|Home | About | Journals | Submit | Contact Us | Français|
GATA1 is a key regulator of erythroid cell differentiation. To examine how Gata1 gene expression is regulated in a stage-specific manner, transgenic mouse lines expressing green fluorescent protein (GFP) reporter from the Gata1 locus in a bacterial artificial chromosome (G1BAC-GFP) were prepared. We found that the GFP reporter expression faithfully recapitulated Gata1 gene expression. Using GFP fluorescence in combination with hematopoietic surface markers, we established a purification protocol for two erythroid progenitor fractions, referred to as burst-forming units-erythroid cell-related erythroid progenitor (BREP) and CFU-erythroid cell-related erythroid progenitor (CREP) fractions. We examined the functions of the Gata1 gene hematopoietic enhancer (G1HE) and the highly conserved GATA box in the enhancer core. Both deletion of the G1HE and substitution mutation of the GATA box caused almost complete loss of GFP expression in the BREP fraction, but the CREP stage expression was suppressed only partially, indicating the critical contribution of the GATA box to the BREP stage expression of Gata1. Consistently, targeted deletion of G1HE from the chromosomal Gata1 locus provoked suppressed expression of the Gata1 gene in the BREP fraction, which led to aberrant accumulation of BREP stage hematopoietic progenitor cells. These results demonstrate the physiological significance of the dynamic regulation of Gata1 gene expression in a differentiation stage-specific manner.
GATA1 is a founding member of the GATA family of transcription factors that harbor two zinc finger DNA binding domains (50). GATA1 is expressed in erythroid cells, megakaryocytes, mast cells, eosinophils, and dendritic cells (8, 17, 30, 53) and in Sertoli cells in the testis (12, 51). GATA1 has been shown to be essential for erythroid cell differentiation in vivo (6, 9, 40). Closer examination of Gata1-deficient mice and cells revealed that without GATA1, erythroid cells fail to mature beyond the proerythroblast stage (29, 47). Indeed, missense mutations in the N-terminal zinc finger domain of GATA1 have been identified in patients with anemia and thrombocytopenia (4, 5, 20, 22). An altered expression level of GATA1 appears to be related to idiopathic myelofibrosis (11, 44). Truncation mutations in the N-terminal domain are closely linked to transient myeloproliferative disorder and acute megakaryoblastic leukemia in Down syndrome patients (46, 49). Consistent with these observations in clinical hematology, targeted knockdown of Gata1 gene expression in mice to 5% of wild-type levels (the Gata1.05 allele) resulted in erythroid leukemia in heterozygous female mice due to a mechanism involving random inactivation of the X chromosome in vivo (31, 32).
Within the erythroid differentiation cascade, GATA1 expression was initially detected in common myeloid progenitors, but the expression level sharply increased when cells differentiated into the proerythroblast stage (15, 38). GATA2 is highly expressed in hematopoietic stem cells and early progenitors, but its expression declines quickly upon commencement of GATA1 expression (2, 7, 25, 48). From the proerythroblast stage onward, the expression level of GATA1 decreases en route to maturation into red blood cells (38). This dynamic change in Gata1 gene expression seems essential for normal erythropoiesis, since constitutive expression of high levels of GATA1 was lethal to transgenic mice due to defective erythroid cell maturation (3, 48).
The mouse Gata1 locus is composed of two noncoding first exons, termed IT and IE, and an additional five coding exons (12, 42). The distal IT promoter mainly directs Gata1 gene expression in testis Sertoli cells, whereas Gata1 gene expression is directed in hematopoietic cells by the proximal IE promoter (12). The 8-kbp region spanning 3.9 kbp 5′ of the IE exon to the second exon contained sufficient regulatory elements for erythroid expression of green fluorescent protein (GFP) or β-galactosidase reporter in both yolk sac-derived primitive erythroid cells and fetal-liver-derived definitive erythroid cells in a transgenic-mouse reporter assay (18, 26). This 8-kbp region is referred to as the Gata1 hematopoietic regulatory domain (G1HRD) (21). When GATA1 cDNA was linked to the G1HRD and expressed in transgenic mice, this G1HRD-GATA1 cDNA transgene sustained hematopoiesis and rescued GATA1-deficient mice from embryonic lethality, indicating that the G1HRD contains regulatory elements that sufficiently support hematopoiesis in the mouse in vivo (33, 41).
An important application of the G1HRD is the identification of erythroid progenitors. While a reliable method using Ter119 and CD71 antibodies for separating erythroblasts has been established (34), no method for isolating more immature proerythroblasts had been developed before our G1HRD approach. Indeed, the erythroid progenitors, or proerythroblasts, could only be detected retrospectively by means of a colony-forming assay (36). The two types of erythroid progenitors identified by colony-forming assay are burst-forming units-erythroid (BFU-E) and CFU-erythroid (CFU-E). We addressed this issue using transgenic mouse lines expressing GFP reporter under G1HRD control and identified two erythroid progenitor fractions in mouse bone marrow cells called late erythroid progenitors (LEP) and early erythroid progenitors (EEP) (38). The LEP fraction was c-Kit and CD71 double positive and contained an abundance of CFU-E, while the EEP fraction was c-Kit positive/CD71 negative and contained BFU-E (38).
