Leukemic transformation of normal murine erythroid ... - Nature

Oncogene (2001) 20, 3651 ± 3664 ã 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

Leukemic transformation of normal murine erythroid progenitors: v- and c-ErbB act through signaling pathways activated by the EpoR and c-Kit in stress erythropoiesis Marieke von Lindern1,5, Eva Maria Deiner2,5, Helmut Dolznig3, Martine Parren-van Amelsvoort1, Michael J Hayman4, Ernst W Mullner3 and Hartmut Beug*,2 1

Institute of Hematology, Erasmus Medical Centre Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands; 2Institute of Molecular Pathology, Dr. Bohrgasse 7, A-1030 Vienna, Austria; 3Institute of Medical Biochemistry, Vienna Biocenter (VBC), Dr. Bohr-Gasse 9, A-1030 Vienna, Austria; 4Department of Molecular Genetics & Microbiology, Life Sciences Bldg, State University of New York at Stony Brook, Stony Brook, New York, NY 11794, USA

Primary erythroid progenitors can be expanded by the synergistic action of erythropoietin (Epo), stem cell factor (SCF) and glucocorticoids. While Epo is required for erythropoiesis in general, glucocorticoids and SCF mainly contribute to stress erythropoiesis in hypoxic mice. This ability of normal erythroid progenitors to undergo expansion under stress conditions is targeted by the avian erythroblastosis virus (AEV), harboring the oncogenes vErbB and v-ErbA. We investigated the signaling pathways required for progenitor expansion under stress conditions and in leukemic transformation. Immortal strains of erythroid progenitors, able to undergo normal, terminal di€erentiation under appropriate conditions, were established from fetal livers of p537/7 mice. Expression and activation of the EGF-receptor (HER-1/ c-ErbB) or its mutated oncogenic version (v-ErbB) in these cells abrogated the requirement for Epo and SCF in expansion of these progenitors and blocked terminal di€erentiation. Upon inhibition of ErbB function, di€erentiation into erythrocytes occurred. Signal transducing molecules important for renewal induction, i.e. Stat5- and phosphoinositide 3-kinase (PI3K), are utilized by both EpoR/c-Kit and v/c-ErbB. However, while v-ErbB transformed cells and normal progenitors depended on PI3K signaling for renewal, c-ErbB also induces progenitor expansion by PI3K-independent mechanisms. Oncogene (2001) 20, 3651 ± 3664. Keywords: erythropoiesis; signal transduction; cellular transformation Introduction In hematopoiesis, the pluripotent stem cell has to decide between self-renewal (long-term proliferation without lineage commitment) and entering commit-

*Correspondence: H Beug, E-mail: [email protected] 5 The ®rst two authors contributed equally to this work Received 25 January 2001; revised 21 March 2001; accepted 2 April 2001

ment/di€erentiation pathways. During commitment, the cells undergo a limited number of cell divisions, followed by terminal di€erentiation. General concepts state that the stem cell's ability for self-renewal is irreversibly lost during lineage commitment (for review see Keller, 1992; Till and McCulloch, 1980). However, in leukemia, progenitors derived from both pluripotent stem cells and multi- or unipotent progenitors can undergo long-term proliferation without entering terminal di€erentiation. The recent observation, that multipotent (Nutt et al., 1999), B-lymphoid (Rolink et al., 1991) and erythroid progenitors (Steinlein et al., 1995) can be continuously propagated as undi€erentiated cells in vitro strongly suggests that committed progenitors may have a large, or even unlimited, renewal potential under particular physiological conditions. However, the mechanisms responsible for this `renewal' of normal or leukemic committed progenitors are unknown. Transformation by the retrovirus AEV (avian erythroblastosis virus) enables erythroid progenitors to undergo prolonged renewal. AEV encodes v-ErbB, a truncated and mutated version of the avian EGFreceptor, EGFR/c-ErbB (Downward et al., 1984), which is sucient for in vitro transformation and leukemogenesis in chicks (Beug et al., 1985). Subsequently, nontransformed committed erythroid progenitors capable of extended renewal were identi®ed (Pain et al., 1991), which express endogenous c-ErbB (Hayman et al., 1993). These cells undergo renewal in response to ligands for c-ErbB (TGFa) and estrogen- and glucocorticoid receptors (ER, GR, Wessely et al., 1997a,b). A second, apparently distinct avian erythroid progenitor can be expanded in the presence of Erythropoietin (Epo) and stem cell factor (SCF), again in cooperation with ligands for the GR and ER (Wessely et al., 1999). This type of progenitor has also been identi®ed in human bone marrow, cord blood (von Lindern et al., 1999) and mouse fetal liver (Reichardt et al., 1998). Interestingly, expression of the v-ErbB oncogene or the avian EGFR in Epodependent avian progenitors was found to substitute for the combined action of the EpoR and c-Kit to

ErbB induced transformation of erythroid progenitors M von Lindern et al

3652

induce long-term erythroblast proliferation (Bauer et al., 2001; Beug et al., 1996; Gandrillon et al., 1999). Thus, leukemic transformation by v-ErbB appears to use mechanisms that can be activated by physiological stimuli to induce expansion of normal progenitors. Recently, the physiological relevance of erythroid progenitor expansion in vitro in response to the activation of plasma membrane- and nuclear receptors became clear. Mice lacking the EpoR fail to generate erythrocytes (Wu et al., 1995b), while c-Kit-de®cient animals have early defects in hematopoiesis. However, adult mice can live without activation of c-Kit. Mice lacking Stat5 or the GR show no apparent erythropoiesis defects (Reichardt et al., 1998; Teglund et al., 1998). However, the GR and c-Kit are indispensable for stress erythropoiesis (Bauer et al., 1999; Broudy et al., 1996). Likewise, a potential function for Stat5 in stress hematopoiesis was recently identi®ed using genetically modi®ed mice, both in the myeloid and the erythroid lineage (Kieslinger et al., 2000; Deiner et al. (2001) in preparation). Notably, STAT5-de®cient mice failed to develop certain types of leukemia (Levy and Gilliland, 2000). Thus, despite its apparent redundancy in standard hematopoiesis, STAT5 is required both for stress erythropoiesis and leukemic transformation. In stress erythropoiesis, erythroid progenitors are required to rapidly proliferate without signi®cant di€erentiation in order to replenish cells within a lineage. The notion that certain signaling molecules are primarily involved in stress erythropoiesis lends itself to a hypothesis that these signaling molecules could be important both in renewal and in leukemic transformation. We have addressed this hypothesis in the present paper. We established a novel in vitro model to study renewal and terminal di€erentiation of murine erythroid progenitors. Subsequently we use this model to demonstrate that the EGFR (c-ErbB) or its mutated oncogenic version (v-ErbB) can abrogate the requirement for Epo and SCF in erythroid renewal divisions. Signal transduction analysis of these erythroblasts indicates that renewal induction by the EGF-R utilizes signaling pathways activated by the EpoR and c-Kit (Stat5 and PI3K). In addition the EGFR is able to induce distinct pathways contributing to renewal. VErbB, on the other hand, induces renewal by the same signaling pathways employed by EpoR/c-Kit.

