NF-E2p18/Mafk Is Required in DMSO-Induced Differentiation Of

NF-E2p18/mafK is required in DMSO-induced differentiation of Friend erythroleukemia cells by enhancing NF-E2 activity C Francastel, V Poindessous-Jazat, Y Augery-Bourget and J Robert-Le´ze´ne`s

INSERM U268, Hoˆpital Paul Brousse, 94800 Villejuif, France

When Friend murine erythroleukemia (F-MEL) cells are induced (LCR).6 The ␤-globin complex LCR (␤-LCR) include four to differentiate by dimethylsulfoxide (DMSO), erythroid-specific erythroid-specific DNasel hypersensitive sites (designated HS- genes are transcriptionally activated. The erythroid transcrip- 1 to HS-4) and acts as a powerful enhancer of transcription tion factor NF-E2 is essential for enhancer activity of the globin 7,8 locus control regions. NF-E2 functions as a heterocomplex of globin promoters. While all four HS have LCR activity in consisting of a 45-kDa subunit (NF-E2p45) and a 18-kDa sub- transgenic mice, only one region (HS-2) is crucial for unit (NF-E2p18). The larger subunit NF-E2p45 is tissue-restric- enhancer activity when assayed in transient transfection ted and is believed to play a role in globin gene expression in systems.9–11 Within HS-2, the enhancer activity for globin F-MEL cells. The expression of the smaller subunit NF-E2p18, expression requires a tandem repeat of AP-1-like which is a Maf family member (MafK), is cell type- and develop- 12,13 mental stage-specific. We have investigated the possible role elements. This tandem motif was shown to bind two tran- of NF-E2p18 in Friend erythroid differentiation by stably trans- scription factors: the ubiquitous AP-1 protein and the lineage fecting either sense and antisense p18 constructs into differen- limited NF-E2 protein.14,15 The NF-E2 complex is a hetero- tiation-sensitive 745A and partially defective-differentiation dimer of two b-Zip proteins of 45 and 18 kDa.16,17 While NF- TFP10 cell lines. Overexpression of NF-E2p18 induced E2p45 expression is restricted to cells of the erythroid megaka- expression of globin transcripts in both cell lines and increased ryocyte and mast cell lineage, the expression of the smaller their sensitivity to erythroid differentiation when exposed to subunit NF-E2p18 is cell type- and developmental stage-spe- DMSO. Conversely, inhibition of p18 expression by antisense 18 transcripts resulted in the inhibition of DMSO-induced differen- cific. Since p45 homodimers bind very poorly to NF-E2 tiation in both cell lines. These results indicate that NF-E2p18 motifs, NF-E2p45 protein binds as an obligate heterodimer is necessary for globin expression in F-MEL cells and that it is with NF-E2p18 protein,16 to form the NF-E2 complex which the predominant gene of the Maf family involved in DMSO- has been described as the transcriptional activator of NF-E2 induced erythroid differentiation. Moreover F-MEL clones over- enhancer elements.19 The smaller subunit NF-E2p18 protein expressing NF-E2p18 showed an increase in specific NF-E2 DNA-binding activity whereas this activity was decreased in is encoded by a Maf family gene, the mafK gene, and lacks 20,21 clones expressing antisense p18. Finally, studies using transi- the putative trans-activator domain. Enhancer activity of ent transfection assays showed that p18 activated NF-E2 site- the ␣-globin locus HS-40 element is also dependent on NF- dependent transcription in F-MEL cells. These data suggest E2/AP-1 sequences.22 Beside ␣- and ␤-globin LCR, NF-E2/AP- that NF-E2p18 can participate in DMSO-induced erythroid 1 binding sites are also present in the promoters of three genes differentiation of F-MEL cells by enhancing NF-E2 activity. encoding heme biosynthetic enzymes, porphobilinogen Keywords: Friend erythroleukemia; NF-E2p18; Maf protein; 14 23 erythroid differentiation; globin expression deaminase, 5-aminolevulinate synthetase and ferrochelat- ase24 suggesting a broad role for the protein NF-E2 in erythroid-specific gene expression. The role of the NF-E2 complex and its subunits p45 and Introduction p18 in the DMSO- or HMBA-induced erythroid differentiation of F-MEL cells is unclear. It was reported that both NF-E2 DNA Friend murine erythroleukemia (F-MEL) cells are virus-trans- binding and NF-E2 site-dependent transcriptional activities formed erythroid precursors cells blocked at an early stage increase during the induction of MEL differentiation.13,14,25,26 in erythroid differentiation. This arrest can be overcome by Recent studies on one Friend erythroleukemia cell line, which exposure to a variety of chemical inducers, including DMSO lacks NF-E2 p45 due to Friend virus integration, showed a and HMBA.1,2 Variants of F-MEL cells have been obtained drastic reduction in expression of the ␣- and ␤-globin genes.27 which differ in their response to chemical inducers. These cell Reintroduction of the NF-E2p45 gene into this cell line par- models are useful for studying the mechanism of erythroid dif- tially restores globin RNA expression suggesting that it may ferentiation since induced F-MEL cells undergo phenotypic be critical for differentiation of F-MEL cells. The presence of changes that resemble the final changes of normal erythropo- NF-E2p18 in the NF-E2 complex increases binding specificity iesis. Induction of erythroid differentiation of MEL cells by for the bases lying outside the AP-1 core of the NF-E2 consen- DMSO or HMBA is characterized by the cessation of DNA sus binding site, suggesting its possible role in transcription synthesis and by the expression of hemoglobin in the cells.3 activation of the globin gene.17 Igarashi et al28 have recently The most striking feature during the process of induced differ- reported that overexpression of mafK induces spontaneous entiation of F-MEL cells is the increased synthesis of globin erythroid differentiation of a Friend cell line. However, the chains which results from the activation of ␣ and ␤ globin molecular basis of this induction of differentiation by gene transcription.4,5 p18/mafK, specially the involvement of NF-E2 activity, has not Expression of mammalian globin genes depends on been clarified. upstream regulated elements, termed locus control regions In order to gain better understanding of the function of NF- E2p18/MafK during the DMSO-induced differentiation of F-MEL cell, we have introduced NF-E2p18 expression vectors Correspondence: J Robert-Le´zeńe`s INSERM U268, Hoˆpital Paul- Brousse, 14 avenue Paul Vaillant Couturier, 94807 Villejuif cedex, in both sense and antisense orientations into differentiation- France sensitive 745A and partially defective TFP10 Friend cell lines. Received 26 July 1996; accepted 24 October 1996 Sense p18 vectors increase the sensitivity to DMSO-induced NF-E2p18 and erythroid differentiation C Francastel et al 274

