Genetic Analysis of Hierarchical Regulation for Gata1 and NF-E2 p45 Gene Expression in Megakaryopoiesis

Abstract

GATA1 and NF-E2 p45 are two important regulators of megakaryopoiesis. Whereas GATA1 is known to regulate the p45 gene, details of the GATA1 contribution to the spatiotemporal expression of the p45 gene remain to be elucidated. To clarify the relationship between GATA1 and p45, we performed genetic complementation rescue analysis of p45 function in megakaryocytes utilizing the hematopoietic regulatory domain of the Gata1 gene (G1HRD). We established transgenic mouse lines expressing p45 under G1HRD regulation and crossed the mice with p45-null mice. Compound mutant mice displayed normal platelet counts and no sign of hemorrhage, indicating that G1HRD has the ability to express p45 in a spatiotemporally correct manner. However, deletion of 38 amino acids from the N-terminal region of p45 abrogated the p45 rescue function, suggesting the presence of an essential transactivation activity in the region. We then crossed the G1HRD-p45 transgenic mice with megakaryocyte-specific Gata1 gene knockdown (Gata1^ΔneoΔHS) mice. The G1HRD-p45 transgene was insufficient for complete rescue of the Gata1^ΔneoΔHS megakaryocytes, suggesting that GATA1 or other factors regulated by GATA1 are required to cooperate with p45 for normal megakaryopoiesis. This study thus provides a unique in vivo validation of the hierarchical relationship between GATA1 and p45 in megakaryocytes.

Cell differentiation is uniquely regulated by lineage-specific transcription factors. Efforts have been made to understand the function of each transcription factor from the viewpoint of context-dependent and hierarchical relationships with other transcription factors. Recent studies have revealed critical regulatory networks among key transcription factors regulating differentiation of hematopoietic cells (8, 51). However, such interrelations between transcription factors are usually validated in cell culture but not by in vivo experimental systems. Recent studies have shown that the latter approach is more informative than in vitro or in transfecto approaches because various physiological parameters are present for validation.

Several key transcription factors regulating megakaryocytic differentiation have been described. c-Myb is a critical regulator at the bifurcation of erythroid and megakaryocytic differentiation from the megakaryocytic-erythroid bipotential progenitor (MEP). Decreased c-Myb activity in MEP was found to enhance megakaryocytic differentiation (27). The other factors regulating megakaryocytic differentiation, especially after lineage commitment, include GATA1, GATA2, SCL/Tal1, Runx1, and the Ets family factors Fli-1 and TEL (32, 38, 41). In contrast, terminal maturation of megakaryocytes depends heavily on NF-E2, a heterodimer of the cap'n'collar (CNC) transcription factor p45 and a small Maf protein (24, 31, 34, 36).

GATA1 has been shown to be an indispensable regulator of erythroid and megakaryocytic cell differentiation. Disruption of the Gata1 gene in mouse results in lethality at the midgestation stage due to the failure of primitive hematopoiesis (10, 42). In contrast, a megakaryocyte-specific knockdown of the Gata1 gene (Gata1^ΔneoΔHS/Y) was reported not to be lethal but to result in severe thrombocytopenia and accumulation of immature megakaryocytes (37). The temporal and spatial expression pattern of Gata1 conforms with these contributions of GATA1 to erythroid and megakaryocytic cell differentiation (43). Regulatory regions recapitulating endogenous Gata1 gene expression in hematopoietic cells have been delineated (30), and we refer to the regulatory region as the Gata1 gene hematopoietic regulatory domain (G1HRD).

It has been shown that impaired function of NF-E2 affects terminal maturation of megakaryopoiesis, resulting in the accumulation of mature megakaryocytes with higher ploidy (34, 36) and defective proplatelet formation (16, 31). As the small Maf proteins lack any canonical transactivation domains, the transcription activation ability of NF-E2 resides solely in the p45 subunit (3, 24, 29), suggesting that p45 abundance may be a primary determinant of NF-E2 activity. In this regard, the N-terminal half of p45 has been identified as a transactivation domain (3, 29), which recruits TAF_II130 and CBP (1, 12). Another structural feature of the p45 N-terminal region is the presence of two WW domain-binding motifs (or PPXY motifs), which are necessary for β-Globin gene transcription (15, 22). In addition to the PPXY motifs, the very end of the N-terminal region is also necessary for β-Globin gene transcription (3). It should be noted that all of these studies examined the domain function of p45 in immortalized erythroid cells but not in the megakaryocytic lineage.

Three lines of evidence suggest that GATA1 directly activates p45 gene expression in megakaryocytes. First, genetic analyses revealed that GATA1 dysfunction causes a reduction of p45 expression in megakaryocytes (35, 49). Second, a well-conserved tandem palindromic GATA-binding site has been found in the p45 gene 1b promoter, which was shown to be functional in a reporter assay using K562 cells (21). Third, GATA1 appears to be required at earlier stages of megakaryopoiesis than p45 (5, 36, 37, 49). In addition to GATA1, it has also been shown that other factors, such as GATA2 and SCL, participate in p45 gene regulation in megakaryocytes (4, 19).

To validate the GATA1-p45 regulatory axis in megakaryocytes, we adopted a transgenic complementation rescue approach. We examined whether G1HRD directs sufficient expression of p45 to sustain normal megakaryopoiesis. We generated transgenic mouse lines expressing p45 under the control of G1HRD, and these lines were crossed into the p45-null background. The compound mutant mice showed normal platelet counts and no sign of hemorrhage, indicating that G1HRD-driven p45 rescued the defective megakaryopoiesis and thrombogenesis of p45-null mice. In contrast, a p45 mutant lacking 38 amino acids of the N-terminal region could not rescue thrombogenesis in p45-null mice, suggesting the presence of essential transactivation activity in this region. We also evaluated the contribution of p45 to the GATA1-directed regulatory hierarchy by crossing the G1HRD-p45 mice with Gata1^ΔneoΔHS mice. Whereas Gata1^ΔneoΔHS/Y megakaryocytes express reduced amounts of p45, the p45 transgene rescued p45 levels to normal levels in megakaryocytes from compound mutant mice. Of the phenotypes observed in Gata1^ΔneoΔHS/Y megakaryocytes, decreased expression of platelet genes and an abnormal increase in immature megakaryocytes were partially restored by the transgene-derived p45 expression, but thrombocytopenia in peripheral blood was not substantially improved. This study is a unique in vivo validation of the hierarchical relationship between GATA1 and p45, demonstrating that the GATA1-p45 regulatory axis is operative in megakaryopoiesis.

MATERIALS AND METHODS

Generation of transgenic mouse lines and mating with p45-null and Gata1 knockdown mice.

G1HRD-polyAH was first generated to construct G1HRD-p45. A XhoI-NotI fragment of IE3.9int-LacZ (30) was inserted into the XhoI-NotI site of pIM-LacZ (23) to make G1HRD-polyAH. The mouse p45 cDNA was inserted into the NotI site of G1HRD-polyAH to generate G1HRD-p45. To construct G1HRD-ΔN38 p45, a DNA fragment encoding ΔN38 p45 (deletion of amino acids 2 to 38) was amplified by PCR and cloned into the pGEM-T Easy vector (Promega, Madison, WI). The NotI-NotI fragment containing the cDNA encoding ΔN38 p45 was then inserted into the NotI site of G1HRD-polyAH. The XhoI-HindIII fragments of G1HRD-p45 and G1HRD-ΔN38 p45 were purified and injected into fertilized eggs using standard procedures as described previously (23).