Results Mortal and immortal murine erythroid progenitors Primary avian and human erythroblasts undergo sustained proliferation in the presence of Epo, SCF and Dex. Trials to apply the same approach to primary murine erythroid progenitors from fetal livers yielded only short-term cultures unsuitable for extensive biochemical characterization (Reichardt et al., 1998). To overcome this problem we employed a serum-free Oncogene

medium (Stem Pro 34TM) and the ®nding that oncogene-transformed hematopoietic cells from p537/7 mice rapidly and spontaneously immortalized (Metz et al., 1995) Fetal livers (E12.5 ± 14.5) from p537/7 and wt control mice were cultivated in serum-free medium supplemented with Epo, SCF and Dex. In several experiments, cultures derived from p53 wt control mice proliferated for 14 ± 16 days, yielding between 107 and 108 cells. Thereafter, these cultures became stationary, showing increased terminal di€erentiation and cell death (Figure 1a). Cells from p537/7 fetal livers proliferated similar to wt-cells for the ®rst 15 days. Thereafter, p537/7 erythroblasts continued to proliferate, although at a reduced rate and accompanied by spontaneous di€erentiation (Figure 1a). Mature cells were removed by Percoll/Ficoll puri®cation at regular intervals until the cultures resumed their initial growth rate and behaved like immortalized cells (Figure 1a). Cells from the established p537/7 cultures expressed high levels of c-Kit (CD117) and transferrin receptor (CD71), low-levels of Ter119 and failed to express markers for myeloid progenitors (Mac-1, GR1), lymphoid progenitors (B220) or multipotent cells (CD34, Sca-1; Figure 1b and data not shown). This, together with their morphology characterizes them as proerythroblasts. The p537/7 erythroblasts retained their strict dependence on Epo, SCF and Dex. If any of the three factors was removed, the cells ceased to grow and underwent either di€erentiation (following the withdrawal of SCF or Dex) or apoptosis (following the withdrawal of Epo; Figure 1c). To induce di€erentiation, the `renewal factors' Epo, SCF and Dex were replaced by the `di€erentiation factors' Epo and insulin. The murine p537/7 erythroblasts underwent 3 ± 4 cell divisions within 72 h and then ceased to proliferate (Figure 2a, top). Their volume decreased from 500 femtoliters (¯) to below 100 ¯ (Figure 2a bottom). Furthermore, cell cycle length decreased from 24 to 11 h within 12 ± 15 h and stayed that short until 36 h after di€erentiation induction. This was mirrored by a similar shortening of the G1 phase from 11 to 4 h as measured by DNAstaining and FACS-analysis. The S and G2 phases stayed roughly constant (Figure 2b). After 48 h, the G1 phase and thus also cell cycle time increased dramatically (Figure 2b). In contrast, control cells maintained under renewal conditions proliferated exponentially with a constant cell cycle and G1/S/G2 phase, and maintained their large size (Figure 2a and data not shown). This is similar to what we previously observed in avian erythroid progenitors (Dolznig et al., 1995). Analysis of cytospins showed an increased number of mitoses after 36 h, polychromatic erythroblasts after 55 h and enucleated red cells after 72 h (Figure 2c). Essentially all cells underwent terminal di€erentiation, the frequency of undi€erentiated cells was below 1% (data not shown). In addition, the cells accumulated hemoglobin to levels resembling those in normal, peripheral murine erythrocytes (Figure 2d).

ErbB induced transformation of erythroid progenitors M von Lindern et al

3653

Figure 1 Long-term/unlimited proliferation of primary murine erythroblasts from fetal liver cells of wild-type- and p537/7 mice. (a) Single cell suspensions from fetal livers of p537/7 mice or wild-type control mice were seeded in StemPro34TM serum-free medium supplemented with Epo, SCF and Dex and cumulative cell numbers were determined daily. (b) p537/7 cells obtained at day 32 (arrow in a) represent murine proerythroblasts. Cells were subjected to FACS analysis (black curves), using PE-labeled antibodies against transferrin receptor (Tf-R, CD71) c-Kit (CD117) and the late erythroid marker Ter119. Gray curves represent staining with an irrelevant control antibody. (c). Factor dependence of immortalized p537/7 erythroblasts. Cells from day 32 (arrow in a) were seeded in Stem-Pro supplemented with Epo, SCF and Dex (Control), similar medium with 2 mM of the speci®c c-Kit inhibitor ExBW50 instead of SCF (no SCF), a 1 : 50 dilution of Epo-neutralizing antibody instead of Epo (no Epo), 3 mM of the GR antagonist ZK112.993 instead of Dex (no Dex). Cells stop to proliferate and undergo either di€erentiation (no SCF, no Dex) or apoptosis (no Epo, not shown) unless supplied with Epo, SCF and Dex at the same time

In conclusion, immortal cultures of p537/7 erythroid progenitors, capable of factor-dependent expansion or terminal di€erentiation can be established in a highly reproducible fashion. P537/7 erythroblasts can be reproducibly cloned without loss of genetic stability For further studies it was important to isolate and characterize stable, single cell-derived cultures. p537/7 erythroblasts from mass cultures were seeded in serum-free semisolid medium (StemPro-Methocult).

Cloning eciency was *10%. After 10 days, more than 100 large colonies were isolated and expanded in serum-free medium. Sixty-four immature clones were analysed for factor dependence and terminal di€erentiation. Twenty-six of these grew slowly and half of them showed a considerable number of aneuploid cells. Therefore, these cells were discarded. The remaining 37 clones behaved closely similar to the cells from mass cultures before cloning for all parameters tested. Their size and hemoglobin content following di€erentiation induction di€ered less than 20%. Upon FACS analysis, they retained DNA pro®les characteristic of normal, Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3654

of chromosome spreads showed a diploid number of chromosomes (39 ± 40) without visible chromosome abnormalities. In conclusion, erythroid progenitors from p537/7 mice combine an unlimited life-span with primary erythroid progenitor behavior. Genetic instability caused by the lack of p53 can be avoided, if exposure of the cells to stress conditions is prevented. The human EGF-receptor (c-ErbB) and its oncogenic variant v-ErbB transform p537/7 erythroblasts The serum-free cultures of established p537/7 erythroblasts lend themselves to study signaling pathways involved in the control of renewal versus terminal di€erentiation and to determine, whether signaling pathways involved in stress-induced renewal are targeted by oncogenes. Therefore, we determined whether the expression of c-ErbB and its leukemic counterpart v-ErbB would transform (i.e. induce sustained renewal in) murine erythroblasts and whether v- or c-ErbB still required cooperation with the ligandactivated GR.

Figure 2 p537/7 erythroblasts recapitulate the normal di€erentiation program of primary erythroid progenitors. (a) P537/7 erythroblasts from day 32 (see Figure 1) were switched to serumcontaining di€erentiation medium supplemented with Epo (10 U/ ml), insulin and the GR-inhibitor ZK112.993 and analysed for various cell parameters at the time intervals indicated (a ± d). Results from three di€erent experiments (diamonds, circles, triangles) are shown. Proliferating control cells (squares) were maintained in StemPro plus Epo, SCF, and Dex. Note increased rate of proliferation accompanied by a severe size decrease in the di€erentiating cells. (b) Cells obtained around day 40 were induced to di€erentiate and cumulative cell numbers determined at the times indicated. Tangents applied to the growth curve obtained served to determine total cell cycle time at any given time point. In the same samples DNA content was determined by FACS analysis, allowing the determination of the relative cell cycle phase length of the G1, S and G2/M phases. (c) Aliquots from proliferating (top panel) cells and cells induced to di€erentiate for the times indicated were analysed for morphology and hemoglobin content on cytospins. (d) Cells cultured under proliferation (P) or di€erentiation (D) conditions for 72 h were subjected to quantitative hemoglobin determination. Hemoglobin content is given in relative units (R.U.) and normalized to both cell number and cell volume

diploid cells (data not shown). Clone p537/7 I/11 was shown to retain a diploid DNA content, strict factor dependence and unaltered kinetics of terminal di€erentiation capacity during 418 months. Analysis Oncogene

