Inactivation of a C-Myb/Estrogen Receptor Fusion Protein In

Inactivation of a c-Myb/estrogen receptor fusion protein in transformed primary cells leads to granulocyte/macrophage dierentiation and down regulation of c-kit but not c-myc or cdc2

Annette Hogg1,4, Sabine Schirm2,5, Hideki Nakagoshi3, Paul Bartley1, Shunsuke Ishii3, J Michael Bishop2 and Thomas J Gonda1

1Hanson Centre for Cancer Research, Institute of Medical and Veterinary Science, Frome Road, Adelaide, South Australia 5000, Australia; 2GW Hooper Foundation, University of California, San Francisco, California 94143-0552, USA; 3Laboratory of Molecular Genetics, Tsukuba Life Science Center, The Institute of Physical and Chemical Research (RIKEN), Tsukuba, Ibaraki 305, Japan

Primary murine fetal hemopoietic cells were transformed tent with this notion, both truncated and wild type Myb with a fusion protein consisting of the ligand-binding can transform murine myeloid cells both in vivo and in domain of the estrogen receptor and a carboxyl- vitro (Ferrao et al., 1995; Macmillan and Gonda, 1994; terminally truncated c-Myb protein (ERMYB). The Gonda et al., 1989a,b). Furthermore, antisense c-myb ERMYB-transformed hemopoietic cells exhibit an im- oligonucleotides block the proliferation of leukemic and mature myeloid phenotype when grown in the presence of normal hemopoietic cells and conversely, overexpres- b-estradiol. Upon removal of b-estradiol, the ERMYB sion of Myb blocks the induced dierentiation of cells display increased adherence, decreased clonogeni- erythroid and myeloid cells (Yanagisawa et al., 1991; city and dierentiate to cells exhibiting granulocyte or Anfossi et al., 1989; Clarke et al., 1988). Mice with a macrophage morphology. The expression of the c-myc,c- homozygous de®ciency of c-myb die during gestation kit, cdc2 and bcl-2 genes, which are putatively regulated and fail to develop a normal hemopoietic system, by Myb, was investigated in ERMYB cells grown in the demonstrating that Myb is essential for hemopoietic presence or absence of b-estradiol. Neither c-myc nor cell development (Mucenski et al., 1991). cdc2 expression was down-regulated after removal of b- Myb binds to and transactivates promoters contain- estradiol demonstrating that dierentiation is not a ing a consensus recognition sequence T/CAACG/TG consequence of decreased transactivation of these genes (Nakagoshi et al., 1990; Weston and Bishop, 1989; by ERMYB. While bcl-2 expression was reduced by Nishina et al., 1989). The amino terminal DNA 50% in ERMYB cells grown in the absence of b- binding domain of Myb, which is related to the estradiol, there was no increase in DNA laddering, helix-turn-helix motif, is essential for transactivation suggesting that Myb was not protecting ERMYB cells of promoters containing the Myb recognition sequence from apoptosis. In contrast, a substantial (200-fold) (Ogata et al., 1994). Analysis of truncated Myb decrease in c-kit mRNA level was observed following proteins led to the de®nition of two additional dierentiation of ERMYB cells, and c-kit mRNA could domains: a transactivation domain located between be partially re-induced by the re-addition of b-estradiol. amino acid (aa) residues 241 ± 325 and a negative Furthermore, a reporter construct containing the c-kit regulatory domain (NRD) situated at the C-terminus promoter was activated when cotransfected with a Myb (aa 326 ± 500) (Kanei-Ishii et al., 1992; Sakura et al., expression vector, providing further evidence of a role for 1989). Deletion or mutation of the NRD results in a Myb in the regulation of c-kit. higher level of transactivation and transformation by Myb (Hu et al., 1991; Kanei-Ishii et al., 1992). A Keywords: Myb; dierentiation; target genes; bcl-2; correlation between transactivation and transformation apoptosis of murine hemopoietic cells has led to the notion that the target genes which Myb regulates are critical to the transforming process (Hu et al., 1991). Putative cellular targets of Myb transactivation Introduction include c-kit, CD34, mim-1, neutrophil elastase and CD13 aminopeptidase, all of which are expressed on The murine c-myb gene encodes a transcription factor immature hemopoietic cells (Oelgeschlager et al., 1996; which has a critical role in hemopoiesis. c-myb mRNA Melotti and Calabretta, 1996; Shapiro, 1995; Melotti et levels are high in undierentiated hemopoietic cells and al., 1994; Ness et al., 1993; Burk et al., 1993; Ratajczak decrease following dierentiation, suggesting that Myb et al., 1992). Activation by Myb of T-lymphocyte is important for the proliferation of immature speci®c genes such as CD4, T cell receptor d and lck hemopoietic cells (Gonda and Metcalf, 1984). Consis- tyrosine kinase and of genes which function in cell growth control such as cdc2 and c-myc has also been demonstrated (Hernandez-Munain and Krangel, 1995; Correspondence: TJ Gonda McCracken et al., 1994; Cogswell et al., 1993; Ku et Present addresses: 4Peter MacCallum Cancer Institute, Locked Bag 1, al., 1993; Siu et al., 1992; Nakagoshi et al., 1992; Evans A `Beckett St, Melbourne, Victoria 3000, Australia; 5Berlex et al., 1990). The bcl-2 gene, which serves an anti- Biosciences, 15049 San Pablo Avenue, Richmond, California 94804 ± 0099, USA apoptotic function in various cell types, has very Received 7 January 1997; revised 4 August 1997; accepted 5 August recently been reported as a Myb target gene 1997 (Salomoni et al., 1997; Taylor et al., 1996; Frampton Differentiation following inactivation of ER-Myb AHogget al 2886 et al., 1996). Other transcription factors can cooperate encodes the DNA binding domain and transactivation with Myb to achieve maximal levels of transcriptional domains of Myb but lacks the negative regulatory activation (Oelgeschlager et al., 1996; Dai et al., 1996; domain and is one of the most potent transforming and Hernandez-Munain and Krangel, 1995; Shapiro, 1995; transactivating forms of Myb (Kanei-Ishii et al., 1992; McCracken et al., 1994; Burk et al., 1993; Ness et al., Hu et al., 1991; Sakura et al., 1989). The ERLBD was 1993), and thus may enhance transactivation of target cloned in frame into the gag sequence located upstream genes and direct cellular speci®city. of CT3Myb in RED(CT3Myb) to create Studies of putative Myb-regulated genes have relied RED(ERMYB) (Hu et al., 1991; Gonda et al., predominantly upon the activation of arti®cial reporter 1989a,b) (Figure 1a). The structure of the gene constructs in cell lines transiently transfected with c- RED(ERMYB) proviral DNA, the expected mRNA myb (Oelgeschlager et al., 1996; Shapiro, 1995; and protein products and a restriction enzyme map of McCracken et al., 1994; Burk et al., 1993; Ness et al., the construct is shown in Figure 1. A full length 1993; Ku et al., 1993; Cogswell et al., 1993; Evans et al., genomic mRNA and a spliced sub-genomic mRNA are 1990). The expression of a conditionally active Myb produced by the RED series of retroviral vectors (Hu et protein in primary cells, however, would more closely al., 1991; Gonda et al., 1989a,b) (see Figure 1b). In mimic the situation in vivo and may allow the de®nition addition to producing a full length genomic RNA of of the physiological targets of Myb which are important 4.7 kb, two sets of potential splice sites (SD1, SA1, SD2, for myeloid cell proliferation and dierentiation. SA2) in RED(ERMYB), generate a 3.4 kb subgenomic Conditionally active forms of a number of transcription mRNA species. The RED(ERMYB) construct was factors have been constructed by fusion with a portion of designed to produce an ERMYB protein initiating at the estrogen receptor containing the ligand binding the ATG codon in the gag sequence (Figure 1a and b). domain (but not the DNA binding domain) (Lyon and Thus, the ERMYB fusion protein would be expected to Watson, 1995; Boehmelt et al., 1992; Burk and contain an additional 20 amino acids (aa) of gag Klempnauer, 1991; Eilers et al., 1989). Accordingly, we sequence at the amino terminus. have developed a conditional system for transformation The fusion of the ERLBD to Myb should confer by Myb, by expressing a C-terminally truncated Myb estrogen-dependent activation of Myb transcriptional protein fused to the estrogen binding domain of the activity. In order to examine the eect of estrogen on human estrogen receptor in primary murine hemopoietic cells. Cells transformed by the estrogen receptor/Myb fusion protein (ERMYB) were dependent on b-estradiol for growth, and dierentiated to mature granulocytes and macrophages on withdrawal of b-estradiol. Thus, this system allowed us to investigate the regulation of putative