DNA Encoding Granulocyte Colony-Stimulating Factor Receptor and Protein Thereof

This invention relates to an isolated DNA encoding granulocyte colony-stimulating factor receptor. More particularly, it relates to an isolated DNA encoding a receptor peptide capable of specifically binding to a colony-stimulating factor (hereinafter, referred to as G-CSF), an expression vector containing said DNA, a transformant transformed by said vector, and a process for the production of said receptor by culturing said transformant. The present invention also relates to a recombinant G-CSF receptor prepared according to the present process.

Proliferation and differentiation of hematopoietic cells are regulated by hormone-like growth and differentiation factors designated as colony-stimulating factors (CSF) (Metcalf, D. Nature 339, 27-30 (1989)). CSF can be classified into several factors according to the stage of the hematopoietic cells to be stimulated and the surrounding conditions as follows: granulocyte colony-stimulation factor (G-CSF), granulocyte-macrophage colony-stimulation factor (GM-CSF), macrophage colony-stimulation factor (M-CSF), and interleukin 3 (IL-3). G-CSF participates greatly in the differentiation and growth of neutrophilic granulocytes and plays an important role in the regulation or blood levels of neutrophils and the activation of mature neutrophils (Nagata, S., "Handbook of Experimental Pharmacology", volume "Peptide Growth Factors and Their Receptors", eds. Sporn, M. B. and Roberts, A. B., Spring-Verlag, Heidelberg, Vol.95/1, pp.699-722 (1990); Nicola, N. A. et al., Annu.Rev.Biochem. 58, pp.45-77 (1989)). Thus, G-CSF stimulates the growth and differentiation of neutrophilic granulocytes through the interaction between cell-surface receptors on precursors of neutrophilic granulocytes to give mainly the neutrophilic granulocytes (Nicola, N. A. & Metcalf, D., Proc. Natl. Acad., Sci. USA, 81, 3765-3769 (1984)).

G-CSF has various biological activities in addition to those mentioned above. For example, G-CSF prepared by recombinant DNA technology has proven to be a potent regulator of neutrophils in vivo using animal model systems (Tsuchiya et al., EMBO J. 6 611-616 (1987); and Nicola et al., Annu. Rev. Biochem. 58, 45-77 (1989)). Recent clinical trials in patients suffering from a variety of hemopoietic disorders have shown that the administration of G-CSF is beneficial in chemotherapy and bone marrow transplantation therapy (Morstyn et al., Trends Pharmacol. Sci. 10, 154-159 (1989)). It is also reported that G-CSF stimulates the growth of tumor cells such as myeloid leukemia cells.

Despite the biological and clinical importance of G-CSF, little is known about the mechanism through which G-CSF exerts its effects. Therefore, it has been needed to elucidate such mechanism to establish more effective treatment and diagnosis for G-CSF-related disorders. For this purpose, the biochemical characterization of specific cell-surface receptors for G-CSF and the evaluation of interaction between G-CSF and the receptor must be performed.

Several reports suggested that the target cells of G-CSF is restricted to progenitor and mature neutrophils and various myeloid leukemia cells (Nicola and Metcalf, Proc. Natl. Acad. Sci. USA, 81, 3765-3769 (1984); Begley et al., Leukemia, 1, 1-8 (1987); and Park et al., Blood 74, 56-65 (1989)). Human G-CSF is a 174 amino acid polypeptide while murine G-CSF consists of 178 amino acids. Human and mouse G-CSFs are highly homologous (72.6%) at the amino acid sequence level, in agreement with the lack of species-specificity between them (Nicola et al, Nature 314, 626-628 (1985)). What makes the research in G-CSF more interesting is that G-CSF receptor has also recently been found in non hemopoietic cells such as human endothelial cells (Bussolino et al., Nature 337, 471-473 (1989)) and placenta (Uzumaki et al., Proc. Natl. Acad. Sci. USA, 86, 9323-9326 (1989)).

As can be seen from the above, the elucidation of the interaction between G-CSF and its receptor should greatly contribute to the development of the treatment or prophylaxis of various diseases including hematopoietic disorders using G-CSF, whereby providing more effective and proper treatments on such diseases. Thus, such elucidation is important not only academically but also clinically. On the other hand, the receptor itself can be useful. For instance, a soluble form of the G-CSF receptor may be useful clinically to inhibit the proliferation of some G-CSF-dependent human myeloid leukemia cells (Santoli et al., J.Immunol. 139, 3348-3354 (1987)). The investigation into the expression of G-CSF receptor in tumor cells such as myeloid leukemia may be beneficial to establish an effective clinical application of G-CSF. Accordingly, owing to the various academic and practical usefulness, a stable supply of a G-CSF receptor-encoding gene and the G-CSF receptor has been demanded.

Recently, the technology of genetic engineering has been used for the production of various physiologically active substances. The production by the genetic engineering is generally carried out by cloning DNA encoding desired polypeptide, inserting said DNA into a suitable expression vector, transforming an appropriate host cell such as microorganism or animal cell by the expression vector, and making the transformant express the desired polypeptide.

To apply the genetic engineering technique to the production of G-CSF receptor, cloning of DNA encoding G-CSF receptor is firstly required. However, cloning cDNA encoding G-CSF receptor was hampered by the low number of receptors present on the cell surface (at most hundreds to 2,000 receptor per cell).

FIG. 1(A-C) depicts the nucleotide sequence and deduced amino acid sequence of murine G-CSF receptor (SEQ ID NOS:1-2).

FIG. 2(A-C) depicts a schematic representation and restriction map of murine G-CSF receptor cDNAs (pI62, pJ17 and pF1), and hydropathy plots thereof.

FIG. 3a depicts the saturation binding of murine .sup.125 I-G-CSF to COS cells.

FIG. 3b depicts scatchard plot of binding data of murine .sup.125 I-G-CSF to COS cells.

FIG. 3c depicts saturation binding of murine .sup.125 I-G-CSF to NFS-60 cells.

FIG. 3d depicts scatchard plot of binding data of murine .sup.125 I-G-CSF to NFS-60 cells.

FIG. 4 depicts specific binding of murine G-CSF to recombinant murine G-CSF receptor expressed by COS cells.

FIG. 5 depicts crosslinking of murine G-CSF receptor expressed in COS cells and those expressed by NFS-60 cells with .sup.125 I-G-CSF.

FIG. 6 depicts northern hybridization analysis of murine G-CSF receptor mRNA.

FIG. 7(A-D) depicts alignment of amino acid sequence of murine G-CSF receptor and those of other growth factors and schematic representation of murine G-CSF receptor.

FIG. 8a depicts the nucleotide sequence and deduced amino acid sequence of plasmids pHQ3 and pHG12 (SEQ ID NOS:3-4).

FIG. 8b depicts the nucleotide sequence and deduced amino acid sequence in plasmid pHQ2, said sequence being different from that of pHQ3 and corresponding to the sequence downstream from nucleotide 2,034 in said pHQ3 (SEQ ID NOS:5-6).

FIG. 8c depicts the nucleotide sequence and deduced amino acid sequence of the insertion present in pHG11, said sequence being inserted in pHQ3 (SEQ ID NOS: 7-8).

FIG. 9 depicts a schematic representation and restriction map of pHQ3, pHG12, pHQ2 and pHG11 described in FIG. 8.

FIG. 10a depicts saturation binding of murine I-CGF to COS cells.

FIG. 10b depicts scatchard plot of G-CSF binding data.

FIG. 11 depicts northern hydridization analysis of human G-CSF receptor mRNA.

FIG. 12 depicts detection of human G-CSF receptor mRNA by PCR.

FIG. 13 depicts southern hybridization analysis of human G-CSF receptor gene.

FIG. 14 depicts schematic representation map of expression vector pEF-BOS.

The finding that human and mouse G-CSFs are highly homologous (72.6%) lacking species specificity lead to an presumption that these G-CSFs probably cross-react. Thus, present inventors have studied for the purpose of obtaining a G-CSF receptor which can be used in the research as well as the diagnostic analysis, and succeeded in the purification of the receptor as a protein with a molecular weight (M.W.) of 100,000 to 130,000 from a solubilized mouse G-CSF receptor from mouse myeloid leukemia NFS-60 cells. The purification can be carried out by, for example, extracting cell membrane suspension with CHPAS (3-[(3-cholamidepropyl)dimethylammonio]-1-propanesulfonic acid), treating the extract with G-CSF-affinity chromatography prepared by binding recombinant human G-CSF to the gel resin, and purifying by gel filtration.