c-ErbB/EGFR p537/7 clone I/11 cells were infected with a murine retrovirus expressing the human EGFR (HER-1). After infection, the cells were selected for their ability to proliferate in EGFR-ligands plus Dex, in the complete absence of Epo and SCF (see Materials and methods). This selection yielded proliferating mass cultures (HER1-I/11 cells) that retained a DNA pro®le indicative of diploid cells (data not shown). The selected cells expressed high levels of cell-surface cErbB/HER-1, detected by FACS-analysis using an antibody to the c-ErbB extracellular domain. (Figure 3a). The HER1-I/11 cells were strongly responsive to the HER-1 ligands TGFa, EGF and heparin-binding EGF, while uninfected I/11 control cells showed no growth response to these EGF-R ligands (Figure 3b and data not shown). HER1-I/11 cells retained cell surface c-Kit expression and a proliferation response to SCF and Epo in the absence of EGF (Figure 3b). This indicates that exogenous expression of the EGFR did not a€ect c-Kit and EpoR function. When activated by EGF, the EGFR was able to induce and maintain long-term proliferation in I/11 cells. Proliferation was still dependent on an active GR, since a GR antagonist (ZK 112.993, a complete antagonist speci®c for the GR and progesteron receptor; Wessely et al., 1997a) e€ectively inhibited proliferation (Figure 3c). Both removal of EGF or inhibition of EGFR kinase activity by the low MW c-ErbB-speci®c inhibitor PD153.035 (Fry et al., 1994) caused a rapid and complete growth arrest, indicating that EGF requires the EGFR kinase activity. Finally, the cells continued to proliferate upon inhibition of the EGFR and addition of Epo, SCF and Dex. (Figure 3c) During this switch, the cells transiently proliferated at a reduced rate and many cells di€erentiated, suggesting that endogenous c-Kit and EpoR receptor levels had to adapt during the switch period (Figure 3c). Interest-

ErbB induced transformation of erythroid progenitors M von Lindern et al

3655

Figure 3 The human EGFR can substitute for Epo plus SCF in induction of sustained proliferation in clone I/11 erythroblasts. (a) HER-1 transduced I/11 cells analysed by FACS using an antibody against mammalian EGFR at the surface of living cells (red curve) or with an antibody to CD 117 (c-Kit, blue curve). The gray curve shows the negative control ¯uorescence after staining with an irrelevant antibody. Note high expression of HER-1 in the retrovirally transduced cells. As expected, HER-1 was absent in control I/11 cells strongly positive for c-Kit only (data not shown). (b) HER-1 transduced I/11 cells were tested for their responsiveness to EGF-receptor ligands (EGF, TGFa, heparin binding (hb) EGF) as well as to SCF and Epo, measuring [3H] thymidine incorpation. In contrast to control I/11 cells, the c-ErbB transduced cells showed a strong response to all EGFR ligands. Concentrations of ligand stock solutions: Epo 10 U/ml, SCF 100 ng/ml, TGFa 5 ng/ml, EGF 20 ng/ml, heparin-binding EGF 20 ng/ml. (c,d). HER-1 transduced I/11cells were seeded in StemPro medium supplemented as indicated, counted daily and cumulative cell numbers determined. Cells proliferated exponentially in StemPro medium supplemented with Dex and the EGFR ligands EGF, TGFa hbEGF (EGF, Dex). Withdrawal of EGFR ligands plus inhibition of the EGFR kinase by the c-ErbB speci®c inhibitor PD153.035 (PD, Dex) abolished proliferation, as did the GR antagonist ZK112.993 (EGF, ZK). After inhibition of c-ErbB by PD153.035, cells still actively proliferated in Epo, SCF and Dex, after an adaptation period characterized by slower growth lasting *5 days (PD, Epo, SCF, Dex). Note a tighter di€erentiation block in the EGFR-driven cells (d, top) as compared to partially hemoglobin positive (brown) cells driven by EpoR plus c-Kit in the absence of an active EGFR (d, bottom). (e) c-ErbB1-I/ 11 erythroblasts were induced to di€erentiate by withdrawal of proliferation factors and switch to Epo/Insulin/high transferrin in the presence of PD153.035. Note identical di€erentiation parameters (left, middle panels) and similar maturation to enucleated erythroctyes of the c-ErbB1 expressing- and control I/11 cells (right panel)

ingly, the activated, exogenous c-ErbB caused a much tighter di€erentiation arrest than endogenous c-Kit

plus EpoR, as shown by lower proportion of mature cells in cytospins (Figure 3d). Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3656

To determine whether retroviral transduction of cErbB and selection for EGF-dependent growth had caused irreversible changes, terminal erythroid di€erentiation was investigated. HER-1 expressing cells were withdrawn from c-ErbB ligand and exposed to di€erentiation factors. According to all di€erentiation parameters tested, they di€erentiated into enucleated erythrocytes like uninfected I/11 control cells. (Figure 3e, see also Figure 2). Taken together, these results indicate that c-ErbB is able to induce renewal if allowed to cooperate with an activated GR and that this renewal induction is completely reversible if cErbB signaling is shut o€. v-ErbB I/11 cells were infected with a retrovirus expressing the v-ErbB oncoprotein of the ES4 avian leukemia virus (gp74v-ErbB; von Rueden et al., 1992). The cells were then selected for proliferation in the absence of cytokines, providing only Dex. Cells from the mass culture were cloned in Methocel and clones showing rapid proliferation in the presence of Dex only were selected. All clones showed similar behavior and one of these clones was used in further experiments (vErbB I/11 #10). v-ErbB (68 ± 74 kD) could be immunoprecipitated from 35S-methionine-labeled cell lysates with a v-ErbBspeci®c antibody (data not shown). Proliferation of vErbB-I/11 cells was dependent on an active v-ErbB kinase and an activated GR, since both the c-ErbB inhibitor PD153.035 and the GR-antagonist ZK112.993 e€ectively and rapidly arrested proliferation (Figure 4a). Finally, the cells could be `switched' to proliferation induction by Epo and SCF upon inactivation of v-ErbB by PD153.035, showing that the cells maintain expression of endogenous EpoR and c-Kit able to drive renewal, despite transformation by the v-ErbB oncogene (Figure 4a). To test if transformation by the v-ErbB oncoprotein would alter the normal response of I/11 cells to Epo and/or SCF, these factors were titrated on v-ErbB-I/11 #10 cells in the presence of the ErbB inhibitor PD153.035 (Figure 4b). Epo was 50% active at 0.1 units and maximally active at 1 ± 2 units, while SCF was maximally active at 50 ± 100 ng/ml, results indistinguishable from those obtained in uninfected I/11 cells and primary mouse erythroblasts (data not shown). Finally, v-ErbB-I/11 cells could be e€ectively induced to di€erentiate into enucleated erythrocytes when v-ErbB activity was turned o€ by PD153.035. Di€erentiation of the v-ErbB I/11 #10 involved similar growth kinetics, size changes and hemoglobin accumulation as observed for uninfected I/11 cells (Figure 4c). In conclusion, v-ErbB can cause reversible transformation of murine I/11 p537/7 erythroblasts, requiring only the cooperation with an activated GR. Signal transduction by v-ErbB and c-ErbB compared to EpoR/c-Kit To investigate whether induction of stress erythropoiesis by Epo/SCF uses the same mechanism as transforma-

