopoiesis and References spermatogenesis. Dev Dyn 2003; 228: 606–616. 22 Schwartzbauer G, Schlesinger K, Evans T. Interaction of the 1 Yang LV, Nicholson RH, Kaplan J, Galy A, Li L. Hemogen is a erythroid transcription factor cGATA-1 with a critical auto- novel nuclear factor specifically expressed in mouse hematopoie- regulatory element. Nucleic Acids Res 1992; 20: 4429–4436.

Leukemia Transcriptional control of the hemogen promoter by GATA1 LV Yang et al 425 23 Wall L, deBoer E, Grosveld F. The human beta-globin gene 30 GATA-3 are independent sequence-specific DNA binding do- enhancer contains multiple binding sites for an erythroid-specific mains. EMBO J 1997; 16: 2874–2882. protein. Genes Dev 1988; 2: 1089–1100. 27 Testa U. Apoptotic mechanisms in the control of erythropoiesis. 24 Ko LJ, Engel JD. DNA-binding specificities of the GATA Leukemia 2004; 18: 1176–1199. transcription factor family. Mol Cell Biol 1993; 13: 4011– 28 Fujiwara Y, Chang AN, Williams AM, Orkin SH. Functional 4022. overlap of GATA-1 and GATA-2 in primitive hematopoietic 25 Cantor AB, Orkin SH. Transcriptional regulation of erythropoiesis: development. Blood 2004; 103: 583–585. an affair involving multiple partners. Oncogene 2002; 21: 29 Fujiwara Y, Browne CP, Cunniff K, Goff SC, Orkin SH. Arrested 3368–3376. development of embryonic red cell precursors in mouse embryos 26 Pedone PV, Omichinski JG, Nony P, Trainor C, Gronenborn AM, lacking transcription factor GATA-1. Proc Natl Acad Sci USA Clore GM et al. The N-terminal fingers of chicken GATA-2 and 1996; 93: 12355–12358.

Supplementary Information is available on the Leukemia website (http://www.nature.com/leu/)

Leukemia