Signaling by the Erythropoietin receptor in retrovirally transduced normal hematopoietic progenitors: A conserved proline-rich motif is required for association with JAK2.
Gregory D. Longmore 1,3,*, J. William Tanner 2,3, Grant A. McArthur 4, Pamela N.Pharr 5 ,
Gregory R. Johnson 4, and Andrey S. Shaw 2,3   
 In: Zander AR et al. (eds) Gene Technolgy, Stem Cell and Leukemia Research, Nato ASI Series H: Cell Biology, Vol 94,
Springer-Verlag, Berlin Heidelberg New York London, pp 321-331

1 Departments of Medicine and Cell Biology, 2 Pathology, and
3 Center for Irnmunology, Washington University School of Medicine, St. Louis, MO 63110 USA
4 The Walter and Eliza Hall Institute of Medical Research, 3050 Victoria, Australia
5 Ralph H. Johnson Department of Veterans Affairs Medical Center, Charleston, SC 29401 USA
*corresponding author: Telephone 314-362-8800; FAX 314-362-8826;


The receptor for erythropoietin is restricted to cells of mature erythroid and possibly megakaryocyte lineages. Studies in cell lines have suggested that cytokine receptors share a conserved signaling pathway for proliferation. We retrovirally transduced the erythropoietin receptor or a constitutively activated form of the EPOR into normal hematopoietic progenitors, including blast cell colonies. The EPO-R was able to support the proliferation and differentiation of early erythroid, early megakaryocytic, and macrophage progenitors, but not granulocyte progenitors. Blast cell colonies transduced with the EPO-R proliferate in response to EPO but the development of erythroid cells was not favored over other lineages. These results with normal cells suggest that some but not all cytokine receptors exhibit shared signaling pathways, and that EPO signaling alone is not sufficient to drive erythroid development. The Janus family of cytosolic tyrosine kinases mediate cytokine initiated mitogenic signals. We have determined that the EPO-R box 1 cytoplasmic motif is required for the binding and activation of JAK2. However, sequences outside the box 1 domain most likely regulate the specificity of JAK kinase association.