Oncogene

tion and renewal-induction by ErbB, signaling pathways employed by the EpoR and c-Kit were compared to those activated by c- and v-ErbB. Whole cell lysates from HER1-I/11 and vErbB-I/11 cells were subjected to Western blot analysis using phosphotyrosine antibodies. Ligand activation of the various receptors induced tyrosine phosphorylation of distinct proteins (Figure 5a) In case of v-ErbB expressing cells, we compared cells treated with the ErbB inhibitor PD153.035 with cells left untreated. As a consequence, weak, constitutive signaling is analysed for v-ErbB, in contrast to the strong activation of c-ErbB, EpoR and c-Kit caused by acute, short-term factor stimulation. Some proteins that were phosphorylated in response to Epo, EGF and active vErbB appear similar while others are only phosphorylated in response to a speci®c ligand. The latter appear to represent mostly the speci®c receptors. To analyse whether ligand-induced tyrosine phosphorylation of the di€erent receptors or their interaction partners was receptor-speci®c or whether cross activation occurs, the EGFR, v-ErbB, c-Kit, the EpoR, as well as Stat5 and JAK2, were immunoprecipitated and tested for tyrosine phosphorylation induced by speci®c ligands (Figure 5b). In HER1-I/11 erythroblasts, EGF, SCF and Epo induced phosphorylation of their cognate receptor. Only Epo induced phosphorylation of the EpoR-associated kinase JAK2. The same was true for v-ErbB erythroblasts, in which v-ErbB (active in the absence of PD153.035) displayed constitutive autophosphorylation (Figure 5b). Comparable to non-transfected I/11 cells (data not shown), tyrosine phosphorylated Stat5 co-immunoprecipitated with the EpoR upon Epo-stimulation of HER1-I/11 cells. Surprisingly, activation of the EGFR by ligand not only caused tyrosine phosphorylation of Stat5, but also induced its binding to a complex containing non-tyrosine- phosphorylated EpoR. This complex was of similar abundance as the complex between the tyrosine-phosphorylated EpoR and Stat5 seen after Epo-stimulation of both I/11 and HERC-I/ 11 erythroblasts (Figure 5b and data not shown). The same complex did not, or hardly, contain the EGFR itself. Using anti-STAT5 antibodies to probe the EpoR-immunoprecipitate showed that STAT5 is not associated with the EpoR when left non-stimulated or when stimulated with SCF. Following EGF-stimulation, tyrosine phosphorylated STAT5 was not coimmunoprecipitated with the EGFR. Tyrosine phosphorylation of Stat 5 by the EGFR could be con®rmed using phosphospeci®c Stat5 antibodies in Western blots (Figure 5b). In v-ErbB-I/11 cells, formation of a complex of tyrosine-phosphorylated Stat5 with the EpoR could not be detected. However, the active v-ErbB did cause a weak though signi®cant tyrosine phosphorylation of Stat5, suggesting that v-ErbB activates Stat5 (Figure 5b). The relatively weak Stat5 phosphorylation signal obtained as compared to EpoR activation is probably explained by the fact, that the signals from the constitutively active v-ErbB are strongly attenuated compared to acutely induced signals.

ErbB induced transformation of erythroid progenitors M von Lindern et al

3657

Figure 4 Avian v-ErbB cooperates with the GR to induce factor-independent renewal in p537/7 clone I/11 erythroblasts. (a) v-ErbB-I/11 erythroblasts (#10) were seeded in StemPro-medium supplemented as indicated, counted daily and cumulative cell numbers determined. Cells proliferate exponentially in the presence dexamethasone only (Dex). Proliferation is abolished by either inhibiting v-ErbB kinase activity PD153.035 (Dex, PD) or by replacing dexamethasone by the GR-antagonist ZK112.993 (ZK). However, cells proliferate after an adaptation period of slow growth, if v-ErbB (inhibited by PD153.035) is substituted for by the activated Epo-R plus c-Kit (Epo, SCF, Dex, PD). (b) Responsiveness of v-ErbB-I/11 cells (#10) to Epo and SCF was determined by thymidine incorporation assay (see legend to Figure 3). During this assay, v-ErbB activity was inhibited by PD153.035. Cells show half-maximal proliferation responses at the expected factor concentrations (Epo, 0.3 U; SCF, 5 ng/ml). (c) v-ErbB-I/11 cells (#10) were induced to di€erentiate by exposure to Epo/insulin plus PD153.035 to block constitutive v-ErbB activity. Proliferation and size decrease during di€erentiation (left panel), hemoglobin accumulation (middle panel; relative units, R.U.) and production of enucleated erythrocytes all proceeded similarly as in untransformed I/11 cells (Figure 2)

Activation of PI3-K is required for erythroblast proliferation driven by EpoR plus c-Kit, but is dispensable for EGFR-induced proliferation It is still unknown which signaling pathways are crucial to induce expansion of erythroid progenitors. In Friend virus transformed cells, PI3K-activation appeared crucial for progenitor expansion (Nishigaki et al.,

2000), while others proposed that synergistic activation of ERK1/2 by Epo and SCF is important for expansion of human primary erythroid progenitors (Sui et al., 1998). In our cells, the MEK1-inhibitor PD98,059 did not a€ect the balance between renewal and di€erentiation divisions, when used at concentrations suf7®cient to inhibit Epo- and/or SCF-induced ERK1/2 phosphorylation (50 mM, data not shown). Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3658

Figure 5 Signal transduction in I/11 cells expressing the human EGFR (HER-1/c-ErbB) or v-ErbB. (a) Analysis of total tyrosinephosphorylated proteins. c-ErbB (left panel) or v-ErbB (right panel) expressing I/11 cells were withdrawn from factors and restimulated with human recombinant Epo (hr-Epo, 10 U/ml) murine SCF (mSCF, 1 mg/ml) and human EGF (h-EGF, 200 ng/ml) or left unstimulated (no factor). Lysates were subjected to Western blot analysis using the 4G10 anti-phosphotyrosine antibodies. Numbered arrows indicate phosphorylated proteins for which a potential identity is suggested in the results section. (b) Analysis of signal transduction complexes containing the EpoR, Stat5, Jak-2 and c-Kit in I/11 cells expressing HER-1 or v-ErbB. Cells were factor withdrawn and restimulated as in (a). Lysates were immunoprecipitated with antibodies to the EGFR, v-ErbB, EpoR, c-Kit, Stat5 or Jak-2. The immunoprecipitates were then separated by SDS ± PAGE and processed for Western blot analysis, using 4G10 phosphotyrosine antibodies to detect phosphorylated proteins present in the immunoprecipitated complexes, followed by analysis with antibodies revealing immunoprecipitated, nonphosphorylated proteins Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

Therefore, proliferation of I/11-EGFR and I/11-vErbB was examined in the presence of the PI3Kinhibitor LY294002, combined with appropriate `renewal factors'. In multiple experiments, LY294002 arrested proliferation of the I/11 cells and induced di€erentiation as measured by hemoglobin accumulation (Figure 6a) and analysis of cell morphology (data not shown). In contrast, HER-1-I/11 cells continued to proliferate in response to EGF irrespective of the presence of LY294002 and accumulation of hemoglobin was not detected (Figure 6b). To test whether HER-1-I/11 erythroblasts were independent of PI3K in general, the cells were switched to media containing the ErbB-inhibitor PD153.035 as well as Epo and SCF plus Dex for 6 days (see Figure 4). In contrast to cells grown in the presence of EGF, the cells grown in the presence of Epo/SCF/Dex were as sensitive to LY294002 as parental I/11 cells (Figure 6b). The morphology of cells exposed to LY294002 was similar to di€erentiated cells shown in Figure 3e. These results indicate that HER-1 signaling speci®cally renders I/11 erythroblasts independent of PI3K-activation. Surprisingly, LY294002 blocked expansion of

v-ErbB-I/11 cells and induced hemoglobin accumulation, when grown with active v-ErbB plus Dex, or when grown in the presence of the ErbB-inhibitor PD153.035 and Epo, SCF and Dex (Figure 6c). The morphology of cells exposed to LY294002 was similar to di€erentiated cells shown in Figure 4c. This suggests that v-ErbB-induced renewal depends on mechanisms used by the EpoR/c-Kit complex, while c-ErbB can activate additional pathways to induce expansion and block di€erentiation. LY294002 did not a€ect di€erentiation kinetics, which indicates that inhibition of PI3K is not toxic in erythroid progenitors (data not shown). To verify the speci®city of LY294002, its e€ect on ligand-induced phosphorylation of multiple receptor targets was analysed. In HER1-I/11 cells, we detected weak phosphorylation of PKB by Epo and strong phosphorylation by both SCF and EGF (Figure 7). In both cases, PKB-phosphorylation was completely inhibited by the PI3K inhibitor LY294002 (Figure 7). Thus the failure of LY294002 to inhibit EGF-induced expansion is not due to insucient inhibitor concentrations in cells expressing high levels of EGFR.

