Mafb Is an Interaction Partner and Repressor of Ets-1 That Inhibits Erythroid Differentiation

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Cell, Vol. 85, 49–60, April 5, 1996, Copyright 1996 by Cell Press MafB Is an Interaction Partner and Repressor of Ets-1 That Inhibits Erythroid Differentiation

Michael H. Sieweke, Hildegard Tekotte, which differ in their N-terminal transactivation domain. Jonathan Frampton, and Thomas Graf p68ets-1 is of particular interest because it represents European Molecular Biology Laboratory the cellular homolog of the v-Ets oncoprotein, which is Meyerhofstrassel 69117 encoded as a fusion protein with v-Myb by the E26 Heidelberg leukemia virus (Leprince et al., 1983; Nunn et al., 1983). Federal Republic of Germany Within the hematopoietic system, v-Ets appears pre- dominantly to affect the growth and differentiation of erythroid cells. Thus, it transforms erythroid cells and Summary enhances the erythroid transformation potential of v-ErbA and v-Myb (Metz and Graf, 1991, 1992). Ets-1 is Using a yeast one-hybrid screen with a DNA-bound expressed in erythroblasts transformed by the v-erbA Ets-1 protein, we have identified MafB, an AP-1 like and v-erbB-containing avian erythroblastosis virus protein, as a direct interaction partner. MafB is specifi- (Ghysdael, 1986) and in erythroblasts of early yolk sac cally expressed in myelomonocytic cells and binds to of the chick embryo (Que´ va et al., 1993). Together with the DNA-binding domain of Ets-1 via its basic region the finding that Ets-1 is capable of activating promoters or leucine–zipper domain. Furthermore, it represses of erythroid genes (Seth, 1993; J. F. et al., unpublished Ets-1 transactivation of synthetic promoters con- data), these data suggest that it plays a role in early taining Ets binding sites and inhibits Ets-1–mediated erythroid differentiation. transactivation of the transferrin receptor, which is It is unknown whether Ets-1 activity is regulated by known to be essential for erythroid differentiation. Ac- direct interactions with other transcription factors in he- cordingly, overexpression of MafB in an erythroblast matopoietic cells. In other systems, Ets-1 has been cell line down-regulates the endogenous transferrin shown to cooperate with various transcriptional activa- receptor gene and inhibits differentiation without af- tors in regulating gene activity (Ge´ gonne et al., 1993; fecting cell proliferation. These results highlight the Giese et al., 1995), such as with the basic region–leucine importance of inhibitory interactions between tran- zipper (b–Zip) AP-1 family (Wasylyk et al., 1990). In addi- scription factors in regulating lineage-specific gene tion, a number of promoters contain functionally impor- expression. tant AP-1 and Ets–responsive elements in close proxim- ity (Gutman and Wasylyk, 1990; Nerlov et al., 1991; Wasylyk et al., 1991). It has also been shown that the Introduction Drosophila Ets homolog pointed and the AP-1 protein D-Jun both lie at the endpoint of the Ras/MAP kinase The phenotype of a differentiated cell is ultimately deter- signal transduction pathway (Bohmann et al., 1994; mined by the set of its active genes. Since the execution Brunner et al., 1994; O’Neill et al., 1994) and cooperate of a particular gene-expression program is only rarely in the induction of R7 photoreceptor cells (Treier et al., triggered by a single “master gene,” it has been pro- 1995). AP-1–like factors have recently also been impli- posed that this usually requires the combinatorial action cated in erythroid differentiation. Thus, several erythroid of several transcription factors (Ness and Engel, 1994; promoters have been shown to contain functionally rele- Orkin, 1995). It has been assumed that this is due to the vant binding sites with a core AP-1 recognition se- simple additive effect of the factors present, and also quence (Higgs and Wood, 1993). These sites are recog- that it involves protein–protein interactions that can fur- nized by the heterodimeric transcription factor NF-E2, ther modify the activities of the individual partners and which consists of two bZip-type factors, a hematopoi- thus increase the options for activation of diverse ge- etic-specific 45 kDa subunit (Andrews et al., 1993a; Ney netic programs. Such regulatory interactions with other et al., 1993) and a broadly expressed 18 kDa subunit that factors are well described for the Ets proteins, a family of transcription factors containing a characteristic helix– belongs to the AP-1–related Maf family of transcription turn–helix DNA-binding domain (Janknecht and Nord- factors (Andrews et al., 1993b; Igarashi et al., 1994). heim, 1993; Treisman, 1994). For example, complex for- In the present study, we have searched for direct mation of the Ets protein Elk-1/SAP-1 with serum protein interaction partners of Ets-1 that might modify response factor dimers is critical for regulation of the its activity, using a yeast one-hybrid screen. We have c-fos promoter (Dalton and Treisman, 1992); the Ets- identified MafB, a bZip protein of the Maf family, as such related protein GABP␣ forms heterotetramers with a partner. This protein is highly expressed in myelomo- GABP␤ to activate immediate early promoters of HSV-1 nocytic but not in erythroid cells. Its overexpression in (LaMarco et al., 1991); and the association of the Ets erythroblasts inhibits transactivation of the transferrin family member Pu.1/Spi-1 with NF-EM5/Pip1 is impor- receptor gene by Ets-1 and represses erythroid differen- tant for the regulation of immunoglobulin light chain tiation. These results demonstrate a functional interac- enhancers (Eisenbeis et al., 1995). tion between members of the Ets and Maf protein fami- Much less is known about protein partners of Ets-1, lies in the hematopoietic system and suggest that the the founding member of the family. Ets-1 occurs in two repression of Ets-1 by MafB plays a role in the regulation alternatively spliced isoforms, p54ets-1 and p68ets-1, of lineage-specific gene expression. Cell 50

of specifically binding DNA (Lim et al., 1992). To obtain a reporter plasmid, we inserted five multimerized Ets binding sites (EBS) from the polyoma virus enhancer into a minimal cyc-1 promoter controlling a lacZ gene (Figure 1A, map 3). We constructed a cDNA library, as a source of potentially interacting proteins, from the quail cell line QT6 in a yeast expression vector tagged with the transcriptional activation domain from VP16 of HSV (Figure 1A, map 4). We chose QT6 cells because transient transfection experiments suggested that they contain cofactors that can modulate Ets-1 activity (Lim et al., 1992). The screen was performed by transforming the library into a yeast strain containing both the reporter and the Ets⌬TA constructs. To avoid counterselection of colonies in which protein interactions would result in a growth disadvantage, cells were initially plated on glucose medium, where the transcription of the bait and the cDNA clones was repressed until small colonies had formed. These were then shifted to galactose medium to induce expression of the Ets⌬TA and the library constructs (which were both controlled by gal-responsive promoters). Of 1.2 ϫ 106 primary transformants screened, 58 clones were scored positive after two re- screens. In subsequent plasmid loss experiments, we identified 5 clones that were dependent on both the Ets-1 construct and the library plasmid for reporter activity. One of these clones (99) was selected for further characterization. Analysis of the other clones is in prog- ress and will be published elsewhere.