Figure 1 Expression of NF-E2p45 and NF-E2p18 mRNAs in F-MEL cells (strains 745A and TFP10), following stimulation with DMSO. Poly(A)+ mRNA (5 ␮g) was isolated at the time points indicated (in hours) after addition of 2% (v/v) DMSO. Northern blots were hybridized sequentially to probes homologous to NF-E2p45, NF-E2p18 and GAPDH followed by autoradiography. The different transcripts migrate as follows: NF- E2p45 1.8 kb; NF-E2p18 3.1 kb; GAPDH 1.4 kb.

Figure 2 NF-E2p45 protein expression and NF-E2 DNA complex formation in F-MEL cells following stimulation with 2% of DMSO. (a) Western blot analysis of total proteins using a NF-E2p45-specific antibody. Total protein 50 ␮g were separated by SDS-PAGE and blotted onto nitrocellulose. The blots were developed using the alkaline phosphatase system. The time in hours of DMSO treatment is indicated. (b) DNA- binding activity of NF-E2 in nuclear extracts proteins from F-MEL cells before DMSO treatment. The oligonucleotide probe containing NF-E2 consensus sequence was incubated with 10 ␮g of nuclear extract protein and EMSA were performed as described in Materials and methods. Specificity of binding was confirmed by competition with excess unlabeled oligonucleotide probe (not shown). Where indicated, 1 ␮l of anti- NF-E2p45 antiserum was added before addition of probe. (c) DNA-binding activity of NF-E2 in nuclear extracts proteins from F-MEL cells after DMSO treatment (4 days).

erythroid differentiation of both cell lines. Conversely, trans- increasing both NF-E2 DNA-binding and NF-E2 transcrip- fection with antisense p18 vectors inhibits the induction of tional activities. F-MEL erythroid differentiation by the DMSO suggesting that NFE2p18/MafK is necessary to induce globin expression in F- MEL cells during the process of differentiation. In an effort to Materials and methods elucidate the molecular mechanism that underlie the relationship between p18/MafK and F-MEL differentiation, we show a Plasmid constructs speciﬁc increase of the NF-E2 DNA-binding activity in clones overexpressing p18. In addition, we demonstrate that NF- The expression vector used for the construct was pcDNA3 E2p18 induces an increase of NF-E2 site-dependent transcrip- (Invitrogen, San Diego, CA, USA). Murine NF-E2p18 cDNA tional activity in both F-MEL cell lines suggesting that it can containing one ORF (open reading frame) was excised from be a positive regulator in erythroid cells which contain NF- its plasmid vector (gift from N Andrews, Dana-Farber Cancer E2p45. Taken together, these ﬁndings indicate that p18/MafK Institute, Boston, MA, USA) as 0.51-kb EcoRI fragment and can participate in erythroid differentiation of F-MEL cells by ligated into EcoRI-digested pcDNA3 with the sense and anti- NF-E2p18 and erythroid differentiation C Francastel et al 275

Figure 3 Construction of NF-E2p18 sense and antisense vectors (pcDNA3-p18S and pcDNA3-p18AS). A 510-bp EcoRI/EcoRI cDNA fragment encompassing the entire coding region was inserted into the corresponding sites of the expression pcDNA3 vector (Invitrogen). sense orientation.17 The resulting plasmids contain the p18 agarose gels and visualized by ethidium bromide staining. To sense (pcDNA3-p18S) or antisense (pcDNA3-p18AS) down- conﬁrm speciﬁcity in some cases, the agarose gels were stream of the CMV promoter and the neomycin gene as a Southern blotted on Biodyne nylon membranes (Pall, Ilfra- selective marker. combe, UK) and hybridized with p18 probe.

