NO180544B - Process for Preparation of Human IL-3 (Pro-8), Expression Vector, Vector and Transformed Living Host Cell - Google Patents

Abstract

A cDNA encoding human interleukin-3, also designated herein hIL-3 or hmulti-CSF, has been disposed in expression systems and caused to produce hIL-3 substantially free of impurities normally accompanying this protein as it is produced $i(inter alia) by peripheral blood lymphocytes in nature. The resulting hIL-3 can be produced in practical amounts and is useful in a variety of therapeutic and diagnostic protocols.

Description

The present invention relates to a method for the production of human IL-3 (Pro-8) in a host cell, expression vector, vector and transformed living host cell.

Hemopoiesis involves the active process of proliferation and differentiation of pluripotent precursor cells into all types of mature blood cells and some specialized tissue cells. Production of functional blood cells is regulated by specific proteins, hemopoietic growth factors (HGF). Some HGFs control the maturation of a specific maturing cell lineage, while others stimulate the proliferation and differentiation of progenitors in multiple pathways. Much of our knowledge regarding haemopoietic differentiation processes has been provided from in vitro and in vivo mouse studies, using purified growth factors. The murine growth factor interleukin-3 (mIL-3), which is also called Multi-CSF, mast cell growth factor, stem cell activation factor or several other factors, stimulates the proliferation of early developed, multipotent cells (CFU-S) which are detected by the spleen colony assay, resulting in the production of precursor cells in the erythroid, megakaryocyte, granulocyte/macrophage, osteoblast and several other cell lines. mIL-3 has been implicated in the replication of pluripotent stem cells, particularly in synergism with other hFGFs.

In recent years, several groups have cloned mIL-3 cDNA. No results have so far been reported regarding the identification of homologous sequences in human DNA using mIL-3 DNA as a probe. The human gene has presumably diverged considerably from the mIL-3 gene or it has lost its function during primate evolution. Human white blood cells have been shown to produce an HGF(s) that can replace mIL-3 in supporting the proliferation of murine CFU-S. It has thus been postulated that there exists a human HGF which has similar biological properties to mIL-3 and which may therefore be the human homologue. Yang, Y-C, et al, Cell

(1986) 47:3-10, dated October 10 discloses cDNA encoding a protein having IL-3-like activity from gibbon T cells, and recovery of a genomic DNA encoding the human counterpart. The sequence of a cDNA encoding human IL-3 can be deduced from the human gene sequence published by Yang et al. Said article neither describes nor shows a method for isolating a cDNA that codes for human IL-3, production of hIL-3 was not achieved either. The invention mainly describes the isolation of a cDNA which includes the entire coding sequence of human IL-3.

Human IL-3 protein has never been produced in purified form, its characteristics, other than activity in certain in vitro proliferation assays and deduced primary structure, have been described.

The present invention therefore relates to a method for the production of human IL-3 (Pro-8) in a host cell, characterized in that it includes: introducing into said host cell a DNA expression vector containing an expression cassette which, in the direction of transcription, consists of a transcription initiation regulatory region that is functional in said host cell; a DNA sequence encoding human IL-3; (Pro-8) (Fig. 1); and a transcription termination regulatory region that is functional in said host cell;

culturing said host cell containing said DNA construct in a nutrient medium under suitable culture conditions, whereupon human IL-3 is produced; and

recovery of the human IL-3 product.

The expression vector according to the present invention is characterized by the fact that it contains and can express a DNA sequence that codes for human IL-3 with 1 Pro in position 8 of the mature protein molecule (Fig. 1).

The present invention also relates to vectors characterized by the fact that they are selected from the group consisting of pGBL/IL-300 to pGB/IL-319, and of pLB4 and pLB4/BPV.

A transformed living host cell has also been described, characterized by the fact that it contains and can express recombinant DNA material that codes for human IL-3 with 1 Pro in position 8 of the mature protein molecule (Fig. 1).

As indicated above, the present invention describes the isolation of a cDNA containing the entire coding sequence for human IL-3. Low degree of homology between the DNA sequences encoding murine and human IL-3 does not allow the provision of a cDNA for hIL-3 by hybridization with the mIL-3 coding sequence. The hIL-3 cDNA clone could unexpectedly be isolated by exploiting the high degree of homology in the 3' non-coding part of the cDNAs. The availability of the cDNA clone allows the production of hIL-3 from many host organisms. At the same time as large-scale production, the protein can be purified and used therapeutically.

The present invention allows the production of recombinant human IL-3 protein in many host cells by transcription and translation from a cDNA sequence that codes for the human IL-3 protein. Production of the protein is illustrated in several hosts, including E. coli, COS cells, C127 cells, B. subtilis and B. licheniformis, S. cerevisiae and K. lactis, herein below. Production in other hosts using appropriate expression systems is also possible by providing intronless cDNA. The availability of anti-human IL-3 antibodies that allow the identification of colonies exhibiting successful production of the recombinant protein aids in the production of human IL-3 from any recombinant system.

Figure 1 shows a comparison of DNA and protein sequences from human multi-CSF and mouse IL-3. hmulti-CSF protein and DNA sequence (clone Dll, top line) was listed along with mIL-3 DNA (11, 35) and protein sequence (30). Identical nucleotides are indicated by a vertical line, and identical amino acids are shown in boxes. Black dots indicate a polyadenylation signal sequence and horizontal bars mark ATTTA repeat units.

Figure 2 shows construction of pLB4 containing human IL-3 cDNA. E = EcoRI, Sm = SmaI, B = BamHI, S = SstI, K = KpnI. Figure 3 shows the biological activity of C0S/pLB4 CM on human bone marrow progenitors. The mean numbers of erythroid (BFU-E), granulocyte-macrophage (CFU-GM), granulocyte (CFU-G), eosinophil (CFU-Eo), macrophage (CFU-M) and mixed (CFU-MIX) colonies (± SD) are shown for duplicate cultures stimulated with measured volumes of COS/pLB4 CM. Figure 4 shows induction of AML proliferation by C0S/pLB4 CM as seen in a colony culture assay (panel A) and in a DNA synthesis (3H-TdR incorporation) assay (panel B). Figure 5 shows a construction diagram of the E. coli expression vectors pGB/IL-301, GB/IL-302, pGB/IL-303, pGB/IL-304, pGB/IL-305 and pGB/IL-306. In this figure, X stands for Xhol, E for EcoRI, B for BamHI and A for Aval site. Figure 6 shows the sequence of the multicloning site in pTZ18R (Pharmacia) and its derivative pT1. Figure 7 shows a schematic presentation of hmulti-CSF expression clones. For the eukaryotic expression plasmids pLB4 and pLH1, only the hmulti-CSF cDNA insert is shown. Leader peptide ({/////// I) and mature hmulit-CSF protein (fll^HB ) coding regions are indicated in boxes. Bacterial expression clones of hmulti-CSF (derived from pLH1) contain lacZ and the multi-linker protein coding region ( w /////// A ). the 5'-terminal non-coding region of hmulti-CSF (I')) and the hmulti-CSF coding region. The arrows mark the ATG start codon used in the particular vector. Figure 8 shows the sequences of the fusion regions in lacZ/hmulti-CSF DNA for different bacterial expression vectors. The sequence of clones is given from the beginning of lacZ protein in either pUC8 or pTZ18R (lower case) and of the hmulti-CSF DNA sequence up to the ClaI site at position 158. Mutations in the hmulti-CSF DNA sequence are underlined, which result in trp1<3> -+ arg1<3> (pGB/IL-302); leu<9> -* pro9 and trp<3> -» arg<13 >(pGB/IL-303); met<3> -> thr<3> and a change that is not expressed (pGB/IL-304). The numbers written higher up denote the amino acid residue number of the mature (final) protein. Figure 9 shows polyacrylamide gel electrophoresis of bacterial hmulti-CSF produced from bacteria containing pGB/IL-301 and pGB/IL-302. Figure 10 shows titration of hmulti-CSF fusion protein on AML blast cells. Figure 11 shows a Western blot demonstrating the IL-3 specific reaction of rabbit antisera raised against the 21 kd protein isolated from a lysate of E. coli transformed with pGB/IL-301. Figure 12 shows the effect of antisera from Figure 11 on IL-3 activity. Figure 13 shows a schematic representation of plasmid pGB/IL-307. The human IL-3 coding sequence is indicated therein. The N-terminal amino acids of the fusion protein are listed below the drawing. Figure 14 shows a schematic representation of plasmid pGB/IL-308. The nucleotide sequence of the promoter region is written below the drawing. Figure 15 shows construction of plasmid pGB/IL-309. A part of the human IL-3 sequence, i.e. signal sequence plus 20 amino acids of the mature protein is indicated therein. Portions of the 3' non-coding region of the IL-3 cDNA sequence are also shown. Figure 16 is a schematic representation of plasmid pGB/IL-310.

Figure 17 shows the nucleotide sequence of plasmid pBHAl.