The G1HRD-p45 and G1HRD-ΔN38 p45 transgenes were identified by PCR using the primer pair 5′-TGT CTC ACA ACC CTT TCT GTC C-3′ and 5′-GCT TGG GTT GCC CCA CTG GC-3′ to amplify the transgene. p45 heterozygous mice (36) and Gata1^ΔneoΔHS (20) mating pairs were purchased from Jackson Labs (Bar Harbor, ME). p45 heterozygous mice were mated with G1HRD-p45 and G1HRD-ΔN38 p45 transgenic mice to generate p45^+/−:G1HRD-p45 (p45^+/−:Tg) and p45^+/−:G1HRD-ΔN38 p45 (p45^+/−:ΔNTD Tg) mice, respectively. p45^+/−:Tg and p45^+/−:ΔNTD Tg mice were further crossed with p45^+/− mice to obtain p45^−/−:Tg mice and p45^−/−:ΔNTD Tg mice, respectively. Genotypes were determined by 2 weeks after birth. Gata1^ΔneoΔHS/X female mice were mated with G1HRD-p45 transgenic male mice to generate Gata1^ΔneoΔHS/Y:Tg mice. Genotyping of the Gata1 locus for the ΔneoΔHS allele was performed as previously described (20). All analyses using transgenic mice were performed with two independent lines, and results obtained from one line are shown as representative. All animal experiments were carried out under the permission of the animal center of Tohoku University.

Hematological analysis.

Whole blood was collected from the retro-orbital sinuses of anesthetized mice, and hematopoietic indices were measured using an automatic blood cell analyzer (Nihon Koden, Tokyo, Japan). Peripheral blood smears were stained with Wright-Giemsa stain.

Histological analysis.

Spleens were fixed in 3.7% formalin in phosphate-buffered saline (PBS). The samples were embedded in paraffin and stained with either hematoxylin and eosin (HE) or silver impregnation. The microscopic images were captured with a DP71 digital camera system (Olympus, Tokyo, Japan).

Proplatelet formation.

Megakaryocytes were isolated from the bone marrow of 2-month-old mice, and the projection of proplatelets was observed as described previously (31).

Primary culture of megakaryocytes.

Whole livers were collected from mouse fetuses at embryonic days 13.5 (E13.5) and E14.5, and single-cell suspensions were prepared by successive passage through 25-gauge needles. Fetal liver cells were maintained in RPMI 1640 (Wako, Osaka, Japan) supplemented with 20% charcoal-stripped fetal bovine serum (FBS), 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 ng/ml recombinant human thrombopoietin (TPO) (generously provided by Kyowa Hakko Kirin Pharma). From a day 3 culture, megakaryocytes were harvested for RNA purification. All samples processed from primary megakaryocytes were compared within littermates.

RNA purification and quantitative reverse transcription (RT)-PCR.

CD41⁺ cells, Ter119⁺ cells, and Mac-1⁺ cells were selected from a day-3 culture of E14.5 fetal liver cells and bone marrow cells of adult wild-type mice using Dynabeads (Dynal Biotech ASA, Oslo, Norway). Fetal liver cultures and bone marrow cells were incubated with biotinylated anti-CD41 antibody (clone MWReg30; Serotec), biotinylated anti-Ter119 antibody (Ly-76; BD Pharmingen), and biotinylated anti-Mac-1 antibody (M1/70; BD Pharmingen), followed by incubation with streptavidin-coupled Dynabeads. Total RNA was purified from the sorted cells using RNAiso (Takara), and cDNA was synthesized from the RNA using random hexamers. Real-time PCR was performed using an ABI7300 sequence detection system. The reaction was carried out for 40 to 60 cycles of 30 s at 95°C and 1 min at 60°C using qPCR Mastermix (Eurogentec). rRNA control reagents (Applied Biosystems) were used as an internal control. To examine the molar ratio of 1b transcript against 1a transcript, quantitative real-time PCR was performed using qPCR Mastermix for SYBR green I (Eurogentec). The molar ratio was calculated by the following equation (44): molar ratio = [L_a × (1 + E_a)^CTa]/[L_b × (1 + E_b)^CTb]. L_a and L_b indicate lengths of the amplicon for 1a and 1b transcripts, respectively. E_a and E_b indicate the amplification efficiency of a primer set for 1a and 1b transcripts, respectively. CT_a and CT_b indicate the numbers of threshold cycles for the 1a and 1b transcripts, respectively. Sequences of primers and probes used for quantitative real-time PCR are shown in Table 1.

TABLE 1.

Sequences of primers and probes used for quantitative real-time PCR

Primer name	Sequence
p45-F	5′-TTGGCACAGTATCCGCTAAC-3′
p45-R	5′-TCCAGTTTTCTCTTGCGACA-3′
p45-P	5′-FAM-TTCGGGACATCCGTCGAC-TAMRA-3′
Txas-F	5′-AATTGGAACTCCGAGAGCGA-3′
Txas-R	5′-GACAACGTGCATCCGACG-3′
Txas-P	5′-FAM-ACGGGCCTCTGTGTGGGTACTATCTTGG-TAMRA-3′
Slc6a4-F	5′-AGACGTGTCCGAGGTGGC-3′
Slc6a4-R	5′-TGCCTCCGCATATGTGATGA-3′
Slc6a4-P	5′-FAM-AGACGCGGGCCCCAGCCTC-TAMRA-3′
Gp6-F	5′-AGCCCCGAGTGACCCTCTA-3′
Gp6-R	5′-TCCGTGGGTACCTGGCTG-3′
Gp6-P	5′-FAM-TGCTTGTGGTTACTGGACTCTCTGCCAC-TAMRA-3′
Selp-F	5′-ACTGGTCAGATGCCATGCC-3′
Selp-R	5′-ACCCAAGTACGTCAGAGCTTCC-3′
Selp-P	5′-FAM-CTGCCAAGCAGGGACACTGACAATCC-TAMRA-3′
Slamf1-F	5′-AGCCTGCCCCAGAGTCTGT-3′
Slamf1-R	5′-TATGGGTCAGCTCTCTGGCAGT-3′
Slamf1-P	5′-FAM-CAGGAACCAAACCCCACCACAGTTTATG-TAMRA-3′
F5-F	5′-GGCATGCAAACGCCATTT-3′
F5-R	5′-ATATTCCGAAGCCTTGATCTGTG-3′
F5-P	5′-FAM-TGCCAATGGGACTAAGCACTGGTGTCA-TAMRA-3′
p45 promoter 1a-F	5′-CACAGGTGCCTGAAAGGTTG-3′
p45 promoter 1b-F	5′-TGGGAAGAAAGGGCTGAAAT-3′
p45 promoter-R	5′-ACCTGTTCCTGTTCTGCTGA-3′
Gata1-F	5′-CAGAACCGGCCTCTCATCC-3′
Gata1-R	5′-TAGTGCATTGGGTGCCTGC-3′
Gata1-P	5′-FAM-CCCAAGAAGCGAATGATTGTCAGCAAA-TAMRA-3′

FACS analysis.