The present inventors have also succeeded in the isolation and cloning of cDNA encoding murine G-CSF receptor (hereinafter, referred to as G-CSF receptor cDNA) by the reverse transcription of mRNA isolated from NFS-60 cells for the first time. Thereafter, the nucleotide sequence of cDNA and deduced amino acid sequence were determined. The nucleotide sequence and predicted amino acid sequence of murine G-CSF receptor cDNA are shown in FIG. 1, and are identified in the Sequence Listing as SEQ ID NO:1 and SEQ ID NO:2. When COS cells were transfected with an expression cloning vector containing murine G-CSF receptor cDNA, said cells expressed a receptor which has similar properties to that of the native G-CSF receptor on NFS-60 cells. Comparison of the amino acid sequence predicted from murine G-CSF receptor cDNA with those of other members belonging to growth factor receptor family revealed that the murine G-CSF receptor possesses many properties commonly found in these members (FIG. 7). A probe prepared from thus obtained murine G-CSF receptor DNA hybridized with human G-CSF receptor, demonstrating that human and murine G-CSF receptors are highly homologous. Therefore, murine G-CSF receptor can serve as so-called "an intermediate" for the preparation of human G-CSF receptor.

As the next step, the present inventors studied with a purpose of providing sufficient amount of human G-CSF receptor and succeeded in the isolation and cloning of cDNA encoding human G-CSF receptor by isolating total mRNA from human placenta and U937 cells, constructing cDNA library, and screening said library using a probe prepared from murine G-CSF receptor cDNA.

When COS cells were transfected with a plasmid encoding human G-CSF receptor of the present invention, said cells expressed a receptor which has similar specific binding properties to that of the native human G-CSF receptor.

Accordingly, the present invention provides an isolated DNA encoding G-CSF receptor. The present invention further provides an expression vector containing G-CSF receptor DNA. The present invention also provides a method for producing G-CSF receptor which comprises transforming a host cell by the expression vector, growing said transformant in a medium, and recovering G-CSF receptor.

For purposes of the invention, the term "G-CSF receptor peptide" herein used refers to both the mature G-CSF receptor peptide and peptide fragments thereof, said fragments having an ability to bind specifically to G-CSF.

The human G-CSF receptor, which has been hardly obtained heretofore, can be easily prepared by the genetic engineering by virtue of the present invention, and in turn used for many studies directed to, for example, the elucidation of the mechanism through which the G-CSF and/or G-CSF receptor exerts the effect, the clinical (for diagnosis and treatment) application, as well as for the promotion of the practical application. Additionally, probes obtained from G-CSF receptor cDNA can facilitate the detection of G-CSF receptors on tumor cells such as leukemia cells before the clinical application of G-CSF during the treatment of patients suffering from these diseases. Therefore, said cDNA is useful to perform an effective clinical application of G-CSF. Furthermore, proteins or compounds capable of binding to G-CSF receptors can be developed by investigating into tertiary structure of soluble-form G-CSF receptor which are prepared in large scale by the DNA recombinant technology.

Cloning of DNA encoding murine G-CSF receptor was carried out as follows. Thus, G-CSF receptor was initially purified from mouse myeloid leukemia NFS-60 cells which have relatively higher expression of the G-CSF receptor and determined the molecular weight of about 100,000 to 130,000 dalton. Total RNA was prepared from NFS-60 cells by the guanidine isothiocyanate/CsCl method, and poly(A) RNA was selected, which was then used for the synthesis of double-stranded cDNA using the reverse transcriptase, DNA polymerase and the like. A cDNA library was constructed in the mammalian expression vector CDM8 (Seed, Nature 329, 840-842 (9187)), as 884 pools of 60 to 80 clones. Plasmid DNAs from each pool were prepared and introduced into COS-7 cells. Two positive pools I62 and J17 which showed significant binding of radioiodinated G-CSF were selected. From these pools were identified plasmids pI62 and pJ17 which have higher binding activity with G-CSF. When plasmids pI62 and pJ17 were transfected into COS-7 cells, resultant cells expressed receptors capable of binding to G-CSF.

The determination of nucleotide sequences of resultant plasmids pJ17 and pI62 revealed that the two cDNAs contain the complete coding sequence of G-CSF receptor but lack the poly(A) tract and poly (A) additional signal. The cDNA library was, therefore, rescreened by colony hybridization using the 2.5 kb HindIII-XbaI fragment of pJ17 as a probe. Among positive clones was selected pF1 which had 603 bp 3' non-coding region and contained two overlapping poly(A) addition signals. The composite nucleotide sequence of the three cloned cDNAs (pI62, pJ17 and pF1) is presented in FIG. 1 together with the predicted amino acid sequence, also identified in the Sequence Listing as SEQ ID NO:1 and SEQ ID NO:2. The schematic representation and restriction map of the hydropathy plot of them are presented in FIG. 2.

The murine G-CSF receptor cDNA cloned by the present invention has the following characteristics.

There is a long open reading frame starting from the initiation codon ATG at nucleotide position 180-182 and ending at the termination codon TAG at position 2691-2693 (2,511 nucleotides). At the 5' upstream from the long open reading frame, three other potential initiation codon ATGs can be found at petitions 73, 105 and 126. All of these are followed by short open reading frames. Deletion of these ATG codons from the cDNA by digesting the plasmid pI62 with HindIII did not increase or decrease the expression level of the recombinant G-CSF receptor in COS cells.

The long open reading frame starts with a stretch of hydrophobic amino acids which seems to serve as a signal sequence. By comparing the 5' portion of the sequence with typical signal peptide cleavage sites, the N-terminal 25 amino acids were assigned as the signal sequence.

The mature murine G-CSF receptor thus consists of 812 amino acids with a calculated molecular weight (M.W.) of 90,814. This M.W. is 5,000 to 35,000 daltons smaller than the M.W. (95,000 to 125,000) estimated from the .sup.125 I-G-CSF cross-linking experiment (Example 2 (2), FIG. 5), or the M.W. of the purified murine G-CSF receptor. The difference is probably due to the attachment of sugar moieties to some of the 11 putative N-glycosylation sites (Asn-X-Thr/Ser) which are found on the extracellular domain of the G-CSF receptor (FIG. 1).

According to the hydropathy plot (FIG. 2, B) (Kyte and Doolite, J.Mol.Biol., 157, 105-132 (1982)) of the amino acid sequence of the mature G-CSF receptor, there exists a stretch of 24 uncharged amino acids extending from Leu-602 to Cys-625, which is followed by three basic amino acids. These properties are consistent with those observed in the membrane-spanning segments of many proteins.

The mature G-CSF receptor thus appears to consist of an extracellular domain of 601 amino acids, a membrane-spanning domain of 24 amino acids (single trans-membrane domain), and a cytoplasmic domain of 187 amino acids. The NH.sub.2 -terminal half of the extracellular domain is abundant in cysteine residues (17 residues in 373 amino acids), which seems to be a feature common to the ligand-binding domain of many receptors (McDonald et al., British Medical Bulletin 455, 54-569 (1989)). As found in erythropoietin receptor (D'Andrea et al., Cell 57, 277-285 (1989)), the G-CSF receptor is rich in proline (80 residues, 9.9%). Furthermore, the content of tryptophan residues in murine G-CSF receptor is relatively high (26 residues, 3.2%).

As mentioned above, the murine G-CSF receptor of the present invention consists of N-terminal signal sequence, single transmembrane domain, extracellular domain at N-terminal region, and cytoplasmic domain at C-terminal region, which are commonly found features among growth and differentiation factors.

Comparison of amino acid sequence of the murine G-CSF receptor with that of other growth factor receptors, such as growth hormone, prolactin, erythropoietin, IL-6, IL-2, IL-4, IL-3, and GM-CSF receptors revealed the following facts (FIG. 7). In the G-CSF receptor, as is often found in the growth factor receptors, the consensus cysteine and tryptophan residues are conserved, and the "WSXWS" motif (Gearing et al., EMBO J. 83, 667-3676 (1989); and Itoh et al., Science 247, 324-326 (1990)) is also found at amino acid residues 294-298, suggesting that the G-CSF receptor belongs to the family (see, FIG. 7(a)). In this comparison of the G-CSF with other hemopoietic growth factor receptors, it may be noteworthy that the similarity (44.6%) of the G-CSF and IL-6 receptors is less pronounced than that of the G-CSF and prolactin receptors. Furthermore, as shown in FIG. 7(b), the amino acid sequence from 376 to 601 in the extracellular domain of the G-CSF receptor has a significant similarity (42.9%) with a part of the extracellular domain of chicken contactin (Ranscht, J.Cell.Biol. 107, 1561-1573 (1988). Contactin is a neuronal cell surface glycoprotein of 130 KD and seems to be involved in cellular communication in the nervous system. Since the region from amino acid residues 737 to 818 of contactin can be aligned with the fibronectin type III segment which participates in binding to cells, heparin and DNA, it is possible that this region plays an important role in cell adhesion.

