Ectopic Expression of Transcription Factor NF-E2 Alters the Phenotype of Erythroid and Monoblastoid Cells*

In this study, regulation of transcription factor NF-E2 was examined in differentiating erythroid and myeloid cells, and the impact of raising NF-E2 concentrations within these cell types was assessed. NF-E2 was expressed in the J2E erythroid cell line, but the levels increased only marginally during erythropoietin-induced differentiation. In contrast, rare myeloid variants of J2E cells did not express NF-E2. Although NF-E2 was present in M1 monoblastoid cells, it was undetectable as these cells matured into macrophages. Compared with erythroid cells, transcription of the NF-E2 gene was reduced, and the half-life of the mRNA was significantly shorter in monocytoid cells. Ectopic expression of NF-E2 had a profound impact upon the J2E cells; morphologically mature erythroid cells spontaneously emerged in culture, but the cells failed to synthesize hemoglobin, even in the presence of erythropoietin. Although proliferation and viability increased in the NF-E2-transfected J2E cells, their responsiveness to erythropoietin was severely diminished. Strikingly, increasing the expression of NF-E2 in M1 cells produced sublines that contained erythroid or immature megakaryocytic cells. Finally, overexpression of NF-E2 in primary hemopoietic progenitors from fetal liver increased erythroid colony formation in the absence of erythropoietin. These data demonstrate that elevated NF-E2 (i) had a dominant effect on the phenotype and maturation of J2E erythroid cells, (ii) was able to reprogram the M1 monocytoid line, and (iii) promoted the development of erythroid colonies by normal progenitors.

Several transcription factors have been shown to play a critical role in the development of various hemopoietic lineages, with different combinations of these regulatory proteins directing lineage fate (1). Increasing evidence is emerging that the concentration of transcription factors influences lineage commitment. Indeed, by manipulating the levels of transcription factors in cell lines, the plasticity of the hemopoietic system has been demonstrated (2,3). The phenomenon of hemopoietic lin-eage switching (4) probably occurs because of alterations to the levels of transcription factors.
NF-E2 is a heterodimeric transcription factor that consists of a hemopoietic-restricted subunit (p45 NF-E2 ) and a ubiquitously expressed subunit belonging to the Maf family of proteins (5,6). NF-E2 is expressed primarily in erythroid cells, but is also found at lower levels in progenitor, megakaryocytic, mast, and granulocytic cells (5). It binds to an extended AP-1-binding site, (T/C)GCTGA(G/C)TCA(T/C) (7), present in the promoters of genes for heme biosynthesis (8 -12) and in the locus control region of globin genes (7,13,14). Loss of NF-E2 expression in murine erythroleukemia cells results in a drastic reduction in ␣and ␤-globin expression (15,16). Thus, in vitro studies have implicated NF-E2 as a major regulator of hemoglobin production during erythropoiesis (17).
NF-E2 knockout mice display an unexpectedly mild diserythropoiesis. Alterations to the erythroid compartment are most pronounced in neonates, where anemia, dysmorphic red cells, and decreased hemoglobin content are observed (18). However, these mice die shortly after birth due to a lack of circulating platelets, indicating that NF-E2 plays a key role in megakaryopoiesis (19). While megakaryocytes are present in the NF-E2 Ϫ/Ϫ mice, they have aberrant distribution of demarcation membranes and platelet fields in their cytoplasm (19). Although megakaryocytes in these mice can respond to thrombopoietin, they fail to produce platelets, suggesting that NF-E2 is required in the late stages of megakaryopoiesis (19 -21). Recently, Levin et al. (22) demonstrated that NF-E2 is required for megakaryocyte proliferation as well as differentiation. To date, NF-E2 has been shown to bind to the promoter and to regulate the expression of only one megakaryocytic gene, viz. thromboxane synthase (23).