3659

Figure 6 p537/7 erythroblasts require PI3K signaling for proliferation, when driven by either Epo plus SCF, or v-ErbB, but not when driven by HER-1 or when induced to di€erentiate. (a) p537/7 I/11 cells (squares) were cultivated in medium containing Epo, SCF and Dex, in the absence (white symbols) or presence of LY294002 (dark symbols). Cells were counted at the times indicated and cumulative cell numbers determined (top panel). At the times indicated, hemoglobin content per cell volume was determined (no LY, light dotted bars, LY, dark dotted bars). The error bars indicate the standard deviations from three indiviual experiments. (b) HER-1 expressing I/11 cells, grown either in EGFR ligands plus Dex (triangles, hatched bars) or in medium containing SCF, Epo, the ErbB-inhibitor PD153.035 and Dex (squares, dotted bars) were cultivated in the absence (light symbols) or presence of LY294002 (dark symbols). Cumulative cell numbers (top panel) and hemoglobin content (bottom panel) were determined as described in (a). The error bars (associated with triangles) indicate the standard deviations from four individual experiments, using cells growing in EGFR ligands plus and minus LY. (c) V-ErbB expressing I/11 cells, grown either in the presence of Dex only (triangles, hatched bars) or in medium containing SCF, Epo, the ErbB-inhibitor PD153.035 and Dex (squares, dotted bars) were cultivated in the absence (light symbols) or presence of LY294002 (dark symbols). Cumulative cell numbers (top panel) and hemoglobin content (bottom panel) were determined as described in (a) Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3660

LY294002 had no e€ect on the phosphorylation of STAT5 (Figure 7), SHC or ERK1/2 (data not shown). PKB phosphorylation by v-ErbB was not detectable, which is most likely due to the fact that only constitutive signals can be obtained. In line with this, PKB phosphorylation was also undetectable in the parental I/11 cells chronically stimulated with Epo plus SCF (data not shown). In conclusion, PI3K is required for the expansion of I/11 cells driven by the cooperation of Epo and SCF, or by v-ErbB. In contrast, c-ErbB/EGFR is able to induce progenitor expansion independent of PI3K. Discussion In this paper, a cell culture approach is introduced to study the control of expansion versus terminal di€erentiation of murine erythroid cells. Erythroid progenitors from p53 wt or de®cient mice are shown to proliferate with an enhanced, or even unlimited lifespan, respectively, using serum-free media supplemented with cytokines and hormones relevant for stress erythropoiesis. We used this novel cell system to demonstrate that oncogenic (v-ErbB) and normal receptor tyrosine kinases (c-ErbB) substitute for the requirement of Epo plus SCF in progenitor renewal. Both c-and v-ErbB were found to activate the same signaling pathways (Stat5 and PI3K) that are recruited through the activation of the EpoR and c-Kit. Activation of PI3K appeared to be crucial for induction of renewal divisions by Epo plus SCF or v-ErbB. In contrast, activation of c-ErbB (HER1) induces PI3Kindependent expansion of erythroid progenitors.

Figure 7 v-ErbB and c-ErbB substitute for distinct signaling pathways activated by the EpoR and by c-Kit. Lysates from HER-1 expressing I/11 cells, factor-withdrawn and restimulated similar as described in the legend to Figure 5, were analysed on Western blots using antibodies to tyrosine-phosphorylated Stat5 and serine-phosphorylated PKB. The blots were reprobed with respective antibodies detecting total Stat5 or PKB. Cells were factor withdrawn and stimulated in the presence or absence of the PI3K-inhibitor LY294002 Oncogene

Immortal murine erythroblasts with primary cell behavior: a novel approach to study signal transduction and oncogenesis Despite the undisputed advantages of in vivo models employing genetically modi®ed mice to study normal and leukemic renewal, our in vitro analysis of progenitor expansion and terminal di€erentiation can reveal control mechanisms that remain hidden in in vivo models due to compensation and feed back mechanisms. In addition, studies at the cellular and molecular level require cell culture systems that allow producing large numbers of progenitors able to faithfully recapitulate the factor dependence and terminal di€erentiation behavior of erythroid progenitors in vivo. The murine cell systems described here ful®l these criteria. Erythroid progenitors from fetal livers of p537/7 mice show unlimited growth potential, similar to oncogene-transformed erythroblasts (Metz et al., 1995). Both p537/7 erythroblasts and their mortal counterparts from p53+/+ mice behave like primary murine erythroid progenitors with respect to progenitor expansion, terminal di€erentiation and cytokine-dependent cell survival. The progenitors execute a physiological `renewal program' typical for stress conditions, which is strictly controlled and limited in time in vivo. Being grown under de®ned conditions in serum free medium, these cells are an ideal source to study signaling pathways involved in regulating the balance between renewal and terminal di€erentiation. Although the data are not shown in this paper, p537/7 and p53 wt cells show similar expansion and di€erentiation kinetics during the ®rst 14 days after isolation of fetal livers (see also Dolznig et al., 2001). Compared to the di€erentiation kinetics shown in Figure 2, pre-crises cells di€erentiate faster: enucleation is completed in 48 h instead of 72 h. However, this is the same for p537/7 and p53 wt cells. Also in signal transduction and cell survival we observed no di€erences between p537/7 and p53 wt cells. Immortal erythroblasts from p537/7 mice (clone I/ 11) continuously proliferated for more than 18 months, without losing a diploid DNA pro®le, dependence on Epo, SCF and Dex, and the ability to undergo synchronous di€erentiation into enucleated erythrocytes. However, a major concern with respect to the use of p537/7 erythroblasts as a model for primary erythroid progenitors was the reported loss of genetic stability in cells de®cient for wild-type p53 (Hinds and Weinberg, 1994; Ozbun et al., 1993; Wong et al., 1999). We observed that a diploid DNA content as well as factor dependence and terminal di€erentiation comparable to primary cells could be ensured by maintaining the cultures under optimal conditions. Retroviral transduction of the (proto)oncogenes v- and c-ErbB genes also did not a€ect genetic stability. Inactivation of the transforming genes showed that the cells had retained factor dependence and full di€erentiation capacity. To our knowledge, no other immortalized erythroid cell system with such properties has been described.

ErbB induced transformation of erythroid progenitors M von Lindern et al

(Proto)oncogene function in renewal control: EGFR and v-ErbB coactivate Stat5 and PI3K, two pathways separately employed by the EpoR and c-Kit The requirement for the cooperation of Epo and SCF in erythroid proliferation has been well established (Muta et al., 1995; von Lindern et al., 1999; Wessely et al., 1999). c-Kit and the EpoR are expressed in close proximity at the cell membrane as shown by FRETanalysis (Broudy et al., 1998). In our system, both receptors can be co-immunoprecipitated, suggesting direct interaction (data not shown). However, we never observed a major phosphorylation of c-Kit by Epo, nor EpoR phosphorylation by SCF (Figure 5b) as described by others in Friend virus transformed erythroid cells (Wu et al., 1995a; Zochodne et al., 2000). However, these experiments were performed in HCD57 expressing an exogenous c-Kit, or in F-Mulvhelper-virus-infected erythroleukemic cell lines overexpressing a rearranged Spi-1/PU-1 gene (Tamir et al., 1999). In contrast, our cells express normal levels of cKit and EpoR and do not express Spi-1/PU-1 at detectable levels (data not shown). It is possible that cross-phosphorylation of the EpoR and c-Kit is dependent on other (genetic) factors in the erythroid progenitors. The aim of our studies was to analyse whether the same mechanisms are targeted by Epo/SCF in stress erythropoiesis and by ErbB in transformation. PI3K appears to be dispensable for terminal erythroid di€erentiation since low molecular weight inhibitors of PI3K failed to a€ect terminal di€erentiation into erythrocytes (Figure 6). In contrast, PI3K is required for expansion of erythroid progenitors in vitro (Figure 6) and for the expansion of leukemic cells. PI3K activation is a hallmark of SFFV induced erythroleukemia (Nishigaki et al., 2000). Inhibition of PI3K abrogates renewal of normal erythroid progenitors and v-ErbB transformed cells (Figure 6). PI3K mediates recruitment of a cohort of signaling molecules to the signaling complex. Its activation generates phosphoinositoles in the plasma membrane, which recruit molecules with a Pleckstrin Homology (PH-)domain (Lemmon and Ferguson, 2000). These include members of the Tec tyrosine kinase family and sca€old proteins of the Gab-, Dok-, and IRS families, reported to be targets of Epo and SCF signaling (Lecoq-Lafon et al., 1999, van Dijk et al., 2000). Another important target of PI3K is PKB, which phosphorylates and controls protein translation regulators, transcription factors of the Forkhead family, and the pro-apoptotic factors BAD and Caspase 9 (Kops and Burgering, 1999). These multiple functions of PI3K make it a likely candidate to act as a signal integrator, i.e. for those emanating from the EpoR and c-Kit. In addition to PI3K, other pathways could also play a crucial role in both stress erythropoiesis and leukemic transformation. Noteably, activation of the EpoR or cErbB and consitutively active v-ErbB all induce phosphorylation of Stat5. Stat5 a/b-/- erythroblasts show a strong proliferation defect under our `renewal'