Cell lines and transfection Western blot analysis

F-MEL cells, strain 745A and TFP10 and derived transfectants Total cellular protein was subjected to SDS-PAGE, electroblot- were grown in RPMI-1640 medium (Gibco, Paisley, UK) con- ted onto a membrane, incubated with a NF-E2-p45 specific taining 10% fetal calf serum (FCS; Eurobio, Les Ulis, France). antibody (generous gift from N Andrews) and developed using For each transfection 2 × 107 log-phase cells were washed in an antirabbit (IgG) alkaline phosphatase conjugated antibody, phosphate-buffered saline (PBS), pelleted and resuspended in as previously described.31 0.4 ml of ice cold PBS/50 mm Hepes containing 20 ␮g of plasmid construct and 50 ␮g of salmon sperm DNA. Electropora- tions were performed using the Bio-Rad gene pulser (BioRad, Electrophoretic mobility-shifts assays (EMSA) Hercules, CA, USA) with conditions of 370 V and 500 ␮F. After electroporation cells were resuspended in fresh medium Nuclear extracts were prepared from 5 × 107 cells using pre- and after 48 h at 37°C were plated in 96-well dishes at viously described protocols32,33 in the presence of protease 2.5 × 104 cells per ml in medium containing 300 ␮g/ml of inhibitors (leupeptin, aprotinin, phenylmethylsulfonyl fluor- geneticin (G418; Gibco-BRL). Wells containing single clones ide, o-phenatroline). Protein was quantitated by the Bradford were selected on day 12–17 and transferred to 24-well dishes method. Nuclear extract protein 10 ␮g were incubated with then to 25 cm2 plates. 30 000 c.p.m. of 32P-labeled probe in a 20 ␮l of reaction mixture containing 1 ␮g of poly(dI-dC) and 2 ␮g of bovine serum albumine (BSA). After incubation on ice for 30 min, samples Benzidine staining were run on a 5% polyacrylamide gel in 0.5 × tris-borate- EDTA buffer for 2 h at 10 V/cm. Gels were dried and exposed Cells were seeded at 1 × 105 cells per ml and DMSO was to the screen of a phosphorimager (Molecular Dynamics, added at a final concentration of 1 or 2% (v/v). The degree of Sunnyvale, CA, USA). When indicated, 1 ␮l of anti-NF-E2p45 differentiation was scored on day 5 by counting the percent- antiserum was preincubated with nuclear extracts for 10 min age of benzidine-positive cells as previously described.29 before the addition of the probe. The following oligonucleotide probe, containing the NF-E2 site was used (top strand): 5Ј-CTG- GGGAACCTGTGCTGAGTCACTGGAGG-3Ј. RNA analysis The top strand fragment was 32P labeled at 5Ј ends with T4 polynucleotide kinase and (␥-32P)ATP (New England Nuclear, Total RNA was extracted by the guanidine/cesium chloride Boston, MA, USA) according to standard procedures, method and analysed by Northern blot or RT-PCR as pre- annealed to the complementary strand, and separated from viously described.30,31 For Northern blot analysis, the follow- the unincorporated nucleotides over a Sephadex G50 column. ing hybridization probes were used: p18, a 0.51-kb EcoRI fragment of cDNA NF-E2 p18; p45, a 1.3-kb EcoRI fragment of cDNA NF-E2p45; ␣ globin, a 1.5-kb PstI fragment human Transient transfection experiments genomic fragment; GAPDH, a 1.3-kb PstI fragment of cDNA of rat GAPDH. For RT-PCR analysis of p18 sense and anti- Transcriptional activation assays were performed by transient sense expression, the upstream primer is located in the T7 transfections of F-MEL or NIH/3T3 cells with a hGH reporter RNA promoter of the pcDNA3 vector (position 862–882). The construct containing a globin TATA box under the control of downstream primers for detecting p18 sense or antisense the ␤-globin tandem NF-E2 sites and a binding site for one expression are located from position 516 to 497 or 185–166 other erythroid transcriptional factor GATA-1. This construct according to the numbering of the NF-Ep18 sense cDNA, (GATA-NF-E2-hGH) was previously described26 and kindly respectively.17 The PCR products were electrophoresed in provided by D Martin (Fred Hutchinson Cancer Institute, NF-E2p18 and erythroid differentiation C Francastel et al 276 Results

Increase of NF-E2 p45 expression and NF-E2 DNA- binding activity during DMSO-induced differentiation of F-MEL cells

