Constitutive Phosphorylation of GATA-1 at Serine26 Attenuates the Colony-Forming Activity of Erythrocyte-Committed Progenitors

We previously reported that IL-3 signaling induces phosphorylation of GATA-1 at the serine26 position, which contributes to IL-3-mediated anti-apoptotic response. Here, we demonstrate that phosphorylation of GATA-1 at serine26 is also transiently induced in cells of the erythroid lineage (primary erythroblasts and erythrocyte-committed progenitors [EPs]) by erythropoietin (EPO), the principal cytokine regulating erythropoiesis. To examine whether phosphorylation of GATA-1 at serine26 would have any impact on erythropoiesis, mutant mice carrying either a glutamic acid (GATA-1S26E) or alanine (GATA-1S26A) substitution at serine26 were generated. Neither GATA-1S26E nor GATA-1S26A mice showed any significant difference from control mice in peripheral blood cell composition under either steady state or stress conditions. The erythroblast differentiation in both mutant mice also appeared to be normal. However, a moderate reduction in the CFU-E progenitor population was consistently observed in the bone marrow of GATA-1S26E, but not GATA-1S26A mice, suggesting that such defect was compensated for within the bone marrow. Surprisingly, reduced CFU-E progenitor population in GATA-1S26E mice was mainly due to EPO-induced growth suppression of GATA-1S26E EPs, albeit in the absence of EPO these cells manifested a survival advantage. Further analyses revealed that EPO-induced growth suppression of GATA-1S26E EPs was largely due to the proliferation block resulted from GATA-1S26E-mediated transcriptional activation of the gene encoding the cell cycle inhibitor p21Waf1/Cip1. Taken together, these results suggest that EPO-induced transient phosphorylation of GATA-1 at serine26 is dispensable for erythropoiesis. However, failure to dephosphorylate this residue following its transient phosphorylation significantly attenuates the colony-forming activity of EPs.


Introduction
Red blood cells differentiate from multi-potential hematopoietic stem cells in the bone marrow (BM) [1]. The committed progenitors are slowly proliferating erythroid burst-forming units (BFU-Es) [2]. These BFU-E cells divide and differentiate through the ''mature'' BFU-E stage into rapidly dividing erythroid colonyforming units (CFU-Es) [2], which differentiate further into erythroblasts, including the proerythroblast, basophilic, polychromatic, and orthochromatic erythroblast stages [3]. The nucleus shrinks and is shed as the erythroblast cells become reticulocytes before differentiating into erythrocytes [3]. As one of the most highly characterized differentiation model systems, erythrocyte differentiation has been shown to be regulated by many transcription factors, including the GATA family proteins [4,5]. GATA-1 is the first identified member of the GATA family transcription factors [5,6] that contains a transactivation domain in the N-terminal region and two zinc-finger domains for dimerization and DNA binding in the C-terminal region [7,8,9]. It is highly expressed in all erythroid cells [10,11] and the functional GATA-binding DNA motif is present in the regulatory regions of virtually all erythroid-specific genes, including the globin gene family and genes coding for heme metabolism enzymes, erythroid transcription factors, and red cell membrane proteins [12,13].
Phosphorylation has been suggested to affect GATA-1 function. MAPK-dependent phosphorylation has been shown to be involved in the regulation of GATA-1 protein stability [14]. It has also been reported that Akt-dependent phosphorylation of GATA-1 at serine 310 is necessary for EPO-induced erythrocyte terminal differentiation in a proerythroblast cell line [15] and for EPOinduced TIMP1 secretion and maturation of fetal liver erythroid cells [16]. Surprisingly, in a knock-in animal model, it was shown that mutation of GATA-1 serine 310 alone or triple mutations at serines 72, 142, and 310 did not significantly influence hematopoiesis in vivo [17].
In our previous study, we reported that MEK/MAPKdependent phosphorylation of GATA-1 at serine 26 was crucial for GATA-1 to mediate IL-3-dependent cell survival [18]. In an IL-3-dependent pro-B cell line, ectopic expression of mutant GATA-1 containing a glutamic acid substitution at residue 26, which mimics the negative charge of phospho-GATA-1, sup-pressed IL-3 deprivation-induced apoptosis and elevated the expression of endogenous anti-apoptotic protein Bcl-X L [18]. In this study, we explored the possibility that phosphorylation of GATA-1 at serine 26 might have some impacts on erythropoiesis, in which GATA-1 expression is essential [19]. We demonstrated that, in both primary erythroblasts and erythrocyte-committed progenitors (EPs), phosphorylation of GATA-1 at serine 26 was transiently induced by EPO. By analyzing mutant mice harboring either a glutamic acid (GATA-1 S26E ) or alanine (GATA-1 S26A ) substitution at serine 26 , we demonstrate that phosphorylation of GATA-1 at serine 26 is dispensable for erythropoiesis. However, failure to dephosphorylate this residue following its transient phosphorylation significantly attenuates the colony-forming activity of EPs. The underlying mechanism for the latter defect was also explored in this study.