The 5′-end sequence of the mouse G1HRD is highly conserved with that of humans. Indeed, in a transgenic-mouse reporter assay, deletion of 1.3 kbp from the 5′ end of the G1HRD markedly abrogated reporter expression in yolk sac and fetal liver erythroid cells (23, 26). We therefore assumed that this 1.3-kbp region corresponds to the Gata1 gene hematopoietic enhancer (G1HE). The G1HE coincides with the tissue-specific DNase I-hypersensitive site HS1, and histone H3/H4 hyperacetylation has been found in the G1HE region in mouse erythroleukemia cells (18, 44). Further dissection revealed that the 235-bp region most 5′ of the G1HE (the G1HE core) plays an essential role in enhancer activity (23, 24). Importantly, the G1HE core contains a highly conserved GATA box that binds GATA factors. Mutation of this GATA box in the G1HRD context completely abolished reporter transgene expression in erythroid cells (23, 45). However, the precise roles that the GATA box, G1HE core, and G1HE play in controlling Gata1 gene expression during erythroid cell differentiation remain elusive.
When we developed the G1HRD transgenic-mouse system, we noticed the existence of apparent limitations to further application of this approach. For instance, similar to the cases for other transgenic-mouse approaches, the G1HRD is affected by position effect variegation (26, 38). A more serious problem is that the G1HRD appears insufficient to recapitulate Gata1 gene expression in EEP (38). Therefore, to circumvent this difficulty, we decided to try the use of a bacterial artificial chromosome (BAC) transgenic-mouse system. Compared to the size of the plasmid-based transgene, in which the G1HRD is approximately 8 kbp, the BAC can host genomic regions of almost 200 kbp surrounding the Gata1 gene. Thus, it seems reasonable to expect that a Gata1 BAC clone retains a comprehensive set of Gata1 gene regulatory elements required for proper spatiotemporal gene expression.
In this study, we identified a BAC clone harboring 196 kbp of the mouse Gata1 gene and established transgenic-mouse lines expressing GFP under its Gata1 gene-regulatory influence. As we expected, GFP reporter expression recapitulated endogenous Gata1 gene expression much more faithfully than did the G1HRD. Importantly, the reporter expression showed copy number dependence, one of the important hallmarks of the locus control region in vivo. Utilizing the Gata1-BAC transgenic-mouse system, we established a novel purification protocol for two distinct stages of erythroid progenitor, and these two fractions were designated BFU-E-related erythroid progenitor (BREP) and CFU-E-related erythroid progenitor (CREP). We examined how the G1HE, G1HE core, and GATA box contribute to enhancer activity or Gata1 gene expression during erythroid differentiation in a quantitative manner. The results clearly demonstrated that the G1HE is indispensable for Gata1 gene expression during erythroid differentiation and that the GATA box is crucial for G1HE activity. Especially in the absence of G1HE activity, the expression of GATA1 from Gata1-BAC was markedly abrogated in BREP containing BFU-E stage progenitors. In contrast, BAC-GFP expression in CREP was relatively preserved in the mutant transgenic animals. In essence, this study demonstrated that Gata1 gene expression is regulated dynamically in a differentiation stage-specific manner and that the GATA box within the G1HE substantially directs the BREP stage expression of Gata1 in vivo.
The 5′ homologous regions of the G1BAC-GFP and G1BAC-Luc targeting vectors contain a fragment that includes part of the first intron up to exon II (SacI-NcoI) (Fig. (Fig.1),1), a GFP or Luc gene, and a poly(A) signal, which were derived from the G1HRD-GFP and G1HRD-Luc constructs, respectively (37, 38). The 3′ homologous region of the G1BAC-GFP targeting vector contains an EcoRI-EcoRI fragment spanning from exon III to exon VI. The 5′ homologous regions of the ΔG1HE, ΔG1HE core, and GATAmut targeting vectors contain a SacI-BamHI fragment, as shown in Fig. S4B and C in the supplemental material. The 3′ homologous region of ΔG1HE contains an EcoRI-HindIII fragment (see Fig. S4B in the supplemental material). The HindIII site of ΔG1HE was created at the 3′ end of the IE exon using PCR (26). The 3′ homologous region of ΔG1HE core contains a StuI-EcoRI fragment (see Fig. S4C in the supplemental material). The 3′ homologous region of GATAmut contains a BamHI-EcoRI fragment bearing point mutations in the GATA box. These point mutations were introduced into the GATA box according to a PCR-based mutation protocol (see Fig. Fig.5A).5A). The genomic fragments were cloned into a vector containing a neomycin resistance gene cassette to generate the respective targeting constructs.
The BAC clone RP23-443E19 containing the mouse Gata1 gene was purchased from the Children's Hospital Oakland Research Institute and introduced into Escherichia coli strain EL250 by electroporation (16). In this study, we used a prophage-based recombination system for BAC mutagenesis (16). For removal of the neomycin resistance cassette, FLP recombinase was induced with l-arabinose (13). Recombinant BAC clones were identified and verified through sequencing and Southern blot analysis.