conditions (Deiner et al. (2001) submitted) indicating that Stat5 is required for erythroid progenitor expansion. Our data suggest that both c-ErbB and v-ErbB exert crosstalk to the EpoR/c-Kit complex. Activation of the EGFR results in Stat5 phosphorylation, which can be co-immunoprecipitated with a non-phosphorylated EpoR, while the presence of tyrosine-phosphorylated EGFR in the immunoprecipitate was negligible. This indicates that Stat5 can associate with the EpoR complex independent of EpoR phosphorylation. Stat5 is not constitutively bound to the EpoR since it cannot be detected in lysates of unstimulated or SCFstimulated cells. It is unknown whether Stat5 can directly associate with nonphosphorylated EpoR or whether a sca€old protein is involved. V-ErbB causes a weak tyrosine phosphorylation of Stat5, and no interaction with the EpoR can be detected. However, may be due to the weaker signal provided by the constitutively active v-ErbB. It is also possible that complex formation between Stat5 and the EpoR requires other players, which are activated by c-ErbB but not by v-ErbB, due to the fact that the mutations in v-ErbB have deleted most sites that function in the EGFR to recruit signaling pathways. Notably, expression of dominant-negative Stat5 variants in avian vErbB-transformed erythroblasts severely interfered with cell proliferation (Mellitzer et al., 1996; G Mellitzer and H Beug unpublished) suggesting that Stat5 signaling is important for renewal in v-ErbB transformed erythroblasts. A major di€erence between the mechanisms utilized by c-ErbB and v-ErbB in activation of renewal is the dependence on PI3K. The PI3K inhibitor LY294002 abrogates progenitor renewal induced by Epo plus SCF or v-ErbB, but has no e€ect on renewal induced by EGF in EGFR-expressing cells. This strongly suggests that v-ErbB activates Stat5 and PI3K, and in doing so directly substitutes for the function of the EpoR/c-Kit complex in the activation of these key pathways. In contrast, the EGFR activates Stat5 and PI3K, but in addition must recruit distinct pathways that are also able to evoke renewal divisions. These pathways are not activated by v-ErbB, probably due to the mutations in this oncogene which restrict the number of signal transduction pathways activated by it (Meyer et al., 1994).

3661

Conclusions In contrast to the currently used erythroid cell models, the cell culture system for normal erythroid progenitors described here does not employ oncogenes to drive progenitor expansion, but uses physiological factor combinations involved in the expansion of erythroid progenitors during stress erythropoiesis. Using this system, we demonstrate that the distinct receptors driving expansion either under hypoxic conditions or in leukemia activate the same signal transduction pathways. In addition to easy retroviral gene transfer and extensive signal transduction analysis in immortal p537/7 erythroblasts, our system also o€ers the Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3662

possibility to expand and molecularly analyse erythroid progenitors from genetically modi®ed mice. We expect these advanced tools to allow a more comprehensive analysis of how receptor signaling causes progenitor renewal.

Materials and methods Cultivation of murine erythroid progenitors Fetal livers of day 11.5 to 13.5 mouse embryos (wt or p537/7 Bl-6/129 mice) were resuspended in 1 ml serum-free stem cell expansion medium (StemPro-34TM; Life Technologies Gibco BRL). Cells were passed through a 70 mm Nylon cell strainer (Falcon #2350, Becton Dickinson), washed and seeded at 46106 cells/ml into Stem-Pro-34TM medium supplemented with human recombinant erythropoietin (Erypo, Cilag AG, Switzerland, 2 U/ml) murine recombinant stem cell factor (SCF, R&D Systems, Minneapolis, MN, USA, 100 ng/ml), dexamethasone (Dex, Sigma, 1076 M) and insulin-like growth factor 1 (IGF-1, Promega, 40 ng/ml). IGF-I increases cell survival, but is not required for expansion. Thus, it was always added in the experiments described and will not be mentioned separately. The mass cultures of erythroblasts were subjected to daily partial medium changes and addition of fresh factors. Suspension cells were removed from the adherent cell layer. On day 5, cells were centrifuged through Ficoll (lymphocyte separation medium, 1078 g/cm3, Eurobio, France) to remove dead and di€erentiated cells. Thereafter, cell density was maintained between 3 and 76106 cells/ml by daily dilution with fresh medium containing 26 factors. Cell numbers and size distributions were determined daily, using an electronic cell counter (CASY-1, SchaÈrfe-System, Reutlingen, Germany). Cells were puri®ed again using Ficoll when containing 440% of dead and/or di€erentiated cells as estimated by size distribution. Establishment of cell strains with unlimited growth potential from p537/7 fetal livers Erythroblasts from p537/7 fetal livers were propagated until day 15 ± 20, when growth rate decreased and di€erentiated erythrocytes increased. During this period, the p537/7 erythroblasts were puri®ed every 2 days for 7 ± 10 days using Ficoll, and reseeded at 36106 cells/ml. This procedure yielded erythroblasts showing unlimited growth potential from 10 out of 10 p537/7 fetal liver cultures. Optimum quality of the StemPro medium was essential. To ensure this, the serum-free medium should be stored for less than 10 months after production date and the serum supplement should be purchased frozen less than 2 months after production date and stored at 7708C avoiding refreezing of thawed vials. Cloning of p537/7 erythroblasts After *30 days in culture, p537/7 erythroblasts were seeded into semisolid medium (Methocel-containing StemPro34 SFM, Life Technologies, Gibco BRL) containing mSCF, Epo, Dex and IGF-I as above. After 7 ± 10 days of incubation, large colonies were obtained with a cloning eciency of 2 ± 10%. One hundred and twenty clones were picked from dishes with well-separated colonies, and Oncogene