Figure 18 shows the construction of the plasmids pGB/IL-311 and pGB/IL-312. The precursor human IL-3 coding region is indicated therein. Figure 19 shows construction of plasmid pGB/IL-313. The sequence at the 5' side of the IL-3 sequence is noted below the drawings. Figure 20 shows a schematic representation of plasmid pGB/IL-317. Figure 21 shows a schematic representation of plasmid pGB/IL-316. Figure 22 shows the nucleotide sequence of plasmid pGB/IL-316 between the unique SacII site in the lactase promoter and the HindIII site after the terminator (residues 4457 to 7204). Figure 23 shows the nucleotide sequence of plasmid pGB/IL-318 between the unique SacII site in the lactase promoter and the HindIII site after the terminator (residues 4457 to 7190). Figure 24 shows the nucleotide sequence of plasmid EF-1a promoter,

SalI-Bglll-Xhol linker and actin terminator as in plasmid pGB/TEFact.

A. Define. ions

"Human IL-3", "hIL-3", "human multi-CSF" and "hmulti-CSF" are used interchangeably here and denote a protein preparation that exhibits the following effects: 1. The protein stimulates colony formation of human hemopoietic precursor cells in which the colonies that are formed include erythroids, granulocytes, granulocyte-macrophages and mixtures. 2. The protein stimulates the DNA synthesis of human acute myelogenous leukemia (AML) blasts, which can be seen, for example, by labeled thymidine uptake.

To fall within the definition of hmulti-CSF, the action in the foregoing assay must not be substantially inhibited by antibodies made in response to, and immunospecific for, GM-CSF, unless those antibodies inhibit those actions of the illustrative hmulti-CSF below.

An illustrative form of hmulti-CSF is shown in Figure 1 as a mature protein of 133 amino acids, with a 19 amino acid long signal sequence. The amino acid sequence in Figure 1 is identical to that described by Yang, Y-C, et al., Cell (1986) 47:3-10 (supra) with the exception of position 8 of the mature protein in which the Ser of the Yang protein is replaced by Pros here. As shown herein, this amino acid sequence is effective in its unglycosylated form. However, it contains two glycosylation sites, and the glycosylated form is also included within the scope of the invention. It has also been discovered that the protein can exist in acid addition salt form, basic salt form, or it can be neutral, depending on the pH of the environment. Derivatization by phosphorylation, acetylation, and so on, to such an extent that the action is not destroyed, also results in a protein.

It has also been discovered that the entire sequence is not required to achieve activity. Parts of the amino acid sequence can be deleted or replaced, while the biological activity is retained. As illustrated herein, alanine at position 1 may be deleted, likewise as many as the first fourteen amino acid residues may be replaced by a residue sequence of a fused peptide sequence. It is also believed that the murine form of the protein only needs the first 79 residues to achieve activity; this roughly corresponds to the first 83 residues of the corresponding human. One should also take into account that the N-terminus of mature hIL-3 is formed by the residues ala-pro-met etc. (see Figure 1). It is known that the protein, when secreted by a yeast host, is in some cases shortened by two amino acids (ala-pro), due to the interaction with a dipeptidyl aminopeptidase (72). hIL-3 without the N-terminal alanine and proline still maintains its biological activity. Yeast strains carrying a null mutation in the X-prolyl dipeptidyl aminopeptidase gene produce complete hIL-3 (amino acids 1-133).

When produced as a mature protein in a prokaryotic host, the coding sequence for the mature protein is preceded by an ATG start codon. The resulting N-terminal methionine may then be removed, or partially removed, by processing within the bacterial host, depending on the nature of the subsequent amino acid sequence. Both forms of hIL-3 are again biologically active.

From what has been described above, it is clear that amino acid changes can be introduced into the human IL-3 protein without affecting its biological function. It has been discovered that minor changes in the amino acid sequences by chemical modification of coded residues, substitution of another residue, or deletion or addition of one or more, preferably only one, residue result in proteins that maintain activity. Especially the naturally occurring allelic variations and other mutations that are non-lethal on the activity.

On the other hand, it should be considered that amino acid changes in the human IL-3 protein may be beneficial to the therapeutic use of the protein. As discovered herein, the mature protein has four conserved domains at residues 15-36, 54-61, 74-91, and 107-118. Proteins containing single and multiple amino acid changes in the non-conserved regions, 1-14 (which are in all cases replaceable by the sequences of host-derived fusion proteins), 37-53-62-73, 92-106 and 119-133 possible. It appears that the cysteine residues at positions 16 and 84 may be necessary for disulfide bridge formation as these are conserved between species. Changes in the conserved domains mentioned above can influence the biological properties of the protein, such as receptor binding and signal transduction. One expects that hIL-3, which has changed properties, is of therapeutic use.

The protein preparation can contain hmulti-CSF peptides in monomeric or aggregated form, if the aggregates maintain the activity as defined above.

"Expression system" refers to a DNA sequence that contains both a coding sequence for which expression is desired and appropriate control sequences that are operably linked to it and that allow expression when the control sequences are compatible with the host in which the expression system is located. "Control sequences" refers to DNA segments that are necessary for or regulate the expression of the coding sequence to which they are operably linked.

The control sequences for all hosts include promoters, which may or may not be controllable by regulation of their environment. Typical promoters suitable for prokaryotes include, for example, the trp promoter (inducible by tryptophan deprivation), the lac promoter (inducible by galactose analog IPTG), the beta-lactamase promoter, and the phage-derived PL promoter (inducible by temperature -variety). In addition, useful promoters particularly for expression in Bacillus include those for alpha-amylase, protease, Spo2 and synthetic promoter sequences. Suitable promoters for expression in yeast include the 3-phosphoglycerate kinase promoter and those for other glycolytic enzymes, as well as promoter regions for alcohol dehydrogenase and yeast phosphatase. Transcription elongation factor (TEF) and lactase promoters are also useful. Mammalian expression generally uses promoters derived from viruses such as adenovirus promoters and SV40 promoter systems, but they also include regulatable promoters such as the metallothionein promoter, which is controlled by heavy metals or glucocorticoid concentration. Virus-based insect cell expression systems are also now available, as are expression systems based on plant cell promoters such as nopaline synthetase promoters.

In addition to the promoter DNA sequence, which is necessary for the transcription of the gene by RNA polymerase, a number of control sequences, including those that regulate termination (for example, resulting in polyadenylation sequences in eukaryotic systems) are also useful for controlling expression. Some systems also contain amplifier elements which are desirable but not necessarily necessary to provide expression.

Translational control involves a ribosome binding site (RBS) in prokaryotic systems, while in eukaryotic systems translation may be controlled by the nucleotide sequence surrounding the AUG codon.

As indicated above, recombinant protein production can be achieved in a large number of hosts, including bacteria (mainly E. coli, Bacillus and Streptomyces), in yeast and fungi (such as Saccharomyces, Kluyveromyces and Aspergillus), and in mammals and other cell cultures, such as COS cells, C127 cells, Chinese hamster ovary cells, Spodoptera frugiperda (Sf9) cells, and so on. The protein may be produced as an intracellular mature protein or fusion protein, or it may be secreted when the DNA encoding an appropriate compatible signal is included in the gene.

The present invention enables for the first time the production of recombinant human IL-3 on a large scale, so that this protein in purified form can now be used as a therapeutic agent. The methods described herein provide means for the production of glycosylated and likewise unglycosylated forms of the protein, which can be purified to substantially pure human IL-3. "Purified" human IL-3 refers to human IL-3 as defined above and which does not contain other proteins that it normally contains.

B. Provision of cDNA encoding human IL-3

Human IL-3 was isolated according to the following strategy:

1. A method was developed that allowed reproducible production of hemopoietic growth factors (EGFs) by human white blood cells. 2. mRNA was prepared from such producing cells and transcribed into double-stranded cDNA. 3. The cDNA was screened with a complete mIL-3 cDNA containing both the coding and untranslated 3' downstream portions to provide DII. 4. The hybridizing cDNA clone DII was inserted into an expression vector pLO to provide pLB4 which was expressed in COS cells to confirm the presence of the sequence encoding human IL-3. Conditioned media from these cells showed the biological activity expected for hIL-3.

Human cDNA could be obtained using this method because, despite the considerable lack of homology with the murine coding sequence, a surprising degree of homology was present in the 3' untranslated region. Applicants are not aware of any prior disclosures of the use of a 3'-untranslated sequence homology to provide another species gene.

Conditioned medium of lymphocytes cultured in the presence of 12-0-tetradecanoylphorbol-13 acetate (TPA) and concanavalin A (Con A) is a suitable source of human HGFs that emerge from analysis of the medium using stimulation of mouse CFU-S in suspension cultures, proliferation of mIL-3 dependent DA-1 cells, human hemopoietic precursor assays by colony formation in vitro, and in vivo stimulation of acute leukemia blasts. A cDNA library from human lymphocytes was constructed in lambda gt-10 phage (20) and screened using the HindIII-XbaI fragment of mIL-3 cDNA, for the presence of mIL-3 related sequences. No hybridizing clones were identified. When complete murine IL-3 cDNA was used as a probe, four clones were identified. Restriction enzyme analyzes of the largest clone (Dll) indicated a 910 bp insert containing an internal EcoRI site (at position 411, Figure 1).

(It was investigated whether this EcoRI site had arisen by ligation of two independent cDNA fragments or whether it was a naturally occurring site. Southern analyzes of restriction enzyme-cleaved human DNA using labeled 5' and 3' EcoRI- fragments of clone Dll as a probe, showed identical DNA fragments after digestion with HindIII (15 kb) and BamHI (4.6 kb). The DNA sequence surrounding the EcoRI site does not correspond to the linker sequence (pCCGAATTCGG) used to insert cDNA into phage DNA, indicating that these EcoRI fragments come from a single mRNA).