Modified BAC clones were purified from 800-ml culture using a NucleoBond BAC 100 kit according to the low-copy-number plasmid protocol (BD Biosciences Clontech). The final pellet was resuspended in sterile injection buffer (100 mM NaCl, 10 mM Tris-HCl [pH 7.5], 0.1 mM EDTA, 3 μM spermine, and 7 μM spermidine). The DNA was diluted with injection buffer to give concentrations from 0.5 to 2 ng/μl and then microinjected into the pronuclei of fertilized eggs harvested from BDF1 mice. Transgenic founder mice were screened by PCR using the GFP-specific primers 5′-CTGAAGTTCATCTGCACCACC-3′ and 5′-GAAGTTGTACTCCAGCTTGTGC-3′. To determine the copy number and integrity of the transgene, high-molecular-weight genomic DNA was prepared from the thymus or spleen of each line, embedded in agarose plugs using a CHEF Genomic DNA Plug Kit (Bio-Rad), and digested with restriction enzymes. After pulsed-field gel electrophoresis, DNA was transferred onto a nylon membrane, and the blots were hybridized at 65°C with 32P-labeled DNA probes containing GFP cDNA. Gata1G1.05/+ mice were described previously (39). Gata1ΔneoΔHS/Y mice (19) were kindly supplied by the Jackson Laboratory (Bar Harbor, ME).
FACS analyses of bone marrow cells were performed using FACS Vantage and FACS Caliber (Becton Dickinson, San Jose, CA). For the analysis of immature hematopoietic cells, mononucleated cell suspensions from bone marrow were incubated with biotinylated monoclonal antibodies recognizing Mac1, Gr1, Ter119, B220, CD4, and CD8. Hematopoietic lineage marker-negative (Lin−) cells were enriched by magnetic separation using streptavidin-conjugated magnetic beads (Polyscience, Warrington, PA) and then stained with allophycocyanin-conjugated anti-c-Kit and phycoerythrin-conjugated anti-Fcγ receptor (FcγR) II/III or anti-CD71 antibodies. All antibodies were obtained from BD Pharmingen (San Diego, CA).
Sorted cells were cultured in 1 ml of 0.8% methylcellulose medium containing 30% fetal bovine serum. For the detection of CFU-E, BFU-E, CFU-eosinophils (CFU-Eo), and myeloid colonies, the medium was supplemented as follows: for CFU-E, 4 U/ml erythropoietin (Epo) (a generous gift from Chugai Pharmaceutical Co.); for BFU-E, 4 U/ml Epo and 100 ng/ml stem cell factor (SCF) (a generous gift from Kirin Brewery); for CFU-Eo, 50 ng/ml interleukin 5 (IL-5) (R&D systems); and for myeloid colonies, 4 U/ml Epo, 100 ng/ml SCF, 100 ng/ml IL-3, and 100 ng/ml IL-6. Colonies were counted after 3 days (CFU-E), 7 days (BFU-E), 9 days (CFU-Eo), or 11 days (myeloid colonies) of culture. To distinguish erythroid cell colonies, the colonies were stained with benzidine before being counted. A CFU-megakaryocytic (CFU-Meg) assay was performed using a MegaCult-C kit (Stem Cell) according to the manufacturer's instructions.
RNA was purified from cells using RNeasy (Qiagen) and reverse transcribed with SuperScript II (Invitrogen) and random hexamers. GATA1 mRNA levels were measured quantitatively with an ABI Prism 7700 (Perkin-Elmer) using the primer pair 5′-CAGAACCGGCCTCTCATCC-3′ and 5′-TAGTGCATTGGGTGCCTGC-3′ and the VIC-labeled oligo-DNA probe 5′-CCCAAGAAGCGAATGATTGTCAGCAAA-3′ (38). GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was used as an internal standard (primer pair, 5′-GAAGGTGAAGGTCGGAGTC-3′ and 5′-GAAGATGGTGATGGGATTTC-3′; VIC-labeled probe, 5′-CAAGCTTCCCGTTCTCAGCC-3′).
In vivo bioluminescence imaging was performed using an IVIS imaging system (Xenogen, Alameda, CA) as previously described (37). For stimulation of erythropoiesis, phenylhydrazine (60 mg/kg) was injected intraperitoneally for two consecutive days. GFP immunohistochemistry on frozen spleen sections was performed as described previously (13).
To examine how the G1HE contributes to proper spatial and temporal expression of the Gata1 gene during erythroid differentiation, we decided to utilize a BAC reporter transgenic-mouse system. To this end, we obtained the BAC clone RP23-443E19 containing a 196-kbp genomic region spanning flanking sequences approximately 71 kbp 5′ and 116 kbp 3′ of the mouse Gata1 gene (Fig. (Fig.1A).1A). To generate a reporter transgene representing endogenous Gata1 gene transcription, we introduced GFP cDNA into the second exon of the mouse Gata1 gene in the BAC clone by means of homologous recombination in E. coli strain EL250. Subsequently, the neomycin resistance cassette was removed by arabinose induction of FLP recombinase (Fig. (Fig.1B).1B). The purified BAC clone was injected into fertilized BDF1 ova, and four independent lines of transgenic mice (lines 357, 415, 392, and 393) were established that expressed GFP under the influence of the Gata1 gene regulatory elements in the BAC clone (G1BAC-GFP).
To determine the transgene copy number in each BAC transgenic-mouse line, we performed genomic Southern blot analysis and compared the signal intensities of the Gata1-BAC-derived bands with those of the endogenous Gata1 locus-derived bands using a VI exon probe (see Fig. S1A in the supplemental material). Lines 357 and 415 carried a single-copy BAC, whereas lines 392 and 393 carried two copies of the G1BAC-GFP transgene (Fig. (Fig.1F1F and data not shown; see Fig. S1B in the supplemental material). A long-range genomic Southern blot analysis using pulsed-field gel electrophoresis showed that lines 357 and 392 harbored at least 65 kbp of a Gata1 gene-flanking region (35 kbp of 5′ and 33 kbp of 3′ genomic sequences) (see Fig. S2A and B in the supplemental material). Therefore, we mainly used line 357 for the subsequent analyses.