Myb-transactivated genes in primary cells in which the only lesion is the presence of a transforming Myb protein. With this in mind, we examined the expression of four putative Myb-regulated genes ± cdc2, c-myc,c-kit and bcl-2 ± which could potentially mediate the proliferative eects of Myb on myeloid cells. c-kit encodes a transmembrane tyrosine kinase with a critical role in the growth of immature hemopoietic cells (reviewed in Galli et al., 1994). cdc2 encodes a protein kinase that plays a pivotal role in cell cycle progression (van den Heuvel and Harlow, 1993; Lee and Nurse, 1987). The c-Myc transcription factor is thought to be a major regulator of cell growth, dierentiation and apoptosis (reviewed in LuÈ scher and Eisenman, 1990; Evan et al., 1992); and the bcl-2 gene inhibits cell death in a variety of cell types (reviewed in Craig, 1995). Our results do not support a role for Myb in the regulation of Figure 1 Structure of the RED(ERMYB) retroviral construct. cdc2 or c-myc in myeloid cells. C-kit and bcl-2 expression The structure of the integrated proviral DNA is shown in (a). The levels, however, did correlate with activation of the coding portion contains the gag viral DNA (®lled box), into ERMYB fusion protein. which the estrogen receptor ligand binding domain (ER) (open box) was cloned and also encodes a C-terminally truncated Myb protein (MYB) (not to scale). The integrated DNA is ¯anked by the viral long terminal repeates (LTR). Protein translation Results initiates at the ATG codon shown in the gag sequence and is terminated at the premature stop codon (TGA) in the MYB sequence. The structure of the viral genomic and spliced mRNA Construction of an estradiol-inducible myb retroviral products (upper) and ERMYB fusion protein (lower) are shown vector in (b) (not to scale). The use of the splice sites SD1, SA1 and SD2, SA2 remove 1.16 kb and 0.2 kb of sequence, respectively, In order to construct a hemopoietic cell line which generating a spliced RNA which encodes the ERMYB fusion expressed a conditionally active Myb protein, the ligand protein. A schematic representation (not to scale) of the proviral binding domain of the estrogen receptor (ERLBD) was RED(ERMYB) DNA showing the locations of restriction enzyme (RE) sites is shown in (c) (S, SacI; H, HindIII; E, EcoRI). The cloned into a retroviral vector RED(CT3Myb) which boxes above the (RE) map represent cDNA probes used to probe expresses a C-terminally truncated Myb protein Southern blots and are derived from the c-myb (stippled) and ER (CT3Myb) (Hu et al., 1991). The CT3Myb protein cDNA (open), respectively Differentiation following inactivation of ER-Myb AHogget al 2887 transactivation by the ERMYB fusion protein, a had not occurred, genomic DNA from RED(ERMYB) plasmid containing six copies of a Myb binding site transformed hemopoietic cells was prepared and fused to a CAT reporter gene, p8MybbCAT, was analysed using Southern blotting. Recognition sites cotransfected with the RED(ERMYB) plasmid into for HindIII and EcoRI are located within the Balb/c 3T3 cells. A pUC13 plasmid into which the v- RED(ERMYB) construct and SacI sites are located myb gene was inserted [pUC13-RSV(v-myb)] was used within the viral long terminal repeats (Figure 1c). To as a non-inducible control. Transfection of pUC13- identify endogenous murine DNA fragments, DNA RSV(v-myb) resulted in a 16-fold induction of CAT from uninfected cells was analysed in parallel. DNA activity compared to pUC13-RSV alone (Table 1). As fragments of 1.1 kb and 0.8 kb were observed when expected, the addition of b-estradiol had no eect on genomic DNA from ERMYB cells was digested with the levels of CAT activity in the presence of pUC13- HindIII and SacI and probed with a 5' c-myb cDNA RSV(v-myb). The addition of b-estradiol to Balb/c 3T3 probe (Figures 1c and 2a). These fragments were cells transfected with the p8MybbCAT reporter absent from the control DNA, indicating that they plasmid and RED(ERMYB), however, resulted in a were derived from the integrated RED(ERMYB) 60-fold induction of reporter gene activity (Table 1). vector (Figure 2a). In addition, a 3 kb EcoRI/SacI Importantly, the levels of reporter gene activity in the DNA fragment from ERMYB cells hybridized to a absence of b-estradiol were similar to that obtained cDNA probe derived from the ligand binding domain with pUC13-RSV alone, indicating that there was of the human estrogen receptor and was absent in negligible transcriptional activity from ERMYB in the control DNA (Figures 1c and 2a). These results absence of b-estradiol induction. indicate that the coding portion of the RED(ERMYB) vector had incorporated intact into the genome of the transformed cells. Generation of ERMYB transformed cells In order to determine the sizes of the genomic and ERMYB transformed hemopoietic cells were generated spliced RNA species produced by the RED(ERMYB) by co-cultivation of RED(ERMYB) transfected C2 vector and to ascertain whether b-estradiol in¯uenced packaging cells with murine fetal liver cells as described the expression of RED(ERMYB), poly(A)+ mRNA previously (Gonda et al., 1989a, 1993; Hu et al., 1991). was prepared from ERMYB transformed cells grown Both GM-CSF and b-estradiol were necessary for the in the presence or absence of b-estradiol. Two growth of ERMYB transformed cells. Initially, the transcripts of 4.7 kb and 3.4 kb were observed when ERMYB transformed cells were established on Northern blots prepared using poly(A)+ mRNA from irradiated ®broblast feeder layers because culture in ERMYB cells were probed with the ligand binding the absence of feeder layers failed to generate a cell domain of the human estrogen receptor (Figure 2b). line. Previous results have shown that the growth of These transcripts corresponded to the sizes expected Myb-transformed cells is density-dependent (Macmil- for the genomic mRNA and sub-genomic spliced lan and Gonda, 1994), and we suspect that due to a mRNA, respectively (Figure 1b), and also hybridized low frequency of infection by RED(ERMYB), the to a c-myb cDNA probe (data not shown). The initial density of transfected cells was probably too low smaller species is unlikely to represent the 3.6 kb to support growth. Since we have previously found endogenous c-myb mRNA since other Myb-trans- that feeder layers stimulate the growth of Myb- formed hemopoietic cells do not produce detectable transformed cells cultured at low density, we used levels of this message (Hu et al., 1991; Lipsick, 1987), feeder layers to allow the initial expansion of ERMYB and the ratio of the 4.7 kb and 3.4 kb transcripts was transformed cells. Once the cultures were well the same when Northern blots were probed with either established on ®broblast feeder layers (6 ± 8 weeks), the 5' c-myb cDNA probe or the ERLBD cDNA the cells were subsequently grown in liquid culture probe. without feeders in the presence of GM-CSF, b- To verify expression of the ERMYB fusion protein, estradiol and charcoal-stripped fetal calf serum. Western blots of whole cell lysates from ERMYB (Charcoal-stripped serum was used to prevent inter- transformed cells were probed with a monoclonal ference from estrogen that may be present in FCS). antibody (5.1) which recognizes the amino terminal region of murine Myb (Ramsay et al., 1989). This antibody detected a 35 kd protein, as expected (Hu et Structure and expression of RED(ERMYB) DNA, al., 1991), in murine hemopoietic cells transformed by mRNA and protein in ERMYB transformed cells the RED(CT3Myb) retroviral vector (Figure 2c). The In order to examine the structure of the integrated ERLBD, which is fused to CT3Myb to generate the RED(ERMYB) vector and ensure that rearrangements ERMYB fusion protein, encodes 287 amino acids and thus the presence of a protein of 65 kd in the ERMYB transformed cells (but not in the CT3Myb transformed cells) was consistent with the predicted molecular Table 1 Inducible activity of the estrogen receptor-Myb protein weight of the ERMYB fusion protein. The additional assayed on a CAT reporter plasmid band present in lysates from both the CT3Myb and Construct b-estradiol fold increase ERMYB cells was also detected in cells not expressing pUC13 RSV 7 1.0 Myb (not shown) and was, therefore, probably due to + 1.0+0.085 a non-speci®cally cross-reacting protein. These results pUC13-RSV (v-myb) 7 16.0+2.45 con®rmed that the ERMYB transformed murine + 14.4+0.55 hemopoietic cells were expressing ERMYB mRNA RED(ERMYB) 7 1.6+0.3 and protein derived from the RED(ERMYB) retroviral + 60.5+0.2 vector. Differentiation following inactivation of ER-Myb AHogget al 2888 a FD ERMYB FD ERMYB