For this study we used two different MEL cell lines, the 745A cell line which is induced to terminal erythroid differentiation by exposure to 2% of DMSO and the TFP10 cell line, the erythroid differentiation of which is partially resistant to this same treatment.29 The expression of p45 and p18 gene was examined by Northern blot analysis using 5 ␮g of poly(A)+ RNA and labeled probes as described in the Materials and methods. As shown in Figure 1 both untreated F-MEL cell lines consti- tutively expressed p45 mRNA. DMSO treatment of 745A induced an early decline in p45 between 4 and 8 h. The levels of p45 mRNA returned to that of untreated by 24 h and pro- gressively increased until day 5 (120 h). A same early biphasic change of p45 expression, followed by an increase after 24 h was observed in DMSO-treated TFP10 cells. The p45 protein expression analyzed by Western blot followed p45 mRNA expression and an increase of p45 protein after 24 h of DMSO treatment was observed in both cell lines (Figure 2a). Similar p18 mRNA levels were observed in both untreated 745A and TFP10 cells (Figure 1). In contrast to p45 mRNA expression, the levels of p18 mRNA were low and were unchanged during DMSO treatment. We next sought to investigate NF-E2 DNA-binding activity in nuclear extracts as determined by a gel mobility shift assay. Figure 2b showed this DNA-binding activity in extracts of 745A and TFP10 cell lines before DMSO treatment. A predominant band was seen corresponding to the specific NF-E2 complex since it disappeared when the nuclear extract was treated by an anti-p45 antibody. The most slowly migrating band appeared to correspond to the AP-1 complex since its Figure 4 NF-E2 p18 and globin gene expression in p18 sense level was higher in c-jun expressing TFP-10 cells than in 745A transfected 745A and TFP10 clones (a) RT-PCR analysis of exogenous cells which do not express c-jun.31 As shown in Figure 2c, a p18 (sense) expression. Total RNAs were isolated from parental cells and stably transfected 745A (745A-p18S) and TFP10 (TFP10-p18S) marked increase of specific NF-E2 DNA-binding activity was clones and subjected to RT-PCR as described in Materials and observed in both 745A and TFP10 cell lines after 4 days of methods. Aliquots of the PCR reactions were separated on agarose DMSO treatment. This increase of NF-E2 DNA-binding gels, transferred to nylon membranes, and hybridized with the p18 activity seen during DMSO induction of F-MEL cells corre- probe. (b) Northern blot analysis of total RNAs from parental 745A sponded to an increase of the NF-E2p45-p18 complex since and TFP10 cells and from clones transfected with p18 sense (745A the anti-p45 antibody eliminated this complex (data not p18-S and TFP10 p18-S) and p18 antisense (745A p18-AS and TFP10 shown). These results showed that during the induction of p18-AS) vectors. Total RNA 20 ␮g was isolated and Northern blots were hybridized sequentially to probes homologous to NF-E2p18, ␣- Friend erythroid differentiation an increase of the NF-E2 globin and GAPDH. The exogenous p18 transcript migrates as 0.5 kb. complex was observed which was probably due to the The larger p18 transcript (1 kb) observed in 745A p18-S clone presum- increased amount of the NF-E2p45 subunit found after ably results from read-through of the polyadenylation signal into the DMSO treatment. neoR gene. The endogenous p18 transcript (3.1 kb) is not shown in the Figure. Establishment of stable transfectants expressing sense Seattle, WA, USA). Supercoiled plasmids were prepared by and antisense NF-E2p18 transcripts purification over a Qiagen tip-500 column. Electroporations were performed using the Bio-Rad gene pulser with conditions To investigate the potential role of the NF-E2p18 subunit in of 400 V and 500 ␮F, 10 ␮g of supercoiled reporter construct conferring DMSO-induced erythroid differentiation, we intro- and 2 × 107 cells in a 0.8 ml of phosphate-buffered duced sense and antisense p18 expression vectors into both saline/50 mm Hepes/10 ␮g salmon sperm DNA. When indi- 745A and TFP10 cells. p18 cDNA was cloned in both sense cated, the transfection mixture also contained 40 ␮g of the and antisense orientations into the pcDNA3 vector supercoiled p18 sense expression vector (pcDNA3-p18S). (Invitrogen) as shown in Figure 3. Cells were transfected by Cells were resuspended in fresh medium and 48 h after elec- electroporation with plasmid DNA from one of the expression troporation, medium was harvested and 10 ␮l (or 100 ␮l for vectors. Cells stably expressing sense or antisense constructs NIH/3T3) was assayed for hGH using an ELISA kit were selected by growth in G418 (Gibco-BRL) and 10 to 20 (Eurogenetics, Tessenderlo, Belgium) according to the manu- individual clones were obtained for each group. Clones facturer’s recommendations. expressing sense (p18-S) or antisense p18 (p18-AS) transcripts NF-E2p18 and erythroid differentiation C Francastel et al 277