Primary megakaryocytes cultured from fetal livers were stained with biotinylated anti-CD41 antibody, allophycocyanin (APC)-conjugated streptavidin, and phycoerythrin (PE)-conjugated anti-CD61 antibody (BD Pharmingen, San Diego, CA). The cells were analyzed using a FACSAria cell sorter (Becton Dickinson, San Jose, CA).

Immunoblot analysis.

Whole-cell extracts were prepared from bone marrow megakaryocytes. To purify megakaryocytes, bone marrow cells were incubated with biotinylated anti-CD41 antibody (clone MWReg30; Serotec), followed by incubation with streptavidin-coupled Dynabeads. Whole-cell extracts were examined by immunoblot analysis using anti-p45 antibody (sc-291X; Santa Cruz) and anti-lamin B (sc-6217; Santa Cruz). Whole-cell extracts were prepared from 293T cells 24 h after transfection and examined in the same way.

Plasmid construction.

To generate pGL3-TATA-p45 1b, a 160-bp fragment of the mouse p45 1b promoter containing a tandem GATA site (from nucleotide position −163 to −4 relative to the transcription initiation site, set as +1) was amplified by PCR using the primers 5′-GGG ACA TGA CCA AAT GAC CT-3′ and 5′-TAG CAA CCC TTC CCT CTC CT-3′. The fragment was cloned into pGEM-T Easy (Promega, Madison, WI), and a NotI-NotI fragment was inserted into the SmaI site of pGL3-TATA. pGL3-TATA was constructed by replacing the BglII-HindIII region containing the simian virus 40 (SV40) promoter of the pGL3 promoter vector with a minimal TATA promoter derived from the rabbit β-globin gene (53). To generate pGL3-TATA-p45 1b-mutGATA, a double GATA site, TGATAA ACCCC TTATCT, was mutated to TttTAA ACCCC TTAaaT by PCR. To generate mammalian expression plasmids of p45 and its truncated mutants, cDNA fragments containing His₆-tagged wild-type p45, ΔN38 p45 (deletion of amino acids 2 to 38), ΔN109 p45 (deletion of amino acids 2 to 109), ΔN153 p45 (deletion of amino acids 2 to 153), and ΔN225 p45 (deletion of amino acids 2 to 225) were amplified by PCR, and their NcoI-SalI fragments were replaced with that of pEF-p45 (13). To generate pIM-His-MafK XB, pET15b-MafK was first generated by inserting a full-length mouse MafK cDNA between the NdeI and BamHI sites of pET15b. The XbaI-BamHI fragment of pET15b-MafK was inserted into the NotI-NotI vector fragment of pIM-lacZ (25).

Transient transfection and reporter assay.

MEG01 cells were maintained in RPMI 1640 (Wako) supplemented with 10% charcoal-stripped FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. To attain cell differentiation, 10 nM 12-O-tetradecanoylphorbol-13-acetate (TPA) was added to the subconfluent MEG01 culture. At 24 h after the addition of TPA, either pGL3-TATA, pGL3-TATA-p45 1b, or pGL3-TATA-p45 1b-mutGATA was electroporated using a MicroPorator-mini instrument (Digital Bio). 293T cells were maintained in Dulbecco's modified Eagle medium (DMEM) (Wako) supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. pRBGP2, a previously described reporter plasmid (13), was introduced into 293T cells along with expression vectors encoding p45 and its truncated mutants and the MafK expression vector pIM-His-MafK XB using the FuGENE6 transfection reagent (Roche) according to the manufacturer's protocol. For both experiments, luciferase activity was measured 24 h after transfection. The firefly luciferase activities of the reporter genes were normalized against the cotransfected sea pansy luciferase activities driven by the elongation factor 1α promoter. The expression of both firefly and sea pansy luciferases were quantified using the dual-luciferase reporter assay system (Promega, Madison, WI) and a luminometer (Berthold Japan, Tokyo, Japan). All samples were prepared in triplicate, and means and standard deviations were calculated.

RESULTS

The p45 gene utilizes both 1a and 1b promoters in megakaryocytes.

The p45 gene possesses two alternative first exons with affiliated promoters, which produce transcripts with different untranslated sequences (21, 33, 45). The distal and proximal first exons are designated 1a and 1b (or 1f) exons, respectively, and organization of the exons is conserved between human and mouse (Fig. 1A). There is a tandem palindromic GATA binding motif within the 1b promoter, and this is conserved between the human and mouse p45 genes. Previous studies have demonstrated that the 1b promoter is utilized in erythroid cells but not in myeloid cells (21), while in contrast, the 1a promoter is widely utilized, including in myeloid cells. These observations imply that the 1a and 1b promoters are differentially regulated.

FIG. 1.

Promoter utilization of the p45 gene in megakaryocytes and contribution of a tandem GATA site to 1b promoter activity. (A) The structures of mouse and human p45 genes. White and gray boxes indicate untranslated regions and protein-coding regions, respectively. The tandem GATA motifs are indicated with arrowheads. Arrows indicate PCR primers for specific detection of each transcript variant. Two different primers corresponding to each first exon were coupled with a common primer corresponding to the second exon. The bar corresponds to 1 kb. (B) Relative abundance of the transcript variants derived from 1a (left panel) and 1b (right panel) promoters. cDNA synthesized from total RNA of CD41-positive, Ter119-positive, and Mac1-positive cells isolated from bone marrow cells was examined. The specific primer sets depicted in panel A were used for detecting the 1a and 1b transcripts. CD41⁺, n = 4; Ter119⁺, n = 4; and Mac1⁺, n = 4. Each value was normalized to rRNA expression. Average values are expressed, and error bars indicate standard deviations. (C) 1b promoter activity of the p45 gene was examined in TPA-induced MEG01 cells. The tandem GATA site contained in pGL3-TATA-p45 1b and the mutated tandem GATA site contained in pGL3-TATA-p45 1b-mutGATA are shown in boldface, and the mutated bases are underlined (top panel). Relative luciferase activities are shown in the bar graph (bottom panel). The average values are shown with standard deviations. A representative result from three independent experiments is shown. The horizontal axis indicates relative luciferase activity. Firefly luciferase activity with the control reporter plasmid pGL3-TATA is set to 1. The Student t test was used to calculate statistical significance (P). *, P < 0.05. (D and E) Comparison of the molar abundances of 1a and 1b transcripts between CD41⁺ (n = 4), Ter119⁺ (n = 4), and Mac1⁺ (n = 4) cells of wild-type mice (D) and between CD41⁺ cells of Gata1^X/Y mice (n = 3) and those of Gata1^ΔneoΔHS/Y mice (n = 4) (E). The stacked bar graph (left panels) shows the relative abundances of 1a and 1b transcripts within the total p45 transcript. The molar ratio of the 1b transcript against the 1a transcript was calculated and is shown in the bar graph (right panels). Average values are expressed, and error bars indicate standard deviations. The Student t test was used to calculate statistical significance (P). *, P < 0.05.