To further examine the effect of NF-E2 in different hemopoietic lineages, we generated J2E erythroid and M1 monoblastoid cell lines that express NF-E2 ectopically. J2E cells are erythropoietin (epo) 1 -responsive and show enhanced proliferation and viability, morphological maturation, and accumulation of hemoglobin when exposed to the hormone (24 -26). However, under adverse conditions, J2E cells have occasionally switched lineage and displayed the phenotype of monocytoid cells (27). In contrast, M1 cells are immature myeloid cells that develop into macrophages in response to interleukin-6 (IL-6) or leukemia inhibitory factor (LIF) (28). Here we show that excess NF-E2 promoted a mature erythroid phenotype in both J2E and M1 cell lines and the appearance of immature megakaryocytes in one clone of the M1 cells. Furthermore, the higher level of NF-E2 in J2E cells interfered with hemoglobin synthesis, proliferation, viability, and responsiveness to epo. Erythroid colony formation by nontransformed progenitors was also enhanced by elevated NF-E2 levels. These data demonstrate that changing the concentration of NF-E2 in hemopoietic cells has a profound effect on their phenotype.

EXPERIMENTAL PROCEDURES
Cell Culture-The J2E (24) and M1 (28) cell lines and the J2E myeloid lines J2E-m1, J2E-m2, J2E-m3, J2E-NR-m1, J2E-NR-m2, and J2E-NR-m3 (27) were maintained in Dulbecco's modified Eagle's medium with 5% fetal calf serum. For differentiation studies, M1 cells were stimulated with either LIF (5 ng/ml) or IL-6 (32 ng/ml), whereas J2E cells were induced with epo (5 units/ml). Cell viability was determined by eosin exclusion (25), and hemoglobin synthesis was determined by benzidine (29) or diaminofluorene (30) staining. Proliferation assays were performed on cells transferred to RPMI 1640 medium (Biosciences, New South Wales, Australia) immediately before the assay. Cultures were then established at 10 3 cells/100 l and stimulated with epo for 16 h before being pulsed with 0.5 Ci of [methyl-3 H]thymidine (Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom) for 4 h. Cell morphology was examined by cytocentrifuging cells onto glass slides and immersion in Wright's stain. For hemoglobin spectra, cells (5 ϫ 10 7 ) were lysed in water for 1 h on ice, and scans were performed between 350 and 700 nm. Expression of the epo receptor was determined by cytocentrifuging cells onto glass slides, fixing in 2.5% paraformaldehyde, and incubating in the presence of an anti-epo receptor polyclonal antibody, followed by a fluorescein isothiocyanate-conjugated anti-rabbit secondary antibody (Amersham Pharmacia Biotech) and examination by immunofluorescence. The rabbit anti-epo receptor antibody was made using the extracellular domain of the epo receptor fused to glutathione S-transferase as the immunizing agent.
Retroviral Infection-The retroviral vector pRuf(tk)Neo (31) was used to express the entire coding region of p45 NF-E2 . The packaging cell line PA317 was then transfected with the retroviral construct by calcium phosphate precipitation (32), and supernatants containing amphotropic retroviruses were used to infect J2E and M1 cells. Cells were selected in Geneticin (Sigma) before cloning in methylcellulose as described elsewhere (26). Six to twelve independent colonies were isolated for each construct, and integration of the construct was confirmed by Southern analysis. The NF-E2 retroviral construct was also transfected in ⌿2 cells, and ecotropic virus-containing supernatants were used to infect day 12 fetal liver cells. These cell were then placed in methylcellulose cultures as described previously (33) and counted 7 days later.
Flow Cytometry-M1 cells (10 6 ) were incubated with Mac-1 antimouse Ig antibody (34) for 30 min on ice, washed, and then incubated with secondary antibodies conjugated to fluorescein isothiocyanate (Silenus Laboratories, Hawthorn, Victoria, Australia) for 30 min on ice. The cells were washed again before analysis on a Beckman-Coulter Epics XL/MCL flow cytometer. Cells incubated in the absence of primary antibody were analyzed as controls.

NF-E2 Levels during Erythroid and Myeloid
Differentiation-To monitor changes in NF-E2 levels during erythroid differentiation, J2E cells were stimulated with epo. Untreated J2E cells expressed the transcription factor, and the levels of NF-E2 transcripts and protein increased slightly during epoinduced differentiation (Fig. 1, A and B). Although the rise was modest (20 -50%), it was reproducible. These observations are consistent with a marginal increase in NF-E2 transcription determined by nuclear run-on assay (data not shown).