Erythropoietin (EPO) is a serum glycoprotein hormone required for the survival, proliferation and differentiation of committed erythroid progenitor cells, and is the principal hormone regulating the level of circulating red blood cells. The administration of recombinant human EPO to anemic patients suffering from chronic renal failure, AIDS, or bone marrow suppression due to chemotherapy has dramatically alleviated their need for blood transfusions. In contrast to many other hematopoietic growth factors, EPO acts primarily on relatively mature erythroid progenitors, and to a significantly lesser extent megakaryocyte progenitors, within the fetal liver and adult bone marrow (reviewed in (Krantz, 1991). The receptor for EPO is normally restricted in its expression to relatively mature cells of the erythroid and megakaryocyte lineage(Youssoufian, Longmore et al., 1993), and has also been reported to be expressed in the placenta(Sawyer, Krantz et al., 1989), by embryonic stem cells(Keller, Kennedy et al., 1993), on endothelial cells(Anagnostou, Lee et al., 1990), and on some neuronal-like celllines(Masuda, Nagao et al., 1993). The functional relevance of this developmentally diverse EPO-R gene expression is not clear, and quite apart from its function in hematopoiesis, EPO-Rs may play other roles in non-hematopoietic cells. In the case of the erythroid lineage, expression of the EPO-R is low or absent on immature progenitors such as BFU-E and is increased by the CFU-E stage of differentiation, before decreasing as the erythroid cells undergo terminal differentiation(Sawada, Krantz et al., 1990). Stimulation of adult BFU-E with EPO as a single stimulus fails to support the formation of erythroid colonies(Oai, Krantz et al., 1991). Whether EPO directly affects differentiation as well as proliferation is not clear. A murine EPO-R cONA, which stimulates EPO-dependent proliferation of several hematopoietic cell lines, was isolated by expression cloning and encodes a type I membrane-spanning protein of 66kO which is a member of the cytokine receptor superfamily(O'Andrea, Lodish et al., 1989). Studies have detected a single binding affinity for EPO (kO = 300-800pM) in heterologous hematopoietic cells, fibroblasts, or COS cells transfected with the EPO-R cONA. Binding studies on human and murine bone marrow and fetal erythroid progenitor cells have detected either one or two affinities for radioiodinated EPO (Youssoufian, Longmore et al., 1993). Two affinities for EPO may suggest the presence of other components which modulate the binding activity of the cloned EPO-R. However, it is unresolved whether there are indeed EPO-Rs of multiple affinities. A constitutively active (cytokine-independent) form of the EPO-R has been isolated from a retroviral transduction system and found to contain a single point mutation, resulting in an Arg to Cys change at residue 129 of the exoplasmic domain (Yoshimura, Longmore et al., 1990). The R129C receptors form disulfide-linked homodimers independent of hormone, yet retain the capacity to bind EPO with a single affinity (kD= 700pM)(Watowich, Yoshimura et al., 1992). Since several members of the cytokine receptor family are active as ligand-induced homo- or heteromultimers, the disulfide-linked dimers most likely mimics the structure of the hormone-bound form of the EPO-R and thus transmit a constitutive proliferative signal. Chimeric c-kit/EPO-R receptors transmit mitogenic signals in response to added c-kit ligand (Ohashi, Maruyama et al., 1994). When an EPO-R lacking the entire cytoplasmic region is coexpressed, in excess, with wild type EPO-R EPO-induced mitogenic signals are blocked ("dominant negative" receptor form) (Watowich, Hilton et al., 1994). Taken together these observations strongly suggest that the functional cell surface form of the EPO-R is a multimeric protein complex consisting of, at least, a homodimer of the cloned cDNA. Although the cytosolic domain of the EPO-R does not contain an obvious protein kinase domain, ligand induced multimerization of the EPO-R stimulates rapid and transient tyrosine phosphorylation of a number of cellular proteins, which are essential for EPO-induced proliferation (Miura, D'Andrea et al., 1991). Sequence analysis, and functional mutagenesis studies of members of the cytokine receptor superfamily of receptors have identified two small conserved motifs, box 1 and box 2, in the membrane proximal cytoplasmic domains, which are required for proliferative signals (Murakami, Narakaki et al., 1991). This membrane proximal domain has been shown to be necessary for binding and phosphorylation of Janus kinase family members (cytosolic, nonreceptor tyrosine kinases). There is some specificity for the JAK kinase member utilized by different receptors. For example, the EPO-R activates only JAK2, whereas the 11-6 receptor signal transducing component, gp130, activates either or both JAK1 and JAK2, depending upon the cell type expressing gp130 (Witthuhn, Quelle et al., 1993)(Stahl and Yancopoulos, 1993).