To gain insight into the transcriptional regulatory mechanisms of the p45 gene in megakaryocytes, we first examined which promoter is utilized in megakaryocytes (Fig. 1B). The 1a exon was transcribed in erythroid and myeloid cells, while the 1b exon was transcribed in erythroid cells but not myeloid cells, which agrees with a previous report (21). In megakaryocytes, the 1b exon was transcribed in addition to the 1a exon.

The critical contribution of the tandem GATA site to 1b promoter activity was shown in an erythroleukemic cell line, K562 (21). The recruitment of GATA1 and GATA2 to the 1b promoter region in K562 cells was also shown in a recent report (9a). To examine the activity of the 1b promoter in megakaryocytes, we used MEG01 cells, which are a human megakaryoblastic cell line and can be induced to differentiate by TPA into a megakaryocytic lineage (28). The 1b promoter was active in TPA-treated MEG01 cells, and mutation of the tandem GATA site completely abrogated the activity (Fig. 1C). Thus, the p45 gene is regulated, at least in part, through a GATA-dependent mechanism.

We found that the 1a transcript is expressed more abundantly than the 1b transcript in erythroid, megakaryocytic, and myeloid lineages (Fig. 1D, left panel). The contribution of 1b is larger in megakaryocytes and erythroid cells than in myeloid cells (Fig. 1D, right panel). While GATA1-deficient megakaryocytes from Gata1^ΔneoΔHS/Y mice displayed dramatic reduction of the 1b transcript compared to the 1a transcript (Fig. 1E, right panel), the amount of 1a transcript was also decreased in the absence of GATA1 (Fig. 1E, left panel). These results demonstrate that GATA1 is a major activator of the 1b promoter and that GATA1 partially contributes to 1a promoter activity, although GATA sites are not found in proximity to the 1a promoter. Since a substantial amount of the 1a transcript was produced in the absence of GATA1, there seems to be a GATA1-independent regulation of the p45 gene in megakaryocytes.

Generation of transgenic mouse lines expressing p45 under regulation of G1HRD.

To address the question of whether GATA1-dependent regulation is sufficient to support p45 function in vivo, we performed a transgenic complementation rescue experiment with p45-null mice with a transgene producing p45 under the regulatory influences of the Gata1 gene. Our rationale was that if GATA1 is the major activator of the p45 gene, the time and place of p45 expression should be contained within those for Gata1 gene expression and the defective megakaryopoiesis observed in p45-null mice would be rescued by p45 expressed under the regulation of the Gata1 gene.

Regulatory domains mediating tissue- and stage-specific expression of the Gata1 gene have been well analyzed in transgenic mouse reporter experiments (30, 39, 40, 43). To express p45 in mice, we utilized G1HRD, an 8.0-kb genomic fragment of the Gata1 gene that is sufficient for recapitulation of endogenous GATA1 expression in erythroid and megakaryocytic lineages. We generated two independent lines of transgenic mice expressing p45 driven by G1HRD (Fig. 2A). G1HRD-p45 transgenic mice (Tg mice) were viable and fertile and did not show apparent hematological abnormalities (data not shown).

FIG. 2.

G1HRD-driven p45 corrected thrombocytopenia and defective megakaryopoiesis of p45-null mice. (A) The structure of the G1HRD-p45 transgene. G1HRD contains the Gata1 genomic locus spanning from the 3.9-kb upstream region of the hematopoietic IE exon to the second exon (II). A cDNA fragment encoding p45 was fused to the 5′ half of the second exon. (B) Expression of p45 in wild-type and p45^−/−:Tg mice. Whole-cell extracts were prepared from bone marrow megakaryocytes and subjected to immunoblot analysis with anti-p45 antibody. Lamin B was used as a loading control. Arrows indicate a doublet of endogenous p45, and an arrowhead indicates transgene-derived p45. (C) Number of pups of each genotype born from matings between p45^+/− mice and p45^+/−:Tg mice. Genotypes were determined at 2 weeks after birth. Two independent transgenic lines (205 and 207) were examined. The expected numbers calculated from Mendelian inheritance are shown in parentheses. Comparison of two proportions was performed to calculate statistical significance (P) between the birth frequencies of p45^−/− pups and p45^−/−:Tg pups. *, P < 0.05. WT, wild type. (D) Wright-Giemsa staining of peripheral blood smears drawn from wild-type (left), p45-null (middle), or p45^−/−:Tg (right) mice. White and gray arrowheads indicate platelets and poikilocytes, respectively. A result from line 205 is shown as a representative. (E) Macroscopic observation of spleens: (i) wild-type, (ii) p45-null, and (iii) p45^−/−:Tg mice at 2 months of age are shown. The scale bar corresponds to 1 cm. Results from line 207 are shown as representative data. (F and G) Hematoxylin and eosin staining (F) and silver impregnation (G) of spleens from wild-type (left), p45-null (middle), and p45^−/−:Tg mice (right). Scale bars correspond to 100 μm (F) and 50 μm (G). Results from line 205 are shown as representative data (F and G).

G1HRD-driven p45 sustains thrombogenesis in the absence of endogenous p45.

We crossed the Tg mice into a p45-null background, and p45^−/−:Tg mice were obtained from the mating pairs of p45^+/− and p45^+/−:Tg mice. The expression level of the transgene was examined by an immunoblot analysis of whole-cell extracts prepared from bone marrow megakaryocytes (Fig. 2B). For both lines 205 and 207, transgene expression in each mouse varies to some extent even though they were obtained from the same line, which was interpreted as position effect variegation of the transgene. The fluctuation of the expression levels of the transgene did not affect the hematological and histological phenotypes of the compound mutant mice (see below). The results of gene expression analysis were interpreted by considering the abundance of transgene-derived p45 (Fig. 3). While wild-type mice contain two endogenous p45 bands, the size of transgenic p45 coincides with the more slowly migrating band. The molecular mechanism for this phenomenon is unknown at present.

FIG. 3.

G1HRD-driven p45 contributes to activation of platelet genes. Expression levels of platelet genes in primary megakaryocytes, including Txas, Slc6a4, Gp6, Selp, Slamf1, and F5, were examined by quantitative RT-PCR. Results of two representative sets of experiments are shown according to the abundance of transgene-derived p45 mRNA. Upper and lower panels are the results of high and low expressors of the transgene, respectively. Average values are shown, and error bars indicate standard deviations. The average values for wild-type megakaryocytes are set to 1. Results from lines 205 and 207 are shown as representative data for high expressors of the transgene (upper panels) or low expressors of transgene (lower panels), respectively. The Student t test was used to calculate statistical significance (P). *, P < 0.05. ns, not significant.

It has been reported that more than 90% of p45-null mice die due to extensive hemorrhage during delivery (36). To examine whether the fatal perinatal hemorrhage of p45-null mice was rescued in p45^−/−:Tg mice, we genotyped an entire litter by 2 weeks after birth (Fig. 2C). Showing excellent agreement with the previous report, the number of p45-null mice born alive was significantly less than that of wild-type mice. In contrast, the number of p45^−/−:Tg mice born alive was similar to that expected from Mendelian inheritance for both Tg lines (Fig. 2C).