Granulopoiesis occurs daily in bone marrow, and the direct interaction of the neutrophilic progenitor cells with the bone marrow stroma cells has been proposed (Roberts et al., Nature 332 376-378 (1988)). The similarity between a part of extracellular domain of the G-CSF and that of contactin may suggest that this region is involved in the communication of neutrophilic progenitor cells and stroma cells.

The cytoplasmic domain of G-CSF receptor, as observed in other growth factor receptors, is rich in serine (12.8%) and proline (12.3%). When the sequences of transmembrane and cytoplasmic domains of the G-CSF receptor are compared with those of IL-4, a significant similarity was found.

As shown in FIG. 7(c), the transmembrane domain and the first 46 amino acids of the cytoplasmic domain of the G-CSF receptor are homologous (50.0%) to the corresponding regions of the murine IL-4 receptor. Furthermore, amino acid residues 672 to 808 of the G-CSF receptor show significant similarity (45.4%) with amino acid residues 557 to 694 on the IL-4 receptor. These results suggest that the signal transduction by G-CSF and IL-4 may be mediated by a similar mechanism.

In the illustrated Examples of the present invention, cDNA was obtained by reversely transcribing the mRNA isolated from murine leukemia NFS-60 cells. However, the same cDNA can be obtained from other murine cells such as WEHI-3B D.sup.+ and bone marrow cells (see, FIG. 6). The cDNA is also homologous to that of other species including human being.

The 3.7 kb mRNA for the G-CSF receptor was detected not only in NFS-60 cells but also in WEHI-3B D.sup.+ cells (FIG. 6), suggesting that the same G-CSF receptor is involved in G-CSF-induced proliferation of NFS-60 cells, and differentiation of WEHI-3B D.sup.+ cells. The different effects of G-CSF on NFS-60 and WEHI-3B D.sup.+ cells, that is, it stimulates the proliferation of NFS-60 cells, while it stimulates the differentiation of WEHI-3B D.sup.+ cells, may therefore be mediated by different signal transduction mechanisms downstream of the receptor. In this regard, it is interesting that the c-myb and evi-1 locus, which appear to participate in differentiation of myeloid cells, are rearranged in NFS-60 cells but not WEHI-3B D.sup.+ cells (Morishita, et al., Cell 54, 831-840 (1989).

The DNA encoding murine G-CSF receptor can be prepared according to the nucleotide sequence presented in FIG. 1, identified in the Sequence Listing as SEQ ID NO:1. It is available from Escherichia coli pI62 by a conventional method, which was originally deposited as a domestic microorganism deposit (FERM P-11353) at the Fermentation Research Institute, Agency of Industrial Science and Technology, Ministry of International Trade and Industry, Japan, on Mar. 9, 1990 and converted into an international one (FERM BP-3312) on Mar. 16, 1991 under the provision of Budapest Treaty. It also can be prepared, for example, by chemical synthesis, or by probing a genomic library or cDNA library using a probe of about 30 nucleotide synthesized on the basis of the sequence of FIG. 1. Such libraries can be constructed from any species expressing G-CSF receptor, for example, human, mouse, rat and the like. The synthesis of a DNA fragment used as a probe, construction of genomic or cDNA library, and the hybridization procedures are well-known to persons ordinary skilled in the art.

Cloning of cDNA encoding human G-CSF receptor was carried out as follows. Poly (A)RNA was selected from total RNA isolated from human placenta cells and U 937 cells (human histiocytic lymphoma, ATCC CRL 1593), which was used for the synthesis of double-stranded cDNA using a reverse transcriptase, DNA polymerase and the like. A cDNA library was prepared using the mammalian expression vector pEF-BOS (FIG. 14). The cDNA library was screened by colony hybridization or plaque hybridization using a probe prepared from the above-mentioned DNA encoding murine G-CSF receptor and positive clones were selected.

From the cDNA library constructed from mRNA prepared from U937 cells, 5 positive clones (pHQ1-pHQ5) were identified. From the cDNA library constructed from mRNA prepared from human placenta cells, more than 100 positive clones were identified and 6 clones were isolated among them, and digested with EcoRI. The resultant EcoRI fragment was subcloned in pBluescript SK(+). The isolated cDNA clones were analyzed by restriction enzyme mapping and DNA sequence analyses, which revealed that they can be divided into three classes. Most of cDNA clones isolated from U937 and placenta cDNA libraries belonged to class 1.

Class 1: plasmids pHQ3 and pHG 12 (isolated from U937 and placenta cDNA library, respectively).

Clones of this class contain a large open reading frame that encodes a protein consisting of 836 amino acids. The nucleotide sequence and deduced amino acid sequence of this cDNA clone is given in FIG. 8A, identified in the Sequence Listing as SEQ ID NO:3 and SEQ ID NO:4. The hydropathy analysis of the predicted amino acid sequence has indicated that the N-terminal 23 amino acid residues correspond to the signal sequence, and following 604, 26 and 183 residues constitute the extracellular, transmembrane and cytoplasmic domains, respectively. The schematic representation and restriction map of plasmid pHQ3 (same as that of pHG12) is given in FIG. 9.

The calculated M.W. (89,743) of mature human G-CSF receptor (813 amino acids) encoded by said plasmid differs from that reported for the native human G-CSF receptor by 30,000-60,000 daltons. This difference may be explained by N-glycosylation at some of 9 potential N-glycosylation sites on the extracellular domain of the receptor.

The overall similarity of human G-CSF receptor to the murine G-CSF receptor is 72% at the nucleotide sequence level and 62.5% at the amino acid sequence level. The amino acid sequence homology is relatively constant over the entire region of the polypeptide. In the extracellular domain of human G-CSF receptor, there are 17 cysteine residues of which 14 are conserved between human and mouse receptors. Furthermore, the "WSXWS" motif conserved in members of a cytokine receptor family can be found in the extracellular domain, which indicates that human G-CSF receptor is one of such members.

Class 2: plasmid pHQ2 (isolated from U937 cDNA library)

The nucleotide sequence of plasmid pHQ2 is identical to that of pHQ3 except that it lacks a region consisting of 88 nucleotide from the nucleotide number 2,034 to 2,121 of pHQ3, which encodes the transmembrane domain. The nucleotide sequence and predicted amino acid sequence in pHQ2, which occurs following the deletion is given in FIG. 8B, identified in the Sequence Listing as SEQ ID NO:5 and SEQ ID NO:6. This deletion results in altered translational reading frame that encodes the additional 150 amino acids downstream from the deletion point. Thus, polypeptide coded by pHQ2 seems to be a secreted, soluble form of G-CSF receptor, which consists of 748 amino acids with a calculated M.W. of 82,707.

Class 3: plasmid pHG11 and pHG5 (isolated from placenta cDNA library)

These plasmids have a 81 bp insertion at nucleotide number 2,210 of plasmid pHQ3. The insertion is in the cytoplasmic domain of the G-CSF receptor, and it does not change the translational open reading frame. The nucleotide sequence and deduced amino acid sequence of the insertion is given in FIG. 8C, identified in the Sequence Listing as SEQ ID NO:7 and SEQ ID NO:8. The putative polypeptide coded by this class of cDNA, therefore, is 27 amino acids (M.W. 2,957) larger than that coded by the class 1 G-CSF receptor. The schematic representation and restriction map of cDNA having this insertion is shown in FIG. 9.

The above three classes of cDNA encoding human G-CSF receptor were examined as to the binding specificity to G-CSF. Murine .sup.125 I-G-CSF could bind to COS cells transfected with pHQ3 in a saturating manner with an equilibrium dissociation constant of 550 pM and 3.4.times.10.sup.4 receptor/cell. The dissociation constant of the binding of murine G-CSF to human G-CSF receptor expressed in COS cells was almost similar to that observed in binding of murine G-CSF to murine G-CSF receptor. Since the native human G-CSF receptor expressed on the cell surface of U937 cells can bind human G-CSF with an equilibrium dissociation constant of 424 pM (Park, L. S., Waldron, P. E., Friend, D., Sassenfeld, H. M., Price, V., Anderson, D., Cosman, D., Adrews, R. G., Bernstein, I. D. & Urdal, D. L. Blood 74: 56-65 (1989)), these results suggest that the polypeptide coded by the cDNA in pHQ3 is sufficient to express the high-affinity receptor for human G-CSF.