cultivated in 96-well plates. After *4 ± 6 days, colonies were inspected visually. Clones containing either a high proportion of mature cells, cells of irregular size/shape or exhibiting a slow proliferation rate were discarded. The remaining clones (64) were expanded and subjected to quantitative analysis of proliferation kinetics, size distribution, factor dependence and ability to terminally di€erentiate (see below). Thirty-seven of the 64 selected clones showed optimum growth rate and di€erentiation behavior. One of these clones, clone I/11, was used for further experiments. Generation and culture of p537/7 erythroblasts expressing c-ErbB or v-ErbB Cultures of p537/7 clone I/11 erythroblasts were cocultivated with GP+E packaging cells expressing the human EGF-receptor (HER-1, referred to as c-ErbB) or the avian v-ErbB oncogene from a murine retrovirus vector (see von Rueden et al., 1992). After cocultivation, cells were grown for 2 days in complete medium and then transferred to selection media. c-ErbB/HER-1 infected cells, were selected in medium containing 20 ng EGF, 1076 M Dexamethasone and 40 ng/ml IGF-1. Infection eciency was low (1%), but this 1% grew with similar kinetics as parental cells in presence of Epo and SCF. Within 14 days a large culture of c-ErbB expressing cells had been established. V-ErbB infected cells were selected in medium containing 1076 M Dexamethasone, and 40 ng/ml IGF-1. During the ®rst week of selection 2 U/ml Epo was added to avoid excessive generation of apoptotic cells. Immature cells were puri®ed at intervals as described above and reseeded at 2 ± 46106 cells/ml. Cultivation medium after selection contained 20 ng/ml EGF, 20 ng/ml heparin-binding EGF, 5 ng/ml TGFa, 1076 Dexamethasone, 40 ng/ml IGF-1 in StemPro-34TM for HER1-I/11 and 1076 M Dexamethasone, 40 ng/ml IGF-1 in StemPro-34TM for v-ErbB-I/11. The vErbB-expressing mass culture was then cloned in Methocel containing 1076 M Dexamethasone and 40 ng/ml IGF-1 and clones tested for proliferation rate in liquid medium containing the same factors, as well as for di€erentiation ability in di€erentiation medium (see below) plus PD153.035. Out of multiple fast growing, well di€erentiating clones, v-ErbB-I/11 clone 10 was used in subsequent experiments. Expression of c-ErbB and v-ErbB was veri®ed by FACSanalysis using a monoclonal antibody to the extracellular domain of human EGFR (R1, (Khazaie et al., 1988) as well as by Western blot or immunoprecipitation using respective antibodies (Hayman et al., 1993; Meyer et al., 1994). To induce proliferation of HER-1-I/11 cells or v-ErbB-I/11 cells in response to endogenous EpoR and c-Kit, cells were washed and cultivated in StemPro-34TM containing 2 U/ml Epo, 100 ng/ml SCF, 1076 M Dexamethasone, 40 ng/ml IGF1 and the speci®c erbB inhibitor PD153.035 (5 mg/ml, Fry et al., 1994). Determination of factor/hormone dependence p537/7 erythroblasts (30 ± 40 days in culture) were washed twice in PBS and seeded at 36106 cells/ml in media containing factor combinations or low-molecular weight inhibitors (PD153.035, 5 mM; LY294002, 20 mM; Calbiochem) as indicated. The Epo-deprived medium was supplemented with a neutralizing rabbit polyclonal antibody against human erythropoietin (Genzyme #1541-01, 1 : 50 dilution; (Wessely et al., 1999) and the medium lacking Dex contained 361076 M Dex-antagonist (ZK 112.993). Cells were counted at daily intervals until the cell-proliferation arrested and cells either di€erentiated or died.

ErbB induced transformation of erythroid progenitors M von Lindern et al

Determination of growth factor responsiveness by [3H]-thymidine incorporation

antibody to human EGFR (R1, (Khazaie et al., 1988), followed by FITC-labeled goat anti mouse IgG).

Serial dilutions of Epo (starting at 10 U/ml), SCF (starting at 100 ng/ml) TGFa (starting at 5 ng/ml) EGF (starting at 20 ng/ml) and heparin-binding EGF (starting at 20 ng/ml) in StemPro-34TM medium were prepared in 96 well tissue culture plates. I/11 cells, HER-1 I/11 cells or v-ErbB I/11 cells (20 000 cells per well) were added and thymidine incorporation was measured after 2 days as described (Beug et al., 1995). To test factor responsiveness of HER-1 I/11- or v-ErbB I/11 cells, PD153.035 was added during incubation at 5 mM.

Cell cycle parameters were determined by measuring DNA content by cyto¯uorometry (Dolznig et al., 1995). DNA pro®les were determined using ethanol ®xed cells stained with DAPI (4,6-diamidino-2-phenylindol-hydrochloride) followed by ¯ow cytometry (Partec PAS-II). Percentages of cells in the various cell cycle phases were calculated using the software package from the same manufacturer.

Induction of terminal differentiation

Immunoprecipitation and Western blot analysis

p537/7 cells erythroblasts were washed twice in PBS and seeded at 1.56106 cells/ml in di€erentiation medium (CFU-E medium, Beug et al., 1995; Hayman et al., 1993), supplemented with 10 units/ml erythropoietin (Epo), 461074 IE/ml Insulin (Actrapid HM, Novo Nordisk, Denmark), Dex-antagonist (ZK 112.993; 361076 M, Wessely et al., 1997a) and iron-saturated human transferrin (Sigma, 1 mg/ml). Di€erentiation parameters were determined every 12 h for 5 days and cells were maintained at densities of 2 ± 46106 cells/ml, requiring dilution with fresh medium twice daily between 24 and 50 h. Cell counting and determination of size distribution was performed using a CASY I electronic cell counter (SchaÈrfe Systems), cell morphology was analysed in cytospins stained with histological dyes and neutral benzidine (Beug et al., 1982) and hemoglobin content was quantitated in a photometric assay (see below and Wessely et al., 1999). Terminal di€erentiation analysis of HER-1 I/11- or v-ErbB I/11 cells was performed similarly, except that oncogene activity was inhibited by 5 mM PD153.035

Cultured cells were washed once with PBS and incubated in plain Iscove's medium without FCS or growth factors for 4 h at a density of 46106/ml. Cells were concentrated to 20 ± 406106/ml, stimulated 10 min at 378C with either 10 U/ml rhEpo, 1 mg/ml SCF or 200 ng/ml EGF. In case of v-ErbB cells, the ErbB inhibitor PD153.035 (5 mM) was added for 4 h. In the presence of this inhibitor cells were stimulated with Epo and SCF. V-ErbB signaling was studied by comparing cells treated or left untreated with the inhibitor. Cell lysates were made and processed as described (von Lindern et al., 2000). Immunoprecipitations and Western blots were performed as described previously (von Lindern et al., 2000). Antibodies used were rabbit antisera recognizing JAK2 (Upstate Biotechnology, Lake Placid, NY, USA) rabbit antisera recognizing the human EGF-R, mouse EpoR, c-Kit or p44/p42 MAP kinase (Santa Cruz Biotechnology, Santa Cruz, CA, USA) rabbit antisera recognizing PhosphoPKB (Ser473) or PKB, mouse monoclonal antibodies recognizing phospho-p44/p42 MAP kinase (Thr202/Tyr204, New England Biolabs, Beverly, MA, USA) and the antiphosphotyrosine mouse monoclonal antibody 4G10 (Upstate Biotechnology, Lake Placid, NY, USA). Stat5 antiserum was a kind gift of Dr T Decker (University of Vienna, Austria).

Photometric hemoglobin determination Fifty ml aliquots were removed from the cultures to be analysed and processed for hemoglobin content by photometry (Kowenz et al., 1987). Values obtained from triplicate determinations were averaged and normalized to cell number and cell volume. FACS analysis 0.5 ± 16106 cells were stained with the ¯uorescently labeled antibodies: anti mouse Ter119-Phycoerythrin (PE #09085B), anti mouse CD71 (Transferrin-Receptor; PE, C2, #01595), anti mouse CD117/c-Kit (FITC, 2B8, #01904D). (PharMingen). Cell surface EGFR was labeled with a monoclonal

3663

Determination of cell cycle parameters

Acknowledgments We thank Gabi Litos and Gabi Stengl for expert technical assistance. This work was supported by grants from the Austrian ForschungsfoÈrderungs fonds (FFF, Project No. 802447, HB), the Austrian Science Foundation (FWF, SFB006, HB, EM), the Dutch Cancer Society (EUR 951021), the European Community (HPRN-CT-2000-00083, HB, MvL and ERBFMRX-CT-98-0197, HB, EM) and by a fellowship of the Dutch Academy for Arts and Sciences (KNAW) to MvL.