We found that neonates of p45^−/−:Tg mice were indistinguishable among their littermates, while p45^−/− pups were easily recognized due to subcutaneous hemorrhage that occurred during delivery (data not shown). When the peripheral blood count was examined, the severe thrombocytopenia of p45-null mice was rescued in the compound mutant mice (Table 2). Platelets were observed consistently in the blood smears from wild-type and p45^−/−:Tg mice but not in those from p45-null mice (Fig. 2D).

TABLE 2.

Blood count measurement for compound mutant mice in p45-null background^a

Genotype (nb)	Blood countc
	RBC (104/μl)	Hb (g/dl)	Htc (%)	Plt (104/μl)
WT (7)	971 ± 51	14.6 ± 0.4	46.6 ± 2.2	92.1 ± 11.7
p45−/− (3)	875 ± 59	12.7 ± 1.1	40.6 ± 3.1	10.1 ± 2.6
p45−/−:Tg (5)	959 ± 98	13.8 ± 1.5	43.5 ± 4.1	70.4 ± 12.0
p45−/−:ΔNTD Tg (3)	994 ± 26	13.7 ± 0.6	41.8 ± 1.1	32.2 ± 19.3

^aWild-type, p45-null, p45^−/−:Tg and p45^−/−:ΔNTD Tg mice were examined at 2 months of age. Results from lines 207 and 340 are shown as representatives of the p45^−/−:Tg and p45^−/−:ΔNTD Tg genotypes, respectively. Values are expressed as means ± standard deviations.

^bn, no. of mice.

^cRBC, red blood cells; Hb, hemoglobin; Htc, hematocrit; Plt, platelets.

We further examined histological alterations of the spleen. Surviving p45-null mice displayed significant splenomegaly by 2 months of age, whereas spleens of p45^−/−:Tg mice appeared normal and were macroscopically indistinguishable from those of wild-type mice (Fig. 2E). Microscopic examination revealed remarkable improvement of histological abnormalities in the p45^−/−:Tg spleen compared with results for the p45-null spleen. In p45-null mice, enlarged megakaryocytes were accumulated and dispersed all over the spleen, and the red pulp/white pulp organization of the spleen was completely disrupted (Fig. 2F, middle panel). In addition, severe fibrosis was apparent in the p45^−/− spleen (Fig. 2G, middle panel). In contrast, neither abnormal megakaryocytic accumulation nor fibrotic change was observed in the rescued p45^−/−:Tg mouse spleen, and spleen organization appeared to be similar to that of wild-type mice (Fig. 2F and G, left and right panels).

To assess the platelet production ability of the megakaryocytes, a proplatelet formation (PPF) assay was performed using megakaryocytes isolated from bone marrow. While p45-null megakaryocytes completely lacked PPF ability, megakaryocytes from p45^−/−:Tg mice generated proplatelets. The frequency of PPF in p45^−/−:Tg mice was nearly equal to that for wild-type mice (Table 3).

TABLE 3.

PPF frequencies of megakaryocytes purified from the bone marrow of compound mutant mice

Genotype	PPF (%)a
WT	34.0 ± 5.3
p45−/−	0.0 ± 0.0
p45−/−:Tg	31.3 ± 7.6
p45−/−:ΔNTD Tg	0.0 ± 0.0

^aThe ratios of megakaryocytes forming proplatelets to total megakaryocytes are indicated. Results from lines 205 and 340 are shown as representative data for p45^−/−:Tg and p45^−/−:ΔNTD Tg mice, respectively. Values are expressed as means ± standard deviations.

G1HRD-driven p45 activates platelet genes in primary megakaryocytes.

We examined gene expression in megakaryocytes cultured from p45^−/−:Tg fetal livers. We recently identified several platelet genes that are repressed in p45-null megakaryocytes (26). These genes include Thromboxane A synthetase 1 (Txas), Solute carrier family 6 (neurotransmitter transporter, serotonin) member 4 (Slc6a4), Glycoprotein 6 (Gp6), Selectin P (Selp), Signaling lymphocytic activation molecule family member 1 (Slamf1), and Factor 5 (F5), all of which are assumed to be downstream effectors of p45. When primary megakaryocytes were cultured and purified from fetal livers and their cDNAs were examined by quantitative RT-PCR, the expression of the platelet genes was consistently higher in p45^−/−:Tg megakaryocytes than in p45-null cells (Fig. 3). In contrast, the relative expression of the platelet genes in p45^−/−:Tg megakaryocytes compared to those in the wild-type cells varied from embryo to embryo, probably due to position effect variegation of the transgene. The variation amplitude in the amount of p45 mRNA was similar to that of the p45 protein (see Fig. 2B; also data not shown). Platelet gene expression was lower in the p45^−/−:Tg megakaryocytes expressing smaller amounts of p45 (Fig. 3, lower panels), while the genes were fully activated in those expressing higher levels of p45 (Fig. 3, upper panels). Thus, G1HRD appears to confer spatiotemporal specificity of p45 gene expression, supporting the notion that GATA1 is one of the important upstream regulators of p45 in megakaryocytes.

It should also be noted that the mild anemia observed in p45-null mice was not apparent in p45^−/−:Tg mice (Table 2), and the poikilocytosis observed in p45-null mice was not present in the compound mutant mice (Fig. 2D), suggesting that transgene-derived p45 also compensates for the endogenous function of p45 in erythroid cells. Taken together, we conclude that the defective megakaryopoiesis, platelet production, and erythropoiesis observed in the p45-null mice were fully rescued in the p45^−/−:Tg compound mice. These results indicate that G1HRD-driven p45 substantially compensates for the endogenous function of p45.

The N-terminal region of p45 is necessary for reporter gene activation.

To corroborate that the transactivation ability of p45 is actually required for the rescue of defective megakaryopoiesis and platelet production of p45-null mice by G1HRD-driven p45, we tested a transactivation-defective p45 protein in the same set of transgenic rescue experiments. We first made a series of p45 deletion mutants and transfected them into 293T cells (Fig. 4A). The expression levels of full-length p45 and of each mutant were comparable (Fig. 4B). The expression vectors were introduced into 293T cells together with a MafK expression vector, and the luciferase reporter gene was connected to three copies of the Maf recognition element (MARE), a high-affinity binding sequence of the p45-MafK heterodimer (13). Whereas full-length p45 increased luciferase activity, deletion of 38 amino acids from the N terminus greatly decreased reporter activity. Further deletion of the N-terminal region substantially deprived p45 of its transactivation ability (Fig. 4C). These results are consistent with previous reports stating that the N-terminal half of p45 possesses transactivation ability (3, 29).

FIG. 4.

The N-terminal region of p45 is important for transactivation. (A) A series of N-terminal truncations of p45 is shown. PPXY and bZIP motifs are indicated. (B) Transient overexpression of p45 deletion mutants in 293T cells. Whole-cell extracts were examined by immunoblot analysis using anti-p45 antibody. Lamin B is used as a control. (C) A reporter assay with a luciferase reporter gene in 293T cells. Expression vectors for the N-terminally truncated forms of p45 shown in panel A are introduced into 293T cells with the pRBGP2 reporter plasmid. The average values are shown with standard deviations. A representative result from three independent experiments is shown. The vertical axis indicates relative luciferase activity. Firefly luciferase activity in the absence of effector plasmids is set to 1. The Student t test was used to calculate statistical significance (P). *, P < 0.05.