When binding of murine .sup.125 I-G-CSF to COS cells transfected with pHQ2 of class 2 was examined, a very low level of binding was observed, owing to the deletion present in polypeptide coded by pHQ2. The binding sites per cell were 6.times.10.sup.3 and the dissociation constant was 440 pM. These results suggest that the receptor coded by PHQ2, which lacks the transmembrane domain, is probably the one secreted from cells as a soluble form.

To examine the binding specificity of the third class of G-CSF receptor cDNA, an expression plasmid pQw11 was constructed by inserting 5' half of pHQ3 cDNA and 3' half of pHG11 cDNA into a mammalian expression vector pEF-BOS. The resultant transformants were analyzed as to the binding property to the murine .sup.125 I-G-CSF. The results showed that the 27 amino acid insertion in the cytoplasmic domain of the receptor has little effect on the binding of G-CSF to the receptor.

The nucleotide sequence of the DNA of the invention which encodes human G-CSF receptor is shown in FIG. 8, identified in the Sequence Listing as SEQ ID NO:3-SEQ ID NO:8. The DNAs encoding human G-CSF receptor can be isolated from Escherichia coli pHQ2, pHQ3 and pHG11 by a conventional method, which were originally deposited as a domestic microorganism deposit (FERM P-11566, 11567, and 11568, respectively) at the Fermentation Research Institute, Agency of Industrial Science and Technology, Ministry of International Trade and Industry, Japan, on Jun. 28, 1990 and transferred to an international one (Escherichia coli pHQ2: FERM BP-3313; Escherichia coli pHQ3: FERM BP-3314; and Escherichia coli pHG11: FERM BP-3315) on Mar. 16, 1991 under the provision of Budapest Treaty.

The present inventors further investigated into various human cells for the presence of G-CSF receptor RNAs by Northern hybridization using probes prepared from human G-CSF receptor cDNA. A single band of 3.7 kb was observed in RNAs from U937, placenta and KG-1 cells. It was confirmed that placenta cells express the largest amounts of G-CSF receptor mRNA among them.

PCR (polymerase chain reaction) was carried out to detect which class of receptor is expressed in these cells. It was concluded by the PCR that both U937 and placenta cells express the class 1 G-CSF receptor. In addition, U937 cells express the soluble form of the class 2 G-CSF receptor, while the G-CSF receptor of class 3 containing the insertion in the cytoplasmic domain is significantly expressed in placenta cells.

The number of the gene coding for G-CSF receptor was examined by Southern hybridization and it was proved that there can be a single gene for G-CSF receptor per human haploid genom.

Since the present invention discloses the nucleotide sequence of DNA encoding human G-CSF receptor, the production of recombinant human G-CSF receptor can be easily accomplished by constructing an expression vector functional in an appropriate host systems using said DNA, transforming a microorganism with the resultant expression vector, and cultivating the transformant. Thus produced recombinant human G-CSF receptor can be used for many purposes such as the diagnosis of leukemia and the elucidation of mechanism of the action of human G-CSF and so on.

The nucleotide sequence of cDNA encoding murine G-CSF receptor is shown in FIGS. 1(a), 1(b) and 1(c), identified in the Sequence Listing as SEQ ID NO:1 ad SEQ ID NO:2, and that of cDNA encoding human G-CSF receptor is shown in FIGS. 8(a), 8(b) and 8(c) identified in the Sequence Listing as SEQ ID NO:3-SEQ ID NO:8. Persons ordinary skilled in the art will appreciates that it is easy to obtain derivatives having a similar activities by modifying said sequence using conventional methods, such as site specific mutation of DNA which comprises the insertion, substitution or deletion of nucleotide(s). Thus obtained DNA derivatives also fall within the scope of the present invention.

It is not always required to use the entire molecule of the mature G-CSF receptor polypeptide for the attainment of purposes of the present invention, but a fragment thereof can be preferably used in some cases subject to that said fragment retains the ability to bind to G-CSF. Similarly, a DNA fragment encoding such a poly peptide fragment of mature G-CSF receptor is useful, as well as the DNA fragment encoding the entire mature G-CSF receptor.

Thus, the present invention provides a receptor peptide capable of binding to G-CSF, and a DNA encoding said peptide.

As can be seen from the above, the DNA of the present invention which encodes G-CSF receptor peptide encompasses a DNA encoding mature G-CSF receptor and DNA fragments encoding peptide fragments having a binding activity to G-CSF.

It is possible to isolate a DNA encoding G-CSF receptor from cells of various animals using the DNA of the invention, construct an expression vector containing said DNA, transform said expression vector into an appropriate cultured cell, and make the resultant transformant produce G-CSF receptor.

The construction of expression vectors containing DNA encoding G-CSF receptor of the invention can be carried out by any of various known methods. Therefore, persons ordinary skilled in the art can select an appropriate method from them. Examples of suitable vectors for the expression of G-CSF receptor are those having a promotor which initiate the transcription, nearby upstream from the site where the DNA is inserted. Appropriate promoters are also known to persons ordinary skilled in the art and can be selected depending on the functional specificity in host cells. For instance, mouse metallothionine promotor and SV40 small T antigen promotor can be used for mouse and simian cells, respectively. Bacterial promoters are also useful for the G-CSF expression in bacterial cells. It is desired that a poly (A) signal exists downstream from the site of the insertion of G-CSF receptor sequence. It is also desired that vectors contain a selectable marker such as a drug-resistance. A particularly preferable marker is a neomycin resistant gene.

The construction of expression vectors can be conducted by inserting the DNA encoding G-CSF receptor in a suitable vector. Suitable vector can be selected from those known to persons ordinary skilled in the art by considering various factors such as promotor, poly(A) signal, selective marker and the like. Examples of vectors in which the cDNA of the invention is inserted to yield an expression vector which in turn is used to transform cultured cells for the expression of cDNA are pSV2, bovine papilloma virus DNA and the like.

Any cultured cells may be used for the expression of G-CSF receptor of the invention as long as they are self-replicable and are capable of expressing the DNAs presented in FIG. 1 or 8 and in the Sequence Listing. Examples of host cells include procaryotic microorganisms such as Escherichia coli, eucaryotic cells such as S. cerevisiae, and mammalian cells. Tissue cultured cell lines include cultured cells derived from birds, mammalian such as mouse, rat, ape and the like. Selection of suitable host-vector systems and their use are known to persons ordinary skilled in the art and any systems suitable for the expression of cDNA encoding G-CSF receptor can be selected among them.

The following Examples further illustrate and detail the invention disclosed, but should not be considered to limit the invention.

1) Cells

Murine myeloid leukemia NFS-60 cells (Weistein et al., Proc.Natl.Acad. Sci. USA, 83, 5010-5014 (1986); and provided by Dr. J.Ihle, St. Jude Children's Research Hospital) were grown in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS) and 10 to 20 units/ml of recombinant mouse IL-3.

COS-7 cells were routainly maintained in a Dulbecco's modified Eagle's medium (DMEM) containing 10% FCS.

2) Growth, Proliferation Factors such as Recombinant G-CSF

Human recombinant G-CSF was purified from medium conditioned with mouse C127I cells which were transformed with the bovine papilloma virus expression vector (Fukunaga et al., Proc. Natl. Acad. Sci. USA 81, 5086-5090 (1984)) carrying human G-CSF cDNA (Tsuchiya et al., 1987, ibid.). Mouse G-CSF was produced by using a similar expression system and purified as homogenous protein. Human recombinant G-CSF and M-CSF produced by chinese hamster ovary cells were provided by Chugai Pharmaceutical Co., LTD.

Human recombinant G-CSF produced by E.coli was purchased from Amersham.

Mouse recombinant IL-3 and GM-CSF were generous gifts from Drs. Miyajima and Arai, DNAX Institute.

Mouse recombinant IL-6 and mouse recombinant LIF were generously provided by Dr. Hirano, Osaka University, and N. Nicola, Walter Eliza Hall Institute, respectively.

Rat prolactin was purchased from Chemicon International Inc.

Mouse recombinant G-CSF was radioiodinated by the IODO-GEN method (Fraker and Speck, Biochem. Biophys. Res. Commun. 80, 849-857 (1978)) with a slight modification. Specific radioactivities ranged from 6 to 8.times.10.sup.4 cpm/ng protein (1,200-1,600 cpm/fmole).