To ascertain whether G1BAC-GFP faithfully recapitulates endogenous Gata1 gene expression, the levels of GFP and Ter119 marker were examined in the bone marrow cells of G1BAC-GFP mice by FACS. As anticipated, GFP fluorescence was detected in most of the Ter119-positive cells (Fig. (Fig.1C),1C), which are known to express endogenous Gata1 most abundantly (52). Immunohistochemical analysis of GFP in hematopoietic tissues of the G1BAC-GFP mice revealed strong GFP expression in splenic megakaryocytes (Fig. (Fig.1D).1D). These results support the notion that G1BAC-GFP adopts the expression profile of the endogenous Gata1 gene in erythroid cells and megakaryocytes.
To further verify the specific expression and to exclude ectopic expression, if any, of the G1BAC reporter in the transgenic mice, we also generated transgenic-mouse lines expressing firefly luciferase under the control of the G1BAC regulatory domain (G1BAC-Luc) and analyzed whole-body bioluminescence by in vivo luciferase assay (37). We observed bioluminescent signals derived from the G1BAC-Luc transgene specifically in hematopoietic tissues, including spleen and bone marrow of the femur, spine, and sternum (Fig. (Fig.1E;1E; see Fig. S3 in the supplemental material). Stress erythropoiesis response signals induced by phenylhydrazine were also observed specifically in the hematopoietic tissues of G1BAC-Luc mice (Fig. (Fig.1E),1E), indicating that the G1BAC directs reporter expression specifically in hematopoietic tissues.
The fluorescence intensities in the bone marrow cells of G1BAC-GFP mice nicely coincided with the transgene copy numbers in all four transgenic lines generated (Fig. (Fig.1F),1F), suggesting that the G1BAC reporter is expressed independently of position-effect variegation, presumably due to insulator- or locus control region-like activity in the G1BAC regulatory domain (10).
Using G1HRD-GFP transgenic mice and CD71 marker, we previously identified two types of erythroid progenitor fractions (38). The EEP (Lin− c-Kit+ G1HRD-GFP+ CD71−) population contained BFU-E, whereas the LEP (Lin− c-Kit+ G1HRD-GFP+ CD71+) population contained abundant CFU-E (Fig. (Fig.2A).2A). The expression level of GATA1 increased along with the erythroid cell differentiation from EEP to LEP and decreased toward terminal erythroid maturation (Fig. (Fig.2A).2A). However, while the G1HRD mirrored the expression profile of the endogenous Gata1 gene fairly significantly, this domain was still insufficient to fully recover the Gata1 expression profile. Following two lines of evidence suggested that the G1HRD-driven reporter failed to recapitulate Gata1 gene expression in Lin− c-Kit+ immature erythroid progenitors (38). First, the G1HRD-GFP− Lin− c-Kit+ fraction contained GATA1-immunoreactive cells. Secondly, only one-third of total BFU-E colonies were derived from the G1HRD-GFP-positive fraction, while the rest of the BFU-E colonies arose from the G1HRD-GFP-negative fraction. Therefore, we wished to dissect the Gata1 gene regulatory activity preserved in the G1BAC and compare it with that of the G1HRD.
To delineate whether the G1BAC contains regulatory elements sufficient for reporter expression in immature erythroid progenitors, we first analyzed GFP and CD71 expression in the Lin− c-Kit+ fraction from G1BAC-GFP mouse bone marrow by FACS (Fig. (Fig.2B).2B). We identified three fractions (fractions I-T, II-T, and III-T, where T represents CD71 transferrin receptor). The G1BAC-GFP+ cells were separated into two fractions based on the CD71 expression level (CD71high and CD71low~nega, corresponding to fractions III-T and II-T, respectively). The percentage of the GFP+ CD71low~nega fraction (fraction II-T, containing BFU-E) in G1BAC-GFP mice (17.7% ± 5.5%; n = 10) was threefold greater than that in G1HRD-GFP mice (5.0% ± 1.1%; n = 3) (Fig. (Fig.2C,2C, left), while there was virtually no difference in the percentages of the GFP+ CD71high fraction (fraction III-T, containing CFU-E) between G1BAC-GFP (2.3% ± 1.2%; n = 10) and G1HRD-GFP (2.2% ± 0.2%; n = 3) mice (Fig. (Fig.2C,2C, right).