From hybridization and sequencing experiments, it was determined that the small clones (II, IV and VI) are identical to the 3' nucleotide sequence of clone Dll and that it is derived from the same mRNA species.

Comparison (Figure 1) of Dll cDNA and mIL-3 cDNA using data showed sequence homology in the 5' terminal 100 bp, between nucleotides 236-269 and between nucleotides 598-803 in the 3' terminal region (68$, 71$ and 73$ homology, respectively). The region between nucleotides 706 and 763 is highly conserved (93$ homology) and contains repetitive AT-rich sequences. The low homology in the 5' terminal 600 bp sequence of the human cDNA (52$) precludes detection by hybridization with the HindIII-XbaI fragment of mIL-3.

Analysis of the human cDNA clone for an encoded protein shows an open reading frame up to the termination codon TGA at position 495-497 (Figure 1). The first ATG triplet is likely to be the true initiation codon of the encoded polypeptide. The putatively encoded protein consists of a hydrophobic leader peptide of 19 amino acids, which is probably cleaved between the glycine and alanine residues (22, 23).

Alignment of the putative amino acid residues of human and mouse IL-3 (Figure 1) shows a homology of 50% for the leader peptide (residues -26 to +1) and 28% for the mature protein (residues 1 to 133). Within the leader peptide there are two conserved regions of four amino acids (residues -13 to -10 and -3 to +1), the other of which includes the processing site. The mature protein is 133 amino acids and has a molecular weight of 15 kd. The mature protein has four conserved domains (residues 15-36, 54-61, 74-91 and 107-118) and contains two potential glycosylation sites (residues 15-17 and 70-72). Both cysteine residues present in the human protein (positions 16 and 84) are conserved and may play an essential role in protein folding by disulfide bridging.

To verify that this human cDNA encodes a functional protein similar to mIL-3, the Dll cDNA was inserted into a eukaryotic expression vector (pLO, containing an SV40 transcription unit) to provide expression vector pLB4 and transfected into COS 1 cells. The COS/pLB4 conditioned medium (CM) was tested for (1) its ability to stimulate colony formation with human bone marrow cells, and (2) to stimulate human acute myelogenous leukemia (AML) blasts.

In vitro colony growth of human hemopoietic progenitors lacking myelomonocytic (Vim-2 positive) and T-lymphocytic (T-3 positive) accessory cells was effectively stimulated by C0S/pLB4 CM. The results demonstrate stimulation of progenitors of several hemopoietic differentiation cell lines and of a subpopulation of BFU-E by COS/pLB4 CM.

In another experiment, bone marrow was enriched for progenitor cells by density centrifugation, E-rosette sedimentation to remove T lymphocytes and adherence to remove mononuclear phagocytes and cultured in enriched medium containing fetal calf serum. "Under these "conditions, most colonies obtained by stimulation with C0S/pLB4 CM contained two or more hemopoietic differentiation cell lines: all contained macrophages, about half immature blasts and/or immature erythroid cells and/or neutrophil granulocytes and in addition a small amount of basophils or eosinophilic granulocytes. These results demonstrate the multi-cell line stimulatory properties of the protein encoded by the human cDNA clone Dll and its effect on early developing, multipotent hematopoietic cells.

With regard to AML stimulation, AML blasts from five patients were stimulated with COS/pLB4 CM and analyzed for a response by measuring 3H-TdR incorporation and colony formation. Three of five leukemia cell samples reacted to C0S/pLB4 CM in both assays; characteristic dose-response relationships for colony formation and DNA synthesis of AML blasts from different patients were provided. The response to GM-CSF further demonstrated phenotypic differences among the leukemias that responded to C0S/pLB4 CM.

These data demonstrate that the Dll cDNA clone contains the complete genetic information for a biologically active protein that is secreted into the culture medium in transformed COS cells. Despite the apparent lack of protein sequence homology between the human protein and mIL-3 (only 30$), the proteins are comparable in terms of their biological function. Both proteins act on early developing hemopoietic precursors from different cell lines. The low homology at the amino acid level is also reflected by a low homology in the coding nucleotide sequence. Very unexpectedly, a high degree of homology sufficient to provide the human cDNA clone was detected in the 3<*> untranslated region.

Southern analysis of human DNA showed a single hybridizing gene indicating that this cDNA does not belong to a family of closely related genes.

From the foregoing results, we conclude that the human cDNA insert in Dll encodes the human homolog of mlL-3. We have chosen to use the operational term hmulti-CSF for the protein encoded by cDNA clone Dll because of its main biological effect and analysis.

Identification of hmulti-CSF cDNA clones by hybridization with the 3' terminal region of mIL-3 cDNA was unexpected. While homologous DNA sequences are usually found mainly in the coding region, the hmulti-CSF sequence in this part of the gene was highly diverged (45% homology). Analyzes of the highly conserved domain in the 3 V terminal non-coding region show the presence of 5 ATTTA repeat units that are all conserved in mIL-3 cDNA (Figure 1).

hMulti-CSF and mIL-3 exhibit considerably less protein homology than other murine and human growth factors or lymphokines such as GM-CSF (25), interleukin-2 (25), interleukin-1 (26) and interferons (27-29 ). The biological activity of mature mIL-3 appears to be contained in the first 79 amino acids, including that the cysteine residue at position 17 (30) is an absolute necessity. This cysteine residue is conserved in hmulti-CSF (Fig. 1, pos. 16) and may play an essential role in protein folding. Presence of a potential glycosylation site around this cysteine residue may interfere with disulfide bridging.

C. Production and formulation of hmulti-CSF

Applicants have provided various expression systems that have the ability to produce human IL-3 protein in various forms, as fusion proteins, as mature intracellular proteins, and as secreted proteins. The applicants believe that there is no human IL-3 in preparation that does not contain any proteins that are usually included with this desired protein, within the field of recombinant forms of human IL-3. The human IL-3 protein is therefore in a form that enables adaptation to therapeutic and diagnostic applications.

The human IL-3 can be produced as a fusion protein with sequences heterologous to the human IL-3 amino acid sequence. By "heterologous" is meant a sequence that is not found in human IL-3, but is an unrelated sequence. This heterologous sequence may be derived from a bacterial protein, a yeast protein, a mammalian protein, or any of the various, non-uniform, random encoded sequences such as, for example, those encoded by polylinkers. The results herein below show that at least the first 14 amino acids of the N-terminus of the human IL-3 sequence can be replaced by a heterologous sequence, at least if the fusion protein is extended past the N-terminus.

The protein can also be provided as a mature intracellular protein by constructions in which the ATG start codon is placed directly upstream of the desired N-terminus. These intracellular proteins, whether mature proteins or fusion proteins, can be recovered by lysing the cells and purifying the human IL-3 using standard protein purification techniques.

Protein purification is easier if the human 11-3 is secreted into the medium. When produced in mammalian cells where the native signal sequence is compatible, this native signal sequence can be used to provide secretion into the medium. In bacterial or yeast systems, signal sequences compatible with these hosts, such as penicillinase or alpha-amylase sequences in bacteria or the alpha-factor signal sequence in yeast, can be used.

When produced recombinantly, the human IL-3 does not contain the proteins that usually accompany it, and can thus be purified from the proteins and other materials belonging to the recombinant host using, for example, chromatographic methods, gel filtration, ammonium sulfate precipitation, and so on further.

As described hereinbelow, the protein is useful for therapeutic and diagnostic applications. For therapeutic use, the protein may be formulated using methods commonly used for pharmaceutical compositions used for the administration of proteins. Suitable additives include, for example, physiological saline, Ringer's solution, and so on. Alternative formulations, including solid formulations (eg, lyophilized), may also be used.

D. Production of antibodies

The availability of recombinant IL-3 protein or parts thereof will allow the production of antibodies directed against the protein or parts thereof, as defined herein below. Such antibodies are useful for in vitro detection of colonies producing hIL-3, for therapeutic use, and for purification of both native and recombinant hIL-3.

Areas of application

The nucleotide sequence of all or part of human IL-3 cDNA, or closely related DNA sequences, allows the detection of genetic errors, including genomic rearrangements, restriction fragment-length polymorphisms, mutations and altered gene expression using such techniques as analyzes of chromosomal DNA using by restriction enzymes, DNA and RNA blotting and hybridization techniques (Maniatis et al. 1982) and two-dimensional gel electrophoresis (Fisher and Lerman, 1983).

