Comparative genomics analysis

The expression patterns of GATA-2 in hematopoietic and neuronal tissues are conserved in vertebrates, suggesting that the arrangement and sequence of the GATA-2 genomic locus may be highly conserved. Comparative analyses of a158 kbp sequence spanning the zebrafish GATA-2 locus and approximately 400 kbp sequences of GATA-2 from human, mouse and rat and 190 Kb genomic sequence flanking the fugu GATA-2 locus has revealed a conserved syntenic relationship of GATA-2 with RPN1 (Figure 1A). A similar arrangement of general exon and intron structures and high sequence homology in exons has also been observed in the genomes of the five species (Figure 1B). In addition, we have identified highly conserved non-coding sequences in the genomicregion flanking GATA-2 (Figure 1A and 1B). We identified two conserved non-coding sequences, one at ~13 and the other ~10 kbp upstream of zebrafish GATA-2 start codon (Figure 1A and 1B). These sequences are located in approximately the same location in human, mouse, and rat, but in fugu are located closer to the GATA-2 coding sequence (4.7 Kbp and 3.8 Kbp, respectively). A conserved non-coding sequence was also found downstream of GATA-2 in fugu, mouse, rat, human and zebrafish genomes (Figure 1A and 1B).

The conserved sequence located between12892-12811 bp upstream of the zebrafish GATA-2 start codon (Up1) is 82 bp. Zebrafish Up1 sequence has an 81% sequence identity with human Up1 sequence, an 80% sequence identity with both mouse and rat Up1sequences and an 83% sequence identity with that of fugu (Figure 1C). A conserved sequence located 9887–10110 bp upstream of the GATA-2 start codon (Up2) is 224 bp. Zebrafish Up2 sequence has a 74%, 72%, 72% and 94% sequence identity with Up2 sequences from human, mouse, rat and fugu genomes, respectively (Figure 1C).

An alternative first exon (IS) has been found in the GATA-2 genomic sequences of human, rodents and chick [17–19]. In mouse genome, the conserved Up2 region (230 bp) is located at about 3 Kb 5' upstream of IS, which is approximately 4.6 Kb further upstream of the general exon IG of GATA-2 gene. To determine if alternative first exons also exist in zebrafish, 5' RACE was performed using mRNA from zebrafish embryos at the stages of 22~25 somites. As shown in Supplementary Figure S1 (see Additional file 1), only one PCR band was amplified, indicating that there is no alternative transcriptional product. This finding is consistent with the data from analyzing GATA-2 mRNA of 75% epiboly stage zebrafish embryos reported by Oren et al [4]. Although zebrafish is different, the genomic arrangement of IS among human, rodent and chick genomes are conserved.

Functional analysis of the putative regulatory elements of GATA-2 gene in zebrafish

As noted, during zebrafish embryogenesis, GATA-2 is expressed in the embryonic hematopoietic tissue Intermediate Cell Mass (ICM) and in neuronal cells (Figure 3A). ICM specific gene expression of GATA-2 can be recapitulated in transgenic zebrafish embryos carrying a GFP-modified BAC construct that includes 20 kb genomic sequence upstream of and adjacent to the GATA-2 start codon (Figure 2B and 3B)[9]. In contrast, GFP expression was not observed in the ICM of transgenic zebrafish embryos carrying constructs containing only 7.3 kbp of upstream sequence, although strong GFP expression was seen in neuronal cells in these embryos (Figure 2C and 3C). These results indicate that a hematopoietic regulatory element is located between 7.3 and 20 kbp upstream of the zebrafish GATA-2 start codon.

To determine whether conserved non-coding sequences located in this region, Up1 or Up2 are necessary, alone or in combination, to drive hematopoietic expression, we generated germline stable transgenic zebrafish carrying modified BAC constructs having deletions of Up1 or Up2 or both. In progeny of transgenic zebrafish harboring BACΔUp1/GFP, strong fluorescent expression was observed in the ICM region at 24 hpf (Figure 2D and 3D). GFP expression was stronger in embryos carrying BACΔUp1/GFP than in transgenic zebrafish embryos carrying BAC/GFP (Figure 2B and 3B). GFP expression was not observed in the ICM of embryos carrying BACΔUp2/GFP (Figure 2E and 3E) or BACΔUp1ΔUp2/GFP (Figure 2F and 3F); however, GFP expression in neuronal cells was not affected in these embryos (Figure 3E).