Results and Discussion

To identify features of the EPO-R that mediate its interaction with JAK2, we generated chimeric receptor proteins that contained the cytoplasmic domain of the EPO-R or gp130 fused to the extracellular and transmembrane domain of the VSV G protein. This modification allowed us to immunoprecipitate and immunoblot both EPO-R and gp130 receptors with the same monoclonal antibody and eliminated the variability of using different receptor antibodies for our comparison. Because the kinase activity of JAK2 is not detectable in the absence of cytokine stimulation we generated constitutively active forms of JAK1 and JAK2 by replacing the tyrosine kinase domain of the JAKs with an epitope tagged (myc) tyrosine kinase domain from p59fyn .Thus anti-myc monoclonal antibodies will immunoprecipitate the JAKs. The chimeric receptors and JAK kinases were transiently coexpressed in HeLa cells using the vaccinia-17 expression system to determine if cytokine receptors and JAK kinases could form stable complexes. The membrane proximal domain of the EPO-R was required for association with JAK2, and JAK1 did not associate with the EPO-R (Table 1). Specifically box 1, not box 2, was required for the association between the EPO-R and JAK2. Similarly the membrane proximal box 1 domain of gp130 was required for association with JAK2, but also associated with JAK1. This suggests that box 1 sequences are required for both JAK1 and JAK2 association with cytokine receptors but that sequences outside the box 1 domain regulate the specificity of JAK kinase association. IL-6 and CNTF, cytokines whose receptors utilize gp130, can stimulate the phosphorylation of JAK1, JAK2, and tyk2 depending on the cell line examined(Stahl and Yancopoulos, 1993). In contrast only JAK2 has been shown to associate with the EPO-R, in vivo(Miura, Nakamura et al., 1994). Our results demonstrated that the EPO-R and gp130 behave differently with regards JAK activation, in a heterologous cell line. Thus, receptor interactions with JAK kinases could be regulated in two ways; receptors for cytokines such as EPO encode the specificity of association within receptor protein sequence whereas JAK kinase interaction with receptors like gp130 must be influenced by cell-specific factors. It will be important to determine what structural features of the receptors regulate the specific interaction of JAK2 with the EPO receptor protein. Expression of exogenous EPO-R in some IL-3- and GM-CSF-dependent cell lines (BaF3, DA-1, 32D, and FDCP-1) confers upon these cells the capacity to proliferate in response to EPO(Youssoufian, Longmore et al., 1993). This is not the case for most IL-2-dependent cell lines (CTLL-2, HT-2)(Yamamura, Kageyama et al., 1992). Similarly, studies of the biochemical events following ligand binding to members of the cytokine receptor family have suggested conservation of signal transduction mechanisms between some members of the cytokine receptor family(Youssoufian, Longmore et al., 1993)(Ihle, Witthuhn et al., 1994). One possible interpretation for these observations is that specificity of response to a growth factor is obtained at the level of receptor expression rather than at the level of signal transduction. Given the normally limited expression of the EPO-R, we were interested to determine if the EPO-R was capable of generating a proliferative and differentiative signal in cells normally responsive to the ligands of other members of the cytokine receptor family. Retroviral vectors expressing EPO-R(wt) and EPOR(R129C) were constructed, primary hematopoietic progenitor cells (d12.5 fetal liver and 5-fluorouracil treated adult bone marrow) were infected, and cultured in methylcellulose in the absence or presence of added EPO(McArthur, Longmore et al., 1995)(Pharr, Hankins et al., 1993).

Table 1. The association of IAK2 or JAKl with the EPO-R or gp130

(A) Requirement of the membrane proximal box 1 of the EPO-R for JAK2 association. (8) Boxl but not box2 is critical for JAK association with the cytoplasmic domain of gp130. Open and closed boxes represent the extracellular and transmembrane domain of the VSV G protein, respectively. Hatched boxes represent the intracellular domain of the indicated cytokine receptor. HeLa cells were transfected with cDNAs corresponding to the indicated cytokine receptor chimeras and either JAKl or JAK2. Cells were lysed in 1% 8rij 96 and the detergent soluble extract was immunoprecipitated with anti-VSV G or anti-myc antibodies. Association of the cytokine receptor chimeras with the kinases was determined by in vitro kinase reactions, in the presence of 32p-gATP, upon the immunoprecipitated pellets. The products of the in vitro kinase reactions were separated on a 7.5% acrlamide-SDS gel and analyzed by autoradiography. Expression of cytokine receptor chimeras was confirmed by anti-VSV G protein immunoblot, and JAK kinase by anti-myc IP kinase reactions.