To examine the correlation between G1BAC-GFP fluorescence and endogenous Gata1 gene expression, we sorted the mouse bone marrow cells from G1HRD-GFP and G1BAC-GFP mice and determined the endogenous Gata1 expression by real-time quantitative RT-PCR. A salient finding was that endogenous GATA1 mRNA expression was observed exclusively in the G1BAC-GFP+ fraction (Fig. (Fig.2D,2D, fractions II-T and III-T), but not in the G1BAC-GFP− fraction (fraction I-T). Similarly, BFU-E colonies were formed predominantly from the G1BAC-GFP+ CD71low~nega fraction (Fig. (Fig.2E,2E, fraction II-T). These observations are in stark contrast to the case of G1HRD-GFP, as GATA1 mRNA, albeit at a very low level, was apparently found in the G1HRD-GFP− fraction (Fig. (Fig.2D,2D, fraction I-T) and a significant number of BFU-E arose from the G1HRD-GFP− fraction (38). These results indicate that BFU-E stage progenitors are more efficiently enriched in the G1BAC-GFP+ CD71low~nega fraction than in the EEP (G1HRD-GFP+ Lin− c-Kit+ CD71−) fraction. In contrast, CFU-E stage progenitors were as efficiently enriched in fraction III-T of G1BAC-GFP mice as they were in the LEP fraction of G1HRD-GFP mice (Fig. (Fig.2E,2E, right). These results support our contention that the G1BAC contains an important regulatory element that is missing in the G1HRD, and thus, the G1BAC faithfully recapitulates Gata1 gene expression in Lin− c-Kit+ immature erythroid progenitors.
We examined the morphology of the sorted cells of each fraction by Wright-Giemsa staining and found that 90% of the cells in fraction III-T were proerythroblast-like cells (cell type a) (Fig. (Fig.2F,2F, right, and G). Fraction II-T contained three cell types (Fig. (Fig.2F,2F, middle, and G): those with light-purple cytoplasm bearing no granules (cell type a; 40%), those bearing some azurophilic granules (cell type b; 40%), and mature eosinophils bearing eosinophilic granules (cell type c; 20%). We envisage that cell types b and c of fraction II-T are most likely to differentiate into other myeloid cells or eosinophils. Therefore, type a cells of fraction II-T seemingly correspond to BFU-E cells. Almost all the cells in the G1BAC-GFP− fraction (fraction I-T) showed many azurophilic granules, suggesting that they were non-erythroid progenitors (Fig. (Fig.2F,2F, left).
Since the G1BAC-GFP+/CD71low~nega fraction (fraction II-T) still contained an immature granulocyte population, we tried isolating these nonerythroid cells from the fraction using antibody raised against FcγR II/III. FcγR II/III is a myeloid marker applied to the purification of megakaryocyte-erythrocyte progenitor (MEP) cells (1). Using FcγR II/III expression, we separated bone marrow Lin− c-Kit+ G1BAC-GFP+ cells into three fractions (Fig. (Fig.3A,3A, fractions I-F, II-F and III-F, where F stands for FcγR II/III) and examined the cell morphology in each fraction by Wright-Giemsa staining. We found that 94.8% of FcγRII/IIIhigh (fraction I-F) cells exhibited the morphology of eosinophils and immature granulocytes (cell types b and c) (Fig. (Fig.3B).3B). Conversely, 82.5% of FcγRII/IIIlow (fraction II-F) cells exhibited proerythroblast-like morphology with light-purple cytoplasm lacking granules (cell type a) (Fig. (Fig.3B),3B), and almost all of the cells in fraction III-F (FcγRII/III−) were proerythroblast-like cells (cell type a) (Fig. (Fig.3B).3B). This indicated that eosinophils and immature granulocytes (cell types b and c) were successfully separated into the FcγRII/IIIhigh fraction (fraction I-F).
To assess the differentiation potency of each fraction, we carried out a series of colony-forming assays. BFU-E colonies were mainly observed in fraction II-F (Fig. (Fig.3C),3C), while CFU-E colonies were predominant in fraction III-F (Fig. (Fig.3D).3D). CFU-Meg colonies were most abundant in fraction II-F (Fig. (Fig.3E).3E). Of note, we detected CFU-erythroid/megakaryocytic cells exclusively in fraction II-F (the FcγRII/IIIlow fraction), suggesting that MEP cells were consistently enriched in fraction II-F (Fig. (Fig.3F).3F). On the other hand, fraction III-F cells gave rise solely to erythroid-committed progenitors (Fig. (Fig.3F3F).
Based on these results, we named fraction II-F (Lin− c-Kit+ G1BAC-GFP+ FcγRII/IIIlow) and fraction III-F (Lin− c-Kit+ G1BAC-GFP+ FcγRII/III−) the BREP and CREP fractions, respectively. The BREP fraction appeared to contain MEP, CFU-Meg, and BFU-E stage cells, while the CREP fraction contained CFU-E and proerythroblast stage cells in abundance. In good agreement with the GFP fluorescence intensity, the endogenous GATA1 level was highest in fraction III-F, medium in fraction II-F, and lowest in fraction I-F (Fig. (Fig.3H).3H). We also found that fraction I-F was rich in CFU-Eo (Fig. (Fig.3G).3G). This is consistent with the Wright-Giemsa staining pattern (Fig. (Fig.3B),3B), indicating that fraction I-F contained mature eosinophils and eosinophilic progenitors.
In our analyses of Gata1 gene regulation, we identified four important cis-acting regulatory elements within the G1HRD (explicitly, an 8-kbp genomic region [reviewed in references 14 and 24] that includes the enhancer G1HE). Since the highly conserved 235-bp region located most 5′ of the G1HE (the G1HE core) appeared to be important for transgenic reporter expression in erythroid cells (see Fig. S4A in the supplemental material), we assessed the functional contribution of the G1HE core region to the activity of the G1HE in the context of the G1BAC. G1HE (1.3 kbp) (see Fig. S4B in the supplemental material) and G1HE core (235 bp) (see Fig. S4C in the supplemental material) deletion mutants bearing a G1BAC-GFP backbone were used to generate the respective transgenic ΔG1HE and ΔG1HE core mouse lines. Southern blot analysis revealed that two transgenic-mouse lines each for the ΔG1HE and ΔG1HE core mutants carried a single copy of G1BAC (ΔG1HE lines 662 and 663L and ΔG1HE core lines 424L and 442) (data not shown). As lines 662 and 424L were found to harbor more than 65 kbp of intact G1BAC transgene (see Fig. S2B in the supplemental material), these mouse lines were employed for further analysis.