3) CDM8 cDNA Library

Total RNA was prepared from exponentially growing NFS-60 cells by the guanidine isothiocyanate/CsCl method, and poly(A) RNA was selected by oligo (dT)-cellulose column chromatography. Double-stranded cDNA was synthesized as described [Nagata, S. et al., Nature 319: 415-418 (1986)] using a kit from Amersham except for the reverse transcriptase which was purchased from Seikagaku Kogyo Co.

To the resultant blunt-ended cDNA was added BstXI adaptor and electrophoresed on 1% agarose gel. From the gel were recovered cDNAs longer than 1.8 kb, which were then ligated to BstXI-digested CDM8 mammalian expression vector (Seed, 1987, ibid) and transformed into E.coli MC1061/p3 cells by electroporation [Dower et al, Nucl.Acids Res., 16, 6127-6145 (1988)].

4) Preparation of DNA

A total of 6.times.10.sup.4 bacterial colonies were plated on a 24-well microtiter plate at the density of 60-80 colonies per well. Glycerol cultures were prepared for each pool of colonies. LB broth was inoculated with aliquot from each glycerol culture, and plasmid DNAs were prepared by the boiling method (Maniatis et al., Molecular Cloning: A laboratory Manual, 1982) followed by phenol extraction and ethanol precipitation.

5) Transfection of COS-7 Cells

Monolayers of COS-7 cells were grown in 6-well microtiter plates, and transfection of plasmid DNA into COS-7 cells was carried out by a modified DEAE-dextran method (Sompayrac and Danna, Proc. Natl. Acad. Sci. USA, 78, 7575-7578 (1981)).

In brief, about 50% confluent cells were washed three times with serum-free DMEM, and incubated for 8 hr at 37.degree. C. with 0.6 ml of DMEM containing 50 mM Tris-HCl (pH 7.3), 0.3 mg/ml DEAE-dextran and 1 .mu.g of plasmid DNA. After glycerol shock with Tris-HCl-buffered saline containing 20% glycerol for 2 min at room temperature, cells were washed twice with DMEM, and incubated in DMEM containing 10% FCS.

6) Screening of COS-7 Cells (transformants) Expressing G-CSF Receptor

At 72 hr after the transfection, COS-7 cells were washed with DMEM containing 10% FCS and 20 mM HEPES (pH 7.3) (binding medium), and incubated at 37.degree. C. for 2 hr with 1.7.times.10.sup.5 cpm (200 pM) of .sup.125 I-G-CSF in 0.6 ml of the binding medium. Unbound radioiodinated G-CSF was removed, and cells were successively washed three times with phosphate-buffered saline (PBS) supplemented with 0.7 mM CaCl.sub.2 and 0.5 mM MgCl.sub.2 and once with PBS. Cells were then recovered by trypsinization, and the radioactivity bound to cells was counted using an AUTO-GAMMA 5000 MINAXI .gamma.-counter (Packard). Background binding of .sup.125 I-G-CSF to COS-7 cells transfected with the CDM8 vector was 308.+-.38 (SD) cpm. Two positive pools (I62, J17) were identified, which showed significant binding of radioiodinated G-CSF (500 and 912 cpm) to the transfected COS cells. From each positive pool (I62, J17), 144 independent clones were grown in 24-well microtiter plates (six plates), and subjected to sib selection (Maniatis et al., 1982, ibid) using a matrix of 12.times.12 clones. After a final round of mini-preparation of plasmid and transfection into COS-7 cells, a single clone was identified from each positive pool by the binding with .sup.125 I-G-CSF.

Thus, bacterial clones of pools I62 and J17 were arranged in 12 subgroups of 12 clones each, and assayed as above. Some subgroups gave positive responses, that is, binding reaction of 3,710 to 4,010 cpm of .sup.125 I-G-CSF to COS cells. By assaying single clone from each positive subgroup, two independent clones (pI62 and pJ17) were identified. When plasmid DNAs from pI62 and pJ17 were transfected into COS-7 cells, the binding assay gave values of 30,300 cpm and 31,600 cpm, respectively. When the plasmid DNAs from pI 62 and pJ17 were sequenced, it was found that the two cDNAs contained the complete coding sequence for the G-CSF receptor though, they contained neither the poly(A) tract nor the poly(A) additional signal. The cDNA library was, therefore, rescreened by colony hybridization using the 2.5 kb HindIII-XbaI fragment of pJ17 as a probe, and one positive clone (pF1) was isolated from positive ones. The pF1 clone had 603 bp of 3' non-coding region containing two overlapping poly(A) additional signals. The composite nucleotide sequence of the three cloned cDNAs (pI62, pJ17, and pF1) is presented in FIG. 1 identified in the Sequence Listing as SEQ ID NO:1, together with the predicted amino acid sequence identified in the Sequence Listing as SEQ ID NO:2. Schematic representation and restriction map of three independent cDNAs and hydropathy plot are shown in FIG. 2.

1) Binding activity of the Cloned G-CSF Receptor

The binding of .sup.125 I-G-CSF to COS or NFS-60 cells was examined. COS cells grown on 15 cm plates were transfected with 20 .mu.g of the pI62 or pJ17 plasmid. Cells were split into 6-well microtiter plates at 12 hr after the glycerol shock, and grown for 60 hr in DMEM containing 10% FCS. Cells were washed with binding medium, and incubated at 4.degree. C. for 4 hr with .sup.125 I-G-CSF (10 pM to 1.2 nM range). To determine the non-specific binding of .sup.125 I-G-CSF to cells, a large excess of unlabeled G-CSF (800 nM) was incubated in the assay mixture, and the radioactivity bound to the cells was subtracted from the total binding to yield the specific binding.

For binding of G-CSF to NFS-60 cells, 5.2.times.10.sup.6 cells were incubated at 4.degree. C. for 4 hr with various concentrations of .sup.125 I-G-CSF in 0.3 ml of RPMI-1640 medium containing 10% FCS and 20 mM HEPES (pH 7.3). Results are shown in FIG. 3. As mentioned above, 1.times.10.sup.6 COS cells transfected with the plasmid pJ17 were incubated with various amounts of .sup.125 I-G-CSF with or without an excess of unlabeled G-CSF. In the FIG. 3A, the specific binding (.oval-solid.) is shown as the difference between total (.oval-hollow.) and non-specific binding (.DELTA.). FIG. 3B is the scatchard plot of G-CSF binding data in COS cells, and FIG. 3C is the saturated binding of .sup.125 I-G-CSF to NFS-60 cells, showing total (.oval-hollow.), non-specific (.DELTA.) and specific (.oval-solid.) binding to cells. The specific binding is the difference between the total and non-specific binding. FIG. 3D is the scatchard plot of G-CSF binding data on NFS-60 cells. Said Figure shows that the G-CSF receptor expressed on COS cells contains a single species of binding site with an equilibrium dissociation constant of 290 pM and 3.0.times.10.sup.4 receptors per cell. If the transfection efficiency of COS cells was assumed to be 10 to 20% (Sympayrac and Danna, Proc. Natl. Acad. Sci. USA, 78, 7575-7578 (1981)), the positively transfected COS cells probably expressed the recombinant G-CSF receptor at 1.5-3.0.times.10.sup.5 molecules per cell. Since the native G-CSF receptor on NFS-60 cells has an equilibrium dissociation constant of 180 pM (FIG. 3D), these results suggest that the cDNA coded by the plasmid pJ17 is sufficient to express the high affinity receptor for murine G-CSF.

The binding specificity of recombinant G-CSF receptor expressed by COS cells to G-CSF was then examined. As mentioned above, COS cells transfected with the cDNA for the G-CSF receptor (pJ17) were incubated with 2 ng of .sup.125 I-mouse-G-CSF in the absence or presence of 1 .mu.g of unlabeled murine G-CSF, human G-CSF, murine GM-CSF, human M-CSF, murine IL-6, murine leukemia inhibitory factor (LIF) or rat prolactin. As human G-CSF, human recombinant G-CSFs produced in mouse C127 cells, Chinese hamster ovary cells or E. coli were used. Results are shown in FIG. 4. The radioactivities of .sup.125 I-G-CSF bound to COS cells in each experiment are expressed as a percentage of that obtained without competitor.