Since J2E cells have occasionally generated mutants that have switched lineage and developed a monocytoid phenotype (27), NF-E2 expression was examined in these myeloid variants. Significantly, NF-E2 transcription was down-regulated 20 -95% in these myeloid lines (Fig. 1C); moreover, NF-E2 mRNA and protein were detected only in J2E-m1 and J2E-m2 cells, the least mature of the monocytoid variants (Fig. 1, D and and probed with NF-E2, followed by GAPDH. B, J2E cells were stimulated with epo (5 units/ml), and lysates (100 g) were immunoblotted with anti-p45 NF-E2 antibodies, followed by anti-p42 MAPK antibodies as a loading control. C, nuclear run-on assay was performed on J2E cells and J2E myeloid cell lines showing transcription of NF-E2 and GAPDH; pGEM acted as a negative control. D, shown are the results from Northern analysis of RNA from J2E cells and myeloid derivatives of J2E cells probed with NF-E2, followed by GAPDH. E, J2E and J2E myeloid cell line lysates (100 g) were immunoblotted with anti-p45 NF-E2 antibodies, followed by anti-p42 MAPK antibodies as a loading control. E). These results indicated that NF-E2 expression was markedly down-regulated in cells displaying a more differentiated monocytic phenotype. To examine changes in NF-E2 levels during monocytic/macrophage maturation more closely, the M1 monoblastoid cell line was induced to differentiate with either IL-6 or LIF. Although uninduced M1 cells expressed NF-E2, both the transcript and protein disappeared as the cells differentiated (Fig. 2, A and B).
The NF-E2 gene was transcribed in all myeloid variants of J2E cells (Fig. 1C), but little mRNA was detected in most lines; we therefore predicted that the transcript might be less stable in monocytoid cells. To determine whether differences in NF-E2 mRNA stability existed between erythroid and myeloid cells, actinomycin D was added to cultures of J2E and M1 cells, and the decay of NF-E2 mRNA was monitored. Fig. 2C shows that NF-E2 transcripts had a half-life of ϳ80 min in the J2E erythroid cells, compared with only 25 min in M1 monocytoid cells. This abbreviated half-life did not alter appreciably during M1 maturation (data not shown). Together, the data presented in Figs. 1 and 2 show that NF-E2 levels increased slightly during erythroid differentiation, but declined markedly with monocytic maturation via transcriptional and post-transcriptional means.
Ectopic Expression of NF-E2 in J2E and M1 Cells-To de-termine the consequences of raising NF-E2 levels in erythroid and myeloid cells, J2E and M1 cells were infected with a retroviral vector containing the entire coding region of NF-E2; these clones were called JNF and MNF, respectively. Numerous clones were obtained, and those displaying the highest expression are shown in Fig. 3 (A and B). Although the virally encoded NF-E2 RNA was up to four times more abundant than endogenous transcripts, NF-E2 protein content rose moderately, indicating that translation, or polypeptide stability, may play an important role in determining the final protein content.
Interestingly, there was no appreciable alteration in endogenous NF-E2 transcripts. Although virally produced NF-E2 RNA was present throughout epo-initiated differentiation of JNF cells, both viral and endogenous NF-E2 transcripts decreased during LIF-induced maturation in the M1 transfectants (Fig. 3C). Since transcription from this retroviral vector continues throughout myeloid maturation, 2 this result sup-2 T. J. Gonda, personal communication.

FIG. 2. NF-E2 expression decreases during M1 differentiation.
A, shown are the results from Northern analysis of RNA isolated from M1 cells stimulated with IL-6 (32 ng/ml) or LIF (5 ng/ml) and probed with NF-E2, followed by GAPDH. B, M1 cells were stimulated with IL-6 (32 ng/ml), and lysates (100 g) were immunoblotted with anti-p45 NF-E2 antibodies, followed by anti-p42 MAPK antibodies as a loading control. C, the stability of NF-E2 mRNA in J2E and M1 cells was determined by actinomycin D (10 g/ml) treatment, mRNA isolation over 4 h, and Northern analysis with NF-E2 and GAPDH probes. Levels of NF-E2 mRNA were quantitated relative to GAPDH. Endogenous NF-E2 mRNA levels are represented by black bars, and viral NF-E2 mRNA levels by stippled boxes. B, lysates (100 g) from JNF and MNF lines were immunoblotted with anti-p45 NF-E2 antibodies, followed by anti-p42 MAPK antibodies. Levels of p45 NF-E2 were quantitated using NIH Image Version 1.6.1 software and are expressed relative to p42 MAPK levels. C, JNF26 and MNF4 cells were induced with epo (5 units/ml) or LIF (5 ng/ml), respectively; the mRNA was isolated over 4 days; and Northern blots were probed with NF-E2 and GAPDH. ports the observation of increased turnover of NF-E2 RNA in monocytic cells (Fig. 2C).