Figure 5 Erythroid differentiation of parental F-MEL cells and clones transfected with sense NF-E2p18 in absence and in presence of 1% of DMSO. Untransfected 745A and TFP10 cells and 745A and TFP10 clones stably transfected with p18 sense vector (745A p18-S and TFP10 p18-S) were grown in absence (−) or in presence (+) of 1% DMSO for 5 days at which time terminal differentiation was scored by counting the percentage of benzidine-positive cells. had cell doubling times similar to those of the parental cell pared with the parental 745A cells (75% of benzidine-positive lines (not shown). All clones were examined for their ability to cells). TFP10 clone transfected with p18 antisense (TFP10 differentiate with DMSO, and randomly selected clones were p18-AS) also had reduced ability to differentiate in the pres- examined for RNA expression and for NF-E2 DNA-binding ence of 2% DMSO (10% benzidine-positive cells) as com- activity. pared with the parental TFP10 cells (18% benzidine-positive cells). These data indicate that p18/mafK is required for the DMSO-induced differentiation of F-MEL cells. Effect of p18 on spontaneous and DMSO-induced differentiation of MEL cells Correlation of NF-E2 binding activity and globin In the case of p18 sense-transfected clones, RT-PCR and expression in Friend clones overexpressing p18 Northern analysis demonstrated a wide range of exogenous p18 expression in both 745A and TF-P10 clones (not shown). In order to determine whether p18 expression is correlated to Figure 4 illustrates the expression of exogenous p18 RNA in a change of the NF-E2 complex, we analyzed the NF-E2 DNA- two clones, 745A p18-S and TFP10 p18-S. To determine the binding activity in their nuclear extracts in the absence of differentiation potential of the clones transfected with sense DMSO treatment. As shown in Figure 7a, both 745A and p18, they were stained with benzidine to detect hemoglobin TFP10 clones transfected with the p18 sense vector exhibited synthesis before and after induction with DMSO. As com- an increase of speciﬁc NF-E2 binding activity compared to pared with the parental 745A cells which did not exhibit parental cells. In contrast clones overexpressing p18 showed spontaneous erythroid differentiation, 745A clone expressing no apparent change in the AP-1 complex. A decrease of the p18 transcripts (745A p18-S) showed 5–6% benzidine-posi- NF-E2 binding activity was seen when 745A or TFP10 cells tive cells in the absence of DMSO (Figure 5). These results are were transfected with p18 antisense (Figure 7a). These results in agreement with those reported by Igarashi et al.28 Moreover indicate that overexpression of p18 in F-MEL cells leads to an clone 745A p18-S had an increased ability to differentiate in increase of the NF-E2 DNA-binding complex. response to 1% DMSO. Eighty percent of transfected 745A To clarify further the correlation of globin expression and cells were benzidine-positive as opposed to 10% of cells in NF-E2 binding activity in the p18 sense and antisense clones the control 745A. TFP10 p18-S clone was also able to differ- before DMSO treatment, we determined globin RNA entiate following exposure to 1% DMSO (55% benzidine- expression by Northern blot analysis. Figure 4b showed a high positive cells) as compared to parental TFP10 cells which level expression of ␣ globin mRNA in 745A clone expressing were totally resistant to the same treatment with DMSO exogenous p18 sense (745Ap18-S) whereas no expression of (Figure 5). We conclude that the induction of erythroid globin was detected in 745A parental cells. Low levels of glo- differentiation by the DMSO is dependent on the levels of p18 bin transcripts were also detected in TFP10 clone expressing transcripts in F-MEL cells. p18 sense (TFP10p18-S). As expected, no detectable globin RNA was seen in clones expressing p18 antisense (Figure 4b). These results suggest a good correlation between the globin Transfection with antisense p18 inhibits DMSO- mRNA expression and the increase of NF-E2 DNA-binding induced differentiation activity induced by overexpression of p18 in F-MEL cells.

