Deletion of the major GATA1 enhancer HS 1 does not affect eosinophil GATA1 expression and eosinophil differentiation

The transcription factor GATA1 plays a central role in myelopoiesis. It is expressed in the common myeloid progenitor (CMP) and expression is selectively maintained in red cells, megakaryocytes, eosinophils, and mast cells but extinguished in granulocytes/macrophages.¹  It promotes differentiation of myeloid progenitors to red cells, megakaryocytes, and eosinophils at the expense of granulocytes/macrophages.^1-3  It is required for terminal maturation of red cells, megakaryocytes, and mast cells and at the earliest stages of eosinophil differentiation (reviewed in Cantor and Orkin⁴  and McNagny and Graf⁵ ). These observations argue that defining the molecular mechanisms regulating GATA1 expression will be one important component in dissecting the molecular control of myeloid differentiation.

In mice, 3 cis-acting sequences (cis-elements)—the hematopoietic exon 1 erythroid (IE) promoter (this includes both promoter proximal and distal elements within approximately 900 base pairs from the transcriptional start sites in the exon associated with this promoter), an upstream enhancer HS 1/G1HE (hereafter referred to as HS 1), and an element in the first GATA1 intron (HS 4/5)—direct reporter transgene expression to definitive red cells and megakaryocytes.^6,7  The intron element is not required for primitive red cell GATA1 expression.⁷  Germ line HS 1 deletion almost completely abolishes megakaryocyte GATA1 expression but leaves red cell GATA1 expression unaffected,^8,9  suggesting that additional sequences compensate for HS 1 absence in red cells but not megakaryocytes.

By contrast, little is known about sequences controlling eosinophil GATA1 expression. Although sequences in the IE promoter are important,¹⁰  there are conflicting data on the function of HS 1. In 2 reports, all known GATA1 cis-elements (including HS 1) failed to target transgene expression to eosinophils.^6,11  In contrast, another study suggests that a similar combination of GATA1 regulatory sequences can do so.¹²  One potential reason for these conflicting results is that transgene expression can be modulated by position of integration. Therefore, to directly test the nonredundant function of HS 1 in eosinophil GATA1 expression, we asked whether eosinophil GATA1 expression and eosinophil development are perturbed in mice deleted for HS 1.

ΔneoΔHS⁸  and wild-type littermate female mice were mated with interleukin 5 (IL-5) homozygous transgenic male mice.¹³  Male offspring were genotyped for presence or absence of the HS 1 enhancer by using primers 5′-AATCAGGAATGCAACATCTC-3′ and 5′-ACTCTTGCTCTCTTTTGCAG-3′ and for presence of the IL-5 transgene by using previously published primers.¹⁰  A Pentra 60 (ABX Diagnostics, Montpellier, France) automated counter was used to determine blood counts. Blood smears were stained with May-Grünwald-Giemsa (MGG) stain (Sigma, Poole, United Kingdom). Megakaryocytic cell line L8057 was cultured as previously described.¹⁴ 

Peripheral blood red cells were pelleted by mixing 1 volume of blood with 5 volumes of 1.25% dextran. White blood cells were pooled with splenocytes, and the remaining red cells were lysed by hypotonic shock. Cells were labeled with anti-CD2 (reference 553109), anti-B220 (reference 553084), and anti-Gr1 (reference 553123) (Pharmingen, Oxford, United Kingdom) and removed using magnetic beads coated with goat antirat immunoglobulin G (IgG) and a lactic dehydrogenase (LD) depletion column (Miltenyi Biotech, Bergisch Galdbach, Germany). Unlabeled cell population was routinely more than 95% eosinophils (assessed by MGG staining). Fast Green and Neutral Red stains (Sigma) were performed by using standard protocols. Micrograph images were taken with a BX60 microscope and a × 40/0.75 objective (Olympus, London, United Kingdom), using a Qicam camera (Qimaging, Burnaby, Canada) and Openlab software (Improvision, Coventry, United Kingdom).

cDNA, produced from total RNA by standard protocols, was used in real-time polymerase chain reaction (PCR) according to recommended ABI protocols on a ABI7000 Thermocycler (Applied Biosystems, Warrington, United Kingdom). Primers and probe used included 5′-AGAGAAGCTGAGGCCTACAGA-3′,5′-CAGGAATTCCCTCCATACTGTTGAG-3′, and 5′-FAM-CACTCCCCAGTCTTTC-3′-NFQ (Applied Biosystems). GATA1 signal was normalized to both β-actin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), by using primers and probes from Eurogentec (Romsey, United Kingdom) (RT-CKYD-ACTB and RT-CKYD-GAPD).

From 1 to 3 × 10⁷ cells were used for immunoprecipitation by using conditions suggested by manufacturer's protocol (Upstate Biotechnology, Lake Placid, NY), except for GATA1.¹⁵  Polyclonal antibodies used included anti-acetylated H3 (no. 06-599; Upstate Biotechnology), GATA1, CCAAT/enhancer binding protein ϵ (C/EBPϵ) (no. sc-265 and no. sc-158; Santa Cruz, Calne, United Kingdom). For each cell type and antibody, at least 3 independent chromatin preparations were used for immunoprecipitation. Details of the chromatin immunoprecipitation (ChiP) assays, real-time PCR experiments, and sequences of the primers and probes used in Taqman PCR are set out in the supplemental table at the Blood website (see the Supplemental Table link at the top of the online article).