The N-terminal 38 amino acids are required for p45 function in thrombogenesis.

Because the reporter assay indicated that deletion of 38 amino acids from the N terminus mostly abolished the transactivation ability of p45, we chose the truncated mutant ΔΝ38 p45 to confirm that the transgenic rescue experiment actually depends on p45 activity. For simplicity, the N-terminal 38 amino acids and ΔΝ38 p45 are designated the N-terminal domain (NTD) and ΔΝTD, respectively.

We generated transgenic mice carrying G1HRD-ΔΝ38 p45 (Fig. 5A) and obtained two independent mouse lines. G1HRD-ΔΝ38 p45 transgenic mice (ΔΝTD Tg mice) were viable and fertile and did not show apparent hematological abnormalities (data not shown). These mice were crossed into a p45-null background, and p45^−/−:ΔΝTD Tg mice were generated. Transgene expression was examined using the compound mutant mice generated from each line by immunoblot analysis of whole-cell extracts prepared from bone marrow megakaryocytes (Fig. 5B). The transgene-derived ΔΝTD expression was observed in both lines 311 and 340 and varied from mouse to mouse to some extent even though the mice were obtained from the same line. Similar to the case with the compound mutant mice expressing p45, the fluctuation of the transgene-derived ΔΝTD expression did not affect the hematological and histological phenotypes of the compound mutant mice. Gene expression was analyzed separately depending on the abundance of the transgene-derived ΔΝTD (see below) (Fig. 6). Because the anti-p45 antibody used in the immunoblot analysis recognizes the C-terminal region of p45, the immunoreactivity with the antibody should be the same between full-length p45 and ΔΝTD, which made it possible to compare the molar abundances of p45 and ΔΝTD. The variations in p45 protein levels in p45^−/−:Tg mice and ΔΝTD in p45^−/−:ΔΝTD Tg mice were within a similar range, judging from the band intensities (see Fig. 2B and 5B).

FIG. 5.

The N-terminal 38 amino acids are necessary for p45 function in megakaryopoiesis. (A) The structure of the G1HRD-ΔNTD transgene. A cDNA fragment encoding ΔN38 p45 (ΔNTD) was fused to the 5′ half of the second exon of the Gata1 gene. (B) Comparison of expression levels of transgene-derived ΔNTD and endogenous p45. Whole-cell extracts were prepared from bone marrow megakaryocytes and subjected to immunoblot analysis with anti-p45 antibody. Lamin B was used as a loading control. Arrows indicate a doublet of endogenous p45, and an arrowhead indicates the transgene-derived ΔNTD. (C) Number of pups of each genotype born from mating between p45^+/− mice and p45^+/−:ΔNTD Tg mice. Genotypes were determined by 2 weeks after birth. Two independent transgenic mouse lines (311 and 340) were examined. Expected numbers calculated from Mendelian ratios are shown in parentheses. Comparison of two proportions was performed to calculate statistical significance (P) between birth frequency of p45^−/− pups and that of p45^−/−:ΔNTD Tg mice. *, P < 0.05. (D) Wright-Giemsa staining of peripheral blood smears drawn from p45^−/−:Tg mice (left) and p45^−/−:ΔNTD Tg mice (right). White and gray arrowheads indicate platelets and poikilocytes, respectively. Results from lines 205 and 311 are shown as representative data for p45^−/−:Tg and p45^−/−:ΔNTD Tg mice, respectively. (E) Macroscopic observation of spleens: (i) wild-type, (ii) p45-null, (iii) p45^−/−:Tg, and (iv) p45^−/−:ΔNTD Tg mice at 6 months of age. The bar corresponds to 1 cm. Results from lines 207 and 340 are shown as representative data for p45^−/−:Tg and p45^−/−:ΔNTD Tg mice, respectively. (F) Hematoxylin and eosin (HE) staining (left two panels) and silver impregnation (right two panels) of spleens from p45^−/−:ΔNTD Tg mice at 2 and 6 months of age. The scale bar corresponds to 100 μm and 50 μm in HE and silver impregnation, respectively. Results from line 340 are shown as representative data for p45^−/−:ΔNTD Tg mice.

FIG. 6.

Essential contribution of the N-terminal 38 amino acids to platelet gene activation in megakaryocytes. Expression levels of platelet genes in primary megakaryocytes, including Txas, Slc6a4, Gp6, Selp, Slamf1, and F5, were examined by quantitative RT-PCR. Results of two representative sets of experiments are shown according to the abundance of transgene-derived ΔNTD mRNA. Upper and lower panels are the results from high and low expressors of the transgene, respectively. Average values are shown, and error bars indicate standard deviations. The average values of wild-type megakaryocytes are set to 1. Results from line 340 are shown as representative data. The Student t test was used to calculate statistical significance (P). *, P < 0.05. ns, not significant.

To examine whether lethal perinatal hemorrhage of p45-null mice was observed in p45^−/−:ΔΝTD Tg mice, we genotyped an entire litter at 2 weeks after birth. In contrast to the number of p45-null mice born alive, the number of p45^−/−:ΔΝTD Tg mice born alive was similar to that expected from Mendelian inheritance for both lines (Fig. 5C). The NTD was dispensable as far as the avoidance of perinatal lethality was concerned.

However, we found that the platelet counts of p45^−/−:ΔΝTD Tg mice were much lower than those of wild-type mice, albeit higher than those of p45-null mice (Table 2). In the blood smears, while platelets were observed in p45^−/−:Tg mice, very few platelets were observed in p45^−/−:ΔΝTD Tg mice (Fig. 5D). Poikilocytosis was also observed in p45^−/−:ΔΝTD Tg mice. Thus, ΔΝTD only partially rescued the hematological defects in p45-null mice.

We then examined histological alterations of the spleen. p45^−/−:ΔΝTD Tg mice did not display apparent abnormalities of the spleen at 2 months of age, by which time p45-null mice had already developed significant splenomegaly and fibrosis. However, by 6 months of age, mild splenomegaly became recognizable in the rescued p45^−/−:ΔΝTD Tg mice (Fig. 5E) and was accompanied by mild disorganization of splenic architecture, accumulation of megakaryocytes, and severe fibrosis (Fig. 5F and data not shown). The spleen phenotypes were mitigated but not completely corrected in the p45^−/−:ΔΝTD Tg mice.

Importantly, PPF was severely affected by deletion of the NTD (Table 3). Whereas the PPF frequency of p45^−/−:Tg mice was similar to that of wild-type mice, practically no PPF was observed in megakaryocytes isolated from the bone marrow of p45^−/−:ΔΝTD Tg mice. These results authenticate that rescue of p45-null mice by the G1HRD-p45 transgene is attributable to the activity of transgene-derived p45, for which the NTD is critical.

The NTD is essential for expression of selected genes regulated downstream of p45.