Materials and Methods
Ethics Statement and Generation of GATA-1 S26E and GATA-1 S26A mice To generate GATA-1 gene serine 26 codon mutation alleles, two genomic fragments harboring the GATA-1 locus were isolated from a plasmid pSC3Z-mGATA-1 containing the 129/Svj mouse genomic DNA of interest (kindly provided by Dr. Sjaak Philipsen, Erasmus MC, Rotterdam, The Netherlands), and used to construct the targeting vectors. The 59 fragments were engineered to contain either a glutamic acid (S26E) or alanine (S26A) codon at residue 26. These targeting vectors were constructed by PCRassisted cloning so that a floxed cassette containing the Neo selection marker (neo r ) was introduced into intron 2. Lying outside of the selection cassette were two homology arms (2.5 kb and 4 kb) and the gene encoding the thymidine kinase (TK) for negative selection. These targeting vectors were then electroporated into R1 embryonic stem (ES) cells, and Southern blotting using 59 and 39 probes external of the targeting construct was carried out to select clones that had undergone homologous recombination at the GATA-1 locus. Mutant ES cells were injected into C57BL/6 mouse blastocysts, and the resultant male chimeras were crossed with C57BL/6 females. To delete the neo r cassette, GATA-1 S26E(neo+) and GATA-1 S26A(neo+) heterozygous females were further crossed with EIIa-Cre transgenic mice (originally on the FVB/N background, but backcrossed to C57BL/6 for more than seven generations; a gift from Dr. Ying-Hue Lee, IMB, Academia Sinica, Taipei, Taiwan). For some experiments, GATA-1 S26E mice were crossed with mutant mice carrying the null allele of the Bim or the p21 Waf1/Cip1 gene (purchased from The Jackson Laboratory) to generate compound mutant mice. Throughout this study, only male mice were used for all the analyses. Mice were housed under good animal care practice conditions and all experiments were performed with 8-to 9-weeks old males, under protocols approved by the Institutional Animal Care and Use Committee of the Academia Sinica.

Analysis, Expansion and Purification of Primary Erythroblasts
To measure the percentage of erythroblasts at various developmental stages, total BM cells or splenocytes stained with FITC-labeled anti-CD71 (RI7217) and PE-labeled anti-Ter119 (TER-119) antibodies (BioLegend, San Diego, CA) were analyzed by flow cytometry using FACSCanto (BD Biosciences). Propidium iodide (PI) was added to the staining mixture to monitor cell viability. Purified anti-CD16/CD32 (FccRIII/II) antibody (from 2.4G2 hybridoma supernatant) was also included in all analyses to block non-specific binding of anti-CD71 and anti-Ter119 antibodies to the FccR on the cell surface. Primary erythroblasts were expanded and purified essentially as previously described [20]. Briefly, BM cells were harvested and cultured in StemPro-34 medium (Invitrogen) containing 2.5 U/mL of EPO (Amgen), 100 ng/mL of mouse stem cell factor (SCF), 7.5 mg/mL of insulin, 1 mM dexamethasone, 1 mM beta-estradiol, and 75 mg/mL of human transferrin (Sigma). Twenty-four hours after the culture was initiated, cells were supplemented with 0.5 volumes of fresh medium and cultured for another 24 hours before they were resuspended in 80% fresh medium plus 20% residual conditioned medium. Forty-eight hours later, cultured cells were stained with biotin-conjugated anti-CD71 (RI7217) antibody (Biolegend, San Diego, CA), and CD71 + erythroblasts were purified using antibiotin microbeads and a MACS system (Miltenyi Biotec GmbH).

Genomic DNA Sequencing Analysis
The genomic DNA fragment of the exon 2 of the GATA-1 gene was PCR-amplified from mouse tail DNA using primers p1 (59-CAA CCC CAG TGT TCC CAT GGA TT-39) and p2 (59-GGA GTG TCT GTA GGC CTC AGC TT-39). The conditions for the PCR reaction were 95uC for 5 minutes, followed by 30 cycles of 95uC for 1 minute, 56uC for 1 minute, and 72uC for 40 seconds, and a final step of 72uC for 10 minutes. The PCR products of exon 2 were purified and subjected to automated nucleotide sequencing using aforementioned p1 and p2 primers.

