(Received for publication, October 24, 1996, and in revised form, January 15, 1997)
From the Institute of Life Science, Aikawamachi
2432-3 Kurume 839, Japan, the § Cellular Biochemistry,
Animal Resource Science/Veterinary Medical Science, The University of
Tokyo, Bunkyo-ku, Tokyo 113, Japan, and the ¶ Laboratory of
Molecular and Cellular Science, Department of Biology, Faculty of
Science, Niigata University, Niigata 950-21, Japan
Interaction between erythropoietin (EPO) and its membrane receptor induces the proliferation and differentiation of erythroid progenitors. EPO has been shown to activate the JAK2-STAT5 pathway in various hematopoietic cell lines, although the physiological role of this pathway is unclear. We have previously shown that epidermal growth factor activates a chimeric receptor bearing the extracellular domain of the epidermal growth factor receptor linked to the cytoplasmic domain of the EPO receptor, resulting in proliferation of interleukin-3-dependent hematopoietic cells and erythroid differentiation (globin synthesis) of EPO-responsive erythroleukemia cells. In the present study, we introduced various deletion and tyrosine to phenylalanine substitution in the cytoplasmic domain of the chimeric receptor and expressed these mutant chimeras in an EPO-responsive erythroleukemia cell line, ELM-I-1. Mutant chimeric receptors retaining either Tyr343 or Tyr401 could activate STAT5, judged by tyrosine-phosphorylation of STAT5 and induction of CIS, a target gene of STAT5. These mutants were able to induce erythroid differentiation. However, a chimeric receptor containing both Y343F and Y401F mutations could not activate STAT5 nor induce erythroid differentiation. Thus, Tyr343 or Tyr401 of the EPO receptor are independently necessary for erythroid differentiation as well as STAT5 activation. Moreover, exogenous expression of dominant-negative STAT5 suppressed EPO-dependent erythroid differentiation. These findings suggest that STAT5 plays an important role in erythroid differentiation through the EPO receptor cytoplasmic domain.
Erythropoietin (EPO)1 is a glycoprotein hormone required for the survival, proliferation, and differentiation of committed erythroid progenitor cells (1). The EPO receptor (EPOR) belongs to the cytokine receptor superfamily, which includes receptors for other hematopoietic growth factors such as interleukins, colony-stimulating factors, and growth hormone (2). A novel subfamily of protein tyrosine kinases, known as Janus kinases (JAKs), has been shown to associate with cytokine receptors, including the EPOR, and to play an important role in cytokine-dependent cell proliferation and gene regulation (3). Activated JAKs in turn convert latent cytoplasmic transcription factors, known as STATs (signal transducers and activators of transcription), into their active forms by tyrosine phosphorylation. The tyrosine phosphorylated STATs form homo- or heterodimers and translocate into the nucleus, where they bind to their specific target sequences, and then regulate expression of target genes (4).
The STAT family presently includes six members, each of which functions in a specific cytokine system. STAT5, which was originally identified as mammary gland factor regulated by prolactin (5), is activated by multiple cytokines such as IL2, IL3, IL5, granulocyte-macrophage colony stimulating factor, EPO, and thrombopoietin (6-11). Although IL2 and the other cytokines activate different JAKs, they all activate STAT5. The physiological function of STAT5 in hematopoietic cells, however, remains unclear. The role of STAT5 in cell proliferation is still controversial (9, 12-16), and little is known about its role in differentiation. However, because STAT3 has been shown to play a critical role in macrophage differentiation through gp130 (17-19), it would be interesting to determine whether STAT5 is involved in differentiation. We previously reported that exogenous expression of a chimeric receptor containing the extracellular domain of the EGF receptor (EGFR) and the cytoplasmic domain of the EPOR conferred EGF-dependent erythroid differentiation on EPO-responsive cells (20). Moreover, the membrane-proximal 127 amino acids are competent to induce differentiation. This region contains only one tyrosine residue (Tyr343), but it is sufficient to activate STAT5 (12-15). To clarify the role of STAT5 in the EPOR-mediated differentiation, we constructed various deletion and tyrosine substitution of chimeric EPOR mutants and introduced them into a highly EPO-responsive cell line, ELM-I-1. Tyrosine-mutant receptors that failed to activate STAT5 could not induce erythroid differentiation, suggesting that STAT5 positively regulates EPOR-mediated erythroid differentiation.