An Ets-1 Interacting Clone Encodes a bZip Protein of the Maf Family To confirm that the activation of the reporter construct truly required the interaction of clone 99 with Ets-1, we Figure 1. Interaction of Ets-1 and MafB in a Yeast One-Hybrid System performed control transformations of the original yeast (A) Maps of the constructs used. 1) Coding region of p68ets-1.2) strain with combinations of either the Ets bait, the EBS Galactose-inducible bait construct consisting of a truncated Ets-1 reporter, and the VP16-tagged cDNA, or the correspond- (Ets⌬TA), lacking transactivation domains. DBD, DNA binding do- ing parental control constructs. Reporter activity re- main. 3) Reporter plasmid with the lacZ gene under the control of quired the presence of Ets-1, EBS in the reporter con- 5 multimerized EBS. 4) Galactose-inducible cDNA library expression struct, and VP16-tagged clone 99 (Figure 1B, lane 1). construct tagged with VP16 (closed box). HIS3, TRP1, URA3 indicate Removal of either component resulted in loss of reporter auxotroph markers and (2 ␮, CEN) replication origins utilized in the plasmids. activity (Figure 1B, lanes 2–4). (B) Reporter activity in response to Ets-1 interaction with clone 99 Sequence analysis revealed that clone 99 encoded (MafB). A reporter construct with or without EBS was coexpressed an in-frame fusion of VP16 with the major part of a bZip with Ets⌬TA and VP16-tagged cDNA of clone 99 (MafB) or the corre- protein of the Maf family. cDNA sequences encoding sponding empty vector in the indicated combinations. ␤-galactosi- the missing 28 N-terminal amino acids and the 5Ј un- dase activity was measured in a quantitative enzyme assay. translated region of the gene were cloned by a nested polymerase chain reaction (PCR) approach from the same cDNA library. Sequence comparisons showed al- Results most complete identity to chicken MafB (Kataoka et al., 1994) and high homology to the product of the murine Identification of cDNA Clones Encoding kreisler gene (Cordes and Barsh, 1994) with a 99.0% Ets-1 Binding Proteins with a Yeast and 84.8% sequence identity at the amino acid level One-Hybrid Screen and a 97.7% and 80.9 % identity at the nucleotide level, We designed a yeast one-hybrid screen with an Ets-1 respectively. bait that, unlike those used in standard two-hybrid ap- proaches, does not involve a fusion to a heterologous DNA-binding domain, because we reasoned that a DNA- Ets-1 and MafB Interact in Solution bound Ets-1 molecule would more faithfully detect au- and in Vertebrate Cells thentic interactions with other transcription factors. As To determine whether Ets-1 also directly interacts with a bait, we constructed an Ets-1 molecule of p68ets-1 lack- MafB in solution, a glutathione S-transferase (GST) pull- ing the two transactivation domains (Schneikert et al., down assay was employed. For this purpose, a fusion 1992; Ets⌬TA; Figure 1A, maps 1, 2) that was still capable of Ets⌬TA with the C-terminus of GST was immobilized MafB Is An Interaction Partner and Repressor of Ets-1 51

Figure 3. Mapping of the Interaction Sites in Ets-1 and MafB (A) Deletion analysis of Ets-1. N- and C-terminal deletion mutants of Ets⌬TA fused to GST were tested for association with in vitro translated 35S-labeled MafB and analyzed by SDS-PAGE. Left, sche- matic drawing of constructs. The first and last amino acids from chicken p68ets-1 present in the deletions are indicated. Construct 106 corresponds to GST-Ets⌬TA. Right, Coomassie blue staining and Figure 2. Interaction of Ets-1 and MafB in Solution and in Avian autoradiography of the same SDS–PAGE gel. Cells (B) Deletion analysis of MafB. N- and C-terminal truncations of MafB (A) Interaction in solution. In vitro translated [35S]methionine–labeled were tested for association with GST–Ets⌬TA (construct 106) or p68ets-1 (lane 2) or MafB (lane 3) were incubated with affinity matrix– GST control (construct 109) as described above. Molecular mass bound GST (lanes 4, 8), GST–Ets⌬TA (lanes 5 and 9), GST–Fynunique markers (in kDa) are indicated on the right. domain (lane 6), or GST–Src SH3 domain (lane 7), washed, resus- pended in SDS sample buffer, and analyzed by SDS-PAGE. Coo- massie blue staining and autoradiography of the same gel are shown. Lane 1, unprogrammed reticulocyte lysate labeled with [35S] To demonstrate an interaction in avian cells, we methionine. Molecular mass markers (in kDa) are indicated on the initially attempted to coimmunoprecipitate full-length right. MafB and Ets-1. However, all of MafB and most of Ets-1 (B) Interaction in vivo. QT6 cells were transfected with expression proteins cofractionated with the insoluble fraction of constructs for His6-Ets/118 and HA-MafB⌬N (lane 1) or His6-Ets/ nuclear extracts, suggesting a tight association with 118 (lane 2) or HA-MafB⌬N (lane 3) alone. His6-Ets/118 was isolated chromatin. We therefore used deletion mutants of Ets- from [35S]methionine-labeled cell lysates on a Ni2ϩ resin. Complexed proteins were eluted with 1 M NaCl and, after dilution to physiologi- 1 (deletion #118) and MafB (MafB⌬N) that are extractable cal salt concentrations, HA-MafB⌬N was reprecipitated with an ␣HA under mild conditions and can still bind to each other antibody. Molecular mass markers (in kDa) are indicated on the (Figure 3). For efficient precipitation and detection, the right. Ets-1 mutant was tagged with a His6-stretch and the MafB mutant with an influenza virus hemagglutinin epi- tope (HA). QT6 cells were cotransfected with these con- on a glutathione affinity matrix and incubated with in structs and labeled with [35S]methionine. After mild de- vitro translated [35S]methionine-labeled MafB. As shown tergent lysis and sonication, cell lysates were incubated in Figure 2A, lane 9, MafB bound to GST-Ets⌬TA. In with a Ni2ϩ affinity matrix to immobilize the His6-tagged contrast, GST alone or GST fused to other protein– Ets-1. Proteins complexed with Ets-1 were then eluted protein interaction domains, such as the c-Src SH3 do- with 1 M NaCl, reprecipitated with an ␣-HA antibody main or the Fyn unique domain, failed to retain MafB and analyzed by SDS–polyacrylamide gel electrophore- (Figure 2A, lanes 6–8). Likewise, in vitro translated full- sis (SDS–PAGE). As shown in Figure 2B, MafB could be length Ets-1 did not bind to GST-Ets⌬TA (Figure 2A, reprecipitated when the cells had been transfected with lanes 4 and 5). both Ets-1 and MafB constructs (lane 1), but not when Cell 52

they had been transfected with either of them alone (lanes 2 and 3). Identical results were obtained when MafB was detected by Western blotting of the Ni2ϩ matrix-bound material (data not shown).

Ets-1 and MafB Interact via Their DNA Binding Domains To localize the interaction site on Ets-1, we generated GST fusion constructs of several Ets deletion mutants and tested them for binding of in vitro translated MafB (Figure 3A). Whereas the C-terminal deletions abolished interaction (constructs 116 and 117), the N-terminal deletion of amino acids 282–377 (construct 118) showed full binding activity, indicating that the exon VII domain, Figure 4. Tissue Distributionof mafB and ets-1 Expressionas Deter- encoded by this sequence, is dispensable for the inter- mined by Northern Blotting action. However, further deletion into the DNA binding (A) mafB expression in tissues from a 3-week-old chick. Lane 1, domain of Ets-1 (constructs 119 and 120) abolished in- bone marrow; lane 2, brain; lane 3, bursa; lane 4, intestines; lane 5, teraction. The results of the biochemical assay could heart; lane 6, liver; lane 7, skeletal muscle; lane 8, spleen. also be confirmed in yeast interaction assays. While (B) mafB expression in avian cell lines. Lane 1, QT6 fibroblasts; lane amino acids 282–372 of Ets-1 (exon VII) fused to the 2, RPL-12 B cells; lane 3, NPB-4 T cells; lane 4, MSB-1 T-cells; lane 5, HD3 erythroblasts; lane 6, HD37 erythroblasts; lane 7, HD50 bacterial LexA DNA-binding domain failed to bind the multipotent progenitors; lane 8, HD50–3DA eosinophils; lane 9, VP16–MafB fusion (as monitored by the activation of a HD50–4/8 eosinophils; lane 10, HD13 promyelocytes; lane 11, Lex operon reporter construct), the Ets-1 DNA-binding BM2C2 monoblasts; lane 12, HD11 macrophages. domain alone (amino acids 377–485) did (data not (C) mafB expression in transformed primary myelomonocytic cells: shown). wild-type E26-transformed cells at 37ЊC (lane 1) and at 42ЊC (lane We also further localized the interaction site on MafB 2); ts21E26-transformed myeloblasts at 37ЊC (lane 3) and shifted for 48 hr to 42ЊC (lane 4). by testing in vitro translated deletion mutants for their (D) mafB expression in primary normal chicken cell cultures: lane ability to bind to GST-Ets⌬TA. A deletion of the complete 1, chicken embryo fibroblasts passaged five times; lane 2, peritoneal C-terminal DNA-binding and leucine–zipper domain macrophages; lane 3, bone marrow–derived macrophages; lanes (MafB⌬C) completely abolished the interaction. In con- 4–8, Percoll gradient fractions of bone marrow from a 3-week-old trast, the reciprocal N-terminal deletion upstream of the chick (lane 4 contains cells of a density of approximately 0.6 g/cm3 3 bZip region (MafB⌬N) did not diminish binding to Ets-1 and lane 8, cells of 1.0 g/cm ). (E) ets-1 Expression in myelomonocytic cells; lane 1, BM2C2 mo- (Figure 3B). These results indicate that the basic region– noblasts; lane 2, primary peritoneal macrophages; lane 3, bone mar- leucine zipper domain of MafB interacts with the DNA row–derived macrophages. Hybridization with GAPDH (a, b, d, e) or binding domain of Ets-1. 28S (c) is shown in the lower panels.