We next examined the expression levels of platelet genes in the p45^−/−:ΔΝTD Tg mice. Expression of Txas, Slc6a4, and Gp6 was examined in primary megakaryocytes cultured from fetal livers (Fig. 6). The expression levels of Txas, Slc6a4, and Gp6 in p45^−/−:ΔΝTD Tg megakaryocytes were closer to those of p45-null megakaryocytes, regardless of the mRNA abundance of transgene-derived ΔΝTD. When the high transgene expressors were compared, the expression levels of Txas, Slc6a4, and Gp6 in p45^−/−:ΔΝTD Tg megakaryocytes were consistently less than 15% of those in wild-type cells (Fig. 6, top panel), while their expression levels in p45^−/−:Tg megakaryocytes were comparable to those in wild-type cells, except for Txas (Fig. 3, top panel). When the low transgene expressors were compared, platelet gene levels relative to the p45 level were lower in p45^−/−:ΔΝTD Tg cells (Fig. 6, bottom panel) but comparable in p45^−/−:Tg cells (Fig. 3, bottom panel). Thus, the NTD is required for expression of the Txas, Slc6a4, and Gp6 genes.

Interestingly, we observed a distinct effect of NTD deletion on expression of the Selp, Slamf1, and F5 genes. ΔΝTD, when expressed at a higher level, supported the expression of these genes in the absence of endogenous p45 (Fig. 6, top panel). Thus, the requirement of the NTD was not uniform, and the NTD seemed to be dispensable for expression of some platelet genes.

It should be noted that the NTD does not harbor any canonical functional motifs. Compared with the well-characterized PPXY motifs (15, 22), the functional significance of this region residing more N-terminal to the PPXY motifs is not yet fully understood, especially in megakaryocytes. These results imply a novel mechanism of transcriptional activation by p45 within the GATA1-p45 regulatory axis in vivo.

G1HRD-driven p45 partially corrects defective gene expression and maturation arrest in GATA1-deficient megakaryocytes.

Finally, we assessed how p45 contributes to the GATA1-directed transcriptional hierarchy. Because GATA1 is an upstream regulator of p45, GATA1 function during megakaryopoiesis should at least in part be executed via p45 function. To address this issue, we examined whether G1HRD-driven p45 was able to rescue a GATA1 deficiency in megakaryocytes. It has been reported that Gata1^ΔneoΔHS mice display marked thrombocytopenia as a result of proliferation of immature megakaryocytes and impairment of platelet production, accompanied by reduced expression of p45 (37, 49). Therefore, we crossed G1HRD-p45 male mice with heterozygous female (Gata1^ΔneoΔHS/X) mice, expecting that supplementation of p45 would assist Gata1^ΔneoΔHS/Y mice in generating platelets. Because the Gata1 gene is located on the X chromosome, we analyzed male mice (Gata1^X/Y, Gata1^X/Y:Tg, Gata1^ΔneoΔHS/Y, and Gata1^ΔneoΔHS/Y:Tg) at E13.5 and adult (2 months old) stages.

GATA1 expression levels in Gata1^ΔneoΔHS/Y and Gata1^ΔneoΔHS/Y:Tg megakaryocytes were reduced to 1.2% and 1.1% of those in Gata1^X/Y megakaryocytes, respectively, showing good agreement with previous reports (Fig. 7A) (37, 49). As shown in Fig. 1E, the expression level of p45 in Gata1^ΔneoΔHS/Y megakaryocytes was decreased to approximately half of that in Gata1^X/Y megakaryocytes. Expression of all platelet genes examined was generally reduced in Gata1^ΔneoΔHS/Y megakaryocytes (Fig. 7A).

FIG. 7.

G1HRD-driven p45 does not fully rescue the defective megakaryopoiesis caused by GATA1 deficiency. (A) Gata1, p45, and platelet gene (Txas, Slc6a4, Gp6, Selp, Slamf1, and F5) expression levels of primary megakaryocytes cultured from fetal livers were examined by quantitative RT-PCR. Gata1^X/Y, n = 9; Gata1^X/Y:Tg, n = 6; Gata1^ΔneoΔHS/Y, n = 11; Gata1^ΔneoΔHS/Y:Tg, n = 4. Average values are expressed, and error bars indicate standard deviations. The Student t test was used to calculate statistical significance (P). *, P < 0.05. ns, not significant. (B) Primary megakaryocytes in cultures from Gata1^X/Y (n = 4), Gata1^X/Y:Tg (n = 5), Gata1^ΔneoΔHS/Y (n = 7), and Gata1^ΔneoΔHS/Y:Tg (n = 5) fetal livers were analyzed by flow cytometry for expression of CD41 and CD61. Results from line 205 are shown as representative data. The CD41⁺ CD61⁺ cells were divided between regions I and II, with the former and latter representing cell populations with intermediate and high expression levels of the two surface markers, respectively. The ratios of cell numbers in region II to those in region I are shown in panel C. (C) Average ratios of cell numbers in region II to those in region I from Gata1^X/Y (n = 5), Gata1^X/Y:Tg (n = 5), Gata1^ΔneoΔHS/Y (n = 4), and Gata1^ΔneoΔHS/Y:Tg (n = 7) mice. The Student t test was used to calculate statistical significance (P). *, P < 0.05.

p45 expression in Gata1^ΔneoΔHS/Y:Tg megakaryocytes was comparable to that in Gata1^X/Y megakaryocytes (Fig. 7A). This result indicates that G1HRD is active in Gata1^ΔneoΔHS/Y megakaryocytes. Although expression levels of platelet genes tended to be higher in Gata1^ΔneoΔHS/Y:Tg megakaryocytes than those in Gata1^ΔneoΔHS/Y megakaryocytes, the effects were modest, especially for Slamf1 and F5. In the absence of GATA1, the contribution of p45 to the activation of platelet genes appears to be limited.

To examine whether G1HRD-driven p45 restores the abnormal increase in Gata1^ΔneoΔHS/Y megakaryocytes at an immature stage (37), we performed flow cytometry analysis and characterized megakaryocytes cultured from fetal livers. In Gata1^ΔneoΔHS/Y cultures, CD41^low CD61^low megakaryocytes (Fig. 7B, fraction I) were relatively abundant compared with CD41^high CD61^high megakaryocytes (Fig. 7B, fraction II). The ratio of these two fractions (II/I) was lower in Gata1^ΔneoΔHS/Y cultures and slightly higher in Gata1^ΔneoΔHS/Y:Tg cultures, but the latter ratio was not as high as that in Gata1^X/Y cultures (Fig. 7C). These results indicate that transgene-derived p45 partially restored the number of mature megakaryocytes in the Gata1^ΔneoΔHS/Y background. At the adult stage, platelet production remained low and severe thrombocytopenia was observed in Gata1^ΔneoΔHS/Y:Tg mice (Table 4). These results suggest that p45 expression is not sufficient in fully mature GATA1-deficient megakaryocytes for normal platelet production and that GATA1 needs additional downstream effectors other than p45. Taken together, GATA1 is one of the important upstream regulators of p45, and p45 promotes megakaryopoiesis in cooperation with other factors under the control of GATA1 or GATA1 itself.

TABLE 4.