Recombinant hmulti-CSF will facilitate a detailed analysis of its role in human hemopoiesis, particularly the possible synergism of hmulti-CSF and various other HGFs. hmulti-CSF is further of particular interest due to its use in in vitro diagnostics of human diseases in which hemopoietic progenitor cells are involved, including leukemia, as well as for potential therapeutic use directed at the expansion of hemopoiesis in vivo. The effect of hmulti-CSF on different hemopoietic malignancies with regard to terminal differentiation of the leukemic cells also needs to be explored. In addition, hmulti-CSF may be required for the establishment of a proliferative state in the human stem cells in gene therapy protocols, since stimulation with mIL-3 has been shown to be necessary for successful infection of mouse stem cells with recombinant, replication-defective retroviruses. IL-3 protein can also advantageously be used for the detection of early haematopoietic precursor cells in standardized in vitro cultures (Wagemaker and Visser, 1980; Metcalf et al. 1982; Merchav and Wagemaker, 1984, Metcalf, 1986). The IL-3 protein and variants can further be used for propagation of hemopoietic stem cells in vitro, possibly together with other growth factors, for bone marrow transplantation and for genetic manipulation of stem cells (Lowenberg and Dicke, 1977; Wagemaker and Petem, 1978; Lemischka et al, 1986 ). The IL-3 protein can further be used to determine the response pattern of malignant haemopoietic cells in in vitro tests (Touw and Lowenberg, 1985; Griffin et al, 1986; Griffin and Lowenberg, 1986). The IL-3 protein can also be used for the detection of residual leukemia cells by in vitro methods (Touw and Lowenberg, 1986; Griffin et al, 1986; Griffin and Lowenberg, 1986). The IL-3 protein can further be used in vitro for the treatment and prevention of malignant and non-malignant disorders, either alone or in combination, wherein a specific response elicited by the hemopoietic system may result in a clinical benefit.

These applications include:

- cytopenia and/or immunosuppression due to infections such as AIDS

- cytopenia due to chemotherapy and/or radiation

- bone disorders such as bone fractures and osteoporosis

- immunological defects due to general anesthetic conditions

- recovering from a bone marrow transplant

- accessories for vaccinations and associated infection therapy.

The cloned human IL-3 DNA sequence or closely related DNA can be used for gene therapy in case of genetic deviations from the normal IL-3 gene.

To facilitate the above described assays, a large amount of human IL-3 is required. The easiest way to provide sufficient amounts of the protein is production by microorganisms, especially yeast, bacteria and fungi, e.g. Saccharomyces, Kluyveromyces, Aspergillus, Streptomyces, Bacillus and E. coli species. Production in mammals and other eukaryotic systems, such as C127 cells, Spodoptera cells, and transgenic animals and plants, is also possible for trained personnel. These possibilities are all included within the scope of the invention.

To illustrate how to obtain living cells that produce human IL-3 protein by expression of the hIL-3 cDNA, many plasmids were constructed and transferred into E. coli, B. subtilis, B. licheniformis, S. cerevisiae, K. lactis and C127 cells. Using these host strains, production of recombinant human IL-3 was achieved. The products were tested for their ability to stimulate human AML blasts as described above for the COS/pLB4 conditioned medium. These experiments showed that the proteins made were biologically active.

Example 1

Provision of cDNA encoding human multi-CSF

(hmulti-CSF)

Human white blood cells stimulated with TPA (5 ng/ml) and ConA (10 µg/ml) produced substantial amounts of HGFs as measured in the murine stem cell proliferation assay and various other colony assays. The cells were harvested 24 hours after stimulation, because mRNA production is often short-lived after stimulation with phorbol esters and lectins. Already after 24 h, HGFs were clearly detectable in the CM.

mRNA preparation

The cells were harvested, washed with PBS and homogenized in guanidinium isothiocyanate solution (36). RNA was pelleted using a cesium chloride pad. O1igo(dT)-cellulose chromatography was used for selection of the mRNAs (36).

cDNA synthesis

cDNA was synthesized mainly according to Gubler and Hoffman (37), using oligo(dT) as primer and AMV reverse transcriptase. The second chain was synthesized with RNaseH and E. coli DNA polymerase I. Gap was closed with T4-DNA ligase and the ends were flushed with T4-DNA polymerase. To protect internal EcoRI restriction sites, the cDNA was methylated with EcoRI methylase. The cDNA was then ligated to phosphorylated EcoRI linkers with T4-DNA ligase. After digestion with EcoRI, excess linkers were removed by Sepharose CL-4B chromatography.

The substance recovered in the void volume of the column was larger than 250 bp and was used for construction of the libraries.

Construction of phage cDNA library

cDNA was ligated to lambda gtlO phage arms (20) and packaged with commercial packaging extracts (Gigapack, Vector Cloning Systems). The recombinant phages were propagated in E. coli C600 hfl.

Screening of the subject library

From each plate containing 1-5000 plaques, two nitro-cellulose filter replicates were made according to standard procedures. The filters were then hybridized with radioactively labeled mlL-3 probe from the HindIII-XbaI fragment of mIL-3 cDNA or with the entire mIL-3 cDNA clone radioactively labeled with random primers. The mIL-3 cDNA clone (pLIO1) was isolated from the WEHI-3B cDNA library. WEHI-3B mRNA was isolated using the guanidinium isothiocyanate CsCl method, size fractionated on a sucrose gradient and injected into Xenopus laevis oocytes. The RNA fractions that induce the oocytes to produce a factor capable of supporting murine stem cell proliferation were used for the synthesis of cDNA as described above, a tail with the dC residue was attached to the cDNA and inserted into the Pstl site of pUC9. mIL-3 clones were identified using synthetic oligonucleotides (from published mIL-3 sequence, 11). Inserts in pLIO1 were purified on polyacrylamide gel and used for screening the human cDNA library. Probe DNA was labeled using the random primer method (38). Potentially positive plaques were rescreened and plaque was cleaned. In this way, four clones were identified, including subject Dll.

Sequencing of cDNA clones

Recombinant phages were grown on a large scale and purified, cDNA inserts were removed from phage arms by digestion with EcoRI and purified on polyacrylamide gel. The purified fragments were ligated into M13mpl8 and pTZ18R DNA digested with EcoRI and used for transformation of E. coli JM109. Single-stranded DNA was prepared and sequenced according to established methods (39). Sequence data were analyzed using different computer programs (40-43).

The sequence provided for insertion into phage Dll is shown in Figure 1. This 910 bp sequence contains the entire coding region of hmulti-CSF and its signal sequence, and exhibits high homology to murine clone PLIO1 in the 3' untranslated region. Homology upstream in the coding sequence is relatively more limited. As described above, the protein has a putative 19 amino acid signal sequence followed by a 133 amino acid mature protein containing two glycosylation sites (15-17 and 70-72) and two cysteine residues at 16 and 84.

The deduced amino acid sequence is the same as that encoded by genomic DNA described by Yang, Y-C, et al. (above), except for one amino acid, the one at position 8 of the putative mature protein; where the Yang DNA codes for Ser, and the cDNA herein codes for Pro.

The intronless sequence provided in phage Dll can be used for prokaryotic expression, as well as for expression in eukaryotic systems, as illustrated below.

Example 2

Express. ion in mammalian cells

A. Construction of the eukaryotic expression vector pLB4

Phage Dll (containing the longest cDNA insert) was cleaved with Hindllll and Bglll and subcloned into plasmid pTl (derived from pTZ18R, containing some additional restriction sites in multilinkers, see Example 3A). Clones containing the phage fragment containing the cDNA insert were identified by restriction analysis. The cDNA insert was removed from this plasmid by partial digestion with EcoRI and purified by polyacrylamide gel electrophoresis. The appropriate fragment was inserted into a eukaryotic expression vector (pLO) in an SV40 transcription unit.

pLO includes: EcoRI (filled in) - Pstl to pBR322 (1-755), Pstl-Aval to pBR329 (756-1849), Aval-PvuII adapter (1850-1868), PvuII-Hindlll (filled in) to SV40 (promoter) ( 1869-2211), PvuII-BamHI adapter containing the unique EcoRI site (2211-2251), Mbol "splice fragment" of SV40 (2252-2861), BclI-BamHI (filled) "poly A fragment" of SV40 (2862-3098 ), PvuII-HindIII promoter fragment of SV40 (3099-3440), HindIII-BamHI Eco gpt gene (3441-4501), Mbol "splice fragment" of SV40 (4502-5111) and BclI-BamHI (filled) "poly A fragment" to SV40 (5112-5348).

The Eco gpt transcription unit is not important for transient expression of proteins in COS 1 cells. The resulting hmulti-CSF expression plasmid was designated pLB4 and was purified on CsCl. This plasmid in E. coli was deposited in the Centraal Bureau of Schimmelcultures (CBS), Baarn, The Netherlands, according to the Budapest Treaty on December 12, 1986 under CBS 568.86. This construction is shown in Figure 2.

B. Expression of hmulti-CSF in COS 1 cells and bioassays pLB4 DNA was transfected into COS 1 cells using the calcium phosphate coprecipitation method (45). The cells were cultured for 48-72 hours in alpha medium containing 10% fetal calf serum. The culture medium was collected, filtered and used in analyzes to determine biological activity. Human bone marrow precursor colony assays and acute myeloid blast colony and proliferation assays were performed as follows. Bone marrow was obtained from hematologically normal adult volunteers with posterior iliac crest function according to usual agreement. Mononuclear cells were separated by density gradient centrifugation on a Ficoll gradient (Nijegaard and Co., Oslo, Norway), washed and resuspended in Hank's balanced salt solution (HBSS). Myeloid cells and T lymphocytes were then removed. For this reason, marrow cells were lysed after incubation with monoclonal antibodies OKT-3 (CD3; Ortho, Ravitan, N.Y. ) and Vim 2 (myelo-monocyte cells, 46) at saturating concentrations in the presence of rabbit complement (40$; 30 minutes, 25° C) according to known methods (47). The cells were washed twice in HBSS, resuspended in Iscove's modified Dulbecco's medium (IMDM) and cultured in the presence of autologous plasma according to Fauser and Messner (16), as described before (48), at a concentration of 1.5-3 x 10^/ml. Erythropoietin 1 U/ml (sheep, stage III, Connaught, Willowdale, Canada) and COS/pLB4 CM were added as growth stimulatory activities. The results of standard cultures of phytohemagglutinin-stimulated leukocyte CM (PH-LCM) compared to C0S/pLB4 CM are also given. Sixty percent of the colonies were picked and identified using microscopic analyses. CM from COS cells transfected with vector without insert (pLO) did not stimulate colony formation.