References Bauer A, Tronche F, Wessely O, Kellendonk C, Reichardt HM, Steinlein P, SchuÈtz G and Beug H. (1999). Genes Dev., 13, 2996 ± 3002. Bauer A, Gandrillon O, Samarut J and Beug H. (2001). Hematopoiesis: a developmental approach. In Zon, L. (ed.) Oxford University Press, Oxford, in press. Beug H, Palmieri S, Freudenstein C, Zentgraf H and Graf T. (1982). Cell, 28, 907 ± 919. Beug H, Kahn P, Doederlein G, Hayman MJ and Graf T. (1985). Modern Trends in Human Leukaemia VI. Neth R, Gallo R, Greaves M and Janka K. (eds.) Springer Verlag, Berlin, Heidelberg, Vol. 29, pp. 284 ± 289.

Beug H, Bartunek P, Steinlein P and Hayman MJ. (1995). Oncogene Techniques. Vogt, P.K. and Verma, I.M. (eds) Academic Press Inc., New York and London, Vol. 254, pp. 41 ± 76. Beug H, Bauer A, Dolznig H, von Lindern M, Lobmayer L, Mellitzer G, Steinlein P, Wessely O and Mullner E. (1996). Biochim. Biophys. Acta., 1288, M35 ± M47. Broudy VC, Lin NL, Priestley GV, Nocka K and Wolf NS. (1996). Blood, 88, 75 ± 81. Broudy VC, Lin NL, Buhring HJ, Komatsu N and Kavanagh TJ. (1998). Blood, 91, 898 ± 906.

Oncogene

ErbB induced transformation of erythroid progenitors M von Lindern et al

3664

Oncogene

Dolznig H, Bartunek P, Nasmyth K, MuÈllner E and Beug H. (1995). Cell Growth Di€er., 6, 1341 ± 1352. Dolznig H, Boulme F, Stangl K, Deiner EM, Mikulits W, Beug H and MuÈllner E. (2001). FASEB J., in press. Downward J, Yarden Y, Mayes E, Scrace G, Totty N, Stockwell P, Ullrich A, Schlessinger J and Water®eld MD. (1984) Nature, 307, 521 ± 527. Fry DW, Kraker AJ, McMichael A, Ambroso LA, Nelson JE, Leopold WR, Connors RW and Bridges AJ. (1994). Science, 265, 1093 ± 1095. Gandrillon O, Schmidt U, Beug H and Samarut J. (1999). EMBO J., 18, 2764 ± 2781 Hayman MJ, Meyer S, Martin F, Steinlein P and Beug H. (1993). Cell, 74, 157 ± 169. Hinds PW and Weinberg RA. (1994). Curr. Opin. Genet. Dev., 4, 135 ± 141. Keller G. (1992). Curr. Opin. Immunol., 4, 133 ± 139. Khazaie K, Dull TJ, Graf T, Schlessinger J, Ullrich A, Beug H and Vennstrom B. (1988). EMBO J., 7, 3061 ± 3071. Kieslinger M, Woldman I, Moriggl R, Hofmann J, Marine JC, Ihle JN, Beug H and Decker T. (2000). Genes Dev., 14, 232 ± 244. Kops GJ and Burgering BM. (1999). J. Mol. Med., 77, 656 ± 665. Kowenz E, Leutz A, Doderlein G, Graf T and Beug H. (1987). Modern Trends in Human Leukemia VII. Neth R, Gallo RC, Greaves MF and Kabisch H (eds.), Springer Verlag, Heidelberg, Vol. 31, pp. 199 ± 209. Lecoq-Lafon C, Verdier F, Fichelson S, Chretien S, Gisselbrecht S, Lacombe C and Mayeux P. (1999). Blood, 93, 2578 ± 2585. Lemmon MA and Ferguson KM. (2000). Biochem. J., 350, 1 ± 18. Levy DE and Gilliland DG. (2000). Oncogene, 19, 2505 ± 2510. Mellitzer G, Wessely O, Decker T, Eilers A, Hayman MJ and Beug H. (1996). Proc. Natl. Acad. Sci. USA, 93, 9600 ± 9605. Metz T, Harris AW and Adams JM. (1995). Cell, 82, 29 ± 36. Meyer S, LaBudda K, McGlade I and Hayman MJ. (1994). Mol. Cell Biol., 14, 3253 ± 3262. Muta K, Krantz SB, Bondurant MC and Dai CH. (1995). Blood, 86, 572 ± 580. Nishigaki K, Hanson C, Ohashi T, Thompson D, Muszynski K and Ruscetti S. (2000). J. Virol., 74, 3037 ± 3045. Nutt SL, Heavey B, Rolink AG and Busslinger M. (1999). Nature, 401, 556 ± 563. Ozbun MA, Medina D and Butel JS. (1993). Growth Di€er., 4, 811 ± 819. Pain B, Melet F, Saez J, Flickinger T, Raines M, Peyrol S, Moscovici C, Moscovici MG, Kung HJ, Jurdic P, Lazarides F and Samarut J. (1991). Cell, 65, 37 ± 46.

Reichardt HM, Kaestner KH, Tuckermann J, Kretz O, Wessely O, Bock R, Gass P, Schmid W, Herrlich P, Angel P and SchuÈtz G. (1998). Cell, 93, 531 ± 541. Rolink A, Kudo A, Karasyama H and Kikuchi Y. (1991). EMBO J., 10, 327 ± 336. Steinlein P, Wessely O, Meyer S, Deiner EM, Hayman MJ and Beug H. (1995). Curr. Biol., 5, 191 ± 204. Sui X, Krantz SB, You M and Zhao Z. (1998). Blood, 92, 1142 ± 1149. Tamir A, Howard J, Higgins RR, Li YJ, Berger L, Zacksenhaus E, Reis M and Ben-David Y. (1999). Mol. Cell. Biol., 19, 4452 ± 4464. Teglund S, McKay C, Schuetz E, van Deursen JM, Stravopodis D, Wang D, Brown M, Bodner S, Grosveld G and and Ihle JN. (1998). Cell, 93, 841 ± 850. Till JE and McCulloch EA. (1980). Biochim. Biophys. Acta., 605, 431 ± 459. van Dijk T, Van den Akker E, Parren-van Amelsvoort M, Mano H, Lowenberg B and von Lindern M. (2000). Blood, 96, 3406 ± 3413. von Lindern M, Zauner W, Steinlein P, Mellitzer G, Fritsch G, Huber K, LoÈwenberg B and Beug H. (1999). Blood, 94, 550 ± 559. von Lindern M, Amelsvoort MP, van Dijk T, Deiner E, van Den Akker E, van Emst-De Vries S, Willems P, Beug H and Lowenberg B. (2000). J. Biol. Chem., 275, 34719 ± 34727. von Rueden T, Kandels S, Radaszkiewicz T, Ullrich A and Wagner EF. (1992). Blood, 79, 3145 ± 3158. Wessely O, Deiner EM, Beug H and von Lindern M. (1997a). EMBO J., 16, 267 ± 280. Wessely O, von Lindern M, Levitzki A, Gazit A, Ischenko I, Mellitzer G, Hayman MJ and Beug H. (1997b). Cell. Growth Di€., 8, 481 ± 493. Wessely O, Bauer A, Quang CT, Deiner EM, von Lindern M, Mellitzer G, Steinlein P, Ghysdael J and Beug H. (1999) A novel way to induce erythroid progenitor self renewal: cooperation of c-Kit with the erythropoietin receptor. Biol. Chem., 380, 187 ± 202. Wong KS, Li YJ, Howard J and Ben-David Y. (1999) Loss of p53 in F-MuLV induced-erythroleukemias accelerates the acquisition of mutational events that confers immortality and growth factor independence. Oncogene, 18, 5525 ± 5534. Wu H, Klingmuller U, Besmer P and Lodish HF. (1995a) Interaction of the erythropoietin and stem-cell-factor receptors. Nature, 377, 242 ± 246. Wu H, Liu X, Jaenisch R and Lodish HF. (1995b). Cell, 83, 59 ± 67. Zochodne B, Truong AH, Stetler K, Higgins RR, Howard J, Dumont D, Berger SA and Ben-David Y. (2000). Oncogene, 19, 2296 ± 2304.