Hematological Parameters
Blood was obtained by tail vein bleeding and was collected into dipotassium-EDTA coated tube (BD Biosciences), and then analyzed on an automated blood differential counter (Cell-Dyn 3700, Abbott, IL; ref. 17). To measure the response to acute erythroid stimulation, anemia was induced with phenylhydrazine (PHZ) essentially as previously described [17] with some minor modifications. In brief, mice received an intraperitoneal injection of PHZ (40 mg/kg body weight in 100 ml of saline) on days 1 and 2, and their blood was taken for analysis using Cell-Dyn 3700 on days 0, 3, 6, 9, and 14.  26 to glutamic acid or alanine in exon 2 is indicated by an asterisk (*) and relevant restriction sites are as depicted. The solid triangles represent loxP sites. PCR primers (p1 and p2) used to sequence exon 2 (see below) of the GATA-1 gene are as indicated. (B) Confirmation of point mutations by DNA sequencing. Tail DNA from GATA-1 S26E and GATA-1 S26A hemizygous male mice were amplified with primers p1 and p2 and the PCR products were analyzed by the automatic sequencer. The ''WT'' allele shows the TCA codon for serine; the ''S26E'' allele shows the GAG codon for glutamic acid, and the ''S26A'' allele shows the GCC codon for alanine. (C) GATA-1 S26E and GATA-1 S26A proteins are resistant to EPO-induced phosphorylation. Primary erythroblasts purified from GATA-1 S26E and GATA-1 S26A mutant mice were starved for 16 hours and stimulated with EPO for 10 minutes. Cell lysates were subsequently analyzed using various antibodies as indicated. doi:10.1371/journal.pone.0064269.g002

Colony-forming Assays
To detect CFU-E colonies, 1610 5 BM cells or 5610 3 sorted EPs were plated in each well in semisolid medium (MethoCult M3334; StemCell Technologies) according to the manufacturer's protocol and CFU-E colonies were counted after 2 days in culture. To detect BFU-E colonies, 4610 4 BM cells were plated in each well in semisolid medium (MethoCult M3434) and primitive BFU-E colonies were counted after 7 days in culture.

S Phase Cell Determination and Cell Death Analysis
To detect proliferation activity, purified EPs (1610 5 cells) were cultured for 24 hours in EPO-containing medium before 20 mM 5ethynyl-29-deoxyuridine (EdU) was added to the culture medium for another 3 hours. DNA synthesis was then measured using Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kits (Invitrogen) according to the manufacturer's protocol. Under the same analysis conditions, the fraction of cells that had sub-G1 amount of DNA (detected by PI staining) was taken as percentage of cells that had undergone apoptosis.

Western Blot Analysis
Cells were lysed in lysis buffer (50 mM Tris-HCl at pH 7.4, 0.25% Na-deoxycholate, 150 mM NaCl, 1 mM EDTA, 1% NP- 40) containing 1 mM PMSF, 1 mg/ml of aprotinin, and 1 mg/ml of leupeptin, and 50 mg of lysate was subjected to western blot analysis as described previously [18]. The antibodies used were directed against mouse GATA-1 (N6), pGATA-1(S310), Mcl-1, Bcl-X L , Bcl-2, Bim, p38, Stat5 (Santa Cruz Biotechnology), p-Akt(Ser473), Akt, p-Erk1/2, Erk1/2, phospho-Jak2 (Tyr1007/ 1008), Jak2, phospho-Stat5 (Tyr694) (Cell Signaling Technology), and E4bp4 (Abcam). The polyclonal anti-mouse pGATA-1(S26) antibody was raised in rabbit by Bethyl Laboratories Inc.   After purification, each erythroblast population was analyzed for CD71 and Ter119 expression. Ninety percent or more of the cells were CD71 + with the majority being at the stage I and II (boxed areas, see also Figure 4A). (B) EPO-and JAKdependent growth of primary erythroblasts. Purified erythroblasts (1610 6 cells) were cultured in medium with or without EPO or with EPO plus 25 mM AG490 (AG). Twenty-four hours after culturing, total viable cell numbers were determined. (Three mice were used for each genotype and for each treatment). (C) AG490 inhibits EPO-induced phosphorylation of JAK2 and its downstream target STAT5. Purified erythroblasts were starved in cytokine-free medium containing or not containing AG490 for 16 hours, and then stimulated with or without EPO for 10 min prior to being lysed for Western blotting analysis using various antibodies as indicated. (D) EPO-dependent induction of survival proteins in primary erythroblasts. Purified erythroblasts were starved in cytokine-free medium for 16 hours, and then stimulated with EPO for 6 hours prior to being lysed for Western blotting analysis using various antibodies as indicated. One representative data set was shown from 3 independent experiments.