Table 2 shows the effect of EPO upon erythroid progenitors transduced with wild type EPO-R or an activated form of the EPO-R, EPO-R(R129C). Cultures of cells infected with the EPO-R retrovirus demonstrated erythroid colony formation in response to EPO as a single stimulus when analyzed at day 2 or day 8. Normally stimulation of BFU-E by EPO does not support the development of day 8 colonies (Table 1 NeoR experiment). The finding of frequent day 8 erythroid colonies in EPO stimulated cultures of fetal liver cells transduced with EPO-R(wt) suggest that the signal transduction mechanism allowing fior EPO-induced cell proliferation and differentiation are present in some murine BFU-E. Unstimulated cultures of fetal liver cells transduced with EPO-R(R129C) contained erythroid colonies when analyzed at day 2 but not day 8. However, the same cell population stimulated by EPO did contain day 8 erythroid colonies. This suggests that there may be some functional differences in signals generated by EPOR(R129C) in the unstimulated versus the EPO-stimulated state. Analysis of cell surface EPO-Rs in BaF3 expressing EPO-R(R129C) detects a minority of receptor dimers as opposed to monomers(Watowich, Yoshimura et al., 1992). Thus EPOstimulation may be able to generate additional EPO-R dimers leading to an enhanced signal. When fetal liver or post 5-FU bone marrow cells transduced with EPO-R or . EPO-R(R129C) were stimulated by EPO there was an increased number of megakaryocyte colonies (Table 3). This appeared to be due to the recruitment of additional progenitors cells rather than altered committment of existing progenitors as there was no change in non-megakaryocyte colony numbers. The recruitment of additional megakaryocyte clones was of both the BFU-Meg and CFU-Meg type. These observations suggest the presence of signal transduction mechanisms alowing EPOinduced proliferation and differentiation in cells of both erythroid and megakaryocyte lineages at stages of differentiation prior to the expression of significant numbers of EPO-Rs. The current experiments also demonstrated a direct stimulation of individual clones of the macrophage lineage by EPO, when these cells are transduced with EPO-R or EPO-R(R129C) (Table 3). In contrast, the data did not demonstrate EPO-induced proliferation of isolated clones of the granulocyte lineage. Cells of the granulocyte lineage transduced with a c-fms (CSF-1R) also fail to proliferate in response to CSF-1(McArthur, Rohrschneider et al., 1994). These results, taken together, suggest that granulocyte progenitors may express a restricted range of signal transduction molecules influencing their response to specific cytokines.

Table 2. Erythroid colony formation in cultures of fetal liver cells induced to express EPO-R or EPO-R(R129C)

Day 12.5 fetal liver cells were infected with retroviruses expressing NeoR, EPOR/NeoR, or EPO-R(R129C)/NeoR by cocultivation with retroviral producing cells. Cultures contained 5 x 104 fetal liver cells. All cultures contained 1.5 mg/ml of G418. Results are from quadruplicate cultures and expressed as number of colonies relative to NeoR cultures. Cultures were stimulated by 2 U /ml of hEPO or 10% spleen conditioned media (SCM). Maximal responses are depicted in cultures of NeoR transduced cells containing EPO and SCM.

Hematopoietic cells at different stages in development are thought to have different complements of cytokine receptors. Whether the appearance of specific receptors initiates a particular developmental sequence is not known: Does the acquisition of a lineage-specific receptor induce differentiation? To address this question we utilized retroviral-mediated gene transfer to express the EPO-R in multilineage blast cell progenitors (Pharr, Ogawa et al., 1994) (Table 4). By DNA PCR we demonstrated that pluripotent blast cell clones could be infected with EPO-R(R129C)-expressing retroviruses. Blast cells and their progeny, CFU-GM and CFU-mix, express retrovirally derived EPO-R(R129C) as determined by PCR of cDNA prepared from these colonies (not shown). We observed no evidence that blast cells transduced with EPO-R(R129C) could induce erythroid differentiation, even following the addition of EPO. These results are consistent with in vivo experiemtns in which EPO has been shown to regulate the rate at which committed erythrocyte progenitors become erythroblasts.

Table 3. Colony formation in cultures of post 5-FU bone marrow cells induced to express EPO-R or EPO-R(R129C).