To determine whether these elements are sufficient, alone or in combination, to induce hematopoietic expression, GFP reporter gene constructs that contained Up1 or Up2 or both (Figure 2G–L) were used to generate transgenic zebrafish. The conserved non-coding sequences found 12.9 kb and 10 kb, respectively, upstream of the GATA-2 start codon were inserted into Tol2-transposon cassettes proximal to a GATA-2 minimal promoter/GFP fusion generating a series of construct combinations (Figure 2G~L). These constructs were flanked by the Tol2 cis-acting elements, which are required for genome integration of the construct. Each cassette was microinjected into fertilized eggs containing maternally expressed Tol2 transposase. Tol2-transposon mediated germline transmission of a GATA-2 minimal promoter/GFP reporter gene fusion (mp/GFP, Figure 2G) or of constructs with Up1, and/or Up2, inserts (Figure 2H~L) was determined by PCR. This approach, which typically produces at least 40% germline transgenic zebrafish, resulted in multiple stable transgenic zebrafish lines for each construct analyzed.

In zebrafish transgenic embryos carrying the Up1-mp/GFP construct (N = 3) (Figure 2I), no ICM-specific expression was observed; however, strong fluorescence was detected in the heart at 24 hpf (Figure 3G). In progeny of three independent transgenic zebrafish lines carrying Up2-mp/GFP, fluorescent expression was observed in blastoderm cells at 80% epiboly (data not shown). At 24 hpf, in the same transgenic lines, strong GFP expression was seen in the ICM, and weakly in the spinal cord (Figure 2J and 3H). The ICM expression pattern of Up2-mp/GFP correlates well with GATA-2 expression pattern determined by in situ hybridization (Figure 3A). Two Up2 fragments (112 bp Up2a and 112 bp Up2b, Figure 1C, Figure 2K and 2L) were also analyzed for the ability to drive hematopoietic expression in transgenic zebrafish. In four independent transgenic zebrafish lines, we found that Up2a can drive GFP expression in the ICM although less fluorescence was observed than in the Up2 transgenic and Up2a transgenic zebrafish embryos also had non-specific GFP expression in the notochord (Figure 2K and 3I). We also obtained several lines of transgenic zebrafish carrying a GFP expression vector in which both non-coding sequence elements were linked to GATA-2 minimal promoter (Figure 2H); GFP expression in the ICM and the heart was observed at 24 hpf (data not shown). GFP expression was undetectable in five independent transgenic zebrafish lines carrying GATA-2 minimal promoter construct (Figure 2G).

Conservation of the non-coding sequence, Up2, in five vertebrate species suggests functional conservation. To determine whether this enhancer element is active in mammalian hematopoietic cells, we performed transient transfection assays with a construct containing Up2. We transfected two human myeloid leukemia cell lines (KG-1 and K562) and a non-hematopoietic mammalian cell line NIH 3T3, with the mp/GFP, Up2-mp/GFP and the Up2a-mp/GFP constructs along with a CMV β-galactosidase plasmid to normalize for transfection efficiency. In myeloid cell lines, we observed increased GFP expression and this activity was stronger with the entire Up2 compared to activity of the minimal promoter (Figure 4D). Transfection with the Up2a element also showed increased GFP expression but this activity was less than that of the Up2. In both cases, the differences were found to be statistically significant when compared to the minimal promoter (Figure 4D). No GFP expression was observed in NIH3T3 cells; β-galactosidase expression was observed in this cell line indicating that Up2 and Up2a enhancer activities are specific to hematopoietic cells. Finally, we generated a reporter construct containing the mouse Up2 linked to GFP and analyzed its activity in transient transgenic zebrafish assays. As shown in Supplementary Table S (see additional file 2), mouse and zebrafish Up2 showed a similar enhancer activity to direct hematopoietic expression of GFP.

Analysis of binding sites for potential transacting factors

Transfac6.0 Match and P-Match were used to search databases[20] for potential transcription factor binding sites involved in GATA-2 hematopoietic expression directed by Up2 sequence. Multiple transcription factor binding sites including those for AML1a, GATA-2, GATA-3, En-1, FAC1, HNF-1, E2F-1, LMO2, ARP-1 and HOXA3 were identified. Constructs carrying mutations on the binding sites for transcription factor AML-1, E2F-1, ARP-1, LMO2 and HOXA3 were generated and injected into zebrafish embryo for transient assays of hematopoietic GFP expression. Under a condition that is statistically significant, 19.49% (n = 318), 21.79% (n = 335) and 22.09% (n = 335) of the embryos injected with construct carrying mutation in E2F-1, LMO2 and HOXA3-A binding site, respectively, showed transient hematopoietic GFP expression (Figure 4A and 4B). These percentages are significantly lower than that of embryos injected with the control Up2 construct (35.2%). Other constructs did not show any significant differences (see additional file 2, Supplementary Table S).