Blood counts from mice deleted for HS 1 (ΔneoΔHS),⁸  and normal littermate controls were analyzed (Table 1). As the normal eosinophil count is low, a subtle eosinophil phenotype may be missed. To increase the eosinophil count, ΔneoΔHS were bred to transgenic mice overexpressing IL-5.¹³  Thus, ΔneoΔHS mice heterozygous for the IL-5 transgene and IL-5 heterozygous transgenic littermate controls were studied. ΔneoΔHS mice and ΔneoΔHS/IL-5 transgenic mice have eosinophil counts comparable to control wild-type (WT) and IL-5 transgenic mice (WT/IL-5 Tg), respectively. Morphologic analysis by MGG and Fast Green stains shows comparable eosinophil maturation in wild-type and ΔneoΔHS mice (Figure 1A). Consistent with these observations, GATA1 mRNA expression in purified eosinophils with intact and deleted HS 1 is comparable when quantitated by Real-Time Taqman PCR (Figure 1B). For the first time, these observations demonstrate that both eosinophil GATA1 mRNA expression and eosinophil differentiation are unaffected in absence of HS 1.

Chromatin enriched for acetylated core histones H3/H4 is associated with active cis-elements (reviewed in Bulger et al¹⁶  and Johnson and Bresnick¹⁷ ). Furthermore, cis-element deletion can perturb long-range histone acetylation.¹⁸  Therefore, we quantitated relative H3/H4 acetylation at 11 points in the mouse GATA1 locus from coordinates –8.1 to +22.4 (in kilobases with respect to the transcriptional start of the GATA1 gene in the IE promoter) in purified primary eosinophils with and without HS 1. As H3 and H4 acetylation profiles were similar, only H3 results are presented (Figure 2A). Chromatin isolated from eosinophils with intact HS 1 is enriched approximately 7- to 12-fold for acetylated H3 from the IE promoter through to the regulatory element in the first intron compared with neighboring points in the locus. Contrastingly, chromatin containing HS 1 is enriched only 2- to 3-fold for acetylated H3. The same pattern of acetylation is seen in absence of HS 1. The pattern of acetylation with isotype control antibodies revealed no increased acetylation at regulatory elements compared with neighboring points in the locus (data not shown). These results suggest that histone acetyltransferase activity is mainly recruited to the IE promoter and intron element and show that HS 1 deletion does not perturb long-range GATA1 locus chromatin structure. In contrast, in the megakaryocytic cell line L8057 chromatin associated with HS 1 was enriched in acetylated H3 and H4 (B.G., unpublished observation, December 2003). It is noteworthy that HS 1 is required for megakaryocyte-GATA1 expression.⁸ 

Finally, we determined in vivo binding of transcriptions factors that potentially regulate GATA1 expression in primary purified eosinophils (Figure 2B). We studied GATA1 because it is likely to autoregulate its own expression^19,20  and because critical GATA binding sites have been demonstrated in HS 1 and the IE promoter.^19,21  We also assayed C/EBPϵ and C/EBPα binding in the GATA1 locus as mice null for expression of these genes have impaired eosinophil differentiation.^22,23  As a positive control for GATA1 binding at HS 1, we tested the megakaryocytic cell line, L8057 (Figure 2B). In wild-type eosinophils, GATA1 binding was detected at the IE promoter and intron element but not at HS 1. Interestingly, C/EBPϵ was also detected at the intron element with weak binding at the IE promoter. Consistent with this finding there are 2 C/EBP sites in the IE promoter and one site in HS4/5 within 600 base pairs either side of where the primer pairs are located in the analysis, suggesting that these C/EBP motifs do function in vivo. In contrast, C/EBPα binding was not present in the GATA1 locus. In L8057 cells, strong binding of GATA1 was detected at HS 1 in addition to the IE promoter and the intron element. Importantly, we did not see any binding with isotype control antibodies at any location in the GATA1 locus (data not shown).

For the first time, these data demonstrate in vivo binding of GATA1 and C/EBPϵ at known GATA1 regulatory elements in primary eosinophils. GATA1 colocalizes with C/EBPϵ in a cis-element–specific manner, suggesting it may participate in different protein complexes on DNA. At both the intron element and IE promoter, GATA1 and C/EBPϵ colocalization raises the possibility that they may cooperate and physically interact. In C/EBPϵ-null mice, eosinophil differentiation defect is not as early as that seen with GATA1 loss.^10,22  Although this may suggest that eosinophil GATA1 expression is not C/EBPϵ dependent, eosinophil GATA1 levels in C/EBPϵ-null mice have not been reported. If there was a partial reduction in eosinophil GATA1 expression, this may account for some (or all) of the eosinophil phenotype in C/EBPϵ-null mice.

Taken together, unperturbed eosinophil GATA1 mRNA expression and eosinophil differentiation in mice lacking HS 1, weak enrichment for acetylated H3/H4 at HS 1 in primary eosinophils, and absence of regulator binding at HS 1 suggest that HS 1, in its normal chromosomal context, is not important for mouse eosinophil GATA1 expression. In contrast, our data show for the first time that the intron element is likely to function in regulating eosinophil GATA1 expression.

We thank Dr Orkin and Dr Gerard for providing the ΔneoΔHS and IL-5 transgenic mice.