3.0

1.1

0.8

HindIII + SacI EcoRI + SacI

b +– c ERMYB CT3MYB

ERMYB (65 kd) 4.7

3.4 CT3MYB (35 kd)

Figure 2 Analysis of the structure and expression of the RED(ERMYB) retroviral vector in ERMYB cells. (a) Southern blot analysis of DNA from ERMYB cells and, for comparison, FDC-P1 (FD) cells. The DNA was digested with the enzymes shown below each panel. Southern blots were hybridised with a 5' murine c-myb cDNA probe (left) or the ligand binding domain of the human estrogen receptor (right). The sizes, in kilobases, of novel DNA species which were detected in the ERMYB cells are shown at the right. (b) Northern blot analysis of poly(A)+ mRNA from ERMYB cells grown in the presence (+) or absence (7)ofb- estradiol. Northern blots were probed with a cDNA probe derived from the human estrogen receptor ligand binding domain. The sizes of the mRNA species detected, in kilobases, are shown at the right. (c) Western blot analysis of cellular extracts from ERMYB cells and, for comparison, from CT3MYB cells. CT3MYB cells express the MYB portion, only, of the ERMYB fusion protein. The Western blot was probed with monoclonal c-Myb antibody 5.1 (Gonda et al., 1989a; Ramsay et al., 1989). Proteins of 65 kd and 35 kd were observed (shown at right) corresponding to the predicted sizes of the ERMYB fusion protein and the CT3MYB protein, respectively

a proportion of ERMYB cells (35 ± 40%) grown in the Estradiol-dependent growth of ERMYB transformed presence of b-estradiol became adherent 24 h after cells passaging and no further increase in adherence was To examine the eects of Myb inactivation, cellular observed up to 72 h after plating. growth was examined following various periods of b- The eects of b-estradiol withdrawal on the estradiol deprivation in liquid culture. The ERMYB proliferation of ERMYB cells were further examined cells formed non-adherent clusters in liquid medium in using clonogenic assays. Cells were grown in the the presence of b-estradiol. Upon removal of b- absence of b-estradiol for 48 h then plated in estradiol from the ERMYB cells, however, we methylcellulose in the presence or absence of b- observed a decrease in growth (not shown) and an estradiol. Colonies (430 cells) and clusters (530 increase in the proportion of adherent cells (Figure 3). cells) were counted after 1 week of culture. In the Following 72 h of b-estradiol deprivation, 95 ± 100% of presence of b-estradiol, 133 colonies and 76 clusters ERMYB cells had become adherent. In contrast, only grew. The colonies which grew in the presence of b- Differentiation following inactivation of ER-Myb AHogget al 2889 estradiol consisted of small, rounded, highly refractile Like other murine Myb transformed cells (Gonda et cells typical of immature hemapoietic cells. In contrast, al., 1989a,b, 1993), ERMYB cells expressed cell surface in the absence of b-estradiol, no colonies and only 23 markers indicative of the granulocyte/macrophage clusters were observed. Those cells which grew in the lineage in that they expressed Mac 1, Mac 3 and F4/ absence of b-estradiol were elongated with numerous 80 when grown with or without b-estradiol (data not cytoplasmic processes (not shown) which are features shown). We therefore decided to examine expression of of more dierentiated myeloid cells. These experiments a marker of early hemopoietic cells, on ERMYB cells. demonstrated that the ERMYB transformed cells were We have shown previously that Thy-1, a marker dependent on b-estradiol for proliferation under clonal expressed on hemopoietic stem and progenitor cells, is assay conditions. expressed on Myb-transformed cells, presumably Staining of cytospin preparations from ERMYB re¯ecting an early myeloid phenotype (Gonda et al., cells grown in liquid culture in the presence of b- 1993). Thy-1 was expressed on 25 ± 30% of ERMYB estradiol showed that they resembled immature cells in the presence of b-estradiol, in accordance with myeloid cells typical of other hemopoietic cells these results (data not shown). Thy-1 was not transformed by C-terminally truncated Myb (Ferrao expressed, however, on ERMYB cells grown in the et al., 1995; Gonda et al., 1989a,b). The cells were absence of b-estradiol for 48 h (data not shown), relatively small and rounded with a high nuclear/ re¯ecting the more dierentiated state of the latter cells. cytoplasmic ratio (Figure 4a). After 7 h of b-estradiol We next sought to determine whether enzymes deprivation, an increase in cell size and cytoplasmic speci®c for the granulocyte and macrophage lineages granularity and a decrease in the nuclear/cytoplasmic were expressed on ERMYB cells. Cells were stained for ratio was noted (Figure 4b). Following 24 h of b- naphthol-AS-D-acetate esterase (neutrophil esterase), estradiol deprivation, the nuclear/cytoplasmic ratio an enzyme speci®c for the neutrophil lineage and a- had decreased substantially and cells characteristic of napthylacetate esterase (macrophage esterase) which is various stages of neutrophil maturation, including expressed in macrophages. The proportion of cells myelocytes, metamyelocytes, band neutrophils and expressing NE alone was signi®cantly higher in cells mature segmented neutrophils, were observed (Figure grown without b-estradiol for 1 day compared to cells 4c). Cells in the early stages of macrophage grown in the presence of b-estradiol and decreased development were also present (Figure 4c). By 48 h thereafter (Table 2). The decrease in cells staining of b-estradiol withdrawal, the ERMYB cells com- positive for NE alone after 2 days of b-estradiol prised predominantly mature macrophages and mature withdrawal presumably re¯ects the short lifespan of segmented neutrophils (Figure 4d). After 3 ± 6 days of mature neutrophils and is in agreement with the b-estradiol deprivation, the majority of the cells morphological data. In contrast, the proportion of resembled mature macrophages and very few cells expressing ME alone increased signi®cantly from (510%) mature segmented neutrophils were present 1 ± 3% to 80% following withdrawal of b-estradiol for (Figure 4e, f). Taken together, these observations 2 days and remained high up to 6 days (Table 2). In show that ERMYB cultures were dependent on b- parallel, the intensity of ME staining also increased estradiol to maintain an immature myeloid phenotype markedly (data not shown). Presumably, the dierent and developed characteristics typical of maturing myeloid cells, predominantly of the neutrophil and monocyte/macrophage lineages following b-estradiol deprivation. a b

Figure 4 Morphological characteristics of ERMYB cells. Figure 3 Adherent properties of ERMYB cells. Cells were grown Cytospin preparations of ERMYB cells grown in the presence in the presence (dashed line) or absence (solid line) of b-estradiol (a) or absence of b-estradiol (b±f) and stained with May- for various periods of time (x-axis) and the proportion of GruÈ nwald Giemsa are shown. ERMYB cells were grown without adherent cells was calculated (y-axis) b-estradiol for (b) 7 h, (c)24h,(d)48h,(e) 3 days or (f) 6 days Differentiation following inactivation of ER-Myb AHogget al 2890 Table 2 Expression of neutrophil (NE) and macrophage-speci®c (ME) esterases in ERMYB cells b-estradiol time (days) NE ME NE and ME + 1 34 1 65 7 1 71 3 26 7 2 0 80 20 7 3 5 94 1 7 6 4 86 6 The percentage of cells staining positive for NE or ME is shown