Blood count measurement for compound mutant mice in the Gata1^ΔneoΔHS/Y background^a

Genotype (nb)	Blood countc
	RBC (104/μl)	Hb (g/dl)	Htc (%)	Plt (104/μl)
Gata1X/Y (9)	924 ± 52	15.1 ± 0.6	45.4 ± 2.0	87.7 ± 12.7
Gata1X/Y:Tg (9)	894 ± 83	13.2 ± 0.6	40.1 ± 1.8	88.5 ± 6.3
Gata1ΔneoΔHS/Y (3)	740 ± 32	12.9 ± 0.3	40.0 ± 1.4	12.4 ± 3.0
Gata1ΔneoΔHS/Y:Tg (3)	750 ± 85	12.2 ± 1.0	37.2 ± 2.1	24.9 ± 21.6

^aGata1^X/Y, Gata1^X/Y:Tg, Gata1^ΔneoΔHS/Y, and Gata1^ΔneoΔHS/Y:Tg mice were examined at 2 months of age. Results from line 207 are shown as representative. Values are expressed as means ± standard deviations.

^bn, no. of mice.

^cFor abbreviations, see footnote c of Table 2.

DISCUSSION

In megakaryocytes, GATA1 deficiency reduces p45 expression by approximately 50%, indicating the presence of GATA1-dependent and -independent regulation of the p45 gene. In this study, we have revealed that the GATA1-p45 regulatory axis is one of the hierarchies of transcription factors regulating megakaryopoiesis in vivo based on two contrasting approaches. The first approach revealed that G1HRD-driven p45 substantially rescued the defective megakaryopoiesis in p45-null mice, demonstrating that G1HRD retains the ability to direct spatiotemporally accurate expression of p45 and implying that GATA1-dependent regulation of the p45 gene substantially contributes to the function of p45 in megakaryopoiesis. In contrast, despite being expressed under the regulatory influences of G1HRD, a transactivation-defective mutant, ΔΝTD, did not efficiently rescue the defects of p45-null megakaryocytes, indicating that G1HRD-p45 rescue requires intact p45 activity. The second approach showed that G1HRD-p45 partially corrected the defects in Gata1-deficient megakaryocytes, indicating that p45 is one of the effectors regulated by GATA1 but that GATA1 requires additional effectors to direct the full differentiation process of megakaryocytes. These features of the GATA1-p45 regulatory axis are summarized in Fig. 8.

FIG. 8.

Schematic model of a transcriptional regulation hierarchy surrounding p45. p45 is regulated by GATA1-dependent and -independent mechanisms. GATA1 activates or represses several key transcription factors, including p45. Among these hierarchical relations, the GATA1-p45 regulatory axis is demonstrated in this study. Upon transgenic expression, p45 directly or indirectly activates downstream targets in megakaryocytes, including Txas, Slc6a4, Gp6, and Selp, even in the absence of GATA1.

GATA1 is a prototypical lineage-restricted transcription factor that controls erythroid and megakaryocytic cell differentiation. The transcriptional regulatory cascade governed by GATA1 plays an essential role in promoting differentiation of these lineages. In addition to p45, genes encoding a number of transcription factors, including EKLF, GATA2, c-Myb, c-Myc, and PU.1, have been identified as downstream target genes of GATA1 (2, 6, 11, 18, 50). Of these transcription factors, GATA1 activates p45 and EKLF gene expression while it represses expression of the others. Supplementation of transgene-derived p45 did not fully restore the gene expression and maturation arrest with proliferative tendency of GATA1-deficient megakaryocytes, indicating that regulatory pathways other than p45 are also indispensable for GATA1 to regulate the megakaryopoiesis program.

Our analyses have revealed critical contributions of the GATA1-p45 regulatory axis to the suppression of fibrosis in the spleen (see Fig. 2G and 5F) and bone marrow (data not shown). We found extensive fibrosis in spleens and bone marrow from p45-null mice, but these mice were rescued from fibrosis by G1HRD-p45, suggesting that fibrosis is a consequence of p45 deficiency in GATA1-expressing hematopoietic cells, especially megakaryocytes. This is consistent with the observation that mice suffering from pathological accumulation of megakaryocytes in bone marrow and the spleen, including TPO-overexpressing mice (52), Gata1^ΔneoΔHS (or Gata1^low) mice (48), Bach1-overexpressing mice (46) and small Maf mutant mice (H. Motohashi, unpublished data), often develop myelofibrosis and spleen fibrosis. In this regard, it has been reported that an osteosclerotic phenotype also is present in the p45-null mice, and spleen cells from p45-null mice are able to transfer osteosclerosis as well as megakaryocytosis adoptively to wild-type mice (14). These results support the contention that megakaryocytes in which the GATA1-p45 regulatory axis is operative are integral to bone and connective tissue homeostasis.

It is interesting to note that p45 was able to increase expression of some downstream effector genes in the absence of GATA1. Because Txas is a well-established direct target of p45 (7) and Selp and Slc6a4 are newly identified target genes of p45 (R. Fujita, H. Motohashi, and M. Yamamoto, unpublished observation), the increased expression of these genes in Gata1^ΔneoΔHS/Y:Tg megakaryocytes seems to be interpreted as p45 directly activating transcription of these genes in the absence of GATA1. However, in the case of Slc6a4, we surmise that additional factors regulated by GATA1 or GATA1 itself might cooperate with p45, because expression of Slc6a4 was lower in the Gata1^ΔneoΔHS/Y background despite similar p45 expression levels (Fig. 3 and Fig. 7A).

Another interpretation of the restored expression of platelet genes in Gata1^ΔneoΔHS/Y:Tg megakaryocytes is an indirect effect through altered megakaryocyte differentiation. Considering that expression levels of platelet genes are elevated in the course of megakaryocyte maturation (26) and that the megakaryocyte fraction obtained from Gata1^ΔneoΔHS/Y:Tg mice contained a higher frequency of mature megakaryocytes than that from Gata1^ΔneoΔHS/Y mice (Fig. 7B and C), the increased expression of the platelet genes in the former might reflect a difference in maturation stages between the two populations.

Three lines of evidence suggest that the p45 gene may be finely tuned by transcription factors other than GATA1. First, substantial amounts of the 1a transcript were still detected in GATA1-deficient megakaryocytes (Fig. 1E). While the 1b promoter was heavily dependent on GATA1 and did not seem to be activated by other GATA factors, like GATA2, the 1a promoter was partially dependent on GATA1 and/or might be activated by other GATA factors. We surmise that the 1a promoter may be primarily utilized at progenitor stages, before GATA1 is expressed. Second, GATA1 and GATA2 are reported to play redundant roles in cooperation with FOG1 (4), and knockout mouse analysis revealed an indispensable role for FOG1 in early megakaryopoiesis (47). Because p45 contributes to the commitment and proliferation of megakaryocytic cells (9, 17) and GATA2 is expressed in immature megakaryocytes, GATA2 may also be involved in the regulation of p45 gene expression at early stages of megakaryopoiesis. Indeed, other GATA1 target genes, such as EKLF and PU.1, are also under the control of GATA2 (6, 18). Third, deletion of SCL reduced p45 expression, and conversely, SCL augmented transcriptional activation of the p45 1b promoter by GATA1 (19). Thus, p45 may also be under the influence of SCL.

In summary, our results demonstrate that G1HRD-p45 rescues the defective megakaryopoiesis of p45-null mice, indicating that GATA1-dependent regulation of the p45 gene substantially contributes to p45 function in megakaryopoiesis. The results also show that p45 supports a limited portion of GATA1 function in megakaryopoiesis, indicating that GATA1 needs to regulate other downstream effectors in addition to p45.