Reporter Gene Assays
The 24.6 kb p21 Waf1/Cip1 promoter-luciferase reporter plasmid (24.6 kb) containing 24542 to +117 bp of the murine p21 Waf1/ Cip1 promoter region and the same reporter plasmid in which the consensus binding site for Sp1/KLF-like factors at 221 to 213 bp was mutated from GGGCGG to GTTTTG (221 bp mt) [22] were kindly provided by Arthur Skoultchi. Sorted EPs, cultured in medium containing EPO for 16 hours, were transfected with the wild-type (24.6 kb) or mutant (221 bp mt) p21 Waf1/Cip1 promoterluciferase reporter plasmid along with a control vector expressing the Relina luciferase by Microporator (MP-100). Twenty-four hours after transfection, cell lysates were prepared and analyzed for luciferase activities using luciferase assay substrates according to the manufacturer's protocol (Promega).

Statistical Analysis
All data are presented as mean 6 SD, and were analyzed by one-way analysis of variance (ANOVA) (Tukey's post tests).

GATA-1 Serine 26 Phosphorylation is an Early EPO-induced Event
Our previous study revealed that IL-3 induced marked phosphorylation of GATA-1 at serine 26 [18]. Given that GATA-1 plays a critical role in erythropoiesis and that EPO is essential for erythroid production by preventing committed erythroid progenitors from undergoing apoptosis and allowing them to proliferate and differentiate [23,24,25,26], it was of interest to investigate whether phosphorylation of GATA-1 at serine 26 could be induced by EPO signaling in erythroid cells. To address this issue, freshly prepared BM cells were stimulated by various cytokines and the phosphorylation of ERK and GATA-1 were examined. As shown in Fig. 1A, phosphorylation of ERK1/2 could be detected in cells treated for 10 minutes with EPO, thrombopoietin (TPO), IL-3 or GM-CSF (lanes 2,3,5,6), but not with IL-6 (lane 4).Under the same conditions, serine 26 -phosphorylated GATA-1, detected by a specific anti-pGATA-1(Ser26) antibody [18] and appearing as a slow migrating form of GATA-1 (lane 2, indicated by asterisks), was almost exclusively found in EPO-treated cells. Importantly, phosphorylation of GATA-1 at serine 26 was also prominently detected in cultured primary erythroblasts [20] or EPs (Lin 2 IL-3Ra 2 IL-7Ra 2 c-Kit + Sca-1 2 CD71 + CD41 2 ) [21] following EPO stimulation (Fig. 1B, lane 2, and Fig. 1E below). Consistent with our previous findings [18], EPO-induced phosphorylation of GATA-1 at serine 26 was inhibited by the MEK inhibitor PD98059 (Fig. 1B, lane 4), but not by the PI-3K inhibitor LY294002 (lane 3), suggesting that EPO induces phosphorylation of GATA-1 at serine 26 in erythroid cells via a MAPK-dependent signaling pathway.
Serine 310 phosphorylation of GATA-1 in response to EPO stimulation has been previously reported in the HCD57 murine erytholeukemia cell line and mouse fetal liver cells [15]. We thus compared the phosphorylation kinetics of GATA-1 at serine 26 and serine 310 in primary erythroblast cultures in response to EPO treatment. As shown in Fig. 1C, phosphorylation of GATA-1 serine 26 was rather transient, which peaked at 10 minutes and disappeared by 30 minutes after EPO treatment. In contrast, Aktdependent phosphorylation at serine 310 of GATA-1 [15] was not prominent until 1-2 hour of EPO treatment (Fig. 1C), and such phosphorylation event remained detectable even 7 to 20 hours after stimulation (Fig. 1D, lanes 3 and 4). These results indicate that phosphorylation of GATA-1 at serine 26 in primary erythroblasts is an early, but transient, event following EPO stimulation. Notably, EPO-induced transient phosphorylation of GATA-1 at serine 26 in EPs also manifested a similar kinetics as that observed in primary erythroblasts (Fig. 1E).