temporal patterns of expression of these two enzymes re¯ect the dierent dynamics of the dierentiation of neutrophils and macrophages. Taken together, the data on cell proliferation, adherence, cellular morphology, surface marker expression, and expression of macrophage and neutrophil-speci®c enzymes indicate that ERMYB cells, in the presence of b-estradiol, represent a relatively early stage of committed granulocyte macrophage precursor and are capable of differentiat- ing to mature granulocytes and macrophages upon withdrawal of b-estradiol. This behavior is consistent with that expected of Myb-transformed cells following Figure 5 Estradiol-dependent colony formation by ERMYB loss of Myb activity. cells. ERMYB cells were grown with (dashed lines) or without b- estradiol (solid lines) for up to 72 h. The number of colonies (430 cells) or clusters (530 cells) following plating in Are ERMYB cells reversibly dierentiated in the methylcellulose/1 mM b-estradiol and incubating for 7 (®lled absence of b-estradiol? symbols) or 14 days (open symbols) is shown Clearly, the absence of b-estradiol from the culture medium resulted in the dierentiation of ERMYB cells. In order to determine if the dierentiation process was colonies/clusters were observed (Figure 5). The colonies reversible, the cells were deprived of b-estradiol for 3 ± formed by b-estradiol-deprived ERMYB cells devel- 7 days, following which b-estradiol was re-added to the oped more dierentiated characteristics as the period of culture medium. The cells failed to revert to an b-estradiol deprivation was lengthened. Thus, the immature, transformed phenotype. b-estradiol depriva- ability of the cells to form colonies is probably due tion for up to 24 h, however, lead to the maintenance to the maintenance of proliferative capacity during of at least some immature cells following the re- early stages of dierentiation rather than reversion of addition of b-estradiol (data not shown). dierentiated cells. We conclude, therefore, that Next, we studied the clonogenic potential of following 72 h of b-estradiol deprivation, once com- ERMYB cells from which b-estradiol was removed plete dierentiation has taken place as assessed by and then re-added in order to examine the extent and morphological criteria, the ERMYB cells undergo kinetics of the loss of clonogenicity. ERMYB cells, irreversible processes which prevent the re-acquisition grown in the presence or absence of b-estradiol in of properties of immature hemopoietic cells despite the liquid culture for various periods of time, were plated re-activation of Myb. in methylcellulose in the presence of b-estradiol (methylcellulose/1 m b-estradiol), and the number of M Regulation of putative Myb-target genes in ERMYB colonies/clusters were counted after 7 and 14 days cells (Figure 5). Surprisingly, the number of colonies formed by ERMYB cells grown continuously in b-estradiol We have shown that transactivation by ERMYB is was not constant during the time course. The initial dependent on b-estradiol and that ERMYB cells increase in colony number may re¯ect density- dierentiated upon removal of b-estradiol into mature dependent eects (due to a putative autocrine factor) neutrophils and macrophages. Next, we wanted to on the clonogenicity of Myb-transformed cells, as have examine whether the expression of several putative been described elsewhere (Macmillan and Gonda, Myb-regulated genes did, in fact, correlate with Myb 1994). The subsequent decrease in clonogenicity activity and if they were regulated during the observed following 24 h of liquid culture in the dierentiation process. We concentrated on some of presence of b-estradiol may be a consequence of the those putative Myb-regulated genes which are typically higher proportion of mature cells resulting from the expressed by myeloid cells. Poly(A)+ mRNA was spontaneous dierentiation of ERMYB cells observed isolated from ERMYB cells grown with or without at higher cell densities (data not shown). b-estradiol for 6, 24 or 48 h. Although we had The ability of b-estradiol-deprived ERMYB cells to demonstrated that fully dierentiated ERMYB cells form colonies/clusters in methylcellulose/1 mM b- do not revert to an immature morphology following estradiol, however, decreased rapidly with time. After the readdition of b-estradiol, it was possible that some 24 h of b-estradiol deprivation, colonly formation was Myb-regulated genes might be transactivated upon re- substantially reduced while by 72 h, only occasional addition of b-estradiol to partially dierentiated cells. Differentiation following inactivation of ER-Myb AHogget al 2891 Thus, we also examined poly(A)+ mRNA from cells expression in ERMYB cells grown in the presence or deprived of b-estradiol for 24 h and to which b- absence of b-estradiol (Figure 6). While some variation estradiol was subsequently re-added for 6 or 48 h. in c-myc mRNA levels was observed over the time Northern blots of these mRNA samples probed with course of b-estradiol withdrawal, there was no correla- cdc2,c-myc and c-kit are shown in Figure 6. The tion observed between c-myc mRNA expression and the amount of mRNA at each timepoint was measured presence of b-estradiol. This experiment shows that c- relative to the signal generated using a GAPDH cDNA myc mRNA levels are not signi®cantly aected by probe (Figure 6). ERMYB activity and that a decrease in c-myc mRNA cdc2 levels were relatively constant in cells grown up expression is not associated with the earlier stages of to 48 h without b-estradiol, and showed no increase in granulocyte/macrophage dierentiation of ERMYB 24 h b-estradiol deprived cells to which b-estradiol was cells (see Discussion). Expression of c-kit mRNA was re-added (Figure 6). These data show that activity of the similar in ERMYB cells grown in the presence or ERMYB protein does not have a signi®cant eect on absence of b-estradiol for 6 h. Following 24 h of b- cdc2 mRNA levels in ERMYB cells and that the down- estradiol deprivation, however, a dramatic decrease regulation of cdc2 is not responsible for the loss of (200-fold) in the expression of c-kit mRNA was proliferative capacity observed on withdrawal of b- observed (Figure 6). c-kit mRNA expression remained estradiol from ERMYB cells. We next examined c-myc low in ERMYB cells grown for 48 h without b-estradiol. Furthermore, c-kit mRNA levels increased 18-fold in 24 h b-estradiol deprived ERMYB cells to which b- estradiol was re-added for 48 h. These data support a role for Myb in the regulation of c-kit expression. cdc2 A possible explanation for the decrease in the expression of c-kit mRNA upon inactivation of Myb GAPDH activity in ERMYB cells is that the expression levels of c-kit are related to the dierentiation state of the cells and are indirectly regulated by Myb activity. The low level of c-kit expression observed in dierentiated hemopoietic cells is consistent with this notion (Andre et al., 1989). We wanted to determine, therefore, if the c-kit promoter was regulated by Myb. For this purpose, a reporter plasmid containing the mouse c- c-myc kit promotor linked to the CAT gene (pkit-CAT02N; Figure 7a) was co-transfected with a vector expressing GAPDH wild-type or mutant c-Myb proteins (Sakura et al., 1989) lacking either the DNA-binding domain (DDB) or the transactivation domain (DTA) into CV-1 cells. DDB and DTA mutant c-Myb proteins did not activate the c-kit reporter gene construct. In contrast, c-Myb activated c-kit reporter gene transcription 12-fold relative to cells transfected with the empty control plasmid (Figure 7b). Examination of the sequence of c-kit the c-kit promoter between 72710 and +1 revealed six potential Myb binding sites ((Deng et al., 1996); Figure GAPDH 7a). These data suggest that Myb is capable of transactivating the c-kit promoter directly, lending further support for a role of Myb in the regulation of c-kit expression.

Apoptosis in ERMYB cells Myb up-regulates bcl-2 and suppresses apoptosis in at least some myeloid and T cells (Salomoni et al., 1997; Taylor et al., 1996; Frampton et al., 1996). To establish whether ERMYB cells were undergoing apoptosis upon inactivation of Myb, the viability of ERMYB cells was examined following b-estradiol withdrawal. Viability was rapidly reduced in the absence of GM-CSF in a b- estradiol-independent manner demonstrating the Figure 6 Expression of cdc2,c-myc and c-kit mRNA in ERMYB growth factor dependence of the ERMYB cell line, as cells. The analysis of 2 mg of poly(A)+ mRNA from ERMYB cells grown with (+) or without (7) b-estradiol for 0, 6, 24 or expected (Figure 8a). When ERMYB cells were grown 48 h (lanes 1 ± 7) and from cells grown without b-estradiol for in GM-CSF in medium lacking b-estradiol a slight 24 h and to which b-estradiol was subsequently re-added for 6 or decrease in viability (10 ± 15%) was also observed. 48 h (lanes 8, 9) is shown. Northern blots were probed with cdc2, In order to determine if apoptosis accompanied the c-myc or c-kit cDNA probes (upper panels) and re-probed with decrease in viability observed upon inactivation of GAPDH (lower panels) in order to quantitate expression levels. The mRNA levels relative to GAPDH levels are shown below Myb, DNA laddering was examined in ERMYB cells each pair of Northern blots grown with or without b-estradiol. DNA laddering was Differentiation following inactivation of ER-Myb AHogget al 2892 observed in ERMYB cells grown without GM-CSF least 4 mg poly(A)+ RNA were required for its (Figure 8b). In the presence of GM-CSF, however, no detection on Northern blots (unpublished observa- signi®cant DNA laddering was observed either in the tions). A 50% reduction was observed, however, in presence or absence of b-estradiol (Figure 8b). We next ERMYB cells grown without b-estradiol for 24 h. We examined if decreased bcl-2 expression accompanied conclude that decreased bcl-2 expression correlates, to the reduced viability of ERMYB cells. The levels of a degree, with loss of Myb activity but not with an bcl-2 appear to be very low in ERMYB cells, since at increase in apoptosis, as measured by DNA laddering.