We examined the Lin− c-Kit+ fraction of the bone marrow cells by FACS using GFP and FcγR II/III. To our surprise, the GFP+ FcγRII/IIIlow fraction (fraction II-F) was dramatically reduced in the ΔG1HE (2.3% ± 0.9%; n = 3) and ΔG1HE core (1.3% ± 0.7%; n = 3) mutant mice to less than 20% of the control (11.2% ± 1.6%; n = 3) (Fig. 4A and B). These results show that the G1HE core is indispensable for Gata1 expression in BREP stage cells, thus supporting our contention. On initial observation, fraction III-F was seemingly preserved in the ΔG1HE and ΔG1HE core mutants (Fig. 4A and B). However, closer examination revealed that the mean fluorescence intensity of fraction III-F in the ΔG1HE mutants was reduced to 50% of the control (Fig. 4A and C), suggesting that the G1HE core also contributes to the intensity of Gata1 gene expression in CREP stage cells.
It has been reported that CD71 expression decreases, whereas Ter119 expression increases during erythroblast maturation. We therefore examined the expression of GFP in the Ter119+ CD71high and Ter119+ CD71low fractions of the bone marrow of ΔG1HE and ΔG1HE core mutant transgenic mice. Ter119+ CD71high and Ter119+ CD71low fractions contain basophilic/polychromatic erythroblasts and more mature orthochromatophilic erythroblasts, respectively (34). Although the GFP-positive population was preserved in both deletion mutants, the mean fluorescence intensity of GFP was markedly decreased (Fig. 4D and E), suggesting that the G1HE is required for full expression of Gata1 in Ter119+ stage erythroblasts.
Through studies addressing the molecular basis of the G1HE activity, a GATA box residing in the G1HE core has been found to be essential for G1HRD activity in a transgenic-mouse plasmid reporter assay. Available lines of evidence suggest that this GATA box mediates both autoregulation and GATA factor network regulation of Gata1 gene expression (23, 45). To assess the contribution of the GATA box to the activity of the G1HE core or to explore any discrepancies between the consequences of the GATA box and G1HE deletions, we generated transgenic-mouse lines harboring GATA box mutations (TTATCT to TGGCTT) (Fig. (Fig.5A)5A) in the G1BAC-GFP backbone (GATAmut lines). We analyzed GFP, FcγR II/III, and CD71 expression by FACS in the Lin− c-Kit+ immature fraction or Ter119+ mature fraction of bone marrow cells.
The GATAmut mice showed a GFP expression pattern quite similar to those of the ΔG1HE and ΔG1HE core mice (Fig. (Fig.4A4A and and5B).5B). The GFP+ FcγRII/IIIlow fraction (fraction II-F) in the GATAmut mice was reduced to 25% of control (Fig. (Fig.5C).5C). Meanwhile, there was practically no reduction in the number of cells in the GFP+ FcγRII/III− fraction (fraction III-F) in the GATAmut mice, although the mean fluorescence intensity of these cells was repressed to approximately 50% of control (Fig. (Fig.5D).5D). Furthermore, the mean GFP fluorescence intensity of the Ter119+ fraction in the GATAmut mice was also suppressed compared to that of control mice (Fig. 5E and F). Thus, the GATA box in the G1HE core is responsible for the G1HE activity, especially in BREP stage progenitors, and this GATA box activity also contributes to Gata1 expression in CREP stage cells and in further differentiated erythroblasts.
Although the lack of GATA1 causes erythroid differentiation arrest at the proerythroblast stage that corresponds to CREP, the consequence of the loss of GATA1 function at the earlier erythroid differentiation stage has never been analyzed precisely. To address this issue, we crossed Gata1.05 knockdown mice with G1BAC-GFP transgenic-mouse lines and analyzed the GFP expression in both the BREP and CREP populations in the Gata1 knockdown background. The Gata1.05 allele confers attenuated expression of Gata1 at approximately 5% of the control level throughout hematopoietic development (40). Since Gata1 is an X chromosome gene, hematopoietic cells in female heterozygous mutants (Gata1G1.05/+) contained both normal GATA1-expressing cells and Gata1.05 knockdown cells due to random X chromosome silencing, allowing their survival to adulthood (39). Surprisingly, the GFPhigh FcγRII/IIIlow BREP population (fraction II-F) was significantly increased in the Gata1G1.05/+ mouse bone marrow compared to the wild-type control, whereas there was not much difference in the GFPhigh FcγRII/III− CREP fraction (fraction III-F) (Fig. 6A and B).