MafB Is Specifically Expressed expression increased dramatically after temperature in Myelomonocytic Cells shift. The specific expression of mafB in myelomono- To uncover the functional consequences of the interac- cytic cells was also confirmed in normal cells. Both peri- tion of MafB with Ets-1, we first analyzed the expression toneal and bone marrow–derived primary chicken mac- pattern of mafB mRNA in various chicken tissues. We rophages strongly expressed mafB (Figure 4D, lanes 2 found high expression levels in bursa, gut, liver, and and 3). In addition, mafB message was detected in den- spleen; intermediate levels in bone marrow; low levels sity gradient–separated bone marrow fractions that in brain; and none in skeletal muscle and heart (Figure were enriched for myelomonocytic cells (Figure 4D, 4A). Whereas mafB was expressed in the QT6 cell line lanes 6 and 7). Finally, MafB protein could also be de- (Figure 4B, lane 1), it could not be detected in primary tected in HD11 macrophages by Western blotting using chicken embryo fibroblasts (CEF; Figure 4D, lane 1). We cross reacting anti c-Maf antibodies (data not shown). also analyzed several chicken cell lines representing Since Ets-1 expression in myelomonocytic cells has various hematopoietic lineages. Whereas mafB mes- not yet been reported, we analyzed whether Ets-1 was sage was found to be highly expressed in the monoblast co-expressed with MafB in these cells. As shown in BM2C2 and the macrophage HD11 cell line, it could not Figure 4E, Ets-1 is indeed expressed in both the mo- be detected in B cell, T cell, erythroblast, multipotent noblast cell line BM2C2 and in peritoneal and bone mar- progenitor, eosinophilic, or promyelocytic cell lines (Fig- row–derived primary macrophages. ure 4B). Weak expression was also found in primary E26-transformed myeloblasts (Figure 4C, lanes 1–3). To MafB Inhibits Transactivation by Ets-1 determine whether changes in expression levels of mafB To elucidate whether MafB has an effect on the trans- occur during myelomonocytic differentiation, we ana- activation capacity of Ets-1, we performed transient co- lyzed myeloblasts transformed by a mutant E26 virus transfection assays in QT6 cells, using a reporter with (ts21E26) carrying a temperature-sensitive allele of the five multimerized Ets binding sites and an expression v-myb gene. These cells can be induced to differentiate construct of p68ets-1 driven by a cytomegalovirus (CMV) into macrophages by shift to the nonpermissive temper- enhancer–promoter. MafB repressed the transactiva- ature (Beug et al., 1984). As shown in Figure 4c, mafB tion by Ets-1 but had no effect on the basal activity of MafB Is An Interaction Partner and Repressor of Ets-1 53

Figure 5. Repression of Ets-1 Transactivation Potential by MafB (A) Effect of MafB on Ets-1 transactivation in avian cells. QT6 cells were cotransfected with a luciferase reporter construct containing five multimerized EBS together with expression constructs for p68ets-1 (plus, 1 ␮g) or MafB (plus, 1 ␮g), as indicated. Luciferase activities are expressed as percentage of transactivation obtained with Ets-1 only. Assays were performed in triplicate and normalized to ␤-galactosidase activity from a cotransfected LTR–lacZ vector. (B) Effect of MafB on Ets-1 transactivation in yeast. S. cerevisiae was cotransfected with a lacZ reporter plasmid containing three multimerized EBS together with p68ets-1 or MafB (without VP16 tag) expression constructs as indicated. ␤-galactosidase activities were expressed as percentage of transactivation obtained with Ets-1 only. Assays were performed in triplicate and normalized to cell number. (C) Effect of MafB on v-Myb transactivation in avian cells. QT6 cells were cotransfected with a luciferase reporter construct containing three Myb binding sites (MBS) together with expression constructs for v-Myb (plus, 1 ␮g) and MafB (plus, 1 ␮g) as indicated. Luciferase activities are expressed as percentage of transactivation obtained with v-Myb only. Assays were performed intriplicate and normalized to ␤-galactosidase activity from a cotransfected LTR–lacZ vector. Bars indicate standard error of the mean. the reporter (Figure 5A). This effect was not due to an together with combinations of expression constructs for influence on the CMV enhancer–promoter, since a lucif- p68ets-1 and MafB. As shown in Figure 6A, MafB almost erase reporter driven by this promoter was not affected completely suppressed the activation by Ets-1 but had by MafB cotransfection and no changes in Ets-1 protein no effect on the basal activity of the TfR construct. MafB levels were observed by Western blotting (data not also repressed the activity of endogenous Ets-1 on the shown). A repressiveeffect of MafB on Ets-1transactiva- TfR promoter. Transfection of increasing amounts of tion was also observed in yeast. The activation of an MafB expression plasmid into the Ets-1-expressing EBS reporter plasmid by p68ets-1 was repressed 3- to (Ghysdael et al., 1986) erythroblast cell line HD3 re- 4-fold by MafB coexpression (Figure 5B). Furthermore, pressed a cotransfected TfR reporter in a dose-depen- the repression by MafB was not due to a general dent manner (Figure 6B). A promoter with a mutation in repressive effect on transcriptional activation, since the Ets binding site that abolished Ets-1 DNA binding MafB showed no effect on transactivation by v-Myb (data not shown) had a dramatically reduced activity in (Figure 5C). HD3 cells. This basal activity was not further repressible by MafB, emphasizing the Ets-1-specific effect of MafB MafB Represses the Transactivation of the (Figure 6B). Finally, to investigate whether MafB also Transferrin Receptor Gene by Ets-1 repressed the endogenous transferrin receptor gene, Ets-1 is expressed in early erythroid cells (Ghysdael et we stably transfected HD3 erythroblasts with a MafB/ al., 1986; Que´ va et al., 1993) and activates erythroid neoR expression plasmid followed by G418 selection. As genes such as the transferrin receptor (J. F. etal., unpub- shown in Figure 6C, four randomly chosen HD3 clones lished data; Seth et al., 1993). To investigate whether overexpressing MafB showed approximately a 2.5-fold the inhibitory effect of MafB on Ets-1 activity plays a lower level of TfR mRNA than four randomly chosen role in the repression of erythroid genes, we studied the control clones (transfected with a neoR control vector). transferrin receptor (TfR). This protein is expressed at This was also reflected in reduced cell–surface immuno- very high levels in erythroblasts to satisfy the need for fluorescence staining of TfR (data not shown). iron in hemoglobin synthesis (Chan et al., 1989; Horton, 1983; Hu et al., 1977; Iacopetta et al., 1982; Sieff et al., Overexpression of MafB in Erythroblasts 1982) but only at low levels in myelomonocytic cells Suppresses Erythroid Differentiation (Omary et al., 1980; Sieff et al., 1982). To test whether It has been demonstrated that interfering with heme MafB affects the transactivation of the TfR by Ets-1, we synthesis inhibits erythroid differentiation. Thus, inhibi- transfected QT6 cells with a luciferase reporter gene tion of transferrin receptor recycling blocks the differen- driven by a Ϫ117 to ϩ55 fragment of the TfR promoter tiation of both primary and established erythroblasts Cell 54