The results are shown in Figure 3. As shown in the figure, the average number of erythroid (BFU-E), granulocyte macrophage (CFU-GM), granulocyte (CFU-G), eosinophil (CFU-Eo), macrophage (CFU-M) and mixed (CFU-MIX) colonies (±SD) shown by duplicate cultures stimulated with fixed volume C0S/pLB4 CM.

Induction of AML proliferation (see Figure 4)

AML blasts were purified using a bovine albumin (BSA) density gradient. Remaining T lymphocytes were removed from the AML samples by E rosette sedimentation (17, 49, 50). AML (patient 1) colony formation was determined not only in the established PHA white blood cell feeder (PHA l.f) system, but also in a modified version of the technique in which the white blood cells were replaced with C0S/pLB4 CM, which allows the determination of colony- stimulatory activity (17, 18, 49, 50) as shown in Figure 4A. All experiments were performed in triplicate. DNA synthesis of AML blasts (patient 2) was analyzed by thymidine uptake as described (51) and the results are shown in Figure 4B. Both assays showed a dose-dependent relationship with added C0S/pLB4 CM. Addition of control COS medium did not affect AML proliferation in any of the assays.

C. Construction of eukaryotic expression vector PLB4/BPV

To establish stable cell lines expressing human IL-3, C127 cells (ATCC CRL 1616) transfected with a derivative of pLB4. This derivative was constructed by inserting the entire BPV-1 genome (69) into pLB4 using the following strategy. The BVP-1 BamHI fragment was excised from vector pdBVP-MMTneo(342-12) (70). BamHI sticky ends (overhanging end) were filled in using Klenow polymerase. Next, vector pLB4 was cleaved at the unique EcoRV site within the Eco gpt gene. Then, the BPV-1 blunt-ended fragment was cloned into EcoRV cleaved pLB4, resulting in vector pLB4/BPV capable of replicating in C127 cells. pLB4/BPV was transfected into C127 cells using the calcium phosphate precipitation method (45). The transfected cells were cultured for 16 days, and foci were picked from the culture dishes. Several independent cell lines were established. pLB4/BPV vector is stably maintained in the cells, as judged by Southern blotting of the Hirt extracts (71) from several cell lines. Conditioned culture medium was tested for IL-3 activity using the AML proliferation assay. The stable cell lines actively produce human IL-3.

Example 3

Construction of E. coli expression vector

A. Construction of pGB/IL-301 (see figures 5, 6, 7 and 8)

For construction of E. coli expression vectors, the following modifications were performed according to standard procedures (36). 1. The 3'-terminal non-coding sequence between the Aval site (position 541) and the XhoI site (position 856) in pLB4 was deleted by fusion of the DNA fragments after filling in the sticky ends with Klenow enzyme (Figure 5 ). 2. To introduce the hmulti-CSF insert into a bacterial expression vector, the following steps were performed. The pLH1 vector was digested with AvalI and the overhanging ends were filled in with Klenow polymerase. After ligation of a Bglll linker (CAGATCTG), The DNA was digested with Bglll and BamHI. The Bglll-BamHI hmulti-CSF fragment was purified on polyacrylamide gel and subcloned into the Bglll site of pT1, a derivative of pTZ18R (Pharmacia) modified in the multiple cloning site (see Figure 6). Two clones were obtained, which had the insert in the opposite orientation with respect to the lacZ promoter (see Figure 5). The inserts of these two clones were isolated on polyacrylamide gel after digestion with Bglll and EcoRV and subcloned into pTl cleaved with Bglll and Hindll. The connection with Bglll linker and hmulti -CSF DNA was verified by sequence analysis and showed a fusion of the linker to the Avall site located at nt 1 of the cDNA clone (this Avall site had arisen by ligation of the EcoRI linker to the cDNA molecule).Since this construct (pGB/IL-300) was not in phase with the lacZ protein, the Bglll-EcoRV insert was subcloned into BamHI and Hindll cleaved pUC8 (52). The resulting construct (pGB/IL-301, see Figures 5, 7 and 8) was tested for production of a lacZ/hmulti-CSF fusion protein.

B. Construction of pGB/IL-302. pGB/IL-303. pGB/IL-304 and pgb/il-305 (see Figures 5, 7 and 8).

Several base changes were introduced into the coding sequence for the N-terminal part of the fusion proteins by introducing synthetic oligonucleotides into pGB/IL-300. The new expression vectors, designated pGB/IL-302, pGB/IL-303 and pGB/IL-3024, were constructed as follows: The HindIII-HindIII fragment of pGB/IL-300 was isolated on agarose gel and ligated to synthetic oligonucleotides containing the nucleotides 99- 137 to hmulti-CSF and a 5'-terminal SalI recognition sequence and inserted into pTZ18R cleaved with SalI and HindIII. The sequence of several clones was determined. Several base changes were observed, resulting in modifications of the hmulti-CSF protein. The inserts from several clones were transferred into pUC8 for expression of the lacZ fusion protein (pGB/IL-302, pGB/IL-303). Clone pGB/IL-304 was made in phase with lacZ by ligation of the SalI site after filling in overhangs with Klenow. The construct was verified by pVUI cleavage. Several clones lacked a synthetic oligonucleotide and were found to be fused in frame to the lacZ protein. An example of these clones is designated pGB/IL-305.

C. Construction of pGB/IL-306 (see Figures 5, 7 and 8).

An expression vector encoding a protein lacking lacZ N-terminal amino acids was made from pGB/IL-300 by deletion looping as described (53). The synthetic oligonucleotide contained 22 nucleotides upstream of the pTZ lacZ gene including the ATG start codon and the first 24 nucleotides encoding mature IL-3. This plasmid was designated pGB/IL-306 (Figures 5, 7 and 8).

The E. coli strains containing plasmids pGB/IL-300, pGB/IL-301 and pGB/IL-302 were deposited with CBS on July 13, 1987 under CBS 377.87, CBS 379.87 and CBS 378.87, respectively.

Figure 8 shows the sequence of the fusion regions for the different constructed plasmids. The sequence of the clones is given from the beginning of the lacZ protein coding region in either pUC8 or pTZ18R (lowercase letters) and for the hmulti-CSF coding region (uppercase letters) up to the ClaI site at position 158. Mutations in the hmulti-CSF DNA sequence , is underlined, resulting in trp<*3> -»arg13 (pGB/IL-302); leu<9> -» pro<9> and trp<13> -+ arg <13> (pGB/IL-303); met<3> -» thr<3> and a silent change (pGB/IL-304).

In priority application EP 87201322.2, filed July 13, 1987, other designations were used for these plasmids: pGB/IL-300 = pT-hIL3;

pGB/IL-301 = pUC/hmulti;

pGB/IL-302 = pUC/hmultiAlA;

pGB/IL-303 = pUC/hmultiAlB;

pGB/IL-304 = pUC/hmultiAlC;

pGB/IL-305 = pUC/hmultiA2;

pGB/IL-306 = pTZ/hmulti;

D. Expression of lacZ/hmulti-CSF fusion proteins and mature

hmulti-CSF in E. coli.

The E. coli strains (JM 109) containing different expression vectors were grown in LB medium containing 50 µg/ml ampicillin at 37°C until an optical density of 0.5 at 550 nm was obtained. Then IPTG (isopropyl beta-D-thiogalactoside, Pharmacia) was added to the culture to a final concentration of 1 mM and the incubation was continued for 3-4 hours.

Plasmids pGB/IL-306 and pGB/IL-302 were also transformed into E. coli DH1 (wild-type lacZ operon). These strains were grown in LB medium or 2 x TY medium containing 50 µg/ml ampicillin at 37°C for 16 hours.

Bacteria were collected by centrifugation and tonicized in buffer containing 0.1 M Tris/HCl, pH 8.0; 5 mM EDTA 0.2$ Nonidet P40 (NP-40) and 1 mM phenylmethylsulfonyl fluoride (PMSF) and centrifuged for 30 min. at 20,000 x g. Polyacrylamide gel electrophoresis of the pellet and supernatant fractions showed that the majority of the hmulti-CSF proteins are stored in the bacteria in an insoluble form.

The pellet was again extracted with 0.5% NP-40 buffer and finally dissolved in 8 M urea 0.1 M Tris/HCl, pH 8.0 and 5 mM dithiothreitol. In this way, an extensive purification of the fusion proteins was provided (Figure 9).

As shown in the figure, inclusions from the bacteria (E. coli) containing pGB/IL-301 and pGB/IL-302 were isolated as described. Files 1 show 0.2$ NP40 supernatant (sample corresponds to 0.1 ml of the original bacterial culture). Files 2 show 0.5$ NP40 supernatant (0.2 ml) and files 3 show the pellet dissolved in 8M urea buffer (A: 0.05 ml; B: 0.2 ml). The proteins were separated on a 13.5$ SDS-polyacrylamide gel and stained with Coomassie Brilliant Blue. Molecular weights (in kd) of the marker proteins (file M) are listed on the right. The human multi-CSF fusion proteins are indicated by the arrows. The fusion protein encoded by pGB/IL-301 has an expected MW of approximately 20 kd; and that produced from pGB/IL-302 of approximately 16 kd.