To examine the contribution of Gata1.05 knockdown cells to the BREP and CREP populations, we sorted the fractions with FACS and examined the GATA1 mRNA expression levels by means of real-time RT-PCR analysis. Although the endogenous GATA1 mRNA expression levels were reduced in both fractions II-F and III-F of the Gata1.05 knockdown mice, the reduction in GATA1 mRNA relative to the wild-type control was much more significant in fraction II-F (9.8% of control) than in fraction III-F (54.4% of control) (Fig. (Fig.6C).6C). This observation suggests that the accumulation of Gata1.05 knockdown cells is more significant in the BREP fraction than in the CREP fraction. Presumably, the majority of Gata1.05 knockdown cells failed to differentiate to the CREP stage and accumulated in the BREP stage (Fig. (Fig.6D).6D). Thus, in these mice, a large part of the fraction III-F population seemed to be composed of normal GATA1-expressing cells, expanding and compensating for the suppressed differentiation of the Gata1.05 knockdown cells.
Finally, we examined whether targeted deletion of G1HE from the mouse chromosomal Gata1 locus leads to the suppression of endogenous Gata1 gene expression. To this end, we utilized a previously published Gata1 ΔneoΔHS allele in which an 8-kbp genomic region containing G1HE is deleted from the chromosomal Gata1 locus (19). By crossing ΔneoΔHS mice with G1BAC-GFP mice, we identified and separated the BREP and CREP fractions of the Gata1ΔneoΔHS/Y mutant. In the analysis of erythroid progenitors, we found that fraction II-F (the BREP population) was significantly expanded in the Gata1ΔneoΔHS/Y mouse bone marrow compared to that of the wild-type control (Fig. 7A and B). This pattern of erythroid progenitor expansion is similar to that observed in the Gata1G1.05/+ mouse bone marrow (Fig. 6A and B). The GATA1 mRNA level in the accumulated BREP stage cells was more severely suppressed than that in the CREP fraction, which is also consistent with the suppressed GFP expression in the BREP fraction of ΔG1HE, ΔG1HE core, and GATAmut G1BAC-GFP transgenic mouse bone marrow cells (Fig. (Fig.7C).7C). These results further support our contention that G1HE plays a critical role in Gata1 gene expression in the BREP stage cells and that GATA1 expression in cells in this stage is essential for the BREP-to-CREP differentiation of erythroid progenitor cells (Fig. (Fig.7D7D).
In this study, we examined how the stage-specific expression of the Gata1 gene is attained during erythroid differentiation using the G1BAC-GFP transgenic mouse. We found that the G1BAC-GFP transgenic-mouse system gave rise to stable expression of the GFP reporter in vivo. Therefore, by exploiting this system, we identified the two important erythroid progenitor populations BREP and CREP, containing mainly BFU-E stage progenitors and CFU-E stage progenitors, respectively. Deletion of the G1HE from the G1BAC-GFP construct (i.e., ΔG1HE) resulted in a significant reduction in the GFP-positive progenitor population in the BREP fraction. In contrast, GFP-positive progenitors in the CREP fraction and Ter119-positive erythroblasts were practically preserved in the ΔG1HE mice, albeit the GFP intensity was reduced to almost half of the G1BAC-GFP level. We also found that targeted deletion of G1HE from the chromosomal Gata1 locus (the ΔneoΔHS mutant) markedly suppressed the GATA1 mRNA level in the BREP fraction. These results demonstrate that the G1HE significantly contributes to Gata1 gene activation in BREP stage cells and to the differentiation of BREP cells to CREP stage cells and later erythroblasts.
To further dissect the mechanisms for this decrease of Gata1 gene expression in GFP-positive BREP cells, we deleted the G1HE core from the G1BAC-GFP construct or introduced mutations into the GATA box therein. Transgenic-mouse lines harboring these two mutant BAC constructs showed quite similar decreases in GFP-positive cells in the BREP fraction compared to that in the control mice, indicating that the GATA box is an indispensable motif in the G1HE. Thus, the Gata1 gene is regulated distinctly in each differentiation stage of erythroid cell development. This mode of regulation is summarized in Fig. Fig.7D7D.
In order to monitor the endogenous Gata1 gene expression profile, in this study we established a novel reporter transgenic-mouse system using a 196-kbp G1BAC. We previously developed and analyzed the plasmid-based Gata1 gene vector G1HRD, which covers approximately 8 kbp of the Gata1 gene (26). Except for the difficulty in achieving copy number-dependent transgene expression, which is inherent in the plasmid-based transgenic-mouse approach, the G1HRD substantially recapitulated Gata1 gene expression in the mouse (23). Furthermore, GATA1 cDNA expressed under the regulatory influence of the G1HRD nicely rescued the GATA1-deficient mouse from embryonic lethality (33, 41). These results strongly supported the G1HRD containing virtually all the essential regulatory elements for hematopoietic Gata1 gene expression. However, we noticed that the G1HRD had an apparent limitation in fully recapitulating Gata1 gene expression and that the G1HRD could not direct reporter gene expression in BREP or BFU-E stage progenitors (38). To this end, we examined the G1BAC-based reporter system.
In this study, we found that G1BAC-GFP recapitulated the endogenous Gata1 gene expression in the BREP stage more faithfully than the G1HRD. Importantly, each transgenic-mouse line exhibited reproducible copy number-dependent GFP reporter expression. Thus, the G1BAC construct retained sufficient regulatory information for Gata1 gene expression during erythroid differentiation. We envisage that the G1BAC may harbor an insulator activity somewhere outside the G1HRD region that acts to circumvent the chromosomal position effect variegation at the transgene integration site (reviewed in reference 10). In this regard, the recent identification of DNase I-hypersensitive sites located beyond the G1HRD region is noteworthy and may provide clues for further study of Gata1 gene regulation (43).