Adult mice had their bone marrow harvested 4 days after treatment with 5fluorouracil, 150 mg/kg. Cells were infected by cocultivation with retroviral producing cells. Cultures contained 5 x 104 post 5FU bone marrow cells. All cultures contained 2 mg/ml G418. Cultures were stimulated by 2 U/ml of hEPO, and 1000 U/ml of mIL-3. Results are from triplicate plates, scored after 7 days of culture, and reported as colony numbers relative to NeoR infected plates. G, granulocytes; GM, granulocyte-macrophage; M, macrophage; MEG, megakaryocyte.

The initiation of differentiation could be induced by an exogenous stimulus or could be a spontaneous random event with survival of committed cells dependent on the availability of a supportive milieu. Metcalf analyzed the progenitor content of developing blast cell colonies and found that GM-CSF or IL-3 with SCF increased the relative frequency of granulocytic progenitors. Others have demonstrated that primitive progenitors derived from human cord blood divide asymmetrically and that this was not affected by different combinations of cytokines. The results presented here are more consistent with a supportive role for EPO in the differentiation of red cells. Thus expression of lineage-specific receptors for EPO may be a consequence rather than a cause of differentiation.

Table 4. Effect of EPO-R(R129C) on the composition of mixed colonies derived from infected blast cells

Spleen cells were harvested from mice treated with 5-FU (150 mg/ml) 4 days earlier. These were cultured in methylcallulose. Blast cell colonies, identified on day 6 or 7, were picked, washed and resuspended in retroviral supernatant or control. Following infection samples of blast cells were plated at 50-100 per plate. Mixed colonies were identified, picked and cytospin preparations counted after Wrights staining. acultures contained Stem Cell Factor (SCF) (3 U/rnl), IL-3 (100 U/ml), and EPO (1 U/ml) and were scored on day 9. G, granulocytes; M, macrophage; E, erythroid; MEG, megakaryocyte.


Anagnostou, A., E. S. Lee, N. Kessimian, R. Levinson and M. Steiner .1990. Erythropoietin has a mitogenic and positive chemotactic effect on endothelial cells. Proc. Natl. Acad. Sci. USA. 87: 5978-5982.

D'Andrea, A. D., H. F. Lodish and G. G. Wong .1989. Expression cloning of the murine erythropoietin receptor. Cell 57: 277-285.

Dai, C. H., S. B. Krantz and K. M. Zsebo .1991. Human burst-forming units erythroid need direct interaction with stem cell factor for further development. Blood. 78: 2493-2499.

Ihle, I. N., B. A. Witthuhn, F. W. Quelle, K. Yamamoto, W. E. Thierfelder, B. Kreider and 0. Silvennoinen .1994. Signaling by the cytokine receptor superfamily: IAKs and STATs. Trends in Biochem. Sci. 19: 222-227.

Keller, G., M. Kennedy, T. Papayannopoulou and M. V. Wiles .1993. Hematopoietic commitment during embryonic stem cell differentiation in culture. Mol. Cell. BioI. 13: 473-486.

Krantz, S. B. .1991. Erythropoietin. Blood. 77: 419-434.

Masuda, S., M. Nagao, K. Takahata, Y. Konishi, F. Gallyas, T. Tabira and R. Sasaki .1993. Functional erythropoietin receptor of the cells with neural characteristics. I. BioI. Chem. 268: 11208-11216.

McArthur, G. A., G. D. Longmore, K. Klinger and G. R. Johnson .1995. Lineage restricted recruitment of immature hematopoietic progenitor cells in response to erythropoietin after transfection of normal hematopoietic cells with the erythropoietin receptor. Exp.Hematol..1995

McArthur, G. A., L. R. Rohrschneider and G. R. Johnson .1994. Induced expression of c-fms in normal hematopoietic cells shows evidence for both conservation and lineage restriction of signal transduction in response to M-CSF. Blood. 83: 972-980.