The wild type EGFR-EPOR chimera
(designated wt chimera in this study) and a truncated chimeric receptor
(EGFR-EPORH, designated 108 in this study) consisting of the
extracellular domain of the EGFR and the cytoplasmic domain of the EPOR
have been described previously (20, 21). These chimeric cDNAs were
subcloned into the mammalian expression vector pcDNA3 (Invitrogen)
using KpnI and XbaI sites. Deletion of
181 was
created by adding stop codon and an hemagglutinin tag (20) at the
BamHI site of the EPOR cytoplasmic domain, according to the
procedure employed for
108 deletion. Deletion mutants
55,
76, and
135 were created by adding stop codon at the indicated position in
the EPOR cytoplasmic domain by the polymerase chain reaction (PCR)
using synthetic oligonucleotide primers as follows. The forward primer,
CGCCGTACGCTGCAGCAGAAGATCT (includes the Spl I site (21)).
The reverse primers (include the EcoR I site and the stop
codon), AGAATTCACTTCAAGTGAGGTGGAG (for
55), TGAATTCAGGGGTCCAGGATGGTGT
(for
76), and AGAATTCACTTATCCAATACCAAGT (for
135).
To substitute phenylalanine for tyrosine in the 76 deletion
(designated -76YY), the SphI/EcoRI fragment,
which includes Tyr343 and Tyr401 was mutated by
PCR using primers containing Tyr to Phe substitutions. The forward
primer for Y343F was TGAGCATGCCCAGGACACCTTCTTG, and the reverse primer
for Y401F was TGAATTCAGGGGTCCAGGATGGTGAACTCA. PCR fragments were
digested with SphI and EcoRI and then exchanged for the SphI/EcoRI fragment of
76YY. The
resulting mutant containing Y343F and Y401F was designated
76FF.
Similarly, the
76 deletion mutant containing Y343F alone was created
and designated
76FY. The
76 deletion containing Y401F alone was
designated
76YF. The
135 deletion containing Y343F (
135F) was
created by using the reverse primer, AGAATTCACTTATCCAATACCAGAAGGTG. The
integrity of the sequences was confirmed by sequencing. The resulting
constructs are listed in Fig. 1.
Myc-tagged sheep wt and dominant-negative (dn) STAT5s were created by PCR. The dnSTAT5 lacking the 110 C-terminal amino acids was created according to the method of Mui et al. (16). PCR with the forward primer GGAGAATTCGGGCTGGATCCAGGCC and reverse primer CAGACTAGTAGTACTTGGAGAAGAC (for dnSTAT5) or AACACTAGTTCAGGAGAGCGAGCCT (for wtSTAT5) was performed, and the products were then subcloned into pCS-MT+ (22) to add Myc tag (MEQKLISEEDLNE, six times repeated) at the N-terminal. The HindIII/XbaI fragments containing Myc tag and cDNA were recloned into mammalian expression vector pcDNA3.
Cells, Transfection, and Differentiation AssayMurine EPO-responsive erythroleukemia cell line ELM-I-1, which was established from x-ray-induced erythroblastic leukemia, was maintained in DMEM containing 10% fetal calf serum (23). The plasmids carrying chimeric receptors were introduced into ELM-I-1 cells by electroporation, as described previously (20). After selection with G418 (1 mg/ml), stable transformants expressing the chimeric receptors were identified by flow cytometer analysis using a monoclonal anti-EGFR extracellular domain antibody (Genzyme). Transformants expressing wt or dnSTAT5 were identified by immunoblotting with anti-Myc antibody (22). The transformants expressing chimeric receptors or dnSTAT5 were maintained in DMEM containing 0.5 mg/ml G418. For differentiation assay, cells were cultured in EGF (1 µg/ml) or EPO (10 unit/ml) for 3 days. Hemoglobin-positive cells were scored after benzidine staining, and globin content was measured by immunoblotting, as described previously (20).
Phosphorylated tyrosine residues of the EPOR have been
shown to be important for recruiting signal transducing molecules
containing the SH2 domain. For example, phosphatidylinositol 3-kinase
binds to phosphorylated Tyr479 (Y8 in Fig.
1)(24), a hematopoietic cell-specific tyrosine
phosphatase to Tyr429 (Y3) (25, 26), and Syp,
another phosphatase to Tyr401 (Y2) (27).