Figure 6. Effect of MafB on the Potential of Ets-1 to Transactivate the TfR Gene (A) Effect of MafB on exogenous Ets-1 activity on a TfR promoter construct. QT6 cells were cotransfected with a luciferase reporter gene driven by the TfR promoter (-137 to ϩ55) together with expression constructs for p68ets1 (plus, 0.5 ␮g) and MafB (plus, 0.5 ␮g) as indicated. Luciferase activities are expressed as percentage of transactivation obtained with Ets-1 only. Assays were performed in triplicate and normalized to ␤-galactosidase activity from a cotransfected LTR–lacZ vector. (B) Effect of MafB on a TfR promoter construct in erythroblasts. HD3 erythroblasts were cotransfected with a luciferase reporter construct driven by a minimal TK promoter, a TfR promoter (TfR wild-type), or a TfR promoter with a 3 nt substitution in its EBS (TfR mut) together with increasing amounts (plus, 0.5 ␮g; double plus 1 ␮g) of MafB expression plasmid. Activities are expressed as percentage of values obtained with TfR only. Assays were performed in triplicate and normalized to ␤-galactosidase activity obtained with a cotransfected LTR–lacZ vector. Bars indicate standard error of the mean. (C) Transferrin receptor mRNA levels of HD3 control clones compared with clones stably expressing MafB. Total RNA (20 ␮g) from each of four clones were subjected to Northern analysis with a TfR probe, and the bands obtained were quantified with an image analyzer. The band intensities were normalized to the band intensities from a reprobing of the same gel with GAPDH. transformed by a ts mutant of the avian erythroblastosis domain of Ets-1. MafB is specifically expressed in the virus (Killisch et al., 1992; Schmidt et al., 1986). There- myelomonocytic lineage of the hematopoietic system fore, we examined whether the observed MafB-medi- and represses Ets-1–mediated transactivation of the ated repression of Ets-1 activity on the TfR resulted in transferrin receptor gene. Consistent with the known an inhibition of differentiation. As shown in Figure 7A, requirement for transferrin receptor activity for erythroid expression of both ␣- and ␤-globin was dramatically differentiation, overexpression of MafB in erythroblasts reduced in the MafB–HD3 clones compared with the inhibits their differentiation into erythrocytes. These vector-transfected control clones, suggesting that MafB results raise the possibility that MafB represses Ets-1– indeed represses erythroid differentiation. To explore responsive erythroid genes in the myeloid lineage. this possibility further, HD3 cells were induced to differentiate terminally by shifting them from 35ЊC–42ЊC, the nonpermissive temperature (Beug et al., 1982a, 1982b; Mechanism of Ets-1 Repression by MafB Graf et al., 1978). As shown in Figure 7B, in the five Several mechanisms have beenimplicated in the repres- control clones tested, more than 60% of the cells differ- sion of transcriptional activation (Johnson, 1995): first, entiated into hemoglobin-positive cells within 80 hr at interference with DNA binding of an activator protein; 42ЊC. By contrast, the five MafB clones tested showed second, interference with the general transcriptional delayed kinetics of hemoglobin expression and exhib- machinery; and third, interference with the activator ited only 30% positive cells after 80 hr of temperature function of a DNA-bound activator. The first mechanism shift. This was reflected in the reduced levels and de- can be excluded, since activation of the reporter gene layed kinetics of ␣- and ␤-globin expression in two ran- in the yeast one-hybrid assay requires DNA binding of domly selected MafB clones compared with two control Ets-1. MafB also does not act on the general transcrip- clones (Figure 7C). tion machinery, since it does not ablate basal promoter Since MafB has been reportedto transform fibroblasts activity or affect the transactivation by v-Myb. The most (Kataoka et al., 1994), the inhibition of erythroid differen- plausible mechanism, therefore, is that the activator tiation by MafB could simply have been due to an in- function of DNA-bound Ets-1 is masked by the interac- creased proliferative rate of the transfected cells. This tion with MafB. explanation could be excluded, however, since no differ- DNA binding of MafB, on the other hand, does not ence in the growth rate between MafB-expressing HD3 appear to be necessary for the interaction with Ets-1 or clones and the control clones was found (Figure 7D). for its repressor function, as demonstrated by the in vitro interaction assays and the absence of Maf recognition Discussion elements in the studied promoters. However, the possibility remains that MafB would also repress Ets-1 on a In the present study, we have identified MafB as a direct promoter where both proteins bind to DNA. interaction partner of Ets-1. This interaction occurs be- Besides the repressor function demonstrated in this tween the bZip domain of MafB and the DNA-binding study, MafB has also been shown to activate synthetic MafB Is An Interaction Partner and Repressor of Ets-1 55

Figure 7. Effect of MafB on Erythroid Cell Differentiation and Growth (A) Northern blot analysis of ␣- and ␤-globin gene expression in control and MafB-expressing HD3 clones grown at 35ЊC. Clone numbers are indicated. (B) Hemoglobin-positive cells in control or MafB-expressing HD3 clones at different times after shift from 35ЊC–42ЊC (closed symbols) or maintained at 35ЊC (open symbols). Circles, MafB clones; triangles, control clones. Each point represents the average of five clones measured at various timepoints after temperature shift. Bars indicate standard error of the mean. (C) Time course of globin gene expression after temperature shift from 35ЊC–42ЊC as detected by Northern blotting. Times after shift (hr) are indicated on top. (D) Growth curve of control and MafB-expressing clones. Each point represents the average cumulative cell numbers of the same five clones as in (B), maintained at 35ЊCor42ЊC. Bars indicate standard error of the mean. reporter constructs weakly with multimerized Maf re- however, indicate that AP-1 factors do not always syner- cognition elements (Kataoka et al., 1994). Although a gize with Ets proteins but can also be repressive. The transactivation potential on authentic promoters still re- inhibitory action of MafB represents a novel strategy for mains to be determined, it is conceivable that MafB the regulation of Ets activity within the Ets/AP-1 protein may have a dual function as a repressor or activator network, which also differs from the repression of the depending on the promotercontext, the interaction part- Ets protein pointed by the dominant negative Ets factor ner, or both; as is well documented for several other yan in R7 cells of the Drosophila eye (Brunner et al., transcription factors (Bengal et al., 1992; Ponta et al., 1994; Lai and Rubin, 1992; O’Neill et al., 1994). 1992; Roberts and Green, 1995). Mechanism of MafB-Induced Inhibition Implications for the Cross-Talk Between of Erythroid Differentiation Ets and AP-1 Family Proteins Our results indicate that the observed inhibition of ery- As reviewed in the Introduction, AP-1 and Ets proteins throid differentiation by MafB is a consequence of the functionally cooperate according to genetic and molec- repression of Ets-1 activity. More trivial explanations, ular criteria in several systems. In this study, we have such as induction of proliferation or a general block of now demonstrated a direct physical interaction between differentiation, can be excluded, since MafB did not members of the two transcription factor families. Inde- affect growth of the HD3 erythroblasts and MafB expres- pendently, Bassuk et al. (1995) have also found a direct sion increases dramatically during differentiation of association of Ets and Jun family proteins. Our results, ts21E26-transformed myeloblasts into macrophages. Cell 56