Human G-CSF competes with mouse G-CSF for binding to mouse WEHI 3B D.sup.+ cells (Nicola et al., 1885, ibid). Accordingly, unlabeled recombinant human G-CSFs produced either by mammalian cells or E.coli could compete well with labeled mouse G-CSF for binding to COS cells transfected with the plasmid pJ17 (FIG. 4). On the contrary, no inhibition of binding of .sup.125 I-G-CSF to COS-7 cells was observed in the presence of unlabeled recombinant murine GM-CSF, murine IL-3, murine IL-6, murine LIF, rat prolactin or human M-CSF.

2) Cross-Linking Reaction

The chemical cross-linking reaction of .sup.125 I-G-CSF to the receptor expressed in COS cells was performed as follows.

As mentioned above, 8.times.10.sup.5 of COS cells (on 3.5 cm plate) tranfected with the plasmid pI62 were incubated at 4.degree. C. for 2.5 hr with 1.2 nM of the radioiodinated G-CSF in the presence or absence of 1.5 .mu.M of unlabeled G-CSF in 0.6 ml of the binding medium. The cells were scraped from the plate using a cell lifter and washed three times with 1 ml of PBS. Cross-linking reaction was carried out on ice for 20 min in 1 ml of PBS containing 150 .mu.M disuccinimidyl suberate (DSS) and 150 .mu.M disuccunimidyl tartrate (DST). The reaction was terminated by the addition of 50 .mu.l of 1M Tris-HCl (pH 7.4) and cells were collected by centrifugation and were lysed with 15 .mu.l of 1% Triton X-100 containing a mixture of protease inhibitors (2 mM EDTA, 2 mM (p-aminophenyl)methanesulfonylfluoride hydrochloride, 2 mM O-phenanthroline, 0.1 mM leupeptin, 1 .mu.g/ml pepstatin A and 100 units/ml aprotinin). After centrifugation, the clear lysate (10 .mu.l) was analyzed by electrophoresis on a 4-20% gradient polyacrylamide gel in the presence of SDS (Laemmli, Nature 227, 680-685, 1970) and exposed to X-ray film at -80.degree. C. for 2 days with intensifying screens. In the electrophoresis, .sup.14 C-labelled molecular weight standard (Rainbowmarker, Amarsham) was applied as size markers in parallel. Results are shown in FIG. 5. In the figure, lane 2 shows the result obtained by cross-linking reaction in the presence of an excess of unlabeled murine G-CSF, and lanes 3 and 4 show the results obtained in the absence of the same (in the reaction for lane 3, DSS and DST were also omitted). Mouse NFS-60 cells were similarly incubated with .sup.125 I-G-CSF with (lane 5) or without (lane 6) an excess of unlabeled G-CSF which both were cross-linked with DSS and DST. As size markers, .sup.14 C-labeled molecular weight standards (rainbow marker, Amersham) were electrophoresed in parallel (lanes, 1 and 7), and sizes of standard proteins are shown in kd.

The cross-linking reaction of the G-CSF receptor on NFS-60 cells with labeled mouse G-CSF (M.W. 25,000) yielded a band which has an apparent M.W. of 125,000-155,000 (lane 6), indicating that the M.W. of the murine G-CSF receptor on NFS-60 cells is 100,000-130,000. Similarly, cross-linking reaction of .sup.125 I-mouse G-CSF to the receptor expressed on COS cells gave a major band of M.W. 120,000-150,000 (lane 4), which is slightly smaller than that detected on NFS-60 cells. These bands were not observed when the cross-linking reaction was carried out in the presence of unlabeled G-CSF (lanes, 2 and 5) or when the cross-linking agents were omitted (lane 3). The slightly different M.W. observed in COS cells may be explained by the differential glycosylation of receptor in these cell lines.

3) Hybridization

Colony hybridization and Northern hybridization were carried out as described (Maniatis et al., Cold Spring Harbor, N.Y.; Cold Spring Harbor Laboratory (1982)). As a probe, the 2.5 kb HindIII-XbaI fragment of clone pJ17 was labeled with .sup.32 P by the random primer labeling method (Feinberg and Vogelstein, Anal.Biochem. 132, 6-13 (1983)). Results of Northern hybridization are shown in FIG. 6.

Total RNA or poly(A) RNA was prepared from mouse cell lines; L929 (lane 1), NFS-60 (lanes, 2 and 3), FDC-P1 (lane 4), and WEHI-3B D.sup.+ (lane 5) or mouse tissues; brain (lane 6), lung (lane 7), spleen (lane 8), bone marrow (lane 9), liver (lane 10), and kidney (lane 11). Total RNA (30 .mu.g; lanes 1, and 3-11) or 2 .mu.g of poly(A) RNA (lane 2) was electrophoresed on a 1.3% agarose gel containing 6.6% formaldehyde, and analyzed by Northern hybridization as described in the above literature.

1) Determination of Nucleotide Sequence

DNA sequencing was performed by the dideoxynucleotide chain termination method using T7-DNA polymerase (Pharmacia) and .alpha.-.sup.35 S dATP.alpha.S (Amersham). Results are shown in FIGS. 1 and 2. FIG. 1 represents the nucleotide sequence of murine G-CSF receptor cDNAs (pI62, pJ17 and pF1) and deduced amino acid sequence. Numbers above and below each line refer to nucleotide position and amino acid position, respectively. Amino acids are numbered starting at Cys-1 of the mature G-CSF receptor. On the amino acid sequence, the signal sequence and the transmembrane domain are underlined. Two overlapping poly(A) additional signals (AATAAA) are also underlined. Potential N-glycosylation sites (Asn-X-Ser/Thr) (11 in the extracellular domain and 2 in the cytoplasmic domain) are boxed.

In the FIG. 2 for murine G-CSF receptor cDNA, A gives the schematic representation and restriction map of three independent cDNAs (pI62, pJ17 and pF1) for murine G-CSF receptor. The box represents the open reading frame. The dotted and filled regions indicate the signal sequence and the transmembrane region, respectively. The cleavage sites for restriction enzymes are shown. B is the hydropathy plot of the amino acid sequence of murine G-CSF receptor. The hydropathy plot was obtained by the method of Kyte and Doolite (1982) using a window of 10 residues. The numbers under the figure indicate positions of the amino acid residues of the precursor protein.

2) Comparison of the Amino Acid Sequence of the G-CSF Receptor with that of Other Growth Factor Receptors (FIG. 7)

FIG. 7(a): alignment of the G-CSF receptor with prolactin and growth hormone receptors. The amino acid sequence from 96 to 317 of murine G-CSF receptor is aligned with rat prolactin and human growth hormone receptors to give maximum homology by introducing several gaps (-). Identical residues in two sequences are enclosed by solid lines, and residues regarded as favored substitutions are enclosed by dotted lines. Favored amino acid substitutions are defined as pairs of residues belonging to one of the following groups: S,T,P,A and G; N,D,E and Q; H,R and K; M,I,L and V; F,Y and W. Amino acids conserved in 9 members of the growth factor receptor family (G-CSF, prolactin, growth hormone, erythropoietin, GM-CSF, IL-2.beta., IL-3, IL-4, and IL-6) are shown under each line with or without brackets. The residues without brackets are conserved in more than 8 members, while the residues with brackets are conserved in 5-7 members in the family.

FIG. 7(b): alignment of the murine G-CSF receptor with contactin. The amino acid sequence from 376 to 601 of the mouse G-CSF receptor is aligned with the amino acid sequence of chicken contactin as described in FIG. 7(a).

FIG. 7(c): alignment of the G-CSF receptor with the IL-4 receptor. The amino acid sequence from 602 to 808 of the mouse G-CSF receptor is aligned with two corresponding regions of mouse IL-4 receptor as above.

FIG. 7(d): schematic representation of the mouse G-CSF receptor. The box indicates the mature G-CSF receptor. "TM" represents the transmembrane domain. Region "A" indicates a domain (222 amino acids) which has similarity to other growth factor receptors including prolactin and growth hormone receptors, and contains the "WSXWS" motif. Region "B" (226 amino acids) of the mouse G-CSF receptor shows similarity to chicken contactin. Region "C" (211 amino acids) includes the transmembrane domain (underlined) and the cytoplasmic domain of the G-CSF receptor, and is similar to two regions of the mouse IL-4 receptor.

(1) Isolation of Human G-CSF Receptor cDNA Clones

The preparation of poly(A) RNA from U937 cell and the synthesis of double-stranded cDNA were carried out according to a method described in a literature [Nagata et al., Nature 319: 415-418 (1986)] using a cDNA synthesis kit from Amersham except for the reverse transcriptase which was purchased from Seikagaku Kogyo. To the resultant blunt-ended cDNA was added BstXI adaptor and electrophoresed on 1% agarose gel. From the gel were recovered cDNAs longer than 2.5 kb, which were then ligated to mammalian expression vector pEF-BOS (FIG. 14) and transformed into E.coli DH1 cells by electroporation [Dower et al, Nucl.Acids Res., 16: 6127-6145 (1988)].