b DB TA Control c-Myb WT ∆ ∆

1234

Figure 7 Activation of the mouse c-kit promoter by c-Myb. (a) The structure of the reporter plasmid pkit-CAT02N containing the mouse c-kit promoter is shown schematically (B, BamHI; E, EcoRI; H, HindIII; X, XhoI). The sequences of six potential consensus Myb binding sites (Yasuda et al., 1993) in the c-kit promoter are shown below. (b) CAT cotransfection assays. CV-1 cells were transfected with pkitCAT02N and a plasmid expressing wild type c-myb (c-Myb WT) or the plasmid backbone (control) or mutant c-myb constructs lacking the DNA binding domain (DDB) or the transactivation domain (DTA) and the internal control plasmid pact-b-gal. CAT activity is measured and is represented as fold activation Differentiation following inactivation of ER-Myb AHogget al 2893 a

bc 2 2 2 2 2 + GM-CSF E + GM-CSF – E – GM-CSF + E – GM-CSF E 2 + E 24 Hrs – E

9.49 kb — 2.8 kb — 7.46 kb — Bcl-2

4.4 kb —

1.16 kb —

0.98kb —

0.72 kb — GAPDH 1.35kb —

0 24 Hours

Figure 8 Apoptosis and Bcl-2 expression in ERMYB cells. (a) The percentage of viable ERMYB cells is shown after growth in medium containing GM-CSF and b-estradiol (E2), GM-CSF, b-estradiol (E2) or medium alone for the indicated times. (b) DNA laddering is shown in ERMYB cells at the start of the experiment (0 h) grown previously in the presence of b-estradiol (E2) and for cells grown for 24 h in the presence (+) or absence (7) of GM-CSF and E2. The sizes of molecular weight markers are shown at the left. (c) bcl-2 mRNA expression in ERMYB cells. The analysis of 4 mg of poly(A)+ mRNA from ERMYB cells grown with (+E2) or without b-estradiol for 24 h (24 h 7E2) is shown. Northern blots were hybridized with bcl-2 (upper panel) or GAPDH (lower panel) probes. Arrows indicating bcl-2 and GAPDH mRNA transcripts are shown at the right. The sizes of molecular weight markers are shown at the left

Discussion display, in addition to an immature myeloid morphology, many of the characteristics of immature Primary murine fetal liver hemopoietic cells constitu- hemopoietic cells such as anchorage independent tively expressing Myb become immortalized and growth and the ability to form colonies in semi-solid Differentiation following inactivation of ER-Myb AHogget al 2894 medium (Macmillan and Gonda, 1994; Gonda et al., that dierent cooperating transcription factors are 1993). We have used a Myb/estrogen receptor fusion expressed in murine as compared to avian cells, and protein (ERMYB) in order to conditionally transform that the absence of one such factor prevents the re- primary murine hemopoietic cells and to investigate the activation of Myb-dependent genes necessary for the regulation of Myb target genes which may mediate the reversion of dierentiated murine ERMYB cells. ability of Myb to transform hemopoietic cells. Alternatively, very high levels of Myb expression may Although Myb/ER fusion proteins have been ex- also be necessary in order to re-transform ERMYB pressed in cell lines (e.g. Lyon and Watson, 1995; cells. Burk and Klempnauer, 1991), this is the ®rst report (to The correlation between transactivation and trans- our knowledge) of the transformation of mammalian formation by Myb (Lane et al., 1990; Kanei-Ishii et al., primary cells using a Myb/ER fusion protein. (Very 1992; Hu et al., 1991) suggests that the ability of Myb recently, Lipsick and colleagues (Engelke et al., 1997) to transform hemopoietic cells is dependent on the have reported transformation of avian yolk sac cells by transactivation of genes critical for this process. We a v-Myb-ER fusion). We have shown that if this have investigated the regulation of several putative conditionally active form of Myb is inactivated, a Myb regulated genes in the ERMYB transformed cells process of dierentiation along the macrophage/ and have found that cdc2 and c-myc mRNA levels do neutrophil pathway is observed, proliferation is not correlate with activation of Myb. Endogenous Myb inhibited and the expression of the early cell surface is not detected in the ERMYB cells (data not shown), marker, Thy-1, is ablated. That the transformed thus excluding the possibility that endogenous Myb phenotype is mediated by Myb and not estrogen is protein could mask the eect of inactivating the fusion demonstrated by the lack of estrogen receptors on Myb protein. Furthermore, we ®nd that the mRNA levels of transformed cells (data not shown), the lack of any two other putative Myb regulated genes (c-kit and bcl- discernible eect of b-estradiol on cells transformed by 2) do correlate with Myb activity. CT3Myb (data not shown) and the consistency of the It has been postulated that the regulation of ERMYB phenotype in the presence of b-estradiol with hemopoietic cell proliferation by Myb is related to Myb transformed cells described previously (Gonda et cell cycle control. In support of this notion, c-myb al., 1989a,b, 1993). antisense oligonucleotides block T-lymphocytes at the The re-addition of b-estradiol to ERMYB cells G1/S transition (Gewirtz et al., 1989). c-myb and cdc2 induced to dierentiate by b-estradiol withdrawal mRNAs are induced when quiescent T lymphocytes are resulted in full retention of their proliferative stimulated with mitogens and can be blocked by c-myb capacity, as assessed by clonogenicity in methylcellu- antisense oligonucleotides, suggesting that cdc2 and c- lose, only when the period of b-estradiol withdrawal myb regulation might be linked (Furukawa et al., was no longer than 4 h. After this time, the ability of 1990). In fact, Myb has been shown to bind to and the cells to regain their proliferative capacity was transactivate the cdc2 promoter in ®broblasts (Ku et progresssively lost over the next 72 h. The ERMYB al., 1993). However, no decrease in cdc2 expression was cells which continued to proliferate when b-estradiol observed in ERMYB cells induced to dierentiate up was re-added presumably represent a population which to 48 h following removal of b-estradiol, by which time had undergone only partial dierentiation rather than maturing granulocytes and macrophages were evident indicating a reversion of dierentiated cells. This and colony formation in methylcellulose was markedly premise is based on the lack of reversion once the reduced. It is possible that a decrease in cdc2 levels entire population had undergone complete differentia- occurs at later stages of dierentiation but clearly, tion as assessed by the morphology of cells in regulation of cdc2 is not responsible for the early stages methylcellulose, Giemsa staining, and the expression of growth inhibition and dierentiation observed upon patterns of granulocyte and macrophage-speci®c inactivation of Myb. enzymes. The latter criteria demonstrate that 1 ± 6 c-myc expression stimulates cell proliferation and days were required for fully mature neutrophils and exerts a negative eect on terminal dierentiation (Holt macrophages to develop, implying that some less et al., 1988; Homan Liebermann and Liebermann, mature cells were present at earlier times. 