Therefore, the repressive effect on the TfR gene offers Potential Roles of MafB in Lineage-Specific the simplest explanation for the inhibition of erythroid Gene Expression differentiation by MafB. While the metabolism of all euk- In the hematopoietic system, MafB is expressed specifi- aryotic cells requires iron, hemoglobin synthesis in ery- cally in macrophages and their precursors, in contrast throid cells accounts for a vastly increased need for iron with all other Maf family members described so far. The and high levels of transferrin receptor expression (Chan observed expression in other tissues such as bursa, gut, et al., 1989; Horton, 1983; Hu et al., 1977; Iacopetta et liver, and spleen may, therefore, to a large extent be al., 1982; Sieff et al., 1982). These high levels are essen- a consequence of the macrophages present in these tial for erythroiddifferentiation, since a 2-to 3-fold inhibi- organs. This assumption is supported by in situ hybrid- tion of iron uptake with ␣-TfR antibodies has been shown ization experiments in chick embryos, which showed to be sufficient to block the differentiation of HD3 eryth- specific expression of MafB in resident tissue macro- roblasts (Killisch et al., 1992; Schmidt et al., 1986). By phages of several organs (M. H. S. and A. Eichmann, contrast, antibodies that completely block receptor unpublished data). function are cytotoxic, probably reflecting the need for Our results raise the possibility that MafB contributes basic levels of iron in general metabolism. These obser- to the establishment and maintenance of the myelomo- vations are consistent with the notion that the observed nocytic phenotype by preventing erythroid-specific 2- to 3-fold repression of the endogenous TfR gene by gene expression. This notion is supported by several MafB in HD3 cells inhibits their differentiation. Further- observations. First, myelomonocytic cells retain a high more, the finding that the repression of the TfR promoter degree of plasticity in their differentiation potential. is dependent on Ets-1 binding sites strongly suggests Thus, we have recently shown that ectopic expression that the observed down-regulation of the endogenous of GATA-1 in myelomonocytic cell lines can reactivate TfR and the subsequent inhibition of differentiation are the erythroid differentiation program (Kulessa et al., a direct consequence of the binding and repression of 1995). This indicates that erythroid-specific genes are Ets-1 by MafB. still accessible to transcriptional activators and sug- Erythroid differentiation can also be blocked by pre- gests a need to keep them repressed during normal venting heme production with specific inhibitors of por- myelomonocytic differentiation. Second, while Ets-1 is phyrin biosynthesis (Schmidt et al., 1986).In this context, expressed both in erythroid and myeloid cells, MafB it is interesting that MafB can also repress Ets-1 trans- expression is limited to the latter, where it could act as activation of the erythroid-specific promoter of porpho- safeguard against inappropriate Ets-1 activity. Third, in bilinogen deaminase (PBGD), an enzyme in this pathway contrast with its high expression in erythroid cells, TfR (unpublished data). Therefore, it is conceivable that the expression decreases during myelomonocytic differen- repression of other genes required for hemoglobin syn- tiation (Omary et al., 1980; Sieff et al., 1982; J. F. et al., thesis, such as PBGD, may also contribute to the ob- unpublished data), correlating with an increase of MafB served inhibition of erythroid differentiation by MafB. expression. Our results support a model in which lineage choice in Influence on NF-E2 Activity the hematopoietic system is determined by a reciprocal The activity of several erythroid-specific promoters is relationship between the activation of a particular path- dependent on binding sites for NF-E2 (Higgs and Wood, way and the suppression of another. The differentiation 1993). Since the small subunit of this heterodimeric tran- along the myeloid versus the erythroid lineage thus may scription factor is a member of the Maf family (Andrews not only require a distinct combination of transcriptional et al., 1993b; Igarashi et al., 1994), one might speculate activators but also the activity of an appropriate set of that direct interference of MafB with NF-E2 activity could inhibitors, one of which could be MafB. contribute to the inhibition of erythroid differentiation. However, several lines of evidence argue against such Experimental Procedures a mechanism. MafB neither binds to NF-E2 sites (Igara- shi et al., 1994) nor does it heterodimerize with the small Origin of Cell Lines Mafs (Kataoka et al., 1994) or the large NF-E2 subunit, The QT6 cell line was derived from a chemically induced quail tumor the cap ‘n’ collar (cnc) family member p45 (Igarashi et and has been described as fibroblastic (Moscovici et al., 1977). The al., 1994). It also does not bind to Ech (M.H.S., H.T., J.F., HD3 cell line was derived from chick erythroblasts transformed with and T.G., unpublished data), another recently identified a temperature-sensitive mutant of avian erythroblastosis virus with chicken p45–related protein of the cnc family (Itoh et a lesion in v-erbB (Beug et al., 1982a). The origin of the RPL12, REV- transformed NPB-4, MSB-1, HD37, HD50, BM2C2, and HD11 cell al., 1995). Furthermore, whereas the small Maf proteins, lines is described in McNagny et al. (1992). The origin of the HD13 supposedly acting as homodimers, strongly repress NF- and HD50–3DA cell lines is described in Kulessa et al. (1995). The E2 site–containing reporters (Igarashi et al., 1994; Ka- HD50–4/8 eosinophil line has not been published. taoka et al., 1995), overexpression of the small Maf protein MafK does not interfere with erythroid differentiation Construction of the cDNA Library (Igarashi et al., 1995). This indicates that even the over- From a subconfluent culture of QT6 cells, poly(A)ϩ RNA was isolated expression of a Maf protein that is supposed to repress using the protocol of Chomczynsky and Sacchi (1987) followed by NF-E2 sites does not necessarily lead to the inhibition oligo (dT)-cellulose (Pharmacia type 7) affinity purification. The mRNA was reverse transcribed with superscriptTMRNAseHϪ reverse of erythroid differentiation. Taken together, these obser- transcriptase (BRL) using an oligo (dT) primer. The cDNA was in- vations strongly suggest that direct interference with serted into the yeast expression plasmid SD10 (Dalton and Treis- NF-E2 activity cannot account for the inhibitory effect man, 1992) with nonpalindromic BstXI adapters (InVitrogen) and of MafB on erythroid differentiation. electroporated into Escherichia coli strain INVaFЈ(InVitrogen). The MafB Is An Interaction Partner and Repressor of Ets-1 57