In the case of p18 antisense-transfected clones, antisense p18 transcripts were detected by RT-PCR (Figure 6a). The results NF-E2p18 activates NF-E2 transcriptional activity in demonstrated that the p18 antisense was transcribed in both Friend erythroleukemia cells 745A and TFP10 clones. As shown in Figure 6B, 745A clone transfected with p18 antisense (745A p18-AS) showed a dra- To determine whether p18 can activate NF-E2 transcriptional matic decrease of erythroid differentiation induced by 2% activity in F-MEL cells, we then performed transient transfec- DMSO treatment (10% of benzidine-positive cells) as com- tion experiments to analyze NF-E2-dependent reporter gene NF-E2p18 and erythroid differentiation C Francastel et al 278 Discussion

F-MEL cells can be induced to erythroid differentiation by treatment with DMSO and other chemical inducers.1,2,34,35 However, some variants of F-MEL cells differ in their response to DMSO-induced differentiation. Globin gene expression requires the integrity of the NF-E2 sites of LCR HS-2. The role of the NF-E2 complex which binds these sites in globin gene expression, and therefore in the induction of F-MEL erythroid differentiation has been hypothesized.36,37 Thus, the absence of the larger p45 subunit of NF-E2 resulted in a drastic decrease of globin expression in one case of Friend leukemia cells.27 We show herein that differentiation-sensitive 745A cells and partially DMSO-resistant TFP10 cells exhibit similar expression of the NF-E2 components, p45 and p18 before DMSO treatment. Moreover, a similar increase of the p45 component of NF-E2 was observed in both cell lines after DMSO treatment. Therefore, the resistance of TFP10 cells to erythroid differentiation cannot be explained by the amount of NF-E2 complex. Recent studies indicate that protein kinase A may regulate the DNA-binding activity of NF-E2 complex during the HMBA-induced differentiation.25 However, our data show a similar increase in NF-E2 DNA-binding activity in both 745A and TFP10 cell lines following DMSO treatment suggesting the involvement of other regulators than NF-E2 in the regulation of globin expression in these particular Friend cell lines. For example, we have previously reported that the presence of c-jun expression in TFP10 cells can decrease their ability to differentiate by the DMSO.31 The smaller p18 subunit of NF-E2 was described for the first time in F-MEL cells. It was reported that NF-E2 p18 is coded by the mafK gene which belongs to the small Maf family proteins lacking transactivation domains.21 Recently Igarashi et al28 reported that overexpression of p18/mafK could induce Figure 6 Antisense NF-E2p18 inhibits DMSO-induced differen- spontaneous erythroid differentiation of a murine erythroleu- tiation of F-MEL cells. (a) RT-PCR analysis of 745A and TFP10 clones kemia cell line. In this report, we provide new insights into transfected with antisense NF-E2p18. Total RNA of parental 745A and the requirement for NF-E2p18/mafK in the induction of F-MEL TFP10 cells and clones stably transfected with NF-E2p18 antisense cell erythroid differentiation. Overexpression of p18 increases vectors (745A p18AS and TFP10 p18AS) was isolated and subjected the sensitivity to DMSO-induced differentiation of both Friend to RT-PCR as described in Materials and methods. Aliquots of the PCR reactions were separated on agarose gels and visualized with ethidium 745A and TFP10 cell lines. Conversely, inhibition of p18 bromide. The vector pcDNA3-p18AS was used as control. (b) expression by antisense p18 transcripts resulted in the inhi- Erythroid differentiation of parental F-MEL cells and clones transfected bition of DMSO-induced F-MEL differentiation. These results with antisense NF-E2p18 in presence of 2% DMSO. Untransfected indicate a good correlation between the level of p18 745A and TFP10 cells and 745A and TFP10 clones stably transfected expression and the ability of F-MEL cells to produce hemoglo- with p18 antisense vector (745A p18-AS and TFP10 p18-AS) were bin and undergo differentiation. They also suggest that grown in presence of 2% DMSO for 5 days at which time terminal differentiation was scored by counting the percentage of benzidine- p18/mafK is critical for DMSO-induction of F-MEL differen- positive cells. tiation because it cannot be compensated by another small maf family protein in antisense p18 clones. It remains to be determined whether NF-E2p18 could also be involved in the regulation of normal erythroid differentiation in vivo. We found an increase of the DNA-binding activity of NF- expression in the presence of NF-E2p18 sense expression vec- E2 in F-MEL clones overexpressing p18 whereas this activity tor. We have studied the effect of p18 expression on a reporter was decreased in clones expressing antisense p18. In contrast construct described by Walters et al26 which contains a globin to recent studies of Igarashi et al,28 we did not observe new TATA box under the control of the ␤-globin tandem NF-E2 complexes when p18/mafK is overexpressed. This discrepancy sites and a binding site for one other erythroid transcriptional with our results could reflect differences between the Friend factor GATA-1 (GATA-NF-E2-hGH). As shown in Figure 7b, cell lines studied or differences between the probes used for NF-E2p18 overexpression resulted in an increase in hGH testing NF-E2 DNA-binding activity. The increase of binding reporter gene expression, in both 745A (five-fold) and TFP10 activity of the NF-E2 complex induced by overexpression of (2.5-fold) cell lines. In contrast, transfection of NIH/3T3 cells p18 is correlated to the induction of globin mRNAs in F-MEL with the same reporter construct showed that NF-E2 transcrip- cells suggesting that the factor NF-E2 is critical for transcriptional activity was reduced by NF-E2p18 expression, this last tion of globin. Our results are in agreement with recent stud- result being in agreement with the work of Igarashi et al.21 ies36 which show the requirement of the NF-E2 p45-p18 heter- These results indicate that overexpression of p18 can activate odimer for high-level globin gene expression in MEL cells. NF-E2 site-dependent transcription in F-MEL cells. p18/MafK homodimers can bind to NF-E2 sites but they NF-E2p18 and erythroid differentiation C Francastel et al 279