Increased Expression of NF-E2 Promotes Morphological Maturation of J2E Cells in the Absence of Hemoglobin Synthesis-When stimulated with epo, a significant proportion of J2E cells undergo morphological maturation and synthesize hemoglobin (24 -26). An example of epo-induced morphological change is shown in Fig. 4A, where some cells display condensed nuclei on the verge of extrusion, whereas others have the appearance of enucleate reticulocytes. The proportion of cells that retain an immature proerythroblastoid phenotype and those with a more differentiated morphology is summarized in Fig. 4C. Surprisingly, numerous mature erythroid cells and reticulocytes were detected in JNF cells in the absence of epo (Fig. 4, B and C); however, these numbers did not change following epo stimulation (Fig. 4C). Thus, increasing the concentration of NF-E2 in these erythroid cells promoted spontaneous morphological maturation.
The other characteristic feature of erythroid terminal differentiation is hemoglobin synthesis. The presence of hemoglobin in parental J2E cells and JNF19 and JNF26 cultures was assessed initially by benzidine staining. Fig. 5A shows that, as anticipated, J2E cells produced more hemoglobin after epo stimulation (24 -26). In contrast, the JNF19 and JNF26 cells synthesized negligible amounts of hemoglobin in the absence of epo, and these levels did not rise when the hormone was added to the cultures. To confirm this unexpected observation, hemoglobin levels were ascertained by spectral scans. The data shown in Fig. 5B demonstrate that unstimulated J2E cells contain hemoglobin with absorbance peaks appearing at 413, 540, and 577 nm, whereas significantly reduced levels were detected in JNF19 and JNF26 cells. Therefore, elevating NF-E2 levels in J2E cells enhanced morphological maturation, but suppressed hemoglobin synthesis.
NF-E2 Accelerates Proliferation-In addition to morphological changes and hemoglobin synthesis, J2E cells respond to epo with a burst of proliferation and remain more viable under serum-free conditions (24,25,37,45). Therefore, the effect of increasing NF-E2 concentrations on replication rate and cell survival was examined next. Interestingly, raising the levels of NF-E2 in J2E cells increased the incorporation of [ 3 H]thymidine (Fig. 5C) and the rate of cell division (data not shown). Although uptake of labeled thymidine was 3-5-fold greater in JNF cells compared with the parental line, they failed to respond to epo (Fig. 5C). Similarly, cells expressing exogenous NF-E2 remained more viable in the absence of serum, but reacted moderately to the addition of epo (Fig. 5D). Taken together, altering NF-E2 levels in J2E cells had a dramatic impact on their phenotype, as morphological appearance, hemoglobin synthesis, cell division, and viability were all affected by raising the intracellular concentration of NF-E2.
Erythroid Cells Appear in M1 Cultures Expressing Exogenous NF-E2-The immature M1 monoblastoid cell develops into macrophages following exposure to a number of agents, including IL-6 and LIF (Fig. 6A). Despite their apparent commitment to the monocytic pathway, introduction of GATA-1 has been shown to induce the appearance of erythroid and megakaryocytic cells in these cultures (2). Similarly, Fig. 6A shows that increased NF-E2 levels in M1 cells promoted the emergence of erythroid cells in two of three sublines. Both MNF4 and MNF11 cultures contained up to 5% erythroid cells, varying from proerythroblast to orthochromatic erythroblasts and reticulocytes. These cells were also present in cultures exposed to LIF, although their proportion did not change (Fig.  6A). A number of hemoglobin-producing cells (3-7%) were detected in these cultures (Fig. 6B), and surface epo receptors were identified by immunofluorescence (Fig. 6C). In contrast with the MNF4 and MNF11 lines, the MNF12 clone contained some very large cells that had the appearance of megakaryocytic precursors (Fig. 6A). However, these cells did not contain acetylcholinesterase (data not shown) and disappeared from culture after the addition of LIF. These data demonstrate that, like GATA-1 (2), NF-E2 can force M1 cells to alter their appearance and to display an erythroid or immature megakaryocytic phenotype.