EMSA showed that, in vitro, proteins of transcription factors E2F-1 and HOXA3 specifically formed the protein-DNA complexes with the probes containing LMO2 and HOXA3-A binding sites present in Up2, respectively (Figure 4C). With the anti-E2F-1 antibody, a super shift was observed while there was no super shift with a non-specific antibody. The shifted band was not observed with the probes containing a mutation in the E2F-1 and HOXA3-A binding site, respectively.

The function of E2F-1, LMO2 and HOXA3 binding sites were further analyzed in the germline transgenic zebrafish embryos carrying the described base change mutations in the same Up2-mp/GFP construct context. A base change mutation in E2F-1 binding site showed increased non-specific expression in muscle and notochord (Figure 3K) and less severe but notably reduced expression in hematopoietic cells. Mutation within LMO2 and HOXA3-A binding sites resulted in almost elimination of GFP expression in hematopoietic cells, showing a mosaic GFP positive pattern in less than 10% of hematopoietic cells compared to Up2-mp/GFP wild type construct (Figure 3L and see additional file 3, Supplementary Figure S2). As a control, mutation within the ARP-1 binding site showed no effect on GFP expression (Figure 3J). These results revealed that integrity of these putative LMO2 and HOXA3-A binding sites is required for proper regulation of GATA-2/GFP expression in hematopoietic cells.

Identification of highly conserved non-coding sequences in the genomic region bordering GATA-2

Pipmaker software was used to identify conserved non-coding sequences in genomic sequences flanking GATA-2 in mouse, rat, human and fugu and that of the zebrafish genome [34–36]. A 158 kbp zebrafish genomic sequence (Sanger Institute accession number: AL928619, Figure 1) was compared to orthologous regions of the other genomes. The genomic sequences from other species used for comparison were: rat chromosome 4 fragment 122,079,753 to 122,488,528 (~408 kbp); mouse chromosome 6 fragment 88,429,180 to 88,842,527 (~413 kbp); human chromosome 3 fragment 129,480,974 to 129,894,726 (~413 kbp) and fugu scaffold_374 (~189 kbp).

Construction of Tol2 transposon vectors

Highly conserved zebrafish motifs (Up1, and Up2) identified by comparative genomics were amplified by PCR. The primers used in the PCR reactions were:

Up1 5' primer: 5'-ATCCGCTCGAG TTTGACCTTTTCGGAAAACGAGCT-3'

Up1 3' primer: 5'-CGGAATTC TCTGATGGTCTCCACGGCAA-3'

Up2 5' primer: 5'-ATCCGCTCGA G ATCTTCCCCCTCAAATGATGCA-3'

Up2 3' primer: 5'-CGGAATTC GGGGGTAATATTTCATCGGCGT-3'

Up2a 3' primer: 5'-CGGAATTC TGGAAACAGGGTCCAAAGT-3'

Up2b 5' primer: 5'-ATCCGCTCG AG CCTCGCTCGGTCCTGACA-3'

The underlined bases correspond to restriction enzyme sites for Xho I (CTCGAG) or EcoR I (GAATTC); italicized bases are zebrafish genomic sequences. PCR products were digested with restriction enzymes Xho I and EcoR I and inserted proximal to a GATA-2 minimal promoter (mp)/GFP reporter gene contained in a Tol2 transposon vector (Figure 2G~L). The GATA-2 minimal promoter used is a 249 bp genomic sequence 5' proximal to the GATA-2 transcription start site. The Tol2 vector, kindly provided by Kochi Kawakami, was the original construct previously described [37].

Base change or deletion mutations were introduced into the putative binding sites of Up2 for transcription factors E2F-1, ARP-1, AML-1, LMO2 and HOXA3, respectively, using the mutagenesis kit from Strategene, Inc. (Catalog No. 200521). The primers used for the site-directed mutagenesis were:

AML-1 mF: 5'-ATTGAATGTATTTGCGAGAGCGGTTTGAAAAGGCGGCGA-3'

AML-1 mR: 5'-CCGCCTTTTCAAACCGCTCTCGCAAATACATTCAATGCA-3'

ARP-1 mF: 5'-CACCAATTTATTCAGCGCCACGGGAGACCCTGTTTCCAGCCT-3'

ARP-1 mR: 5'-AGGCTGGAAACAGGGTCTCCCGTGGCGCTGAATAAATTGGT-3'