Generation of GATA-1 S26E and GATA-1 S26A mice
To explore the physiological significance of GATA-1 phosphorylation at serine 26 on erythropoiesis in vivo, two knock-in mice bearing a specific amino acid replacement at serine 26 , one to glutamic acid (GATA-1 S26E ) and the other to alanine (GATA-1 S26A ; Fig. 2A), were generated. The glutamic acid substitution was expected to create a mutant protein that would mimic the phosphorylated form of GATA-1, whereas the alanine substitution would generate a GATA-1 mutant that can not be phosphorylated at serine 26 [18]. The introduction of gene mutation at serine 26 in mice was verified by sequencing of the genomic DNA from knockin mice (Fig. 2B). Both GATA-1 S26E and GATA-1 S26A hemizygous males were born with an expected Mendelian frequency. The adult mice were fertile and morphologically normal, and survived for nearly two years without any obvious defect. BM levels of endogenous GATA-1 S26E and GATA-1 S26A proteins in the mutant mice appeared to be very similar to those in the wildtype mice (Fig. 2C), and, as expected, these proteins were refractory to phosphorylation after 10 minutes incubation with EPO (Fig. 2C). Notably, GATA-1 S26E appeared as a slowly migrating form (Fig. 2C, compare lanes 3&4 to lanes 5&6), which is consistent with our previous results obtained from a transient expression study using the Ba/F3 cell line [18].
Normal Peripheral Blood Compositions in GATA-1 S26E and GATA-1 S26A mice To examine whether overall hematopoiesis was affected in GATA-1 S26E and GATA-1 S26A mice, the peripheral blood composition of control and both mutant mice were analyzed. As shown in Table 1, all parameters examined (e.g., white blood cells, red blood cells, platelets, hemoglobin, hematocrit, mean corpuscular hemoglobin concentration, etc) were comparable between control and either mutant mice, suggesting that phosphorylation or dephosphorylation of GATA-1 at Ser 26 does not significantly affect the steady-state hematopoiesis of the animals. We next examined whether erythropoiesis was affected in the mutant mice under stress conditions. To address this issue, acute hemolytic anemia was induced by injection of PHZ on two consecutive days. As shown in Fig. 3, the time course and magnitude of the changes induced by PHZ treatment in the number of red blood cells, percentage of hematocrit, hemoglobin concentration, and platelet number were all comparable in the three genotypes. These results indicate that GATA-1 phosphorylation or dephosphorylation at serine 26 does not significantly impair the ability of the mice to recover from anemic stresses.
The hematopoietic system is known to have remarkable feedback systems that would compensate for abnormalities in the  Figure S3 for one representative contour plot of the flow cytometric analysis shown in panels B and C). (D) Survival advantage of GATA-1 S26E EPs in medium without EPO. Purified EP cells were cultured in medium without EPO and viable cell numbers at various time points were determined. All results shown in panels A-D are mean 6 SD (n = 3-6 for each genotype). (E) Analysis of mRNA expression of genes related to cell growth and/or survival by RT-PCR. Purified EPs with the indicated genotype were cultured in medium without (E) or with (F) EPO for 6 hours, and mRNA expression levels of the indicated genes were analyzed by semi-quantitative RT-PCR analysis (24 or 26 cycles). H and R are negative controls using water or without RT, respectively, in the analysis buffer. W,E and A denote RNA from control, GATA-1 S26E and GATA-1 S26EA EPs, respectively. P21, p21 Waf1/Cip1 . Both (E) and (F) are one representative data set from 3 independent experiments with very similar results (see Figure S4 for quantification of these results). *, P,0.05; **, P,0.01. doi:10.1371/journal.pone.0064269.g007 BM, and maintain normal peripheral blood parameters [27,28]. We thus explored the possibility that earlier stages of erythropoiesis might be affected in these mutant mice. We first examined whether erythroid differentiation in BM or the spleen was affected by using FACS-based profiling of erythroblast subpopulations using CD71 and Ter119 markers [29]. The four erythroblast subpopulations with increasing maturity (I-IV) were defined as follows: stage I, proerythroblasts (CD71 high /Ter119 med ); stage II, basophilic erythroblasts (CD71 high /Ter119 high ); stage III, late basophilic and polychromatophilic erythroblasts (CD71 med / Ter119 high ); and stage IV, orthochromatic erythroblasts (CD71 low /Ter119 high ) [29]. The results shown in Fig. 4 indicate that the proportion of cells in each subpopulation in either BM ( Fig. 4A and 4B) or the spleen (Fig. 4A and 4C) of GATA-1 S26E or GATA-1 S26A mice manifested no significant differences from that in the respective organ of the control mice.