Although it has not been demonstrated that STAT5 directly binds to the
phosphorylated tyrosine residues of the EPOR, Tyr343
(Y1) and Tyr401 (Y2) have been shown
to be independently necessary for STAT5 activation. Thus, specific
tyrosine residues of the EPOR are probably responsible for each signal
transduction pathway. To clarify which tyrosine residues are important
for EPOR-mediated erythroid differentiation, we constructed chimeric
receptors (Fig. 1) bearing the extracellular domain and the
transmembrane domain of the EGFR linked to full-length (wt) or several
deletion mutants of the cytoplasmic domain of EPOR (55,
76YY,
108,
135Y, and
181 in Fig. 1). Previous reports have indicated
that deletion of the 135 C-terminal amino acids can support
proliferation in Ba/F3 cells but that the
181 mutant cannot (21, 28).
Tyrosine substitution mutants were also created as shown in Fig. 1.
These chimeric receptors were introduced into erythroleukemia cell line
ELM-I-1. Previously, we expressed wt and 108 chimeras in a Friend
virus-transformed erythroleukemia cell line, TSA8, and found that these
chimeras could induce erythroid differentiation (20). Whereas EPO
induced globin synthesis in TSA8 cells, EPO alone was incapable of
inducing more mature differentiation, such as hemoglobin synthesis or
extrusion of the nucleus. Moreover, part of the endogenous EPOR may be
activated by gp55 of the Friend virus (29). ELM-I-1, on the other hand,
is retrovirus-free, and EPO induces not only globin but also hemoglobin
synthesis (approximately 10-30% of cells become benzidine-positive
after cultivation in the presence of EPO (30)). Thus, EPO can induce greater terminal erythroid differentiation in ELM-I-1 than in TSA8. In
particular, IL3 strongly potentiates EPO-induced erythroid differentiation (30). This is quite similar to natural bone marrow cell
differentiation in the presence of EPO and IL3. In several artificial
systems, e.g. Ba/F3 sublines expressing exogenous EPOR (31,
32), IL3 suppressed EPO-induced globin synthesis. In this sense,
ELM-I-1 cells seem to be more suitable for the study of erythroid
differentiation in vitro than other
EPO-dependent hematopoietic cells. We therefore introduced
chimeric receptors into this cell line.
The
cDNAs for the chimeric receptors subcloned into a mammalian
expression vector were introduced into the ELM-I-1 cells. Expression of
the hybrid receptors was examined by flow cytometry using anti-EGFR
antibody (data not shown). Positive clones (2-3 clones per chimeric
receptor) were cultured in the presence of EGF or EPO for 3 days and
then stained with benzidine. Without EPO or EGF, the background of
hemoglobin-positive cells was very low. Less than 0.6% of the cells
were benzidine-positive in any transformants without EPO or EGF
stimulation (Figs. 1 and 2, ). Every clone expressing
wt or mutant chimeric receptors was positive (5-48%) after culture in
EPO, thus differentiation potential was retained after chimeric
receptor expression. Representative staining is shown in Fig. 2, and
typical staining results for each transformant are shown in Fig. 1
(right). The globin content of each clone cultured in EPO or
EGF is also compared by immunoblotting (Fig. 3). As
demonstrated in TSA8 cells (20), the
108 deletion possessed differentiation potential in ELM-I-1 cells. The
135 deletion was
still capable of inducing EGF-dependent differentiation,
but
181 could not (Figs. 1 and 3,
135Y and
181). Thus, the
differentiation signal from the EPOR required membrane-proximal 90 amino acids.
The presence of Y1 in the cytoplasmic domain of the EPOR has been shown
to be sufficient to activate STAT5 in the 108 deletion mutant (12).
To determine whether tyrosine residues are important for EPOR-mediated
differentiation, we created Tyr to Phe substitutions. The
135F mutant
contains both deletion of the 135 C-terminal amino acids and the Y343F
mutation. In contrast to the
135Y deletion mutant,
135F did not
induce erythroid differentiation (Figs. 1 and. 3,
135F). Thus, Y1 is
necessary for the differentiation signal when the 135 C-terminal amino
acids have been deleted.