library had a complexity of 2 ϫ 106, and the insert size ranged from were transfected by the lipofectamin method (BRL) with 10 ␮lof 0.4–5 kb, with an average of 1.6 kb. reagent and 15 min incubation time using 0.2 ␮g reporter and the indicated amount of expression plasmids per sample. Cell lysates Construction of the Yeast Ets Reporter Strain were prepared 48–72 hr after transfection and assayed for luciferase and Design of the Interaction Screen activity, as described in deWet et al. (1987). Transfection efficiency The reporter construct was generated by inserting five head-to-tail was normalized by assaying for ␤-galactosidase activity expressed ligated copies of the oligo 5Ј-TCGAGCAGGAAGTTTCG-3Ј, which from 0.5 ␮g of cotransfected RSV–␤-galactosidase plasmid contains the PEA3 Ets binding site from the polyoma virus enhancer (Bonnerot et al., 1987), as described in Herbomel et al. (1984). Each (Martin et al., 1988) into the XhoI site of pLG670Z (Guarente and data point was obtained by averagingof triplicate samples, and each Ptashne, 1981). The plasmid was converted from uracil- to histidine- figure shown is representative for a set of at least two independent conferring auxotrophy by inserting a BamHI- excised HIS3 cassette experiments. MafB was expressed from the avian erythroblastosis from YDp-H (Berben et al., 1991) into the blunted ApaI site in the virus LTR of the pMI3 vector (Introna et al., 1990) and Ets-1 from URA3 marker. The control reporter plasmid contained an equal- the CMV enhancer–promoter in the pCRNCM vector (Lim et al., length insert lacking Ets binding sites. The bait plasmid was con- 1992). The synthetic Ets responsive reporter contained five multi- structed by inserting a PCR-generated fragment coding for amino merized PEA3 elements (Martin et al., 1988) in front of a minimal acids 282–485 of chicken p68ets-1 between the BamHI and EcoRI thymidine-kinase promoter (Lim et al., 1992). A Ϫ117 to ϩ55 frag- sites of the galactose-inducible expression vector pSD.04a (Dalton ment of the human TfR promoter (Owen and Ku¨ hn, 1987) was cloned and Treisman, 1992). Since the ATG from the yeast cyc-1 gene is into the luciferase reporter plasmid pXP2 (deWet et al., 1987). The used in this plasmid, the amino acid sequence MTGS is fused to Ets binding site at Ϫ72 to Ϫ79 in the TfR promoter was mutated the amino terminus of the Ets-1 sequence. The two plasmids were from 5Ј-CAGGAAGT-3Ј to 5Ј-CAGTACTT-3Ј. transformed into the Saccharomyces cerevisiae strain W303–1A (MATa, ho, his3–11,15; trp1–1; ade2–1; leu2–3,112; ura3; can1–100) Protein Interaction Assays in Solution and maintained by selection on SD HisϪ, TrpϪ medium. The library Ets⌬TA or its derivatives were cloned into the GST fusion vector was transformed into the strain by a lithium acetate method as pGEX-2T (Smith, 1993) using available restriction sites or PCR-medi- described by Gietz and Schiestl (1991) with the modification of add- ated strategies. The amino acid junctions of the GST vector with ing 10% dimethyl sulfoxide before heat shock. Cells were plated on the Ets sequence (in bold type) are: constructs 106/116, GSPHMGR; nitrocellulose filters on glucose plates lacking histidine, tryptophan, construct 117, GSPHKFSRG; construct 118, GSPHMLSGSMGPI; and uracil. After 48 hr, when small colonies appeared, the filters construct 119, GSPHKLS; construct 120, GSPHMLSGSSLL. The were transferred to selective plates containing 2% galactose, 40 ␮g/ GST constructs were transformed into E. coli strain Xl-1 Blue (Stra- ml X-Gal and 100 mM Na-phosphate buffer (pH 7.0). Blue colonies tagene). Bacteria were grown to an OD of 0.6–0.8 and shifted to appearing between 36–96 hr after galactose induction were re- 600 30ЊC for 3.5 hr after induction of protein expression with 0.1 mM streaked on selective X-Gal–containing glucose and galactose IPTG (Sigma). To purify GST proteins, we prepared cell lysates as plates. Only galactose-dependent blue colonies were analyzed fur- described in Frangioni and Neel (1993) and Smith (1993) and incu- ther and grown for 48 hr in full YPD medium and then replica-plated bated them with glutathione sepharose 4B resin (Pharmacia) for 1 hr on appropriate selective plates to identify colonies that had lost at 4ЊC, followed by washing with phosphate-buffered saline, 0.05% either the bait(TRP1 marker) or library plasmid (URA3 marker). These Triton X-100 plus protease inhibitor mix (phenylmethylsulfonyl fluo- clones were restreaked on X-Gal indicator plates to eliminate the ride, leupeptin, aprotinin, soy bean trypsin inhibitor). In vitro trans- clones not dependent on the library or bait construct for reporter lated proteins were generated with a rabbit reticulocyte in vitro activity. Plasmid DNA was isolated from the remaining clones by a transcription/translation system (TNT [Promega]) using T7 RNA rapid purification protocol (Hoffman andWinston, 1987) and retrans- polymerase and labeling with 100 ␮Ci [35S]methionine/50 ␮l reticulo- formed into E. coli. Comparison of HaeIII restriction digest patterns cyte lysate. Reticulocyte lysate (5 ␮l) was diluted to 200 ␮l with and Southern blot cross hybridization (Sambrook et al., 1989) of binding buffer (50mM Tris–HCl [pH 7.5], 150mM NaCl, 0.05% Triton library plasmids identified three homology groups, one of which X-100, protease inhibitor mix) and incubated with 10 ␮l of GST was represented by clone 99. Both strands were sequenced for protein–loaded resin for 1 hr at 4ЊC. After five washes in 1 ml of verification. The coding sequence for the 28 N-terminal amino acids binding buffer, the complexes were dissociated in 20 ␮l sample and 5Ј untranslated sequences missing in the initially isolated library buffer (50 mM Tris–HCl [pH 6.8] 2% SDS, 5% ␤-ME, 10% glycerol, plasmid were cloned by PCR with two nested specific primers in 0.02% Bromphenol blue) and separated by 12.5% SDS-PAGE (Har- clone 99 (primer 1, 5Ј-GCAGGCGGGTGCAGTGC-3Ј; primer 2, 5Ј-TG low and Lane, 1988). After staining the gel with Coomassie blue, we CAGTGCCTGCCCGAGCGGTCGTTCC-3Ј) and a primer in VP16 (5Ј- enhanced it with dimethyl sulfoxide/2,5-diphenyloxazole fluorogra- GCGCTCTGGATATGGCCG-3Ј). Several PCR products of different phy (Harlow and Lane, 1988) and exposed it to X-ray film for 4–12hr. lengths were sequenced in the coding region of the N-terminus. The sequence of the 5Ј untranslated region submitted to the European Molecular Biology Laboratory databank corresponds to the longest Protein Interaction Assay in Avian Cells isolated PCR product. Per condition, three 60 mm dishes with 1 ϫ 106 QT6 cells were cotransfected by the Ca-phosphate method (Graham and van der Transactivation Assays in Yeast Eb, 1973) with 2 ␮g each of MafB⌬N or Ets/118 expression con- The W303–1A strain was transformed with the indicated combina- structs or both and, after 48 hr, metabolically labeled with 250 ␮Ci tion of plasmids by the modified lithium acetate method and plated [35S]methionine (1000 Ci/mmol; Amersham) for 5 hr. After washing under His-, TrpϪ, Ura- selective conditions. Single colonies were cells with phosphate-buffered saline, we lysed them in 1.5 ml of restreaked on selective plates, from which triplicate cultures in 1 lysis buffer (50 mM Tris–Cl [pH 7.5], 150 mM NaCl, 1% NP-40) for ml selective synthetic galactose medium for each condition were 20 min and then sonicated them six times at 5 s each at setting 2, inoculated and incubated for 24 hr at 30ЊC. Cells were harvested using a continuous 70% duty cycle of a Branson sonifier. The lysates by centrifugation, washed, permeabilized by freezing on dry ice, were cleared by centrifugation onto a 30% sucrose cushion for 10 and assayed for ␤-galactosidase activity in a liquid enzyme assay, min at 5000 rpm in an HB-4 rotor and then incubated with 30 ␮lof using ONPG (Sigma) essentially as described in Harshman et al. Ni2ϩ agarose beads (Quiagen) for 90 min on a wheel. After three (1988). Enzyme activities were normalized to cell number as mea- washes with lysis buffer, complexed proteins were eluted by incuba- sured by the optical density of the cell suspension at 600 nm. tion with 2 ϫ 100 ml of 50 mM Tris–Cl [pH 7.5], 1 M NaCl, 1% NP- 40, for 30 min each. Eluate (150 ␮l) was diluted with 850 ␮lof50 Transactivation Assays in Avian Cells mM Tris–Cl [pH 7.5], 1% NP-40, and incubated overnight with 10 QT6 cells (2.5 ϫ 105 cells/35 mm plate) were transfected by the ␮l of protein G–sepharose (Pharmacia) that had been preloaded calcium phosphate coprecipitation procedure (Graham and van der with 4 ␮gof␣HA mouse monoclonal 12CA5-I (Babco). After three Eb, 1973) using 0.5 ␮g reporter and the indicated amount of expres- washes with lysis buffer, proteins were eluted from the beads with sion plasmids per sample. HD3 cells (2 ϫ 106 cells/35 mm dish) 20 ␮l of sample buffer and analyzed by SDS–PAGE, as described Cell 58