GATA-1 deficient erythroblasts die at the proerythroblast-like stage [19,30] and our earlier study indicates that phosphorylation of GATA-1 at serine 26 promotes Bcl-X L expression and cell survival [18]. The apparent normal differentiation of erythroblasts in both mutant mice suggests that phosphorylation or dephosphorylation of GATA-1 at serine 26 does not affect the growth or survival of erythroblasts. We thus examined this issue by using cultured primary erythroblasts which were found to be largely at stages I and II of their differentiation process (Fig. 5A). As shown in supplemental Fig. S1, the growth and death responses of these cultured primary erythroblasts in the presence or absence of EPO were very similar between cells from control or either mutant mice. Moreover, like wt cells, both mutant erythroblasts were highly dependent on EPO and the JAK activity for survival, as withdrawal of EPO or treatment with the JAK inhibitor AG490 dramatically reduced their viability (Fig. 5B and 5C). Furthermore, the extent of EPO-induced up-regulation of the antiapoptotic proteins Mcl-1, Bcl-X L and E4bp4 was very similar in cells from all three genotypes (Fig. 5D). Taken together, these results indicate that phosphorylation or dephosphorylation of GATA-1 at serine 26 is dispensable for EPO-dependent growth and survival of cultured primary erythroblasts. The frequency of CFU-E colony in the BM from mice with the indicated genotype was determined as that described in the legend to Figure 6B. (B) Purified EP cells from mice with the indicated genotype were cultured in EPO-containing medium and viable cell number was counted at indicated time points by the trypan blue exclusion assay. The numbers represent mean 6 SD from 3 independent experiments for each genotype. (C) Luciferase reporter assays were performed in wt or GATA-1 S26E EPs in EPO-containing medium with promoter-less control vector (vec), wt (24.6 K) or mutant (221 bp mt) p21 Waf1/Cip1 promoter-reporters as indicated. All results shown in A-C are presented as mean 6 SD. (n = 3-5 for each genotype). *, P,0.05; n.s., P.0.05. doi:10.1371/journal.pone.0064269.g008 Reduced CFU-E in GATA-1 S26E Bone Marrows The erythroblast differentiation appeared to be normal in both GATA-1 S26E and GATA-1 S26A mice. However, this does not exclude the possibility that some defects occur in the BM, but are somehow compensated at later stages of erythropoiesis. We thus examined whether the progenitor population in the BM of these mutant mice was affected. To address this issue, colony forming assay was performed. Fig. 6A shows that BM cells from mice of all three genotypes contained comparable number of BFU-E progenitors and most BFU-E colonies formed from these three different groups were very similar in size and morphology (Fig.  S2A). In contrast, a significant reduction (about 25% compared to wild-type) in the CFU-E progenitor population was found in the BM from GATA-1 S26E , but not from GATA-1 S26A mice (Fig. 6B), albeit most CFU-E colonies formed from the GATA-1 S26E BM cells were also very similar in size and morphology to those formed from the BM cells of wt or GATA-1 S26A mice (Fig. S2A). Next, since the EP population (Lin 2 IL-3Ra 2 IL-7Ra 2 c-Kit + Sca-1 2 CD71 + CD41 2 ) was shown to account for most CFU-E activity in the bone marrow [21], we thus examined whether this population of cells was affected in GATA-1 S26E mice by flow cytometry (Fig. S2B, and Fig. 6C). Surprisingly, the result showed that the frequency of EP population in the BM of GATA-1 S26E mice was actually comparable to that in wild-type or GATA-1 S26A mice (Fig. 6C). Such unexpected result prompted us to compare the CFU-E forming activity of purified EPs from mice of all three genotypes. Fig. 6D shows that, compared to controls, a significantly lower number of CFU-Es was generated from GATA-1 S26E but not from GATA-1 S26A EPs, suggesting that the reduced CFU-E forming activity observed in GATA-1 S26E bone marrow is due to an intrinsic defect in the EP population.
We next employed a regular suspension cell culture system to further dissect the colony forming defect of GATA-1 S26E EPs. As shown in Fig. 7A, total cell number of wild-type or GATA-1 S26A EP was increased for nearly 4 fold after the culture with EPO was initiated for 24 hours. However, under the same experimental conditions, substantially fewer viable cells were observed in culture using purified GATA-1 S26E EPs compared to those using wild-type or GATA-1 S26A EPs. Further addition of SCF to such culture system still failed to rescue the growth defect of GATA-1 S26E EPs (Fig. 7A), excluding the possibility that the observed defect in the CFU-E but not BFU-E activity of GATA-1 S26E BM cells was due to lack of SCF in the former assay system. We next examined whether reduction in the viable cell number in the GATA-1 S26E EP culture was due to defects in cell proliferation or survival or both. To address this issue, purified EPs were cultured in EPOcontaining medium for 24 hrs, pulse-labeled with the thymidine analog 5-ethynyl-29-deoxyuridine (EdU) for additional 3 hours and the percentage of cells in S phase or with sub-G1 amount of DNA (apoptotic cells) was analyzed by flow cytometry. As shown in Fig. 7B, 7C and Fig. S3, compared to wt or GATA-1 S26A cells, GATA-1 S26E EPs manifested a significant reduction of cells in the DNA synthesis (S) phase (,50% of wt level, Fig. 7B), but only a marginal increase of cells that had undergone apoptosis (Fig. 7C). Together, these results suggest that reduced colony forming activity of GATA-1 S26E EPs in EPO-containing medium might be largely due to decreased cell proliferation of these cells.