E2F-1 mF: 5'-CACCAATTTATTCAGC__ACTTTGGACCCTGTTTCC-3'

E2F-1 mR: 5'-GGAAACAGGGTCCAAAGT__GCTGAATAAATTGGTG-3'

LMO2 mF: 5'-CCAGCGGTAAGCTACGCACGATTAGGCCATTTATTATCT-3'

LMO2 mR:5'-GTGAAGATAATAAATGGCCTAATCGTGCGTAGCTTACCG-3'

HOXA3-A mF: 5'-TTTGAAAAGGCGGCGATGGCCTGGGGGAGAGAGATAAG-3'

HOXA3-A mR: 5'-TTATCTCTCTCCCCCAGGCCATCGCCGCCTTTTCAAACC-3'

HOXA3-B mF: 5'-CCAGCGGTAAGCTGATAACGACCCGGCCATTTATTATCT-3'

HOXA3-B mR: 5'-GTGAAGATAATAAATGGCCGGGTCGTTATCAGCTTACC-3'

The underlined bases were the mutated sites. Three bases were deleted from E2F-1 binding site (shown as a line). Constructs carrying site-specific mutations were injected into wild type zebrafish embryos at one cell stage. The transient hematopoietic GFP expression was observed under fluorescent microscope after 24 hpf to calculate the percentage of hematopoietic GFP expression of each mutated construct.

Construction of modified BAC constructs

BACs containing zebrafish GATA-2 identified in previous studies were used in the construction of modified BAC vectors[13, 38]. Deletions of 5' upstream elements of GATA-2 were preformed using the shuttle vector pLD53.SCA_E_B as previously described [11, 16]. In brief, 300~500 bp flanking sequences of Up1, or Up2, respectively, were amplified by PCR. The primers used in deletions of elements Up1 or Up2 or both were:

Up1 A box 5' primer: 5'-GCATGGCGCG CC CCCCAAAAGTCACATCT-3'

Up1 A box 3' primer: 5'-CCGGTTAATTAA TTCCGAAAAGGTCAAACACA-3'

Up1 B box 5' primer: 5'-CCGGTTAATTA A GAGAGCGCGCGCCCGGTGAAGAT-3'

Up1 B box 3' primer: 5'-CCGGATGGCCGGC C TACTTAACAACTCCAACCCA-3'

Up2 A box 5' primer: 5'-GCATGGCGCG CC ACTGGATAAAGGAGGCAA-3'

Up2 A box 3' primer: 5'-ACGTTTAATTAA GATCGCATGGGGTGATAGAGACAG-3'

Up2 B box 5' primer: 5'-CCGGTTAATT AA AGAGAGGGCCCTGGCATCT-3'

Up2 B box 3' primer: 5'-CCGGATGGCCGGC C GTTCAATGGACTCAAAAGGT-3'

The underlined bases correspond to restriction enzyme sites for Asc I (GGCGCGCC), Pac I (TTAATTAA), or Fse I (GGCCGGCC), and italicized bases are zebrafish genomic sequences. PCR products were digested with restriction enzymes, and ligated into pLD53.SCA_E_B lacking the GFP reporter gene. The 5' flanking sequences were termed A boxes while the 3' flanking sequences as B boxes. After modification, the entire shuttle vector was integrated into the GATA-2 BAC clone by homologous recombination through one of the two homologous arms (A box or B box). Selection in LB media containing 5.5% sucrose resulted in deletions of Up1 or Up2 or both in approximately 50% of the BAC clones through recombination with the second homologous arm (B box or A box). Deletions were confirmed by sequencing.

The insertion of GFP into deletion and wild type GATA-2 BAC clones was carried out essentially as previously described [39]. In brief, a 377 bp PCR product, (A box) based on genomic sequence adjacent to and upstream of the zebrafish GATA-2 start codon, was inserted into the targeting vector pLD53.SC₂ adjacent to a GFP reporter gene. The primers used in PCR amplification were:

5' primer: 5'-GATCGGCGCG CC CGGTAGTTATTTGAAATTGCGA-3'

3' primer: 5'-CTCAAGTGTCCGCGCTTAGAAA-3'

Underlined bases in the 5' primer correspond to restriction enzyme site for Asc I (GGCGCGCC) and italicized bases are zebrafish genomic sequences. In the presence of transiently expressed RecA, homologous recombination took place between the A box in the vector and the A box in the BAC clone resulting in integration of the vector insert into the BAC clone, placing the GFP reporter gene under control of the GATA-2 promoter within the clone (Figure 2B, D~F).