E. Determination of biological activity of bacterial hmulti-CSF preparations.

Bacterial protein preparations were diluted in alpha medium containing 1% bovine serum albumin, filter sterilized and analyzed in the AML blast proliferation assay. Diluted samples were added to purified AML blasts and cultured for four days. DNA synthesis was measured using <3>H thymidine as described (51). One unit per ml is defined as the amount of hmulti-CSF required for half maximal proliferation of AML blasts. Figure 10 shows this titration. Various dilutions of urea-extracted protein preparations from bacteria containing plasmid pGB/IL-302 were analyzed for stimulation of AML blast proliferation using <3>H-thymidine. The fusion protein concentration of this protein preparation was 33 µg/ml. Based on the present titration curve, the activity of this preparation is 16,000 units/ml.

The amount of bacterial fusion protein in the preparations was calculated from polyacrylamide gel electrophoresis and used to calculate specific activities.

The results are shown in the following table:

1. Approximate molecular weights are calculated from the DNA sequence of the fusion protein (Figure 8). 2. The IL-3 concentrations were calculated on SDS-polyacrylamide gel and calculated per ml of starting culture. 3. Urea-dissolved protein activity was determined in the AML proliferation assay and is expressed per ml of initial culture.

4. Not determined.

From these results, it was determined that human multi-CSF expressed as a fusion protein in E. coli was provided in a biologically active form. The results show that changes introduced in the N-terminus of the fusion proteins can influence the specific activity of these proteins.

Example 4

Production of antibody preparations that have the ability to immunospecifically react with human IL-3 protein.

A. Polyclonal rabbit anti-human IL-3 antiserum

A preparative gel was made from an E. coli lysate containing plasmid pGB/IL-301. The 20 kd band of IL-3 fusion protein was excised, crushed in saline with a mortar and emulsified in a 1:1 ratio in complete Freund's adjuvant containing 1 mg of Mycobacterium tuberculosis H37RA per ml. New Zealand white rabbits (spf) were immunized with 1 ml emulsion (with ± 100 µg IL-3 fusion protein) divided into 5 injection sites (2 x i.m. in the thighs, 3 x s.c. on the back). Booster injections of the same antigen in incomplete Freund's adjuvant were given at weeks 2, 4 and 6. Serum was collected at week 8 by ear venipuncture.

One volume of serum was absorbed with 9 volumes of sonicated pUC8 containing E. coli (overnight at 4°C) to remove non-specific antibodies. Immunoblotting of all IL-3 constructs made in E. coli, B. licheniformis, B. subtilis, S. cerevisiae and K. lactis showed immunospecific reactions with the absorbed sera at a dilution of 1 to 6500.

Some of these results are shown in Figure 11. The proteins were isolated from the recombinant hosts as described above and then separated on a 13.5% polyacrylamide gel and blotted onto a nitrocellulose membrane. File 1: An E. coli containing pTZ18R (control); File 2: pGB/IL-301; File 3: pGB/IL-301; File 4: pGB/IL-302; File 5: pUC19 (control); File 6: pGB/IL-301; File 7: pGB/IL-302. Files 6 and 7 show the presence of proteins in the pellet after sonication of the bacteria. Files 3, 4 and 5 show the presence of proteins in the pellet after the first washing step. Files 1 and 2 show the final urea-solubilized protein fractions.

Arrows show fusion proteins (of expected size) expressed from pGB/IL-301 and pGB/IL-302. Figure 12A shows inhibition of IL-3 dependent proliferation of AML blast cells by anti-IL-3 antiserum. Figure 12B shows that the preimmune serum does not affect the effect of IL-3 on AML blast cell proliferation. In both panels, A = IL-3 at 10 TJ/ml; P = IL-3 at lU/ml; = control, no addition. Figure 12A shows IL-3 dependent growth in the AML blast proliferation assay (51) was inhibited by sera in a dose-dependent manner; Figure 12B shows that preimmune sera do not have this effect. As a control, GM-CSF dependent growth was not affected by these sera in the same assay (Figure 12A where ^ = GM-CSF at 100 U/ml.)

B. Monoclonal mouse anti-human IL-3 antibodies

Balb/C mice were immunized with 3 x 0.1 ml (s.c.) of the same emulsion used for rabbits. A booster (0.1 ml i.p.) of antigen in incomplete Freund's adjuvant was given at week 2 and three days later splenic lymphocytes were fused with SP2/0 myeloma cells according to standard procedures (65). Hybridoma supernatants were screened in enzyme-linked immunosorbent assay, using a lysate of E. coli pGB/IL-302 (containing the 17 kd IL-3 fusion product) as a positive control and an E. coli pUC8 lysate as a negative control. In total, 29 IL-3 hybridoma cultures secreting antibodies specific for IL-3 were selected and made stable.

Example 5

Construction of Bacillus expression vectors

General cloning techniques were used (36).

A. Construction of pGB/IL-307 (Figure 13)

For construction of pGB/IL-307, the SmaI fragment of pLB4 carrying the hmulti-CSF gene was ligated into PvulI cleaved pUBllO (54). After transformation into competent DB105 cells (56)

(a spo<-> derivative of the protease-deficient strain DB104 (55)), two clones were provided, as expected: the fragment was cloned in both orientations. The plasmid containing the fragment in the correct orientation with respect to the so-called "Hpa II promoter" (57) was designated pGB/IL-307. In this case, a fusion protein will be made (see figure 13).

B. Construction of pGB/IL-310

An hmulti-CSF expression plasmid was prepared as described below.

1. Promoter cloning ( figure 14).

For expression in Bacillus, a synthetic T43 promoter as described (58) is used (the promoter used is designated T55).

Plasmids pPR0M55s (58), the promoter-containing plasmid, and pGPA14 (59) were digested with EcoRI and XbaI. The promoter fragment was ligated into the vector fragment, which had been purified on an agarose gel. After transformation into E. coli (JM 101), the correct plasmid was provided and designated pGB/IL-308 (Figure 14).

2. Introduction of a synthetic oligonucleotide into pGB/IL-308 (figure 15)

A synthetic oligonucleotide including nucleotides 39-158 and 484-546 of hmulti-CSF, a 5' terminal SalI recognition sequence and a 3' terminal XmalII site was ligated into SalI-XmalII cleaved pGB/IL-308. The ligation mixture was introduced into JM101. After analyzes of several transformants, the correct plasmid was found, pGB/IL-309.

3. Introduction of hIL3 ( figure 16)

After transformation into and isolation from B. subtilis DB105, plasmid pGB/IL-309 was digested with XmalII. The recessed ends were filled in with Klenow polymerase, and the plasmid was cleaved with ClaI. Plasmid pGB/IL-307 was digested with Aval and the ends filled in with Klenow and then digested with ClaI. Then, the hmulti-CSF containing fragment was ligated into the pGB/IL-309 fragment and transformed into JM101. The resulting plasmid was designated pGB/IL-310 (Figure 16). This plasmid contained the hIL-3 gene with its own signal sequence. After isolation of the correct plasmid, it was also introduced into B. subtilis DB105. C. Construction of pGB/IL-311 and pGB/IL-312 (Figures 17, 18) pGB/IL-310 was partially cleaved with EindIII and totally with PvuII. The two hmulti-CSF-containing PvuII cleaved with HindiII and Smal.

Figure 17 shows the nucleotide sequence of plasmid pBHAl. The plasmid consists of positions 11-105 and 121-215; bacteriophage FD terminator (double): positions 221-307; part of plasmid pBR322 (positions 2069-2153, respectively): positions 313-768; bacteriophage Fl, origin of replication (positions 5482-5943, respectively): positions 772-2571; parts of plasmid pBR322, respectively the origin of replication and the beta-lactamase gene: positions 2572-2685; transposon Tn903, complete genome: positions 2719-2772; tryptophan terminator (double): positions 2773-3729; transposon Tn9, the chloramphenicol acetyl transferase gene. The nucleotides at position 3005 (A), 3038

(C), 3302 (A) and 3409 (A) differ from the wild-type cat coding sequence. These mutations were introduced to

eliminate the Ncol, Ball, EcoRI, and PvuII sites: positions 3730-3804; multiple cloning site: positions 3807-7264; parts of plasmid pUB11O, respectively replication function and kanamycin resistance gene (EcoRI-PvuII fragment) (66, 67): positions 7267-7331; multiple cloning site. The fragments were assembled using known cloning techniques, e.g. filling in sticky ends with Klenow, adapter cloning, etc. All data are obtained from GenbankR, National Nucleic Acid Sequence Databank, NIH, USA.

After transformation into JM101 and analysis of a number of ampicillin-resistant colonies, two different plasmids were found: pGB/IL-312, which contained the complete gene with complete control sequences, and pGB/IL-311, which contained the complete gene and promoter that missing the -35 region into the other orientation (see Figure 18).

pGB/IL-311 has been transformed into B. subtilis DB105 and B. licheniformis strain T399 (Aamy, spo~, exo-protease negative, rif<r>, see ref. 68, where this strain is designated T9).