M1 cells have also been transfected with SCL, another transcription factor important for erythroid cells (46). Although these transfectants did not display any erythroid features, macrophage differentiation was impeded. To examine whether NF-E2 also inhibited the maturation of M1 cells, cultures were exposed to LIF and monitored for the expression of Mac-1, a macrophage surface marker. Fig. 7A shows that, unlike SCL (46), NF-E2 did not interfere with macrophage maturation. Similar results were obtained at all time points and concentrations of LIF examined (data not shown). The difference between the effects of these two transcription factors may be due, in part, to the degradation of NF-E2 as M1 cells mature (Fig. 3C).
To examine the effect of ectopic NF-E2 expression on other erythroid-restricted transcription factors, Northern blots were probed for the presence of GATA-1 and EKLF mRNAs. Fig. 7B shows that elevated NF-E2 did not induce expression of GATA-1 or EKLF. It is possible that NF-E2 was able to impose an erythroid phenotype on M1 cells without an obvious increase in GATA-1 and EKLF; however, it is likely that subtle changes occurred to GATA-1 and EKLF that were not detected by Northern blotting of total RNA.
NF-E2 Increases Erythroid Colony Formation-Having observed noticeable effects of exogenous NF-E2 expression in cell lines, the consequences of NF-E2 overexpression were examined in primary hemopoietic progenitors. To this end, fetal liver cells were infected with a NF-E2 retrovirus, and colony formation was enumerated 7 days later. Significantly, a 3-4-fold increase in burst-forming units-erythroid (BFU-E) was detected in cultures expressing NF-E2 ectopically (Fig. 8). These colonies were pale and appeared to contain less hemoglobin FIG. 6. Enforced expression of NF-E2 in M1 cells promotes an erythroid or megakaryoblastoid morphology. A, M1 or MNF cells, unstimulated or stimulated with LIF (5 ng/ml) for 48 h, were cytocentrifuged onto glass slides, fixed, and stained using Wright's stain. The arrows indicate the presence of polychromatic erythroblasts, orthochromatic erythroblasts, and reticulocytes. The bar represents 10 m. B, unstimulated MNF cells were stained with diaminofluorene to detect hemoglobin. Note the dark staining in the cells marked by arrows, indicating hemoglobin production. C, MNF cells were cytocentrifuged onto glass slides and fixed, and the presence of the epo receptor was detected by immunofluorescence on cells indicated by the arrows. than controls. Despite the rise in BFU-E with NF-E2, there was no increase in colony formation after treatment with epo. In contrast, exposure to the NF-E2 retrovirus reduced the number of myeloid colonies slightly. These data are consistent with the effects of NF-E2 overexpression in J2E cells (Figs. 4 and 5) and demonstrate that normal erythroid progenitors in particular are influenced by the concentration of NF-E2. DISCUSSION In this study, we have demonstrated that ectopic expression of NF-E2 had a profound effect on erythroid and monoblastoid cells. In the J2E erythroid cell line, this was manifest by markedly increased proliferation rates, spontaneous morphological maturation, suppressed hemoglobin synthesis, and enhanced viability. Responsiveness to epo was also greatly diminished. These effects in J2E cells are compatible with increased epo-independent BFU-E generated by overexpression of NF-E2 in primary erythroid progenitors. In contrast, enforced NF-E2 expression resulted in the appearance of erythroid or megakaryoblast-like cells in the M1 monoblastoid cultures.
The effect of exogenous NF-E2 on J2E cells is remarkably similar to the effects of overexpressing GATA-1 in murine erythroleukemia cells (47). In both cases, cell division was promoted, at the expense of hemoglobin production. Since NF-E2 has been implicated in the regulation of globin gene transcription in vitro (15,16) and neonatal NF-E2 Ϫ/Ϫ mice are anemic due to reduced hemoglobin content (18), the inhibition of hemoglobin production in J2E cells due to excess NF-E2 was unexpected. It is possible that by altering the NF-E2 concentration in the cells, albeit modestly (Fig. 3), the equilibrium with other transcription factors has been modified. NF-E2 is known to heterodimerize with members of the Maf family (6) and to bind the thyroid hormone receptor (48). The severe biological consequences of changing NF-E2 concentration in these cells may be due to disruption of these nuclear complexes.