STAT5 has been shown to be fully activated through either Y1 or Y2 of
the EPOR cytoplasmic domain (13, 15). Thus, we introduced Tyr to Phe
substitution in the 76 deletion mutant containing only two tyrosine
residues (Y1 and Y2) (Fig. 1,
76YY). As shown in Figs. 2 and 3, the
76FY mutant, which contains Y343F substitution but retains Y2, was
able to induce EGF-dependent erythroid differentiation. Similarly,
76YF, which contains Y401F but retains Y1, was also capable of inducing differentiation (Figs. 2 and 3,
76YF). However,
76FF, which contains double mutation of Y343F and Y401F, could not
induce EGF-dependent differentiation (Figs. 2 and 3,
76FF). Thus, one of the two tyrosine residues in the
76 deletion
mutant is independently critical for erythroid differentiation.
To confirm that
the 76FY and the
76YF mutant but not the
76FF mutant can activate
STAT5 in ELM-I-1 cells, we measured STAT5 phosphorylation and
CIS-induction through the mutant chimeric receptors. The
181,
76YY,
76YF,
76FY, and
76FF transformants were stimulated
with EGF, and then JAK2 and STAT5 were immunoprecipitated. Their
tyrosine phosphorylation was detected with anti-phosphotyrosine antibody. As shown in Fig. 4A, EGF stimulated
JAK2 phosphorylation in every mutant chimera. JAK2 phosphorylation was
observed even in the
181 mutant, probably because this mutant
contains box1 and box2 regions, which are putative JAK binding sites
(Fig. 1). Thus, activation of JAK2 does not require tyrosine residues
in the EPOR cytoplasmic domain. In contrast, STAT5 was not
phosphorylated in response to EGF in the
181 nor the
76FF mutant,
whereas
76YY,
76YF, and
76FY were able to induce STAT5
phosphorylation.
CIS is an immediate early gene induced by multiple cytokines
such as IL2, IL3, and EPO in various hematopoietic cells and a direct
target of STAT5 (12, 33). We prepared RNA from EGF- or EPO-stimulated
cells expressing 76YY,
76YF,
76FY, and
76FF. Whereas EPO
induced CIS expression in all transformants, EGF induced CIS in
76YY,
76YF, and
76FY mutant cells but not in
76FF mutant cells (Fig. 4B). These data are consistent
with previous reports that Y1 or Y2 of the EPOR cytoplasmic domain is
sufficient to activate STAT5.
To clarify the critical
role of STAT5 in erythroid differentiation, we expressed wt and dn type
STAT5 in ELM-I-1 cells. The dominant-negative form of sheep STAT5 was
created by deleting the C-terminal transactivation domain as reported
by Mui et al. (16). This construct has been shown to inhibit
STAT5 activation and partially suppress cell proliferation. The Myc
epitope tag was introduced into the N terminus to detect exogenous
protein. Three independent clones from each transfection were tested
for EPO-induced hemoglobin synthesis. As shown in Fig.
5, expression of wtSTAT5 did not affect EPO-induced
benzidine positivity, whereas dnSTAT5 suppressed differentiation to
about 5%. These data support that STAT5 plays a critical role in
EPOR-mediated erythroid differentiation in ELM-I-1 cells. However, we
could not rule out the possibility that dnSTAT5 may suppress the
activation of other signaling molecules through Y1 and Y2, because
dnSTAT5 may tightly bind to these two phosphorylated tyrosine
residues. Extensive study is under way to determine whether STAT5 is
directly involved in differentiation.
Our results contrast strongly with a recent report by Chretien et al. (34). They found that EPO-induced differentiation in a human leukemia cell line, TF-1, correlates with impaired STAT5 activation, although direct evidence that STAT5 suppresses differentiation was not provided. This discrepancy is apparently caused by the difference in cell lines used. ELM-I-1 can grow in the absence of any cytokines, whereas TF-1 requires granulocyte-macrophage colony stimulating factor for growth. These two cells may be derived from different developmental stages of hematopoietic cells, and such differences may cause distinct STAT5 requirements for differentiation. However, the same as natural BFU-E cells, IL3 enhances EPO-induced erythroid differentiation of ELM-I-1 cells, although it does not induce erythroid-differentiation by itself. It is particularly important to examine the role of STAT5 in in vivo differentiation of natural erythroid progenitor cells. Studies are under way using bone marrow and fetal liver cells infected with retrovirus bearing chimeric receptors or dnSTAT5.
We thank Dr. H. Wakao for anti-STAT5 antiserum and H. Ohgusu for technical assistance.