D. Construction of pGB/IL-313 (Figure 19).

To provide a smaller plasmid, with the hmulti-CSF gene after the "Hpall promoter", pGB/IL-312 was digested with BamHI and again ligated. The ligation mixture was transformed into DB105 competent cells. A number of neomycin-resistant colonies were analyzed and the correct plasmid was provided. The plasmid was designated pGB/IL-313.

E. Construction of pGB/IL-317 (Figure 20)

To clone the hmulti-CSF gene after the B. licheniformis alpha-amylase transcription and translation initiation region and signal sequence, one of the previously described pOL5-delta vectors (68) was used, respectively pOL5-2 delta. Besides the alpha-amylase signal sequence (of 29 amino acids), this plasmid contains an amino acid from the alpha-amylase mature sequence (an Ala) followed by a multiple cloning site: EcoRI-Xmalll-Xmal-Sall-Hindlll 868).

The Sall-PvuII fragment of plasmid pGB/IL-310 containing the hmulti-CSF gene was ligated into the Sall-PvuII cleaved pOL5-2 delta vector and transformed into DB105. The resulting plasmid was designated pGB/IL-317 (Figure 20). The hIL-3 gene even contains its own signal sequence in this plasmid. The plasmid was also introduced into B. licheniformis T399.

F. Expression of five expression plasmids in Bacillus strains B. subtilis and B. licheniformis strains carrying the expression plasmids mentioned below were grown in TSB medium containing 20 µg/ml neomycin or 10 µg/ml erythromycin at 37°C (for 16-24 hours ); 300 µg/ml of the culture was centrifuged. The pellet was resuspended in sample buffer and analyzed using polyacrylamide gel electrophoresis followed by Western blotting. The supernatant was TCA-precipitated and the pellet was resuspended in sample buffer. Both the supernatant and the pellet were analyzed for IL-3 protein (see Table 2).

To determine the biological activity of the produced proteins, the following steps were performed: The cell pellets were resuspended in a buffer containing 0.1 M Tris/HCL pH 8.0 and 10 mM MgClg- Lysozyme was added to a final concentration of 1 mg/ ml and PMSF to a final concentration of 1 mM. The solution was incubated for 30 min. at 37° C. Then DNase (final concentration 20 µg/ml) was added and the solution was incubated for 15 min. at 20<C>C. Finally, the biological activity of this preparation and likewise of the supernatant of the cultured cells was determined as described. The results are shown in table 2.

It can be determined that in B. subtilis using pGB/IL-307, a fusion protein is made which has IL-3 activity. When the human IL-3 gene only contains its own signal sequence, no significant secretion of human IL-3 is provided. All IL-3 activity is found intracellularly. In those cases, it appears that mature IL-3 (15 kd) has been formed in the cells in addition to precursor IL-3. Some transport across the membrane may have occurred, but the protein is not transported through the cell wall. When using alpha-amylase regulation and the secretion signals (pGB/IL-317), it appeared that most of the IL-3 activity was secreted in the culture medium. Apart from a degradation product, two proteins were detected in the supernatant, one of approximately 15 kd and one of approximately 17 kd, probably mature IL-3 and precursor IL-3, respectively. These data indicate that both processing sites, respectively alpha-amylase and hmulti-CSF processing site, are used. The product of which there is most in the cell is precursor IL-3 containing the alpha-amylase signal sequence (20 kd protein) which is shown by Western blotting. Sometimes a degradation product is detected.

Example 6

Construction of Kluyveromyces lactis expression vectors

A. Construction of pGB/IL-316

A DNA fragment containing the Tn5 gene (61) conferring resistance to gentamycin G418, under the control of the alcohol dehydrogenase I (ADHI) promoter from S. cerevisiae, similar to that described by Bennetzen and Hall (62), was inserted in the SmaI site of pUC19 (63). An E. coli strain containing the provided plasmid, pUC-G418, was deposited with CBS on December 4, 1987, under CBS 872.87.

Into the XbaI-HindIII cleaved pUC-G418 vector, an XbaI-HindIII fragment from plasmid pGB903 (64) containing the K. lactis lactase promoter and calf prochymosin DNA was inserted, resulting in plasmid pGB/IL-314.

The SalI-HindIII fragment from this plasmid was replaced by a synthetic DNA fragment containing a small multiple cloning site and the lactase terminator (see Figures 21, 22). The resulting plasmid is designated pGB/IL-315.

In the SacII-XhoI cleaved pGB/IL-315 vector, the following fragments were ligated: 1. The SacII-XbaI fragment from pKS105 (TJS Patent Application Ser. 078,539, 64), containing the 3' part of the lactase promoter and the 5' part of alpha- factor signal sequence of S. cerevisiae. 2. A synthetic oligonucleotide containing the 3' part of the alpha-factor signal sequence starting at the XbaI site and the 5' part of the mature hIL-3 cDNA sequence up to the 5' half of the HpaI site (aa residue 14). 3. The HpaI-XhoI fragment containing most of the hIL-3 cDNA sequence (residues 15-133 plus 3' non-coding region). The resulting plasmid, designated pGB/IL-316, is schematically drawn in Figure 21. The complete vector sequence from the SacII site in the lactase promoter sequence up to the HindIII site at the end of the synthetic terminator is given in Figure 22.

Figure 22 shows the nucleotide sequence of plasmid pGB/IL-316 between the unique Sac II site in the lactase promoter and the Hind III site after the terminator (residues 4457 to 7204). Residues 4457 to 6100 contain the lactase promoter sequence. Residues 6101 to 6355 include the alpha-factor signal sequence. Residues 6356 to 7115 include the sequence of the mature human IL-3 plus the 3' non-coding cDNA sequence. Residues 7116 to 7204 include the synthetic terminator sequence.

B. Construction of pGB/IL-318

An expression vector almost identical to pGB/IL-316 was constructed in which the coding information for the alpha-factor signal sequence of S. cerevisiae was replaced by the alpha-factor signal sequence of K. lactis (64). The remaining part of the plasmid is identical to pGB/IL-316. The sequence of pGB/IL-318 between the SacII site in the lactase promoter and the HindIII site after the terminator (residues 4457 to 7190) is given in Figure 23.

Residues 4457 to 6087 include the sequence of the lactase promoter and a small inker sequence. Residues 6088 to 6342 include the K. lactis alpha factor signal sequence. Residues 6343 to 7102 include the sequence for mature human IL-3 plus the 3' non-coding cDNA sequence. Residues 7103 to 7190 include the synthetic terminator sequence.

C. Transformation of Kluyveromyces Lactis and analysis of

secreted hIL- 3

Plasmids pGB/IL-316 and pGB/IL-318 were cleaved at the unique SacII site in the lactase promoter region and used to transform K. lactis strain CBS 2360 (see 64). Integration of the plasmids was thus targeted to the chromosomal lactase gene promoter region. The resulting G418 resistant transformants were grown to saturation in liquid in YEPD medium, and the culture supernatants and cell lysates were assayed for IL-3 activity using the AML cell DNA synthesis assay.

Almost all IL-3 appeared to be secreted into the culture medium, and was found to be active. The proteins from the culture supernatant were precipitated using ethanol and analyzed using denaturing polyacrylamide gel electrophoresis after Western blotting. The major product has an apparent MW of approximately 21 kd, while a distinct band at approximately 15 kd is also observed. The latter product probably corresponds to mature non-glycosylated IL-3, while the 21 kd product is the product containing core glycosylation at the two potential glycosylation sites. Incubation with endoglycosidase H results in a protein that migrates in the 15 kd region. This suggests that all IL-3 is correctly processed during the secretion process and that the majority of the proteins become glycosylated.

Example 7

Construction of a Saccharomyces Cerevisiae expression vector

A. Construction of pGB/IL-319

First, an expression vector designated pGB/TEFact was constructed. On this pTZ18R (Pharmacia)-derived plasmid, the S. cerevisiae translation elongation factor (EF-lalpha) promoter sequence, which was cloned and sequenced as described (73, 74), is linked using a small SalI-Bglll-XhoI linker to the S. cerevisiae actin transcription terminator sequence (75), which was synthesized using an Applied Biosystems DNA synthesizer. The sequence of the expression cassette is given in Figure 24. Residues 1 to 949 include the EF-lalpha promoter. Residues 950 to 967 include the sequence of the SalI-BglII-XhoI linker. Residues 968 to 1113 include the actin terminator sequence.

The unique SmaI site in pGB/TEFact was used to introduce the G418 resistance cassette described in Example 6. The resulting plasmid was designated pGB/TEFactG418.

Finally, the hIL-3 expression vector pGB/IL-318 was constructed by introducing the following DNA sequences into the SalI-XhoI cleaved pGB/TEFactG418 plasmid: - SalI-NruI fragment from pGB/IL-316 containing the S. cerevisiae alpha-factor signal sequence and the hIL-3 coding sequence up to the Nrul site. - A synthetic Nrul-Xhol DNA fragment containing the rest of the nucleotides that code for hIL-3 and the Xhol recognition sequence right after the TGA stop codon.

B. Transformation of Saccharomyces Cerevisiae and analysis of

secreted hIL- 3

Plasmid pGB/IL-319 was cleaved at the unique EcoRI site in the EF-1a promoter. Integration of the plasmid thus targets the chromosomal EF-1a region. S. cerevisiae of wild-type strain D273-103 (alpha; ATCC 25657) was transformed as described for K. lactis (64). G418-resistant colonies were picked and transformants were plated to saturation in liquid YEPD medium. The culture supernatant was analyzed for hIL-3 activity using the AML assay. The protein produced by S. cerevisiae was found to be biologically active.