for the in vitro assays. All experiments were performed at 4ЊC in the Bengal, E., Ransone, L., Scharfmann, R., Dwarki, V.J., Tapscott, presence of protein inhibitor mix. S.J., Weintraub, H., and Verma, I.M. (1992). Functional antagonism between c-Jun and MyoD proteins: a direct physical association. Northern Blot Hybridization Cell 68, 507–519. RNA was isolated either by the method of Chomczynsky and Sacchi Berben, G., Dumont, J., Gilliquet, V., Bolle, P.A., and Hilger, F. (1991). (1987) or by the RNeasy kit (Qiagen). Tissues were flash-frozen by The YDp plasmids: a uniform set of vectors bearing versatile gene immersion into liquid nitrogen immediately after excision andground disruption cassettes for Saccharomyces cerevisiae. Yeast 7, to a powder. After addition to the guanidinium buffer, cell lysates 475–477. were homogenized by being forced several times through a 22G Beug, H., Doederlein, G., Freudenstein, C., and Graf, T. (1982a). needle. Total RNA (20 ␮g) was separated by electrophoresis at 4 Erythroblast cell lines transformed by a temperature-sensitive mu- V/cm through a 1% agarose, 0.66 M formaldehyde gel and trans- tant of avian erythroblastosis virus: a model system to study ery- ferred to N-Hybond membranes (Amersham) by capillary elution throid differentiation in vitro. J. Cell. Physiol. Suppl. 1, 195–207. (Sambrook et al., 1989). Probes were labeled by the random priming method (Feinberg and Vogelstein, 1984) using 10 ng linearized DNA Beug, H., Palmieri, S., Freudenstein, C., Zentgraf, H., and Graf, T. and 50 ␮Ci 32P-␣␦CTP per reaction. Unincorporated nucleotides (1982b). Hormone-dependent terminal differentiation in vitro of were separated by centrifugation through a 1 ml Sephadex G-50 chicken erythroleukemia cells transformed by ts mutants of avian column (Pharmacia). The mafB probe was prepared from a 950 bp erythroblastosis virus. Cell 28, 907–919. fragment covering the full coding sequence; the c-ets probe was Beug, H., Leutz, A., Kahn, P., and Graf, T. (1984). Ts mutants of E26 prepared from a 200 bp fragment of the 3Ј untranslated region or leukemia virus allow transformed myeloblasts, but not erythroblasts a 1.5 kb fragment covering the full coding region of p68ets-1. Adult or fibroblasts, to differentiate at the nonpermissive temperature. ␣- and ␤-globin probes were prepared as described in Kraut et al. Cell 39, 579–588. (1994). The TfR probe was generated from a fragment covering Bohmann, D., Ellis, M.C., Staszewski, L.M., and Mlodzik, M. (1994). nucleotides 1564–2086 of the 3Ј untranslated region of the chicken Drosophila Jun mediates Ras-dependent photoreceptor determina- gene (Chan et al., 1989); the GAPDH probe was generated from a tion. Cell 78, 973–986. fragment spanning the coding region of the chicken cDNA (Dugaic- zyk et al., 1983). Filters were hybridized according to Church and Bonnerot, C., Rocancourt, D., Briand, P., Grimber, G., and Nicolas, Gilbert (1984), washed under high stringency conditions, and ex- J. F. (1987). A beta-galactosidase hybrid protein targeted to nuclei posed to X-ray film. Band intensity was quantified using the National as a marker for developmental studies. Proc. Natl. Acad. Sci. USA Institutes of Health image 1.55 program. 84, 6795–6799. Brunner, D., Ducker, K., Oellers, N., Hafen, E., Scholz, H., and Stable Transfection of HD3 Erythroblasts Kla¨mbt, C. (1994). The ETS domain protein pointed-P2 is a target Plasmid (10 ␮g) containing both LTR–MafB and SV40 early pro- of MAP kinase in the sevenless signal transduction pathway. Nature moter–NeoR cassettes, or the control construct lacking MafB coding 370, 386–389. 7 regions, were electroporated into 10 HD3 cells and plated into Chan, L.-N. L., Grammatikakis, N., Banks, J.M., and Gerhardt, E.M. methyl cellulose-containing medium in the presence of 3.2 mg/ml (1989). Chicken transferrin receptor gene: conservation 3Ј noncod- G418 sulphate (GIBCO BRL). Resistant colonies were picked after ing sequences and expression in erythroid cells. Nucl. Acids Res. 13 days and expanded in medium containing 2.4 mg/ml G418. Ex- 17, 3763–3771. pression of MafB was monitored by Northern hybridization and Chomczynsky, P., and Sacchi, N. (1987). Single-step method of Western blotting. For the differentiation assays, cells were seeded RNA isolation by acid guanadinium–thiocyanate–phenol–chloroform at 3 ϫ 105/ml andincubated at either 35ЊCor42ЊC. At each timepoint, extraction. Anal. Biochem. 162, 156–159. cells were counted in a Coulter counter and stained for hemoglobin expression with benzidine at acid pH (Orkin et al., 1975). A total of Church, G.M., and Gilbert, W. (1984). Genomic sequencing. Proc. 200–300 cells per sample were evaluated. Natl. Acad. Sci. USA 81, 1991–1995. Cordes, S.P., and Barsh, G.S. (1994). The mouse segmentation gene Acknowledgments kr encodes a novel basic domain–leucine zipper transcription factor. Cell 79, 1025–1034. We would like to thank Drs. Fabio Rossi and Kelly MacNagny for their Dalton, S., and Treisman, R. (1992). Characterization of SAP-1, a generous sharing of biologicalsamples and stimulating discussions, protein recruited by serum response factor to the c-fos serum re- ˚ Melanie Weigand and Asa Marknell-Dewitt for technical help, Dr. sponse element. Cell 68, 597–612. Dirk Bohman and Finn-Hugo Markussen for provision of reagents, deWet, J.R., Wood, K.V., DeLuca, M., Helinski, D.R., and Subramani, Drs. Matthias Hentze and Marek Mlodzik for critically reviewing the S. (1987). Firefly luciferase gene: structure and expression in mam- manuscript, and members of the Graf Lab for helpful comments malian cells. Mol. Cell. Biol. 7, 725–737. throughout the project. Michael Sieweke was supported by long- term fellowships from European Molecular Biology Organization, Dugaiczyk, A., Haron, J.A., Stone, E.M., Dennison, O.E., Rothblum, Boehringer Ingelheim Fonds and a grant from the Deutsche K. N., and Schwartz, R.J. (1983). Cloning and sequencing of a deoxy- Forschungsgemeinschaft (SFB 229) to T. G. ribonucleic acid copy of glyceraldehyde-3-phosphate dehydroge- nase messenger ribonucleic acid isolated from chicken muscle. Bio- Received July 27, 1995; revised January 26, 1996. chemistry 22, 1605–1613. Eisenbeis, C.F., Singh, H., and Storb, U. (1995). Pip, a novel IRF References family member, is a lymphoid-specific, PU.1-dependent transcriptional activator. Genes Dev. 9, 1377–1387. Andrews, N.C., Erdjument-Bromage, H., Davidson, M.B., Tempst, Feinberg, A.P., and Vogelstein, R. (1984). Addendum: a technique P., and Orkin, S.H. (1993a). Erythroid transcription factor NF-E2 is for radiolabeling DNA restriction endonuclease fragments to high a haematopoietic-specific basic-leucine zipper protein. Nature 362, specific activity. Anal. Biochem. 137, 266–267. 722–770. Andrews, N.C., Kotkow, K.J., Ney, P.A., Erdjument, B.H., Tempst, Frangioni, J.V., and Neel, B.G. (1993). Solubilization and purification P., and Orkin, S.H. (1993b). The ubiquitous subunit of erythroid of enzymatically active glutathione S-transferase (pGEX) fusion pro- transcription factor NF-E2 is a small basic-leucine zipper protein teins. Anal. Biochem. 179–187. related to the v-maf oncogene. Proc. Natl. Acad. Sci. USA 90,11488– Ge´ gonne, A., Bosselut, R., Bailly, R.-A., and Ghysdael, J. (1993). 11492. Synergistic activation of the HTLV1 LTR Ets-responsive region by Bassuk, A.G., and Leiden, J.M. (1995). A direct physical association transcription factors Ets1 and Sp1. EMBO J. 12, 1169–1178. between ETS and AP-1 transcription factors in normal human T Ghysdael, J., Ge´ gonne, A., Pognonec, P., Dernis, D., Leprince, D., cells. Immunity 3, 223–237. and Stehelin, D. (1986). Identification and preferential expression in MafB Is An Interaction Partner and Repressor of Ets-1 59