1991; Kume et al., 1988; Eilers et al., 1991; reviewed in In contrast, avian hemopoietic cells transformed by LuÈ scher and Eisenman, 1990). Because Myb can also an E26 mutant encoding a temperature sensitive Myb- inhibit the dierentiation of hemopoietic cells Ets protein (E26ts) have a myeloblastic phenotype at (Selvakumaran et al., 1994; Todokoro et al., 1988; 378C and dierentiate to mature macrophages after Yanagisawa et al., 1991; Clarke et al., 1988), it has shifting to 428C (Beug et al., 1987). Unlike the murine been postulated that the eect of Myb on cellular ERMYB cells described here, dierentiated E26ts cells proliferation may be due to the regulation of c-myc can revert to immature cells after returning to 378C. transcription. Indeed, transactivation by Myb of the c- They are irreversibly dierentiated, however, if kept at myc promoter (Evans et al., 1990; Nakagoshi et al., 428C for longer than 4 ± 8 days. Furthermore, AMV 1992; Zobel et al., 1992) and in one case, endogenous v-myb can transform mature avian macrophages, c-myc (Evans et al., 1990) has been reported. Unlike although this ability decreases as the cells become the studies of the regulation of c-myc by Myb reported more terminally dierentiated, and may be due to previously, our studies address the question in primary increased resistance to transformation as they with- myeloid cells in which the only lesion introduced is the draw from the cell cycle (Durban and Boettiger, 1981). expression of the ERMYB fusion protein. We The irreversibility of granulocyte/macrophage differ- demonstrate that c-myc mRNA levels are not down- entiation following withdrawal of b-estradiol from regulated upon dierentiation of ERMYB cells after ERMYB cells may re¯ect dierences between avian 48 h of b-estradiol withdrawal (although it is likely and murine myeloid cells. For example, it is possible that c-myc levels are reduced in more terminally Differentiation following inactivation of ER-Myb AHogget al 2895 dierentiated, non-proliferating cells, as seen in other expressed on hemopoietic progenitor cells and binds systems; e.g. see review by LuÈ scher and Eisenman, stem cell factor (SCF) also known as Steel factor (1990)). Importantly, these results imply that c-Myc is (SLF), which is a potent mitogen for these cells not mediating the proliferative or dierentiation- (reviewed in Galli et al., 1994). Myb-transformed cells blocking eects of Myb in these cells. Consistent with do not express SCF and it is unlikely, therefore, that our results, it has been shown that down-regulation of SCF is acting as an autocrine mitogen for ERMYB c-myc expression is not required for Epo-induced cells (Macmillan and Gonda, 1994). Rather, the dierentiation of erythroleukemia cells (Todokoro et regulation of the expression of c-kit by Myb may al., 1988). The ®nding that Myb blocks myeloid cell model the development of the hemopoietic system in dierentiation at an earlier stage than c-Myc is further vivo. W mutant mice which express a mutant c-Kit evidence that the eect of Myb is not mediated by c- protein have defective hemopoiesis aecting predomi- myc regulation (Selvakumaran et al., 1994). nantly mast cells and erythroid and myeloid lineages It may be signi®cant that the reports of transactiva- (reviewed in Galli et al., 1994). Furthermore, constitu- tion of the chromosomal c-myc (Evans et al., 1990) and tive expression of c-myb in embryonic stem cells cdc2 (Furukawa et al., 1990) genes both pertain to T induces c-kit expression and SCF-enhanced formation lymphocytes. The regulation of c-myb in mature of erythromyeloid colonies (Melotti and Calabretta, lymphoid cells is cell cycle-speci®c, in contrast to 1996). Taken together, these data support the notion earlier T cells (Catron et al., 1992; reviewed in that by regulating c-kit expression, Myb modulates the Thompson and Ramsay, 1995). Thus c-myb may play response of myeloid cells to the c-kit ligand during the a somewhat dierent role in more mature cells, e.g. in development of hemopoietic cells in vivo. We cannot mitogen-induced cell cycle progression, and may rule out, however, that c-Kit may have other as yet therefore act on dierent target genes. unknown functions which contribute to the Myb- The regulation of apoptosis and bcl-2 expression by transformed phenotype; for example, some isoforms Myb has recently been reported (Salomoni et al., 1997; of c-Kit exhibit a low level of ligand-independent Taylor et al., 1996; Frampton et al., 1996). We show a activity (Reith et al., 1991). continuing loss of viability and increased apoptosis in The conditionally transformed hemopoietic cells ERMYB cells when deprived of GM-CSF. In contrast, described here have provided an opportunity to inactivation of Myb in the presence of GM-CSF results examine the mechanism by which Myb transforms in a small decrease in viability and no detectable primary cells and allowed us to study the regulation of apoptosis. Our results do support the above-mentioned putative Myb target genes in myeloid cell differentia- reports of regulation of bcl-2 expression by Myb, tion. Our results do not support a role for c-myc or although the contribution of Myb to total bcl-2 cdc2 in mediating the transforming eects of Myb. expression appears much less in ERMYB cells. However, our results support direct regulation of c-kit However, the levels of bcl-2 expression are very low expression by Myb in this system. In addition, these in the ERMYB cells, and are presumably not sucient data con®rm that bcl-2 expression correlates with Myb to prevent the cell death observed upon growth factor activity but do not support an anti-apoptotic role for withdrawal. The anti-apoptotic eects of Myb appear Myb in this system. The ERMYB cells should provide to depend on the apoptotic stimulus and may also be a valuable reagent for dissecting the function of Myb cell type-speci®c (Salomoni et al., 1997; Frampton et in myeloid cell proliferation and dierentiation and for al., 1996). Thus, we conclude that mechanisms other the identi®cation of other Myb target genes. than the protection of cells from apoptosis are responsible for the transformation of murine myeloid cells by Myb. We show that inactivation of a Myb/estrogen Materials and methods receptor fusion protein in ERMYB cells precedes an Plasmid construction almost complete loss of c-kit mRNA expression. This decrease in c-kit mRNA expression occurs between 6 A retroviral vector expressing an estradiol-inducible Myb and 24 h following inactivation of Myb. Our results protein was constructed using the RED(CT3Myb) retro- demonstrating transactivation of a c-kit reporter viral vector encoding a C-terminally truncated murine Myb construct by Myb but not by mutant Myb proteins protein (CT3Myb) (Hu et al., 1991; Sakura et al., 1989). In order to facilitate cloning of the estrogen receptor ligand lacking the DNA binding or transactivation domains binding domain into the RED(CT3Myb) vector, a BamHI are consistent with direct regulation of the c-kit gene, site was introduced upstream of the gag-myb junction by as we show that Myb transactivates an 8.8 kb region of inserting the oligonucleotides 5'-TCGATCTTGGATC- the c-kit promoter which contains six potential Myb- CAGCC-/3'AGAACCTAGGTCGGAGCT into a XhoI binding sites. These data support and extend other site located 264 bp upstream of the myb coding sequence. studies documenting a correlation between Myb and c- The adaptor introduced a BamHI site and destroyed one of Kit expression (Melotti and Calabretta, 1996; Ratajc- the XhoI sites. In order to construct a vector encoding an zak et al., 1992) but, in addition, demonstrate that estrogen-inducible Myb protein [RED(ERMYB)] the Myb can transactivate the c-kit promoter in vitro. estrogen binding domain of the human estrogen receptor More conclusive proof of a direct eect will require (amino acids 282 ± 576) (Green et al., 1986) was ligated into the BamXI ± XhoI cleaved vectors. A non-hormone mutation of the Myb binding sites in the c-kit inducible control for use in chloramphenicol acetyltrans- promoter. The dierent levels of induction of the c- ferase (CAT) assays was constructed by inserting the viral kit promoter in ERMYB cells (200-fold) compared oncogene v-myb into the pUC13 plasmid downstream of with that seen in CV-1 cells (12-fold) may be due to the long terminal repeat of Schmidt-Ruppin strain of Rous cell speci®c cooperating factors or to the levels of Sarcoma Virus (nucleotide residue 8 ± 376) (Gorman, 1985). active Myb protein achieved in each system. c-kit is The reporter plasmid p8mybbCAT contains the SacI±NcoI Differentiation following inactivation of ER-Myb AHogget al 2896 fragment of the plasmid pb128 (Weston and Bishop, 1989) and resuspended in DMEM/16% FCS. Viable non- inserted 5' to the CAT gene in the CATbasic vector adherent and adherent cells were counted by Trypan blue (Promega). exclusion using a hemocytometer.