Figure 7 Overexpression of NF-E2p18 increases NF-E2 activity in F-MEL cells. (a) NF-E2 DNA-binding activity in F-MEL cells transfected with sense and antisense NF-E2p18. Nuclear extracts were prepared from parental 745A and TFP10 cells and from clones transfected with p18 sense (745A p18S and TFP10 p18S) and antisense (745A p18AS and TFP10 p18AS). Nuclear extract protein (10 ␮g) were incubated with the NF-E2 oligonucleotide probe and gel mobility shifts assays were performed as described in Materials and methods. (b) Effect of NF-E2p18 on NF-E2 transcriptional activity in NIH/3T3 and F-MEL cells. Cells were transfected with a gene reporter construct (GATA-NFE2-hGH) containing GATA and NF-E2 DNA-binding sites upstream of a rabbit ␤-globin TATA box driving expression of the hGH gene. When indicated, GATA- NFE2-hGH was cotransfected with 40 ␮g of plasmid expressing NF-E2p18. Forty-eight hours after electroporation, medium was harvested and was assayed for hGH using an ELISA kit (Eurogenetics) according to the manufacturer’s recommendations. hGH levels in culture medium are indicated. The results consist of the average of three separate experiments and error bars represent standard error (s.e.) of the mean.

repress NF-E2 site-dependent gene expression in non- Acknowledgements erythroid cells, which do not contain p45.21 In contrast, we showed herein that in Friend erythroleukemia cells, NF-E2p18 We thank N Andrews for providing NF-E2p45 and NF-E2p18 increased NF-E2 transcriptional activity suggesting that it can cDNAs and for anti-NF-E2p45 antibody and D Martin for function as an activator of NF-E2 driven gene expresion in GATA-NF-E2-hGH construct. This work was supported in part erythroid cells, expressing NF-E2p45. Since NF-E2p18 homo- by grants from the Association pour la Recherche sur le Can- dimers alone are unable to transactivate NF-E2 sites, this cer (ARC 3035) and from the Association NRB-Vaincre le Can- increase in NF-E2 transcriptional activity induced by NF- cer. C Francastel was supported by grants from the Association E2p18 in F-MEL cells is probably due to an increase in pour la Recherche sur le Cancer. p18/p45 heterodimer formation. Thus the increase of the amount of NF-E2 complex observed in F-MEL cells overexpressing p18/mafK could contribute to activation of NF-E2 References site-dependent globin gene transcription. Finally, experiments including both overexpression and 1 Friend C, Scher W, Holland JG, Sato T. Hemoglobin synthesis in inhibition of selected genes have provided information con- murine virus-induced leukemic cells in vitro: stimulation by dime- cerning which transcription factors play a role in DMSO- thyl sulfoxide. Proc Natl Acad Sci USA 1971; 6: 378–382. induced differentiation of F-MEL cells. Overexpression of 2 Tsiftsoglou AS, Wong W. Molecular and cellular mechanisms of nuclear proto-oncogenes c-myc,c-myb and c-jun inhibits Fri- leukemic hemopoietic cell differentiation: an analysis of the Friend end erythroid differentiation.31,38–40 Similarly the helix-loop- system. Anticancer Res 1985; 5: 81–100. 41 3 Marks PA, Sheffery M, Rifkind RA. Induction of transformed cells helix protein Id inhibits differentiation of MEL cells. In con- to terminal differentiation and the modulation of gene expression. trast SCL, a hematopoietic specific helix–loop–helix protein Cancer Res 1987; 47: 657–666. promotes spontaneous differentiation when transfected into 4 Profous-Juchelka HR, Reuben RC, Marks PA, Rifkind RA. Tran- MEL cells.42 The molecular mechanism by which these differ- scriptional and post-transcriptional regulation of globin gene ent transcription factors, which are not erythroid-specific, accumulation in murine erythroleukemia cells. Mol Cell Biol regulate F-MEL erythroid differentiation remains totally 1983; 3: 229–232. 5 Ramsay RG, Ikeda K, Rifkind RA, Marks PA. Changes in gene unknown. Since we have demonstrated herein a direct role expression associated with induced differentiation of erythroleuke- of p18/mafK in erythroid differentiation by increasing NF-E2 mia: protooncogenes, globin genes, and cell division. Proc Natl activity, it is possible that the interaction between p18 and Acad Sci USA 1986; 83: 6849–6853. other transcription factors may regulate this activity. Recent 6 Orkin SH. Regulation of globin gene expression in erythroid cells. studies indicate that the c-fos protein can inhibit NF-E2 tran- Eur J Biochem 1995; 231: 271–281. 7 Grosveld F, Blom van Assendelft G, Greaves DR, Kollias G. Pos- scriptional activity by forming inactive complexes with small ␤ 43 ition-independent, high-level expression of the human -globin Maf proteins. Analysis of regulators interacting with NF- gene in transgenic mice. Cell 1987; 57: 975–985. E2p18 protein may clarify the mechanism of regulation of 8 Blom van Assendelft G, Hanscombe O, Grosveld F, Greaves DR. MEL differentiation. The ␤-globin dominant control region activates homologous and NF-E2p18 and erythroid differentiation C Francastel et al 280 heterologous promoters in a tissue-specific manner. Cell 1989; 56: 26 Walters M, Martin DIK. Functional erythroid promoters created by 969–977. interaction of the transcriptional factor GATA-1 with CACCC and 9 Sorrentino B, Ney PN, Bodine D, Nienhuis AW. A 46 pair AP-1/NF-E2 elements. Proc Natl Acad Sci USA 1992; 89: enhancer sequence within the locus activating region is required 10444–10448. for induced expression of the gamma-globin gene during erythroid 27 Lu SJ, Rowan S, Bani MR, Ben-David Y. Retroviral integration differentiation. Nucleic Acid Res 1990; 1: 2721–2731. within the Fli-2 locus results in inactivation of the erythroid tran- 10 Caterina JJ, Ryan TM, Pawlik KM, Palmiter RD, Brinster RL, scription factor NF-E2 in Friend erythroleukemias: evidence that Behringer RR, Townes TM. Human ␤-globin locus control region: NF-E2 is essential for globin expression. Proc Natl Acad Sci USA analysis of the 5Ј DNaseI hypersensitive site HS-2 in transgenic 1994; 91: 8398–8402. mice. Proc Natl Acad Sci USA 1991; 88: 1626–1630. 