A total of 3.4.times.10.sup.4 clones of the library was screened by colony hybridization. As a hybridization probe, 2.5 kb HindIII - XbaI DNA fragment of murine G-CSF receptor cDNA was labeled with .sup.32 P using the random oligonucleotide primer labeling method [Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular Cloning: A laboratory Manual, 2nd edition (Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory)]. The murine G-CSF receptor cDNA (pI62) was used for this purpose whose nucleotide sequence and a deduced amino acid sequence are shown in FIG. 1 identified in the Sequence Listing as SEQ ID NO:1 and SEQ ID NO:2. In the FIG. 1, the signal sequence and transmembrane domain are underlined and potential N-glycosylation sites are boxed.

The hybridization was carried out as described [Fukunaga, R., Matsuyama, M. Okamura, H., Nagata, K, Nagata, S. & Sokawa, Y. Nucl. Acids Res. 14: 4421-4436 (1986)] except that the hybridization temperature was lowered to 28.degree. C. and the filter was washed at 37.degree. C. in 150 mM NaCl/15 mM sodium citrate, pH 7.0/0.1% NaDodSO.sub.4. Thus, a replica filter of colony was prepared and each nitrocellulose filter was subjected to hybridization at 28.degree. C. for 12 hr with a probe which had been prepared by heating at 95.degree. C. for 5 min and cooling promptly. The filter was washed as described (ibid) and screened for the presence of desired clones by autoradiography.

A human placental cDNA library prepared in .lambda.gt11 (Clonetech) was screened by plaque-hybridization using murine G-CSF receptor cDNA as a probe as described above.

Thus, about 1.5.times.10.sup.6 clones of phage DNA were transferred onto nitrocellulose filter as described [Benton, W. D. & Davis, R. W., Science, 196: 180-182 (1977)], which was followed by the screening by plaque hybridization. The screening was conducted using the same probe DNA and under the same conditions as those used in the screening of U937 cDNA library as mentioned above.

Five positive clones (pHQ1-pHQ5) were identified and isolated from U937 cDNA library.

The plaque hybridization between human placenta cDNA library (1.5.times.10.sup.6 clones) and murine G-CSF receptor cDNA gave more than 100 clones having positive signal. EcoRI DNA fragments of six positive clones (.lambda.HG4, 5, 11, 12, 14 and 18) were subcloned in pBluescript SK(+) vector to give plasmids pHG4, 5, 11, 12, 14 and 18.

For the DNA sequencing analysis, a series of deletion plasmids, each plasmid containing about 300 bp deletion, was generated using Exonuclease III and mung bean nuclease (Sambrook et al, ibid). The sequencing reaction was performed by the dideoxy chain termination method using T7-DNA polymerase, deaza dGTP and .alpha.-.sup.35 SdATP.alpha.-S.

The DNA sequencing analysis of the isolated cDNA clones from U937 and placental cDNA libraries has revealed that they can be divided into the following three groups.

Class 1: plasmids pHQ3 and pHG 12 (U937 and placenta cells, respectively).

Most of cDNA clones isolated from U937 and placenta cDNA libraries belong to this class. The nucleotide sequence and deduced amino acid sequence of these plasmids is shown in FIG. 8 (a), and in the Sequence Listing as SEQ ID NO:3 and SEQ ID NO:4, and the restriction map of them is given in FIG. 9.

Plasmid of this class 1 contain a large open reading frame that encodes a protein consisting of 836 amino acids. The hydropathy analysis of the predicted amino acid sequence has indicated that the N-terminal 23 amino acid residues correspond to the signal sequence, and following 604, 26 and 183 residues constitute the extracellular, transmembrane and cytoplasmic domains, respectively, as can be seen from the Figure. There are nine potential N-glycosylation sites on the extracellular domain of the receptor.

In the extracellular domain of the peptide encoded by this cDNA, there are 17 cysteine residues and 14 of which are conserved between human and mouse receptors. Furthermore, the "WSXWS" motif conserved in members of a cytokine receptor family can be found in the extracellular domain, which indicates that human G-CSF receptor is one of such members.

The overall homology of human G-CSF receptor (813 amino acids) to the murine G-CSF receptor is 72% at the nucleotide sequence level and 62.5% at the amino acid sequence level. The amino acid sequence homology is relatively constant over the entire region of the polypeptide.

Class 2: plasmid pHQ2 (U937 cell)

The nucleotide sequence of plasmid pHQ2 is identical to that of pHQ3 except that it lacks 88 nucleotide from 2,034 to 2,121. The ends of the deletion in pHQ 2 is shown by filled arrowheads in FIG. 8 (a). As can be seen from the figure, the deleted region includes the transmembrane domain. The nucleotide sequence which occurs following the 88 bp deletion is given in FIG. 8 (b).

The plasmid pHQ2, in which the deletion results in altered translation reading frame that encodes the additional 150 amino acids after the deletion point (FIG. 9), seems to encode a secreted and soluble form of G-CSF receptor. The soluble form of G-CSF receptor consists of 748 amino acids with a calculated M.W. of 82,707.

Class 3: plasmid pHG11 and pHG5 (placenta)

These plasmids have a 81 bp insertion at nucleotide number 2,210 of plasmid pHQ3. The insertion is in the cytoplasmic domain of the G-CSF receptor. The site of insertion is shown by an arrow of thick line in FIG. 8 (a). The translational open reading frame is not changed. The nucleotide sequence and deduced amino acid sequence of the insertion is given in FIG. 8 (c), identified in the Sequence Listing as SEQ ID NO:7 and SEQ ID NO:8. The putative polypeptide coded by this class of cDNA, therefore, is 27 amino acids (M.W. 2,957) larger than that coded by the class I G-CSF receptor.

The nucleotide sequence and deduced amino acid sequence of plasmids of the above 3 classes are shown in FIG. 8, identified in the Sequence Listing as SEQ ID NO:3-SEQ ID NO:8. In FIG. 8A, numbering of the amino acid sequence of pHQ2 starts at Glu-1 of the putative mature G-CSF receptor. The signal sequence and the transmembrane domain are underlined by thick lines and potential N-glycosylation sites (Asn-X-Thr/Ser) are boxed. The "WSXWS" motif conserved in members of a cytokine receptor family is doublely underlined. Filled arrowheads mark the ends of the deletion in pHQ2, while the thick arrows indicate the site of the insertion in pHG11. The thin arrows indicate the oligonucleotide primers used for PCR which will be hereinafter described. On B in FIG. 8(c), identified in the Sequence Listing as SEQ ID NO:5 and SEQ ID NO:6, the sequence following the open arrowhead is the nucleotide sequence in pHQ2 and a deduced amino acid sequence, said nucleotide sequence downstream from nucleotide 2,034 being deleted in pHQ3. C in FIG. 8(c), identified in the Sequence Listing as SEQ ID NO:7 and SEQ ID NO:8 shows the nucleotide sequence and deduced amino acid sequence of the insertion present in pHG11, said insertion occurs following amino acid 657 of pHQ3. The inserted sequence is bracketed.

Schematic restriction map of the plasmids of the above 3 classes are shown in FIG. 9. In the figure, boxes represent open reading frames. The shadowed and filled regions represent the signal sequence and transmembrane domain, respectively. The slashed region in pHQ2 indicates that the amino acid sequence in this region differs from those in other cDNAs as a result of an altered open reading frame. The slashed region in pHG11 shows the 27 amino acids encoded by the inserted sequence.

(2) Detection of G-CSF Receptor mRNA in Human Cells

PCR (polymerase chain reaction) was carried out to detect G-CSF receptor mRNA using primers prepared from the above cDNAs.

The synthesis of single stranded cDNA and PCR was carried out essentially according to the method described by Kawasaki [Kawsaki, E. S. In PCR Protocols, "A guide to methods and application" eds. Innis, M. A., Gelfand, D. H., Shinskly, J. J. & White, T. J. (Academic Press, San Diego, Calif.), pp. 21-27 (1990)]. The results are shown in FIG. 12.

Total RNA (lanes 2 and 5) or poly(A) RNA (lanes 1 and 4) from human U937 cells, or total RNA from human placenta (lanes 3 and 6) was amplified by PCR. Thus, 2 .mu.g of total or poly(A) RNA was subjected to cDNA synthesis in a 50 .mu.l of reaction mixture with 0.5 .mu.g of random hexamer and 80 units of AMV reverse transcriptase as described (Kawasaki, E. S. et al, ibid).