CAT assays Morphological characteristics of ERMYB cells The dependence of ERMyb proteins on b-estradiol was In order to examine the aect of b-estradiol on ERMYB assessed in Balb/c 3T3 ®broblasts grown in DMEM/10% cells, cytospins were prepared following growth in the FCS. DNA was introduced by the calcium phosphate co- presence or absence of b-estradiol. 56103 cells were precipitation method (Weber et al., 1984). The precipitates resuspended in DMEM/50% FCS and cytospun onto contained 10 mg each of the eector and the reporter and glass slides. Cells were dried brie¯y at room temperature, 5 mg of the internal standard plasmids (SV-b-Galactosidase stained with May-GruÈ nwald-Giemsa stain and subjected to vector; Promega). The cells were incubated for 12 ± 14 h microscopic examination. with precipitates, washed and fresh medium was added. Parallel dishes were treated with 1077 M b-estradiol or the solvent ethanol. Cell extracts were prepared 48 h later and Esterase staining aliquots used for CAT assays. The amount of acetylated Cells were resuspended in DMEM/50% FCS, cytospun 3 [ H]chloramphenicol was determined by diusion into onto slides and ®xed in esterase ®xative (2.86 mM scintillation ¯uid (Neuman et al., 1987) and the data were Na2HPO4, 14.7 mM KH2PO4, 45% v/v acetone, 25% v/v normalized with respect to b-galactosidase activity. formalin) for 30 s at 48C. The a-napthyl acetate substrate In order to determine if Myb transactivated the c-kit solution was then prepared by adding 1 ml of 10 mg/ml a- promoter, the CAT reporter plasmid pkit-CAT02N, in which napthyl acetate in acetone and 30 ml of 0.066 M phosphate the 8.8 kb DNA fragment containing the mouse c-kit buer pH 6.3 to a Coplin jar followed by 0.3 ml of 4% promoter was linked to the CAT gene at the SacI site pararosanalin/2 M HCl which had been mixed with 0.3 ml (nucleotide 728 according to the numbering system in which of 4% sodium nitrite. Following three washes in distilled the A of the initiator ATG codon is +1) (Hayashi et al., water, the cells were stained in the a-napthyl acetate 1995). A mixture of 6 mg of pkit-CAT02N, 6 mg of the pact substrate solution for 45 min at room temperature. The expression plasmid encoding either WT c-Myb, DTA Myb, slides were washed three times in distilled water and DDB Myb or no protein (Sakura et al., 1989), and 1 mgof stainedin20mgofFastBluein38mlof0.066M the internal control plasmid pact-b-gal was transfected into phosphate buer pH 7.4 and 5 mg of napthol-AS-D- CV-1 cells, and CAT assays were performed as described chloracetate dissolved in 2.5 ml of N-N dimethylforma- (Sakura et al., 1989). mide for 1 h at room temperature. Finally, the slides were washed three times in distilled water and counterstained in 2% Methyl Green. Cells expressing a-napthylacetate Transformation of hemopoietic cells with the ERMYB retro- esterase (macrophage esterase) were identi®ed by their virus red-brown staining and those expressing naphthol-AS-D- Detailed procedures for the infection of murine hemopoie- acetate (neutrophil esterase) by their blue staining. tic cells by retroviruses have been described elsewhere (Gonda et al., 1989a, 1993; Hu et al., 1991). Brie¯y, viral Colony formation packaging C2 cells were transfected with 10 mg RED(ERMYB) and 0.1 mg pSV2Neo vector using calcium Colony formation was assessed by plating cells at 5000/ml phosphate (Gonda et al., 1989a). Transfected cells were in Iscove's modi®ed Dulbecco medium (IMDM) containing selected in DME/10% FCS containing 400 mg/ml G418. 1.3% methylcellulose, 10% FCS and 400 U/ml GM-CSF Murine fetal liver cells from CBA mice at 14 days of either with or without 1 mM b-estradiol. Colonies 430 cells gestation were infected by co-cultivation with irradiated and clusters 530 cells were scored following 7 and 14 days C2 cells (30 Gy) transfected with RED(ERMYB). Co- of incubation. cultivations were performed in the presence of 400 U/ml GM-CSF, 800 U/ml IL-3 and 1 mM b-estradiol. Analysis of surface marker expression The fetal liver cells infected with RED(ERMYB) (ERMYB cells) were washed twice in phosphate-buered Analysis of cell surface antigens was performed essentially saline (PBS) and resuspended in 5 ml of DMEM containing as described elsewhere (Gonda et al., 1993; and references antibiotics, 16% charcoal-stripped fetal calf serum, 400 U/ml therein). Rat monoclonal antibodies to the surface markers GM-CSF and 1 mM b-estradiol. Initially, the cells were Mac-1, F4/80, Thy-1.2 and Mac-3 were used to stain 105 cultured in the presence of a feeder layer of irradiated ERMYB cells in a volume of 50 ml on ice for 45 min. Cells NIH3T3 ®broblasts, as described previously, until they were were washed twice in DME/5% FCS and incubated with established (Macmillan and Gonda, 1994; Gonda et al., ¯uorescein-conjugated anti-rat IgG (Silenius Laboratories, 1993). After 8 weeks, they were maintained without feeder Melbourne, Australia) on ice for 45 min. Cells were washed layers in DMEM containing antibiotics, 16% charcoal- twice and analysed by ¯ow cytometry using a Coulter stripped fetal calf serum, 400 U/ml GM-CSF and 1 mM b- Prolife II (Hialeah, FL). estradiol. Analysis of cellular RNA and DNA Analysis of adherence of ERMYB cells Total cellular polyadenylated (poly(A)+) RNA or genomic ERMYB cells were washed three times in PBS and DNA was isolated from cells and analysed using Northern resuspended in fresh medium with or without b-estradiol and Southern blotting, respectively. The procedures used at a density of 26105/ml. Following incubation with or for Southern and Northern blotting have been described without b-estradiol, non-adherent cells were decanted and previously (Gonda et al., 1989b). 2 mgofpoly(A)+ mRNA the remaining adherent cells were harvested using or 10 mg of genomic DNA were analysed. Probes were 32P- Lidocaine (Sigma). 4 ml of 0.4% Lidocaine in DMEM labeled using a Multiprime random priming kit (Amer- was added to 75 cm2 ¯asks and the cells were incubated at sham). The 5' c-myb cDNA probe consisted of the 5' 378C for 10 min. After knocking the ¯asks to detach the (0.7 kb) EcoRI fragment of the c-myb clone, pMM49 cells, 1 ml of FCS was added and the cells were centrifuged (Gonda et al., 1985). The human estrogen receptor probe Differentiation following inactivation of ER-Myb AHogget al 2897 consisted of the 0.9 kb BamHI/XhoI fragment of (Schleicher and Schuell) for immunoblotting analysis. RED(ERMYB) comprising the ligand binding domain. Membranes were blocked in TBS-T (50 mM Tris-HCl The cdc2 probewasa0.5kbKpnI/BglII fragment derived pH 7.4, 135 mM NaCl, 0.1% Tween 20) containing 5% from the coding region of human cdc2 cDNa in the skim milk. Myb proteins were detected using the 5.1 plasmid pOB231 (Lee and Nurse, 1987). The c-myc probe monoclonal c-Myb antibody (Ramsay et al., 1989), consisted of a 1.6 kb HindIII/SacI fragment from the horseradish peroxidase-conjugated anti-mouse IgG di- murine c-myc cDNA. The c-kit probe comprised the 3 kb luted 1/1000 (Amersham) and visualized using ECL BamHI/BglII fragment from murine c-kit cDNA and the (Amersham). GAPDH probe consisted of the 0.3 kb HindIII/PstI fragment derived from murine GAPDH in the plasmid Determination of cell viability and DNA fragmentation pmGAP5'.Thebcl-2 probewasa0.92kbEcoRI±TaqI fragment of human bcl-2 cDNA (comprising the coding Cell viability was assessed by Trypan blue exclusion and sequence plus *0.19 kb 3'-untranslated sequence). microscopic examination. DNA fragmentation was determined essentially as described (RoÈ sl, 1992) by labeling cellular DNA with DNA polymerase I (Klenow fragment) Immunoblotting and [32P]-dATP, followed by electrophoresis in a 1.8% Cellular extracts were prepared from 56107 cells. Non- agarose gel and Phosphorimager detection. adherent cells were decanted, washed twice in PBS and collected by centrifugation. 1 ml of lysis buer (2% SDS, 10% glycerol, 0.06 M Tris pH 6.8, 0.01 M dithiothreitol, 2mM phenylmethylsulfonyl¯uoride, 1 mg/ml leupeptin and Acknowledgements 1 mg/ml apportioning) was added to the adherent cells in The authors thank Giresh Chandran and Petranel Ferrao the ¯ask. The lysed cells were removed by scraping and for their contributions to the preliminary stages of this added to the non-adherent cell pellet. The cells were work, and Dr David Horsfall (Flinders Medical Centre, sonicated, boiled for 3 min and centrifuged at Adelaide) for performing the estrogen receptor assays. This 45 000 r.p.m. (TLA-45 rotor; Beckman TL-100 ultracen- work was supported in part by a project grant (to TJG) trifuge) for 45 min at 48C. The supernatants were stored from the National Health and Medical Research Council at 7708C. Cellular extracts were diluted in 1 volume of (NHMRC) of Australia. AH was supported by an sample buer (20% glycerol, 4% SDS, 0.125 M Tris pH NHMRC post-doctoral fellowship, PB is the recipient of 6.8, 0.01% bromophenol blue and 0.7 M b-mercaptoetha- a Dawes postgraduate scholarship from the Royal nol) and separated on 8% polyacrylamide/SDS gels. Adelaide Hospital, and TJG is a Senior Research Fellow Proteins were transferred to nitrocellulose membrane of the NHMRC.