thymic and bursal lymphocytes of a c-ets oncogene-encoded Mr and Griffiths, G. (1992). Characterization of early and late endocytic cytoplasmic protein. Proc. Natl. Acad. Sci. USA 83, 1714–1718. compartments of the transferrin cycle. J. Cell Sci. 103, 211–232. Giese, K., Kingsley, C., Kirshner, J.R., and Grosschedl, R. (1995). Kraut, N., Frampton, J., McNagny, K.M., and Graf, T. (1994). A func- Assembly and function of a TCR␣ enhancer complex is dependent tional Ets DNA-binding domain is required to maintain multipotency on LEF-1-induced DNA bending and multiple protein-protein inter- of hematopoietic progenitors transformed by Myb-Ets. Genes Dev. actions. Genes Dev. 9, 995–1008. 8, 33–44. Gietz, R.D., and Schiestl, R.H. (1991). Applications of high efficiency Kulessa, H., Frampton, J., and Graf, T. (1995). GATA-1 reprograms lithium acetate transformation of intact yeast cells using single- avian myelomonocytic cell lines into eosinophils, thromboblasts, stranded nucleic acids as carrier. Yeast 7, 253–263. and erythroblasts. Genes and Dev. 9, 1250–1262. Graf, T., Ade, N., and Beug, H. (1978). Temperature-sensitive mutant Lai, Z.C., and Rubin, G.M. (1992). Negative control of photoreceptor of avian erythroblastosis virus suggests a block of differentiation development in Drosophila by the product of the yan gene, an ETS as mechanism of leukaemogenesis. Nature 257, 496–501. domain protein. Cell 70, 609–20. Graham, F.L., and van der Eb, A.J. (1973). A new technique for the LaMarco, K., Thompson, C.C., Byers, B.P., Walton, E.M., and assay of infectivity of human adenovirus 5 DNA. Virology 52, 456. McKnight, S. L. (1991). Identification of Ets- and notch-related sub- Guarente, L., and Ptashne, M. (1981). Fusion of Escherichia coli lacZ units in GA binding protein. Science 253, 789–792. to the cytochrome c gene of Saccharomyces cerevisiae. Proc. Natl. Leprince, D., Gegonne, A., Coll, J., de Taisne, C., Schneeberger, A., Acad. Sci. USA 78, 2199–2203. Lagrou, C., and Stehelin, D. (1983). A putative second cell-derived Gutman, A., and Wasylyk, B. (1990). The collagenase gene promoter oncogene of the avian leukemia retrovirus E26. Nature 306, 395–397. contains a TPA and oncogene-responsive unit encompassing the Lim, F., Kraut, N., Framptom, J., and Graf, T. (1992). DNA binding by PEA3 and AP-1 binding sites. EMBO J. 9, 2241–2246. c-Ets-1, but not v-Ets, is repressed by an intramolecular mechanism. Harlow, E., and Lane, D. (1988). Antibodies: A Laboratory Manual. EMBO J. 11, 643–652. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Martin, M.E., Piette, J., Yaniv, M., Tang, W.J., and Folk, W.R. (1988). Press). Activation of the polyomavirus enhancer by a murine activator pro- Harshman, K.D., Moye-Rowley, W.S., and Parker, C.S. (1988). Tran- tein 1 (AP1) homolog and two contiguous proteins. Proc. Natl. Acad. scriptional activation by the SV40 AP-1 recognition element in yeast Sci. USA 85, 5839–5843. is mediated by a factor similar to AP-1 that is distinct from GCN4. McNagny, K.M., Lim, F., Grieser, S., and Graf, T. (1992). Cell surface Cell 53, 321–330. proteins of chicken hematopoietic progenitors, thrombocytes and Herbomel, P., Bourachot, B., and Yaniv, M. (1984). Two distinct eosinophils detected by novel monoclonal antibodies. Leukemia 6, enhancers with different cell specificities coexist in the regulatory 975–984. region of polyoma. Cell 39, 653–662. Metz, T., and Graf, T. (1991). v-myb and v-ets transform chicken Higgs, D.R., and Wood, W.G. (1993). Understanding erythroid differ- erythroid cells and cooperate both in trans and in cis to induce entiation. Curr. Biol. 3, 548–550. distinct differentiation phenotypes. Genes Dev. 5, 369–380. Hoffman, C.S., and Winston, F. (1987). Aten-minute DNA preparation Metz, T., and Graf, T. (1992). The nuclear oncogenes v-erbA and v- from yeast efficiently releases autonomous plasmids for transforma- ets cooperate in the induction of avian erythroleukemia. Oncogene tion of Escherichia coli. Gene 57, 267–272. 7, 597–605. Horton, M.A. (1983). Expression of transferrin receptors during ery- Moscovici, C., Moscovici, M.G., Jiminez, H., Lai, M.M.C., Hayman, throid maturation. Exp. Cell Res. 144, 361–366. M. J., and Vogt, P.T. (1977). Continuous tissue culture cell lines Hu, H.Y., Gardner, J., and Aisen, P. (1977). Inducibility of transferrin derived from chemically induced tumors of Japanese quail. Cell 11, receptors on friend erythroleukemic cells. Science 197, 559–61. 95–103. Iacopetta, B.J., Morgan, E.H.,and Yeoh, G. (1982).Transferrin recep- Nerlov, C., Rorth, P., Blasi, F., and Johnsen, M. (1991). Essential tors and iron uptake during erythroid cell development. Biochim. AP-1 and PEA3 binding elements in the human urokinase enhancer Biophys. Acta 687, 204–210. display cell type–specific activity. Oncogene 6, 1583–1592. Igarashi, K., Kataoka, K., Itoh, K., Hayashi, N., Nishizawa, M., and Ness, S.A., and Engel, J.D. (1994). Vintage redsand whites: combina- Yamamoto, M. (1994). Regulation of transcription by dimerization torial transcription factor utilization in hematopoietic differentiation. of erythroid factor NF-E2 p45 with small Maf proteins. Nature 367, Curr. Opin. Genet. Dev. 4, 718–724. 568–572. Ney, P.A., Andrews, N.C., Jane, S.M., Safer, B., Purucker, M.E., Igarashi, K., Itoh, K., Hayashi, N., Nishizawa, M., and Yamamoto, Weremowicz, S., Morton, C.C., Goff, S.C., Orkin, S.H., and Nienhuis, M. (1995). Conditional expression of the ubiquitous transcription A.W. (1993). Purification of the human NF-E2 complex: cDNA cloning factor MafK induces erythroleukemia cell differentiation. Proc. Natl. of the hematopoietic cell-specific subunit and evidence for an asso- Acad. Sci. USA 92, 7445–7449. ciated partner. Mol. Cell. Biol. 13, 5604–5612. Introna, M., Golay, J., Frampton, J., Nakano, T., Ness, S., and Graf, Nunn, M., Seeburg, P.H., Moscovici, C., and Duesberg, P.H. (1983). T. (1990). Mutations in v-myb alter the differentiation of myelomono- Tripartite structure of the avian erythroblastosis virus E26 trans- cytic cells transformed by the oncogene. Cell 63, 1287–1297. forming gene. Nature 306, 391–395. Itoh, K., Igarashi, K., Hayashi, N., Nishizawa, M., and Yamamoto, O’Neill, E.M., Rebay, I., Tjian, R., and Rubin, G.M. (1994). The activi- M. (1995). Cloning and characterization of a novel erythroid cell- ties of two Ets-related transcription factors required for Drosophila derived CNC family transcription factor heterodimerizing with the eye development are modulated by the Ras/MAPK pathway. Cell small Maf family proteins. Mol. Cell. Biol. 15, 4184–4193. 78, 137–147. Janknecht, R., and Nordheim, A. (1993). Gene regulation by Ets Omary, M.B., Trowbridge, I.S., and Minowada, J. (1980). Human cell- proteins. Biochim. Biophys. Acta 1155, 346–356. surface glycoprotein with unusual properties. Nature 286, 888–891. Johnson, A.D. (1995). The price of repression. Cell 81, 655–658. Orkin, S. (1995). Transcription factors and hematopoietic develop- Kataoka, K., Fujiwara, K.T., Noda, M., and Nishizawa, M. (1994). ment. J. Biol. Chem. 270, 4955–4958. MafB, a new Maf family transcription activator that can associate Orkin, S.H., Harosi, F.I., and Leder, P. (1975). Differentiation in eryth- with Maf and Fos but not with Jun. Mol. Cell. Biol. 14, 7581–7591. roleukemic cells and their somatic hybrids. Proc. Natl. Acad. Sci. Kataoka, K., Igarashi, K., Itoh, K., Fujiwara, K.T., Noda, M., Yama- USA 72, 98–102. moto, M., andNishizawa, M. (1995). Small Maf proteins heterodimer- Owen, D., and Ku¨ hn, L. (1987). Noncoding 3Ј sequences of the ize with Fos and may act as competitive repressors of the NF-E2 transferrin receptor gene are required for mRNA regulation by iron. transcription factor. Mol. Cell. Biol. 15, 2180–2190. EMBO J. 6, 1287–1293. Killisch, I., Steinlein, P., Ro¨ misch, K., Hollinshead, R., Beug, H., Ponta, H., Cato, A.C., and Herrlich, P. (1992). Interference of pathway Cell 60

specific transcription factors. Biochim. Biophys. Acta 1129, 255–261. Que´ va, C., Leprince, D., Stehelin, D., and Vandenbunder, B. (1993). p54c-ets-1 and p68c-ets-1, the two transcription factors encoded by the c-ets-1 locus, are differentially expressed during the development of the chick embryo. Oncogene 8, 2511–2520. Roberts, S.G.E., and Green, M.R. (1995). Dichotomous regulators. Nature 375, 105–106. Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989). Molecular Clon- ing: A Laboratory Manual. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press). Schmidt, J.A., Marshall, J., Hayman, M.J., Ponka, P., and Beug, H. (1986). Control of erythroid differentiation: possible role of the transferrin cycle. Cell 46, 41–51. Schneikert, J., Lutz, Y., and Wasylyk, B. (1992). Two independent activation domain in c-Ets-1 and c-Ets-2 located in non-conserved sequences of the ets gene family. Oncogene 7, 249–256. Seth, A., Robinson, L., Thompson, D.M., Watson, D.K., and Papas, T.S. (1993). Transactivation of GATA-1 promoter with ETS1, ETS2 and ERGB/Hu-FLI-1 proteins: stabilization of the ETS1 protein binding on GATA-1 promoter sequences by monoclonal antibody. Onco- gene 8, 1783–1790. Sieff, C., Bicknell, D., Caine, G., Robinson, J., Lam, G., and Greaves, M.F. (1982). Changes in cell surface antigen expression during he- mopoietic differentiation. Blood 60, 703–713. Smith, D.B. (1993). Purification of glutathione S-transferase fusion proteins. Meth. Mol. Cell. Biol. 4, 220–229. Treier, M., Bohmann, D., and Mlozik, M. (1995). JUN cooperates with the Ets domain protein Pointed to induce photoreceptor R7 fate in the Drosophila eye. Cell 83, 753–776. Treisman, R. (1994). Ternary complex factors: growth factor regulated transcriptional activators. Curr. Opin. Genet. Dev. 4, 96–101. Wasylyk, B., Wasylyk, C., Flores, P., Begue, A., Leprince, D., and Stehelin, D. (1990). The c-ets proto-oncogenes encode transcription factors that cooperate with c-Fos and c-Jun for transcriptional activation. Nature 346, 191–193. Wasylyk, C., Gutman, A., Nicholson, R., and Wasylyk, B. (1991). The c-Ets oncoprotein activates the stromelysin promoter through the same elements as several non-nuclear oncoproteins. EMBO J. 10, 1127–1134.

GenBank Accession Number

The quail MafB sequence determined in this paper is registered under accession number X95611 in the EMBL/GenBank database.