An interesting aspect of this study involved the uncoupling of cytological changes from hemoglobin production in J2E cells.
Here, excess NF-E2 promoted spontaneous morphological maturation, without concomitant hemoglobin synthesis, suggesting that two processes regulated coordinately during erythroid terminal differentiation can be dissected. These observations support previous data in EKLF null mice, in which mature red blood cells lacking hemoglobin are produced (49,50). In addition, down-regulation of EKLF in J2E cells does not prevent reticulocyte formation in the absence of hemoglobin production (51). Significantly, NF-E2 Ϫ/Ϫ mice contain dysmorphic red cells, particularly in neonates (18); moreover, numerous defective red cell fragments are present in NF-E2 Ϫ/Ϫ mice (22). These results suggest that NF-E2 may be an important determinant in the development of morphologically mature erythroid cells.
The expression pattern of NF-E2 during J2E cell differentiation is consistent with several other models of erythropoiesis. Labbaye et al. (52) reported that NF-E2 is expressed throughout the development of normal erythroid cells, and NF-E2 levels remain elevated in chemically induced murine erythroleukemia and K562 cells (53,54). Similarly, EKLF levels are consistently raised during epo-induced differentiation of J2E cells (51), whereas GATA-1 levels decrease late in the maturation process (36). The down-regulation of NF-E2 in myeloid variants of J2E cells as well as in differentiating M1 cells agrees with the lack of NF-E2 in normal myeloid cells (52). The suppression of NF-E2 in myeloid cells was due not only to reduced transcription of the gene, but also to increased turnover of the transcripts (Figs. 1 and 2). The reduction in virally generated NF-E2 in maturing M1 cells suggests that some instability elements may exist within the coding region of the mRNA, as has been described for other transcripts such as c-fos (55). However, NF-E2 levels did not decrease when J2E cells developed an immature monocytoid appearance due to the introduction of the hemopoietic lineage switch gene HLS7 (33), which is homologous to the novel human oncogene MLF1 (56).
The emergence of erythroid cells in M1 cultures expressing NF-E2 is significant. Clearly, this transcription factor is turned off during monocytic terminal differentiation (Fig. 2), and ectopic expression has generated clones containing erythroid cells. This observation is similar to the introduction of GATA-1 into M1 cells, resulting in the appearance of cells with an erythroid phenotype (2). The presence of cells resembling megakaryoblasts in one clone of M1 cells expressing exogenous NF-E2 is also comparable with the data of Yamaguchi et al. (2) and is consistent with the crucial role of NF-E2 in megakaryocyte development identified in knockout mice (19). These experiments demonstrate M1 cells have a propensity to generate erythroid and megakaryocytic cells if the transcription factor ratio is altered. The reprogramming of M1 cells by NF-E2 is also similar to the ability of GATA-1 to generate erythroblasts, thromboblasts, and eosinophils from transformed myelomonocytic cells (3).
Overexpression of NF-E2 in primary hemopoietic cells substantially increased BFU-E numbers in the absence of epo, but only slightly reduced myeloid colony formation. These data reflect the impact of exogenous NF-E2 on J2E and M1 cell lines and suggest that increased NF-E2 expression favors erythroid development over myeloid maturation in agreement with the role of NF-E2 in erythroid gene regulation (8 -17). Although elevated NF-E2 promoted BFU-E formation, the colonies appeared pale and poorly hemoglobinized, which was similar to the increased proliferation and reduced hemoglobin synthesis observed in JNF cells. Thus, as seen with the NF-E2 Ϫ/Ϫ mice (18,22), the correct concentration of NF-E2 is essential for complete development of red blood cells. It is possible that rapid degradation of NF-E2 transcripts in myeloid cells (Fig. 2) may have prevented a more significant effect on colony formation in that lineage.
Taken together, the information presented here on ectopic NF-E2 expression in normal and immortalized cells highlights the importance of transcription factor concentrations during the development and terminal differentiation of hemopoietic cells (1). Interference with this delicate balance can have dramatic ramifications for the production of fully functional blood cells.