Miura, 0., A. D. D'Andrea, D. Kabat and J. N. Ihle .1991. Induction of tyrosine phosphorylation by the erythropoietin receptor correlates with mitogenesis. Mol. Cell. BioI. 11: 4895-4902.

Miura, 0., N. Nakamura, J. N. Ihle and A. Nobuo .1994. Erythropoietindependent association of phosphatidylinositol-s-kinase with tyrosinephosphorylated erythropoietin receptor. I. BioI. Chem. 269: 614-620.

Murakami, M., M. Narakaki, M. Hibi, H. Yawata, K. Yasukawa, M. Hamaguchi, T. Taga and T. Kishimoto .1991. Critical cytoplasmic region of the interleukin 6 signal transducer gp130 is conserved in the cytokine receptor family. Proc. Natl. Acad. Sci. USA. 88: 11349-11353.

Ohashi, H., K. Maruyama, Y.-C. Liu and A. Yoshimura .1994. Ligandinduced activation of chimeric receptors between the erythropoietin receptor and receptor tyrosine kinases. Proc. Natl. Acad. Sci. (USA). 91: 158-162.

Pharr, P. N., D. Hankins, A. Hofbauer, H. F. Lodish and G. Longmore .1993. Expression of a Constitutively Active Erythropoietin Receptor in Primary Hematopoietic Progenitors Abrogates Erythropoietin Dependence and Enhances CFU-E, BFU-E, and GM Progenitor Growth. Proc. Natl. Acad. Sci. USA. 90: 938-942.

Pharr, P. N., M. Ogawa, A. Hofbauer and G. D. Longmore .1994. Expression of an activated erythropoietin receptor or a CSF-1 receptor by pluripotent progenitors enhances colony formation but does not induce differentiation. Proc. Natl. Acad. Sci. USA. 91: 7482-7486.

Sawada, K., S. B. Krantz, C. H. Dai, S. T. Koury, S. T. Horn, A. D. Glick and C. I. Civin .1990. Purification of human blood burst forming units-erythroid and demonstration of the evolution of the erythropoietin receptor. L Cell. Physiol.l42: 219-230.

Sawyer, S. T., S. B. Krantz and K.-I. Sawada .1989. Receptors for erythropoietin in mouse and human erythroid cells and placenta. Blood. 74: 103-109.

Stahl, N. and G. D. Yancopoulos .1993. The alphas, betas, and kinases of cytokine receptor complexes. Cell 74: 587-590.

Watowich, S. S., D. J. Hilton and H. F. Lodish .1994. Activation and inhibition of erythropoietin receptor function: role of receptor dimerization. Mol. Cell. BioI. 14: 3535-3549.

Watowich, S. S., A. Yoshimura, G. D. Longmore, D. Hilton, Y. Yoshimura and H. F. Lodish .1992. Homodimerization and constitutive activation of the erythropoietin receptor. Proc. Natl. Acad. Sci. USA. 89: 2140-2144.

Witthuhn, B. A., F. W. Quelle, 0. Silvennoinen, T. Yi, B. Tang, 0. Miura and J. N. Ihle .1993. JAK2 associates with the erythropoietin receptor and is tyrosine phosphorylated and activated following stimulation with erythropoietin. Cell74: 227-236.

Yamamura, Y., Y. Kageyama, T. Matuzaki, M. Noda and Y. Ikawa .1992. Distinct downstream signaling mechanisms between erythropoietin receptor and interleukin-2 receptor." EMBO I. 11: 4905-4915.

Yoshimura, A., G. Longmore and H. F. Lodish .1990. "Point mutation in the exoplasmic domain of the erythropoietin receptor resulting in hormone-independent activation and tumorigenicity ." Nature. 348: 647-649.

Youssoufian, H., G. Longmore, D. Neumann, A. Yoshimura and H. Lodish. 1993. Structure, Function, and Activation of the erythropoietin receptor. Blood. 81(9): 2223-2236.