An aliquot (5 .mu.l) of the reaction mixture was diluted with 100 .mu.l of PCR buffer containing 50 pmol each of forward and reverse primer, and placed on a DNA thermal cycler (Perkin-Elmer-Cetus) which was preheated at 80.degree. C. The reaction was started by adding 2.5 units of Taq polymerase, and the conditions for the PCR were: 1.5 min at 95.degree. C.; 1.5 min at 70.degree. C.; 1.5 min at 72.degree. C. for 30 cycles.

When the reaction completes, the products (10% of the reaction mixture) were analyzed by electrophoresis on 1.5% agarose gel in TBE buffer and visualized by ethidium bromide fluorescence. As size markers, BamHI and MvaI-digested pBR322 was electrophoresed in parallel, and sizes of DNA fragments (A1, A2, B1 and B2) corresponding to the isolated cDNAs are indicated by arrows.

In FIG. 12, samples in lanes 1-3 were amplified using the first set of forward and reverse primers from nucleotide 1,790 to 1,810 and 2,179 to 2,156, respectively, while samples in lanes 4-6 were amplified using the second set of primers from nucleotide 2,086 to 2,105 and 2,322 to 2,303. The FIG. 12 demonstrates that both U937 and placental cells express the class 1 G-CSF receptor. In addition, U937 cells express soluble G-CSF receptor, and the placental cells express the G-CSF receptor having the insertion in the cytoplasmic domain.

The binding assay was conducted using COS cells transformed as described in Example 4 and murine .sup.125 I-G-CSF. Labeling of recombinant murine G-CSF, transfection of COS cells with an expression vector containing human G-CSF receptor cDNA and the binding assay using the COS cells and .sup.125 I-G-CSF were carried out in accordance with the procedures described in the aforementioned Examples for murine G-CSF receptor.

A full length cDNA having an insertion in cytoplasmic region was constructed using plasmid pHG11. Thus, plasmid pHG11 was digested thoroughly with restriction enzyme NheI (Takara Shuzo) and then partially digested with BstXI (1,425) (Takara Shuzo). The enzymatic reaction was carried out under the conditions indicated by the manufacture.

An expression plasmid pQW11 was then constructed by ligating 1.38 kb BstXI-NheI fragment and 6.9 kb BstXI-NheI fragment of pHQ3 in the presence of T4 DNA ligase (Takara Shuzo). COS cells were transfected by either of expression plasmids pHQ2, pHQ3 or pQw11 and the binding activity of transfectants to .sup.125 I-G-CSF was analyzed.

COS cells grown on 15 cm plate were transfected with 20 .mu.g of pHQ2, pHQ3 or pQW11. Cells were divided into 6-well microtiterplate at twelve hours after the glycerin shock and incubated in 10% FCS-containing DMEM for 60 hr. Cells were washed in binding medium (DMEM containing 10% FCS and 20 mM HEPES (pH 7.3)) and incubated with different amounts of .sup.125 I-G-CSF (10 pM to 1.2 nM) at 4.degree. C. for 4 hr. For the determination of non-specific binding between .sup.125 I-G-CSF and cells, the binding reaction was carried out in the presence of a large excess of non-labeled G-CSF (800 nM). The specific binding of .sup.125 I-G-CSF was determined after subtracting the radioactivity bound non-specifically from the total binding activity. FIG. 10A shows the saturation binding of .sup.125 I-G-CSF to COS cells while FIG. 10B shows the scatchard plot of G-CSF binding data. The binding of G-CSF to COS cells transfected with murine G-CSF receptor cDNA was also examined.

In FIG. 10, closed triangle refers to COS cells transfected with murine G-CSF receptor cDNA. Among COS cells transfected with human G-CSF cDNA, cells transfected with pHQ3, pHQ2 and pQW11 are shown by open circle, closed circle and open triangle, respectively.

As shown in FIG. 10, COS cells transfected with pHQ3 or pHQ11 have a strong affinity for murine .sup.125 I-G-CSF. The dissociation constant of the specific binding is 550 pM and the number of receptor per cell is 3.4.times.10.sup.4. On the other hand, COS cells transfected with pHQ2 cDNA, which encodes a polypeptide having a deletion, can bind to .sup.125 I-G-CSF in very low level (dissociation constant: 440 pM; binding sites: 6.times.10.sup.3 /cell). This strongly indicates that the receptor coded by PHQ2, which lacks the transmembrane domain, is probably secreted from cell and accumulated in medium.

Further, COS cells transfected with plasmid pHQ11 has the similar binding activity as that of COS cells transfected with plasmid pHQ3, indicating that 27 amino acid insertion in the cytoplasmic domain has little effect on the binding of G-CSF to the receptor.

The purification of human G-CSF receptor expressed by transformants can be conducted in the similar manner as that described for the purification of native murine G-CSF receptor.

DNA or RNA encoding human G-CSF receptor was analyzed by Northern or Southern Hybridization.

Total RNA was prepared from various cell lines and fresh human full-term placenta by guanidine isothiocyanate/CsCl method as described, and cellular DNA was prepared from human T lymphocytes as described previously (Fukunaga et al, Nucl.Acids Res.,14: 4421-4436 (1986)). Hybridization for Southern and Northern blots were carried out using a 3 kb XhoI DNA fragment of pHQ3 containing human G-CSF receptor cDNA in accordance with the process described in a literature (Maniatis, T. et al, Molecular Cloning, Cold Spring Harbor Laboratory (1982)).

1) Analysis of G-CSF Receptor Transcripts and Genomic DNA

Northern hybridization was carried out with mRNAs from various cells using murine G-CSF receptor cDNA as a probe. Results are shown in FIG. 11. Cells used are as follows.

Human U937 (lanes 1 and 2), human KG-1 (lane 3), human HL-60 (lane 4), human FL (lane 6), human CHU-2 (lane 7), human placenta (lanes 5 and 8).

U937: human histiocytic lymphoma, ATCC CRL 1593

KG-1: human acute myelogenous leukemia, ATCC CCL 246

HL-60: human promyelocyte leukemia, ATCC CCL 240

FL: human amnion, ATCC CCL 62

Total RNA (20 .mu.g) (lanes 2 to 6) or poly(A) RNA (1 .mu.g) (Lanes 1 and 7) were used.

In the analysis shown in FIG. 11A, human G-CSF receptor cDNA was used as DNA probes, and the filter was exposed to X-ray film for 40 hr except that the lane 8 was exposed for 2 hr.

A single band of 3.7 kb is observed in RNAs from U937, placenta and KG-1 cells. The signal detected with RNAs from placenta is 20 times or over stronger than that detected with RNAs from U937 cells.

In the analysis shown in panel B, blot was rehybridized with .sup.32 P-labeled human elongation factor 1 .alpha. cDNA (Uetsuki, T., Naito, A., Nagata, S. & Kaziro, Y. J. Biol. Chem., 264: 5791-5798 (1990)) and the filter was exposed to X-ray film for 1 hr. In this case, RNAs from various cells give almost the similar signals. These results indicate that the placenta cells express the G-CSF receptor mRNA abundantly, and suggest that G-CSF may have an ability to stimulate the growth and maturation of placenta.

2) Southern Hybridization

The number of the gene coding G-CSF receptor was examined by Southern hybridization.

Ten microgram of human genomic DNA was digested with EcoRI, HindIII, BamHI, BglII, XbaI, PstI, SacI or ApaI, and electrophoresed on a 0.8% agarose gel. DNA was transferred to a nitrocellulose filter, and hybridized using .sup.32 P labeled human G-CSF receptor cDNA. The DNA size marker was electrophoresed in parallel and is given in kilobases. In FIG. 13, fragments of human genomic DNA generated by the digestion with EcoRI (lane 1), HindIII (lane 2), BamHI (lane 3), BglII (lane 4), XbaI (lane 5), PstI (lane 6), SacI (lane 7), and ApaI (lane 8) were analyzed. One or two bands are observed in EcoRI, HindIII, BglII and XbaI-digested DNA, while digested products of DNA with BamHI, PstI, SacI and ApaI yield 4 to 5 bands, respectively. As can be seen from FIG. 9, human G-CSF receptor cDNA contains 3 BamHI, 6 PstI, 2 SacI and 3 ApaI sites. Accordingly, these results suggest that there is a single gene for G-CSF receptor per human haploid genome.