GATA1 plays multiple roles during erythroid cell development (3, 25). We surmise that the molecular basis of these differential contributions of GATA1 to erythropoiesis may be due, at least in part, to the differences in the threshold set by each gene for activation by GATA1. By comparing the expression profiles of GATA1 target genes in erythroid cells from wild-type, Gata1 knockdown, and Gata1 knockout mice, we noticed that these graded levels of GATA1 expression actually influenced the decision of erythroid progenitors to undergo apoptosis, proliferation, or terminal differentiation (28). We found here that the GFP level, and hence the GATA1 expression level, changes in erythroid cells at each stage. We surmise that this stage-specific expression of GATA1 may be essential for erythropoiesis. Indeed, a sixfold overexpression of GATA1 in differentiating erythroblasts led to blocked erythropoiesis (48) and decreased GATA1 expression to approximately 5% of the wild-type level, resulting in erythroleukemia in mice (31, 32). These accumulating lines of evidence support our contention that the stage-specific fine-tuning of Gata1 gene expression is critical to the mechanisms orchestrating erythropoiesis.
We have provided convincing evidence that the conserved GATA box in the G1HE core is required for G1HE activity, especially at the BREP stage. Our next question concerned which GATA factor actually binds to this GATA box at the BREP stage. In the present study, we found that GFP intensity was not decreased in BREP cells harboring the GATA1 knockdown mutation (Fig. (Fig.6A6A and and7A),7A), suggesting that other GATA factors might contribute to the Gata1 gene activation at the BREP stage. In this regard, it is interesting that GATA2 is expressed in immature hematopoietic progenitors and that this GATA2 expression is repressed by GATA1 during erythroid differentiation (7). GATA2 is expressed more abundantly in the BREP fraction than in the CREP fraction or later erythroblasts (data not shown). Moreover, the expression level of GATA2 was higher in BREP cells isolated from the Gata1G1.05/+ mice than in those from wild-type mice (data not shown). It was also reported that GATA2 binds to this GATA box in the early stage of erythroid differentiation (23, 27). Thus, it seems plausible that, in G1HE-mediated Gata1 gene activation, GATA2 works during the early stage of erythroid differentiation or in BREP stage cells but that there is a switch over to GATA1 in the later stage. Considering this situation, it would be intriguing to examine the nature of the transcription factor complex containing GATA2 and its binding to the G1HE core using sorted BREP cells.
One of the important applications of the new cell-sorting strategy for EEP or BREP cells based on the expression of G1BAC-GFP is characterization of the normal erythroid differentiation pathway and disorder of the pathway. We observed an abnormal accumulation of BREP stage cells in the bone marrow of Gata1.05 knockdown heterozygous (Gata1G1.05/+) mice, as well as Gata1ΔneoΔHS/Y mutant mice, in which Gata1 gene expression is markedly suppressed in the BREP stage. These observations suggest that in BREP stage cells, GATA1 is essential for driving normal erythroid differentiation and for suppressing the abnormal proliferation of BREP stage cells. Consistent with this notion, it was recently reported that differentiation of GATA1-deficient ES cells was partially arrested at the MEP stage and that these cells are also contained in the BREP population (35). Since Gata1G1.05/+ knockdown mice develop c-Kit-positive leukemia (31, 32), the abnormally accumulated BREP population might be predisposed to c-Kit-positive leukemogenesis.
To simplify, the molecular basis of cell differentiation can be defined as the inducible changes in gene expression profiles that are controlled by lineage-restricted transcription factors. Therefore, in order to understand the mechanisms by which the cell differentiation process is regulated, it is crucial to delineate the regulation in terms of the expression of lineage-restricted transcription factors. We have described here how the use of Gata1 BAC transgenic mice, coupled with homologous-recombination-based mutation analysis, proved to be an outstanding approach leading to the identification and dissection of GATA regulatory domains, as well as to a detailed understanding of their activities. The present analysis unequivocally demonstrates a novel function of the G1HE, which is essential for BREP stage-specific Gata1 expression. We speculate that the other Gata1 gene regulatory elements also possess such regulatory functions. As genetic mutations in the human GATA1 locus have been linked to human diseases (4, 5, 20, 22), the possibility exists that mutations in the regulatory region might also be responsible for human diseases by affecting the expression level of GATA1. Only through the identification of the enhancer functions and an understanding of their precise activities will we be able to elucidate the complex network of regulatory control underlying erythroid differentiation.
We thank Norio Suzuki, Naoshi Obara, and Harumi Yamamoto-Mukai for help and discussion and Yuko Kikuchi, Reiko Kawai, Naomi Kaneko, and Mitsuru Okano for technical assistance. We also thank Tania O'Connor for critical reading of the manuscript.
This work was supported in part by grants from ERATO-JST (M.Y.), Takeda Science Foundation (T.M.), and Naito foundation (M.Y.) and Grants-in-Aid for Creative Scientific Research (M.Y.), for Scientific Research on Priority Areas (M.Y.), for Scientific Research (T.M. and M.Y.), and for Exploratory Research (M.Y.) from the Ministry of Education, Science, Sports and Culture.
Published ahead of print on 22 December 2008.
†Supplemental material for this article may be found at http://mcb.asm.org/.