Of note, although the survival of GATA-1 S26E EPs in EPOcontaining medium was only marginally reduced compared to that of control cells, such result was somewhat surprising, because our earlier study indicated that ectopic expression of GATA-1 S26E suppressed IL-3-withdrawal-induced apoptosis [18]. We thus compared the survival response of EPs from all three genotypes in medium without EPO. Consistent with our earlier findings in Ba/F3 cells [18], GATA-1 S26E EPs survived better than wild-type and GATA-1 S26A EPs in the absence of EPO (Fig. 7D). Semiquantitative RT-PCR analysis further revealed that, under such culture conditions, GATA-1 S26E EPs expressed higher levels of Bcl-X L and a slightly lower level of pro-apoptotic Bim mRNA compared to cells from the other two genotypes (Fig. 7E, and Fig.  S4A). In contrast, a quite different result was observed when these cells were cultured in medium containing EPO (Fig. 7F and Fig.  S4B). First, Bcl-X L expression in GATA-1 S26E EPs was not significantly different from that in control or GATA-1 S26A cells. Second, the expression of Bim was actually slightly higher in GATA-1 S26E EPs than in wt or GATA-1 S26A cells. Moreover, compared to control or GATA-1 S26A cells, GATA-1 S26E EPs expressed higher levels of the CDK inhibitor p21 Waf1/Cip1 only in the presence but not in the absence of EPO (Fig. 7E v.s. 7F).
Reduced CFU-E in GATA-1 S26E BMs was Mainly due to GATA-1 S26E -mediated Transcriptional Activation of the p21 Waf1/Cip1 Gene Next, we examined whether up-regulated expression of Bim and/or p21 Waf1/Cip1 may account for the reduced CFU-E activity in the BM of GATA-1 S26E mice. To address this issue, compound mutant mice carrying the GATA-1 S26E allele and the knockout allele of the Bim or the p21 Waf1/Cip1 gene were generated. Whereas ablation of Bim did not significantly revert the CFU-E numbers of GATA-1 S26E BM (data not shown), ablation of p21 Waf1/Cip1 completely rescued the colony forming defect of GATA-1 S26E BM albeit knockout of p21 Waf1/Cip1 itself did not affect the CFU-E activity of the mouse BM (Fig. 8A). Consistent with this result, the in vitro culture experiment also showed that ablation of p21 Waf1/ Cip1 restored the growth of GATA-1 S26E EPs in EPO-containing medium to an extent that was very similar to that of WT cells (Fig. 8B). Further reporter gene analysis revealed that, in the presence of EPO, the p21 Waf1/Cip1 gene promoter was significantly activated in GATA-1 S26E EPs (,10-fold, compared to promoterless vector control), but was only marginally activated in WT cells (,3-fold, Fig. 8C). We next examined whether the consensus binding site for the Sp1/KLF-like factor at -21 to -13 bp of the p21 Waf1/Cip1 gene promoter was important for GATA-1 S26Emediated transcriptional activation of the p21 Waf1/Cip1 gene, as such site was shown to be critical to GATA-1-stimulated transcription of the p21 Waf1/Cip1 promoter [22]. The result shown in Fig. 8C indicated that GATA-1 S26E -mediated transcriptional activation was abolished if the consensus binding site for the Sp1/ KLF-like factor at 221 to 213 bp of the p21 Waf1/Cip1 gene promoter was mutated (221 bp mt). Together, these results suggest that the S26E mutation confers on GATA-1 a prominent activity to activate the transcription of the p21 Waf1/Cip1 gene in EPs.