28 Igarashi K, Itoh K, Hayashi N, Nishizawa M, Yamamoto M. Con- 11 Collis P, Antoniou M, Grosveld F. Definition of the minimal ditional expression of the ubiquitous transcription factor mafK requirements within the human ␤-globin gene and the dominant induces erythroleukemia cell differentiation. Proc Natl Acad Sci control region for high level expression. EMBO J 1990; 9: 233– USA 1995; 92: 7445–7449. 240. 29 Robert-Le´zeńe`s J, Meneceur P, Ray D, Moreau-Gachelin F. Pro- 12 Ney PA, Sorrentino BP, McDonagh KT, Nienhuis AW. Tandem tooncogene expression in normal, preleukemic, and leukemic AP-1-binding sites within the human ␤-globin dominant control murine erythroid cells and its relationship to differentiation and region function as an inducible enhancer in erythroid cells. Genes proliferation. Cancer Res 1988; 48: 3972–3976. Dev 1990; 4: 993–1006. 30 Francastel C, Mazouzi Z, Robert-Le´zeńe`s J. Co-induction of c-fos 13 Ney PA, Sorrentino BP, Lowrey CH, Nienhuis AW. Inducibility of and junB during the latent period preceding commitment of Friend the HS II enhancer depends on binding of an erythroid specific erythroleukemia cells to differentiation. Leukemia 1992; 6: 935– nuclear protein. Nucleic Acid Res 1990; 18: 6011–6017. 939. 14 Mignotte V, Wall L, DeBoer E, Grosveld F, Romeo PH. Two tissue- 31 Francastel C, Groisman R, Pfarr CM, Robert-Le´zeńe`s J. Antisense specific factors bind the erythroid promoter of the human porpho- c-jun overcomes a differentiation block in a murine erythroleuke- bilinogen deaminase gene. Nucleic Acid Res 1989; 17: 37–54. mia cell line. Oncogene 1994; 9: 1957–1964. 15 Talbot D, Grosveld F. The 5ЈHS2 of the globin locus control region 32 Dignam JD, Lebovitz RM, Roeder RG. Accurate transcription enhances transcription through the interaction of a multimeric initiation by RNA polymerase II in a soluble extract from isolated complex binding at two functionally distinct NF-E2. EMBO J 1991; mammalian nuclei. Nucleic Acid Res 1983; 11: 1475–1489. 10: 1391–1398. 33 Osborn L, Kunkel S, Nabel GJ. Tumor necrosis factor ␣ and 16 Andrews NC, Erdjument-Bromage H, Davidson MB, Tempst P, interleukin 1 stimulate the human immunodeficiency virus Orkin SH. Erythroid transcription factor NF-E2 is a haematopo- enhancer by activation of the nuclear factor ␬B. Proc Natl Acad ietic-specific basic-leucine zipper protein. Nature 1993; 362: Sci USA 1989; 86: 2336–2340. 722–728. 34 Schaefer A, Dahle M, Radenz G, Steinheider G, Marquardt H. 17 Andrews NC, Kotkow KJ, New PA, Erdjument-Bromage H, Tempst Structure-activity relationship between anthracycline-induced dif- P, Orkin SH. The ubiquitous subunit of erythroid transcription fac- ferentiation and inhibition of glycoprotein synthesis in Friend tor NF-E2 is a small basic-leucine zipper protein related to the v- erythroleukemia cells. Leukemia 1991; 5: 95–100. maf oncogene. Proc Natl Acad Sci USA 1993; 90: 11488–11492. 35 Schaefer A, Dressel A, Lingelbach K, Schmidt CA, Steinheider G, 18 Motohashi H, Igarashi K, Onodera K, Takahashi S, Ohtani H, Marquardt H. Induction of differentiation in Friend-erythroleuke- Nakafuku M, Nishizawa M, Engel JD, Yamamoto M. Mesodermal- mia cells by aclacynomycin A: early transient decrease in c-myc vs neuronal-specific expression of MafK is elicited by different and c-myb mRNA levels. Leukemia 1992; 6: 828–833. promoters. Genes to Cells 1996; 1: 223–238. 36 Kotrow KJ, Orkin SH. Dependence of globin gene expression in 19 Igarashi K, Kataoka K, Itoh K, Hayashi N, Nishizawa M, Yamam- mouse erythroleukemia cells on the NF-E2 heterodimer. Mol Cell oto M. Regulation of transcription by dimerization of erythroid Biol 1995; 15: 4640–4647. factor NF-E2 p45 with small maf proteins. Nature 1994; 367: 37 Trentesaux C, Ngo Nyong M, Aries A, Morceau F, Ronchi A, Otto- 568–572. lenghi S, Jardillier JC, Jeannesson P. Increased expression of 20 Fujiwara KT, Kataoka K, Nishizawa M. Two new members of the GATA-1 and NF-E2 erythroid-specific transcription factors during maf oncogene family, mafK and mafF, encode nuclear b-Zip pro- aclacinomycin-mediated differentiation of human erythroleu- teins lacking putative trans-activator domain. Oncogene 1993; 8: kemic cells. Leukemia 1993; 7: 452–457. 2371–2380. 38 Dmitrovsky E, Kuehl WM, Hollis GF, Kirsch IR, Bender TP, Segal 21 Igarashi K, Itoh K, Motohashi H, Hayashi N, Matuzaki Y, Nakauchi S. Expression of a transfected human c-myc inhibits differentiation H, Nishizawa M, Yamamoto M. Activity and expression of murine of a mouse erythroleukemia cell line. Nature 1986; 322: 748–750. small Maf family protein MafK. J Biol Chem 1995; 270: 7615– 39 Clarke MF, Kukowska-Latallo JF, Westin E, Smith M, Prochownik 7624. EV. Constitutive expression of a c-myb cDNA blocks Friend 22 Strauss EC, Andrews NC, Higgs DR, Orkin SH. In vivo footprinting murine erythroleukemia cell differentiation. Mol Cell Biol 1988; of the human ␣-globin locus upstream regulatory element by 8: 884–892. guanine and adenine ligation polymerase chain reaction. Mol Cell 40 Prochownik EV, Smith MJ, Snyder K, Emeagwali D. Amplified Biol 1992; 12: 2135–2142. expression of three jun family members inhibits erythroleukemia 23 Cox TC, Bawden MJ, Martin A, May BK. Human erythroid 5-amin- differentiation. Blood 1990; 76: 1830–1837. olevulinate synthetase: promoter analysis and identification of an 41 Shoji W, Yamamoto T, Obinata M. The helix–loop–helix protein iron-responsive element in the mRNA. EMBO J 1991; 10: Id inhibits differentiation of murine erythroleukemia cells J Biol 1891–1902. Chem 1994; 269: 5078–5084. 24 Tugores A, Magness ST, Brenner DA. A single promoter directs 42 Aplan PD, Nakahara K, Orkin SH, Kirsch IR. The SCL product: a both housekeeping and erythroid preferential expression of the positive regulator of erythroid differentiation. EMBO J 1992; 11: human ferrochelatase. J Biol Chem 1994; 269; 30789–30797. 4073–4081. 25 Garingo AD, Suhasini M, Andrews NC, Pilz RB. c-AMP-dependent 43 Kataoka K, Igarashi K, Itoh K, Fujiwara KT, Noda M, Yamamoto protein kinase is necessary for increased NF-E2 DNA complex for- M, Nishizawa M. Small Maf proteins heterodimerize with Fos and mation during erythroleukemia cell differentiation. J Biol Chem may act as competitive repressors of the NF-E2 transcription 1995; 270: 9169–9177. factor. Mol Cell Biol 1995; 15: 2180–2190.