The proteins from the supernatant were precipitated using ethanol and then analyzed by polyacrylamide gel electrophoresis followed by Western blotting. Two major products could be distinguished on Western blot, a 21 kd glycosylated product and an unglycosylated product of approximately 15 kd.

REFERENCES

1 Metcalf D., Blood 67, 257-267 (1986).

2 Whetton A.D. and Dexter T.M. TIBS 11, 207-211 (1986).

3 Wagemaker G., In "Bone Marrow Transplantation" (eds. Van Bekkum D.W. and Lowenberg B.) Marcel Dekker Inc. New York 1-72 (1985).

4 Dorssers L. et. al., Exp. Eematol. 12, 357, 1984.

5 To J.E. and McCulloch E.A., Radiat. Res. 14, 213-333

(1961).

6 Hapel A.J., et. al., Blood 65, 1453-1459 (1985).

7 Scheven B.A.A., Nature 321, 79-81 (1986).

8 Garland J.M. and Crompton S. Exp. Hematol. 11, 757-761

(1983).

9 Stanley E.R. a. al., Cell 45, 667-674 (1986).

10 Kreigler A.B. et al., Blood 60, 503-508 (1982).

11 Fung M.C. a. al., Nature 307, 233-237 (1984).

12 Yokata T. et al. al., Proe. Nati. Acad. Pollock. USA, 81, 1070-1074 (1984). 13 Lowenberg B. and Dicke K. A., Exp. Hematol. 5, 319-331

(1977).

14 Wagemaker G. and Peters M.F., Cell. Pee. Kinetic. 11, 45-46

(1978). 15 Ihle J.N., et. al., in Advances in Viral Oncology, vol. 4 (ed Klein G.) 95-137, Raven Press, New York 1984.

16 Fauser A.A. and Messner H.A. Blood 52, 1243-1248 (1978).

17 Lowenberg B. et al. al., Leuk. Res. 4, 143-149 (1980).

18 Lowenberg B. et al. al., Blood 59, 64-645 (1982).

19 Buick R.N. a. al., Blood 54, 95-104 (1979).

20 Huynh TV a. al., in "DNA donation", vol. 1 (Ed. Glover

D.M.) IRL press, Oxford 45-78 (1985).

21 Kozak M. Cell 44, 283-292 (1986).

22 Von Heijne G. Eur J. Biochem 133, 17-21 (1983).

23 Perlman D. and Halvorson HO, J. Mol. Biol. 167, 391-409

(1983).

24 Shaw G. and Kamen R. Cell 46, 659-667 (1986).

25 Schrader J.W., et. al., Proe. Nati. Acad. Pollock. USA 83, 2458-2462 (1986).

26 Maren C.J., et. al., Nature 315, 641-647 (1985).

27 Higashi Y. et al. al., J. Biol. Chem. 258, 9522-9529 (1983).

28 Dijkema R. et al. al., EMBO J. 4, 761-767 (1985).

29 Zwarthoff E.C. a. al., Nucleic Acids Res. 13, 791-804 (1985).

30 Clark-Lewis I. et. al., Science 231, 134-139 (1986).

31 Kindler V. et. al., Proe. Nati. Acad. Pollock. USA 83, 1001-1005 (1986).

32 DeLamarter J.F. a. al., EMBO J. 10, 2575-2581 (1985).

33 Lemischka I.R. a. al., Cell 45, 917-927 (1986).

34 Yu-Chung Yang et al. al., Cell 47, 3-10 (1986).

35 Miyatake S. et al. al., Proe. Nati. Acad. Pollock. USA 82, 316-320 (1985). 36 Maniatis T. et. al., in "Molecular Cloning, A. laboratory manual". Cold Spring Harbor Laboratories, New York (1982).

37 Gubler U. and Hofmann B.J., Gene 25, 263-269 (1983).

38 Feinberg A.P. and Vogelstein B. Anal. Biochem. 132, 6-13

(1983). 39 Sanger F. et. al., Proe. Nati. Acad. Pollock. USA 74, 5463 (1977). 40 Queen C. and Korn L.J. Nucleic Acid Res. 12, 581-599 (1984).

41 Staden R. Nucleic Acid Res. 10, 2951-2961 (1982).

42 Devereux J., et. al., Nucleic Acids Res. 12, 387-395 (1984). 43 Lipman D.J. and Pearson W.R. Science 227, 1435-1441 (1985). 44 Subramani S. and Southern P.J., Anal. Bioch. 135, 1-15

(1983).

45 Vigler M., et. al., Cell 14, 725-731 (1978).

46 Majdic 0., et. al., Int. J. Cancer 33, 617-623 (1984).

47 Lowenberg B. and Bauman J.G.J. Blood 66, 1225-1232 (1984).

48 Delwel R., et. al., Blood 68, 41-45 (1986).

49 Swart K. et al. al., Blood 59, 816-821 (1982).

50 Swart K. and Lowenberg B. Cancer Res. 44, 657-660 (1984). 51 Touw I. et. al., Blood 68, 1088-1094 (1986). 52 Vieira, J. and Messing J., Gene 19, 259-268 (1982). 53 Osinga, K.A. a. al., Nucleic Acids Res. 11, 8595-8608

(1983). 54 Gryczan, T.C. et al.. J. Bacteriology 134, 318-329 (1978) 55 Kawamura, F. and Doi, R.H. , J. Bacteriology 160, 442-444 (1984).

56 Bron, S. and Venema, G., Mutat. Res. 15, 1-10 (1972).

57 Zyprian, E. and Matzura, H., DNA 5, 219-225 (1986).

58 EPA 0224294, published June 3, 1987.

59 EPA 0244042, published November 4, 1987.

60 Stanssens P. et al.. in "Protein Engineering and Site-Directed Mutagenesis". Twenty-Fourth Harden Conference. Program and Abstracts (1985) (Fersht, A.R. and Winter, G.,

eds ).

61 Reiss, B. et ah, EMBO J. 3, 3317-3322 (1984).

62 Bennetzen, J.L. and Hall, B.D., J. Biol. Chem. 257, 3018-3025 (1982).

63 Yanisch-Perron, C. et al.. Gene 33, 103-119 (1985).

64 U.S. Appl. Looking. No. 078,539, filed July 28, 1987.

65 Salfre, S. and Milstein, C, Meth Enz 73, 3-75 (1981).

66 McKenzie, T. et al.. Plasmid 15, 93-103 (1986).

67 McKenzie, T. et al.. Plasmid 17, 83-85 (1987).

68 European Patent Application 87201379.2, filed July 20, 1987.

69 Chen, E.Y. et al, Nature 299, 529-534 (1982).

70 Law, M-F. et aL, Mol. Cell Biol. 3, 2110-2115 (1983).

71 Hirt, B., J. Mol. Biol. 26, 365-367 (1967). 72 Suarez Rendueles, M.P. et al.. FEBS Lett. 131, 296-300

(1981).

73 Najata, S. et aJU, EMBO J. 3, 1825-1830 (1984).

74 Nagashima, K. et al.. Gene 45, 265-273 (1986).

75 Gallwitz, D. and Sures, I., Proe. Nati. Acad. Pollock. USA 77, 2546-2550 (1980).

Claims (12)

1. Process for the production of human IL-3 (Pro-8) in a host cell, characterized in that it includes: introducing into said host cell a DNA expression vector containing an expression cassette which, in the direction of transcription, consists of a transcription initiation regulatory region which is functional in said host cell; a DNA sequence encoding human IL-3; (Pro-8) (Fig. 1); and a transcription termination regulatory region that is functional in said host cell; culturing said host cell containing said DNA construct in a nutrient medium under suitable culture conditions, whereupon human IL-3 is produced; and recovery of the human IL-3 product. 2. Method according to claim 1, characterized in that said host cell is selected from the group consisting of yeast, bacteria, fungi and tissue culture cells. 3. Expression vector, characterized in that it contains and can express a DNA sequence that codes for human IL-3 with 1 Pro in position 8 of the mature protein molecule (Fig. 1). 4. Expression vector according to claim 3, characterized in that the DNA encoding human IL-3 does not contain introns. 5 . Expression vector according to claim 3 or 4, characterized in that the control sequences include a promoter selected from the group consisting of lac promoter, Hpall promoter, 43 promoter, alpha-amylase promoter, EF-1 alpha promoter and SV40 promoter. 6. Vector, characterized in that it is selected from the group consisting of pGBL/IL-300 to pGB/IL-319. 7. Vector, characterized in that it is selected from the group consisting of pLB4 and pLB4/BPV. 8. Transformed live host cell, characterized in that it contains and can express recombinant DNA material that codes for human IL-3 with 1 Pro in position 8 of the mature protein molecule (Fig. 1). 9. Transformed living host cell according to claim 8, characterized in that it is selected from the group consisting of yeast, bacteria, fungi and tissue culture cells. 10. Transformed yeast host cell according to claim 9, characterized in that it is selected from the group Saccharomyces and Kluyveromyces. 11. Transformed bacterial host cell according to claim 9, characterized in that it is selected from the group E. coli and Bacillus. 12. Transformed tissue culture host cell according to claim 9, k a r a k-