Modified GATA-2 BAC clones were identified by PCR using a forward primer (5'-CACGGGAAAAATAAACGCAGGA-3') upstream of the A box, and a reverse primer located in the GFP sequence (5'-GCTGCTTCATGTGGTCGGGGTA-3'). PCR positive clones were further confirmed by comparison of digestion patterns with that of the original BACs, or sequencing, or both.

Gel mobility shift analysis

EST clone BC076469 (cDNA clone MGC: 91782, IMAGE: 7039846), encoding zebrafish HOXA3, was purchased from Open Biosystems, Inc. The ORF of HOXA3 was PCR amplified with primers:

5' primer: 5'-ATCGAGGATCCACATTGGGAAACGGCGAG AT-3'

3' primer: 5'-AACAAAATACACTGCGCCA-3'

The underlined bases in the 5' primer contained the BamH I restriction site. The PCR product was digested with BamH I and cloned into the vector GEX-2T digested with enzyme BamH I and Sma I to generate a HOXA3 in frame fusion protein with GST. Plasmid containing the GST-HOXA3 fusion protein was expressed in BL21(DE3)plyYs. The construct containing fusion protein of GST and human E2F-1 was a gift from Dr. Nevin's lab at Duke University[40]. Protein expression and purification were preformed as previously described[41].

Electrophoretic Mobility Shift Analysis (EMSA) was preformed to confirm the binding affinity of the transcriptional factors with Up2 sequence with non-radioactive gel shift kit (Roche Applied Sciences, Inc. Catalog number: 03353591910). Complementary strand DNA oligoes containing transcription factor E2F-1 and HOXA3-A binding site, respectively, were annealed and DIG labeled. The sequences are as follow:

HOXA3-A probe: 5'-GGTTTGAAAAGGCGGCGATAATCTGGGGGAGAGAGATAA-3'

E2F1 probe: 5'-ACACCAATTTATTCAGCGCCACTTTGGACCCTGTTTCCA-3'

Mouse monoclonal antibody KH95 (Abcom, Inc) was used for super shift analysis and chicken anti-red fluorescent protein (RFP) polyclonal antibody (Chemicon International, Inc. Catalog No. AB3528) was used as non-specific antibody.

Generation of transgenic zebrafish

Tol2 vector-based plasmids were extracted and purified using a Qiaprep Spin Miniprep Kit (Qiagen). Modified BAC clone DNA was extracted using Hi-speed Maxi Prep Kit (Qiagen). BAC DNA was digested with restriction enzyme Not I and dialyzed in 0.5 × TE [13, 14]. DNA concentrations were adjusted to ~100 ng/ul in 0.1 M KCl.

BAC construct DNA was microinjected into wild type zebrafish fertilized eggs as previously described [38], resulting 4–7% germline transmission. Tol2 constructs were injected into transgenic zebrafish fertilized eggs containing maternally expressed Tol2 transposase [37] driven by the zebrafish vasa promoter [42]. F1 generation transgenic zebrafish embryos were identified by PCR. Fluorescent expression patterns were observed using a compound fluorescent microscope (Axioplan Imaging, Zeiss). Photographs were captured using an AxioCam digital camera (Zeiss), and Openlab software (Improvision).

Analysis of GFP reporter gene expression in mammalian cells lines

Transient transfection assays were performed in human myeloid leukemia cell lines, KG1 and K562, and in mouse fibroblasts NIH3T3. KG1 and K562 cells were cultured in Iscoves media containing 10% fetal bovine serum (FBS), 2 mM/L L-glutamine, 100 U/ml Penicillin and 100 mg/mL Streptomycin. NIH3T3 and 293T cells were cultured in DMEM supplemented with 10% FBS, 2 mM/L L-glutamine, 100 U/ml Penicillin and 100 mg/mL Streptomycin. Fifteen micrograms of GATA-2/GFP fusion constructs were co-transfected with 5 ug of the CMV-β-galactosidase plasmids into mammalian cells by electroporation as previously described[43, 44]. Forty-eight hours after transfection, cells were harvested and β-galactosidase activity was measured using the Promega assay kit (Promega). GFP expression levels were determined using flow cytometry. 10⁶ Cells were harvested, washed, and resuspended in phosphate-buffered saline (PBS) and analyzed using a FACS-SCAN cytometer (Beckton-Dickinson). The percentage of GFP positive cells was calculated using cell quest software and expression levels were normalized using CMV-β-galactosidase activity. P values were calculated using the Student-T test method and Jump In software.