References

Andre C, d'Auriol L, Lacombe C, Gisselbrecht S and Evans JL, Moore TL, Kuehl WM, Bender T and Ting JP. Galibert F. (1989). Oncogene, 4, 1047 ± 1049. (1990). Mol. Cell. Biol., 10, 5747 ± 5752. Anfossi G, Gewirtz AM and Calabretta B. (1989). Proc. Ferrao P, Macmillan EM, Ashman LK and Gonda TJ. Natl. Acad. Sci. USA, 86, 3379 ± 3383. (1995). Oncogene, 11, 1631 ± 1638. Beug H, Blundell PA and Graf T. (1987). Genes Dev., 1, 277 ± Frampton J, Ramqvist T and Graf T. (1996). Genes Dev., 10, 286. 2720 ± 2731. Boehmelt G, Walker A, Kabrun N, Mellitzer G, Beug H, Furukawa Y, Piwnica-Worms H, Ernst T, Kanakura Y and Zenke M and Enrietto PJ. (1992). EMBO J., 11, 4641 ± Grin JD. (1990). Science, 250, 805 ± 808. 4652. Galli SJ, Zsebo KM and Geissler EN. (1994). Adv. Immunol., Burk O, Mink S, Ringwald M and Klempnauer KH. (1993). 55, 1 ± 96. EMBO J., 12, 2027 ± 2038. Gewirtz AM, Anfossi G, Venturelli D, Valpreda S, Sims R Burk O and Klempnauer KH. (1991). EMBO J., 10, 3713 ± and Calabretta B. (1989). Science, 245, 180 ± 183. 3719. Gonda TJ, Gough NM, Dunn AR and De Blaquiere J. Catron KM, Purkerson JM, Isakson PC and Bender TP. (1985). EMBO J., 4, 2003 ± 2008. (1992). J. Immun., 148, 934 ± 942. Gonda TJ, Buckmaster C and Ramsay RG. (1989a). EMBO Clarke MF, Kukowska-Latallo JF, Westin E, Smith M and J., 8, 1777 ± 1783. Prochownik EV. (1988). Mol. Cell. Biol., 8, 884 ± 892. Gonda TJ, Ramsay RG and Johnson GR. (1989b). EMBO Cogswell JP, Cogswell PC, Kuehl WM, Cuddihy AM, J., 8, 1767 ± 1775. Bender TM, Engelke U, Marcu KB and Ting JP. (1993). Gonda TJ, Macmillan EM, Townsend PV and Hapel AJ. Mol. Cell. Biol., 13, 2858 ± 2869. (1993). Blood, 82, 2813 ± 2822. Craig RW. (1995). Sem. Cancer Biol., 6, 35 ± 43. Gonda TJ and Metcalf D. (1984). Nature, 310, 249 ± 251. Dai P, Akimaru H, Tanaka Y, Hou DX, Yasukawa T, Gorman C. (1985). DNA Cloning: A practical approach. IRL Kanei-Ishii C, Takahashi T and Ishii S. (1996). Genes Press, Oxford, pp. 143 ± 190. Dev., 10, 528 ± 540. Green S, Walter P, Kumar V, Krust A, Bornert J-M, Argos P Deng QL, Ishii S and Sarai A. (1996). Nucleic Acids Res., 24, and Chambon P. (1986). Nature, 320, 134 ± 139. 766 ± 774. Hayashi S-I, Kunisada T, Ogawa M and Nishikawa S-I. Durban EM and Boettiger D. (1981). J. Virol., 37, 488 ± 492. (1995). DNA Res., 2, 21 ± 218. Eilers M, Picard D, Yamamoto KR and Bishop JM. (1989). Hernandez-Munain C and Krangel MS. (1995). Mol. Cell. Nature, 340, 66 ± 68. Biol., 15, 3090 ± 3099. Eilers M, Schirm S and Bishop JM. (1991). EMBO J., 10, Homan Liebermann B and Liebermann DA. (1991). Mol. 133 ± 141. (Abstract). Cell. Biol., 11, 2375 ± 2381. Engelke U, Wang DM and Lipsick JS. (1997). J. Virol., 71, Holt JT, Redner RL and Nienhuis AW. (1988). Mol. Cell. 3760 ± 3766. Biol., 8, 963 ± 973. Evan GI, Wyllie AH, Gilbert CS, Littlewood TD, Land H, Hu YL, Ramsay RG, Kanei Ishii C, Ishii S, Gonda TJ and Brooks M, Waters CM, Penn LZ and Hancock DC. Kanei-Ishii C. (1991). Oncogene, 6, 1549 ± 1553. (1992). Cell, 69, 19 ± 128. Differentiation following inactivation of ER-Myb AHogget al 2898 Kanei-Ishii C, Macmillan EM, Nomura T, Sarai A, Ramsay Ramsay RG, Ishii S, Nishina Y, Soe G and Gonda TJ. RG, Aimoto S, Ishii S and Gonda TJ. (1992). Proc. Natl. (1989). Oncogene Res., 4, 259 ± 269. Acad. Sci. USA, 89, 3088 ± 3092. Ratajczak MZ, Luger SM, DeRiel K, Abrahm J, Calabretta Ku DH, Wen SC, Engelhard A, Nicolaides NC, Lipson KE, B and Gewirtz AM. (1992). Proc. Natl. Acad. Sci. USA, Marino TA and Calabretta B. (1993). J. Biol. Chem., 268, 89, 1710 ± 1714. 2255 ± 2259. Reith AD, Ellis C, Lyman SD, Anderson DM, Williams DE, Kume TU, Takada S and Obinata M. (1988). J. Mol. Biol., Bernstein A and Pawson T. (1991). EMBO J., 10, 2451 ± 202, 779 ± 786. 2459. (Abstract). Lane T, Ibanez C, Garcia A, Graf T and Lipsick J. (1990). RoÈ sl F. (1992). Nucleic Acids Res., 20, 5243. Mol. Cell. Biol., 10, 2591 ± 2598. Sakura H, Kanei-Ishii C, Nagase T, Nakagoshi H, Gonda TJ Lee MG and Nurse P. (1987). Nature, 327, 31 ± 35. and Ishii S. (1989). Proc. Natl. Acad. Sci. USA, 86, 5758 ± Lipsick JS. (1987). J. Virol., 61, 3284 ± 3287. 5762. LuÈ scher B and Eisenman RN. (1990). Genes Dev., 4, 2025 ± Salomoni P, Perrotti D, Martinez R, Franceschi C and 2035. Calabretta B. (1997). Proc. Natl. Acad. Sci. USA, 94, Lyon JJ and Watson RJ. (1995). Dierentiation, 59, 171 ± 3296 ± 3301. 178. Selvakumaran M, Lin HK, Sjin RT, Reed JC, Liebermann Macmillan EM and Gonda TJ. (1994). Blood, 83, 209 ± 216. DA and Homan B. (1994). Mol. Cell. Biol., 14, 2352 ± McCracken S, Leung S, Bosselut R, Ghysdael J and 2360. Miyamoto NG. (1994). Oncogene, 9, 3609 ± 3615. Shapiro LH. (1995). J. Biol. Chem., 270, 8763 ± 8771. Melotti P, Ku DH and Calabretta B. (1994). J. Exp. Med., Siu G, Wurster AL, Lipsick JS and Hedrick SM. (1992). Mol. 179, 1023 ± 1028. Cell. Biol., 12, 1592 ± 1604. Melotti P and Calabretta B. (1996). Blood, 87, 2221 ± 2234. Taylor D, Badiani P and Weston K. (1996). Genes Dev., 10, Mucenski ML, McLain K, Kier AB, Swerdlow SH, Schreiner 2732 ± 2744. CM, Miller TA, Pietryga DW, Scott Jr, WJ and Potter SS. Thompson MA and Ramsay RG. (1995). Bioessays, 17, (1991). Cell, 65, 677 ± 689. 341 ± 350. Nakagoshi H, Nagase T, Kanei-Ishii C, Ueno Y and Ishii S. Todokoro K, Watson RJ, Higo H, Amanuma H, Kuramochi (1990). J. Biol. Chem., 265, 3479 ± 3483. S, Yanagisawa H and Ikawa Y. (1988). Proc. Natl. Acad. Nakagoshi H, Kanei Ishii C, Sawazaki T, Mizuguchi G, Ishii Sci. USA, 85, 8900 ± 8904. S and Kanei-Ishii C. (1992). Oncogene, 7, 1233 ± 1240. van den Heuvel S and Harlow E. (1993). Science, 262, 2050 ± Ness SA, Kowenz-Leutz E, Casini T, Graf T and Leutz A. 2054. (1993). Genes Dev., 7, 749 ± 759. Weber F, de Villiers J and Schaner W. (1984). Cell, 36, Newman JR, Morency CA and Russian KO. (1987). 983 ± 982. BioTechniques, 5, 444. Weston K and Bishop JM. (1989). Cell, 58, 85 ± 93. Nishina Y, Nakagoshi H, Imamoto F, Gonda TJ and Ishii S. Yanagisawa H, Nagasawa T, Kuramochi S, Abe T, Ikawa Y (1989). Nucleic Acids Res., 17, 107 ± 117. and Todokoro K. (1991). Biochim. Biophys. Acta, 1088, Oelgeschlager M, Nuchprayoon I, LuÈ scher B and Friedman 380 ± 384. AD. (1996). Mol. Cell. Biol., 16, 4717 ± 4725. Yasuda H, Galli SJ and Geissler EN. (1993). Biochem. OgataK,MorikawaS,NakamuraH,SekikawaA,InoueT, Biophys. Res. Commun., 191, 893 ± 901. Kanai H, Sarai A, Ishii S and Nishimura Y. (1994). Cell, Zobel A, Kalkbrenner F, Vorbrueggen G and Moelling K. 79, 639 ± 648. (1992). Biochem. Biophys. Res. Commun., 186, 715 ± 722.