Discussion
In this study, we demonstrate that phosphorylation of GATA-1 at serine 26 is transiently induced by EPO in both erythroblasts and EPs. Mutation analysis indicates that such a transient event, i.e., quick phosphorylation followed by dephosphorylation of GATA-1 at serine 26 , is not essential for EPO-dependent growth or survival of erythroblasts. In contrast, the same mutation study revealed a different effect on EPs. While phosphorylation of GATA-1 at serine 26 is dispensable for the proliferation and survival of EPs, failure to dephosphorylate serine 26 following its transient phosphorylation significantly attenuates such two activities. These results suggest that constitutive phosphorylation of GATA-1at serine 26 interferes with the development of the erythroid lineage in a stage-specific manner. We previously showed that GATA-1 phosphorylation at serine 26 plays an important role in the survival and expression of Bcl-X L and E4bp4 in IL-3-dependent Ba/F3 cells [18]. However, the expression of these proteins was not affected by the GATA-1 serine 26 mutations in cultured primary erythroblasts, suggesting that serine 26 phsophorylation affects survival protein expression in a cell-type and/or cell-context specific manner. Some transcriptional co-factors, which are sensitive to serine 26 phosphorylation, may be present in cells like Ba/F3 and EPs. However, some co-factor(s), which are insensitive to serine 26 phosphorylation, may be present in cells like erythroblasts. A recent study reported that EPO protects neuronal cells from apoptosis via the NF-kB signaling pathway [31]. This result further suggests that EPO may induce the expression of survival proteins and maintain the viability of erythroblasts via an alternative pathway such as the one involving NF-kB. More experiments are required to address this possibility. p21 Waf1/Cip1 is not essential for normal erythropoiesis, as mice deficient of this protein did not manifest any abnormal phenotype in the erythroid lineage (data not shown). One interesting finding from this study is that the reduced CFU-E activity of GATA-1 S26E EPs is mainly due to an increased expression of p21 Waf1/Cip1 following EPO treatment, as ablation of p21 Waf1/Cip1 restored the CFU-E number of GATA-1 S26E mice. Notably, in the absence of EPO, the p21 Waf1/Cip1 expression was not activated in GATA-1 S26E EPs, suggesting that GATA-1 S26E -mediated transcriptional activation of the p21 Waf1/Cip1 gene is dependent on an EPOinduced co-factor. One candidate co-factor would be the protein that recognizes the consensus binding site for the Sp1/KLF-like factor at 221 to 213 bp of the p21 Waf1/Cip1 gene promoter, since mutation of this site abolished GATA-1 S26E -mediated transcriptional activation of the p21 Waf1/Cip1 gene in EPO-treated EPs. More experiments would be required to test this possibility.
The reduced CFU-E activity of GATA-1 S26E EPs appeared to be compensated in vivo, since GATA-1 S26E mice didn't manifest any abnormal phenotype at later stage of erythroid development. This is reminiscent of the phenotype of knock-in mutant mice bearing the triple-alanine substitution of GATA-1 [17], which results in a reduction in the numbers of both BFU-E and CFU-E progenitors in the bone marrow, but does not perturb later stage of erythropoiesis. One possible explanation for this phenomenon is that CFU-E progenitors may respond to additional growth factor(s) other than EPO in the bone marrow, in which GATA-1 is not required for its signaling. More experiments would be required to test this possibility. Analysis of the frequency of GATA-1 S26E and GATA-1 S26A EPs in the bone marrow. BM cells were stained with various cell surface markers as described in the ''Materials and Methods'' section of the text. To analyze the frequency of the EP population in the BM, the lineage 2 /IL-3Ra 2 /IL-7Ra 2 triple negative population was first gated out, followed by gating for the hematopoietic progenitors (Lin 2 IL-3Ra 2 IL-7Ra 2 c-Kit + Sca-1 2 , I). Finally the percentage of the Lin 2 IL-3Ra 2 IL-7Ra 2 c-Kit + Sca-1 2 CD71 + CD41 2 population (II), defined as EPs, was analyzed. Representative contour plots are shown from 5 independent experiments. Numbers indicate the percentage of the indicated population in the BM. (TIF) Figure S3 Impaired proliferation of GATA-1 S26E EPs in EPO-containing medium. EPs purified from mice with the indicated genotype were cultured for 24 hours in EPO-containing medium and labeled for additional 3 hours with EdU. After EdU labeling, cells in S phase (EdU + ) or with sub-G1 amount of DNA (detected by PI staining) were determined by flow cytometry. Data shown here are one representative set of contour plots from three independent experiments with very similar results. Numbers indicate the percentage of the indicated population. (TIF) Figure S4 Relative mRNA expression levels of Bcl-X L , Bim and p21 as described in the legend to Fig. 7E and 7F of the text. The intensity of each PCR product (the 26 cycle) was quantified and normalized to that of the GAPDH signal. The results (mean 6 SD) are plotted as a relative level to that of the wt, which is set as 1. *, P,0.05; **, P,0.01; ***, P,0.001 (N = 3 for each genotype, one-way ANOVA). (TIF)