EP0238655A1 - Human granulocyte-macrophage colony stimulating factor-like polypeptides and processes for producing them in high yields in microbial cells - Google Patents

Abstract

Polypeptides similar to stimulation of granulocyte-macrophage colonies having a specific activity of at least 1 × 10 8 Units / mg and their methods of manufacture. DNA sequences and recombinant DNA molecules and microbial hosts transformed by them, which produce significant quantities of human granulocyte-macrophage colony stimulating factor polypeptides, and methods of making these polypeptides. The polypeptides of this invention similar to the granulocyte-macrophage colony stimulating factor can be used for the treatment of people suffering from cancer to regenerate leukocytes after radiation or chemotherapy treatment, and to increase the blood cell count white to reduce the likelihood of viral, bacterial, fungal and parasitic infection, especially in patients with compromised immunology, such as those with AIDS.

Description

HUMAN GRANULOCYTE-MACROPHAGE COLONY STIMULATING FACTOR-LIKE POLYPEPTIDES AND PROCESSES FOR PRODUCING THEM IN HIGH YIELDS IN MICROBIAL CELLS

TECHNICAL FIELD OF THE INVENTION

This invention relates to human granulocyte- macrophase colony stimulating factor-like polypeptides (hGM-CSF), DNA sequences, recombinant DNA molecules and processes for producing hGM-CSF. More particularly, the invention relates to hGM-CSF-like polypeptides with a high specific activity, and DNA sequences, recombinant DNA molecules and processes that permit the production of hGM-CSF-like polypeptides in high yields in microbial cells.

BACKGROUND OF THE INVENTION

Granulocyte-Macrophage Colony Stimulating Factor (GM-CSF) is one of four classes of hemopoietic growth factors known as colony stimulating factors ("CSF's"). GM-CSF is a growth factor which regulates the proliferation and differentiation of multipoten- tial cells [E. R. Stanley and P. T. Jubinsky, "Factors Affecting the Growth and Differentiation of Haemo- poietic Cells in Culture," Clinical He atology, 13, pp. 329-48 (1984)]. GM-CSF has also been shown to stimulate the formation of clones of neutrophilic granulocytes and mononuclear phagocytic cells from ! -2- single bone marrow cells in vitro [A. W. Burgess et al., "Purification and Properties of a Colony Stimulating Factor from Mouse Lung-Conditioned Medium," J. Biol. Chem., 252, pp. 1998-2003 (19.77)]. Accordingly, the GM-CSF's of this invention should ^• be useful in the recovery of white blood cells after irradiation or chemotherapy. They also should acti¬ vate white blood cells to combat infections of bacteria, fungi, and parasites and to accelerate the maturation of leukemic cells and thereby stop the regeneration of leukemic cells.

Mouse GM-CSF (also known as mouse CSF-2), a 24-26,000 molecular weight glycoprotein which con¬ tains no subunits, has been purified from endotoxin- injected mouse lung-conditioned medium [A. . Burgess, supra; N. A. Nicola et al., J. Biol. Chem., 254, pp. 5290-99 (1979)]. It is unable to stimulate the formation of colonies of erythrόid, eosinophil, mega- karyocyte cells or of T-and B-lymphocytes in suitable culture conditions, which implies that it is highly selective in its proliferative effects on hemopoietic cells [A. W. Burgess and D. Metcalf, "The Nature and Action of Granulocyte-Macrophage Colony Stimulating Factors," J. Amer. Soc. Hema. , 56, pp. 947-58 (1980)]. The nucleotide sequence of a mouse lung GM-CSF cDNA is known and has been expressed in animal cells at low yield [N. M. Gough et al., "Molecular Cloning of cDNA Encoding a Murine Haematopoietic Growth Regula¬ tor, Granulocyte-Macrophage Colony Stimulating Factor, " Nature, 309, pp. 763-67 (1984); N. M. Gough et al., "Structure and Expression of the mRNA for Murine Granulocyte-Macrophage Colony Stimulating Factor, " EMBO Journal, 4, pp. 645-53 (1985)].

Human GM-CSF is also a glycoprotein (24-26,000 daltons). It has been cloned and the cDNA sequence reported [G. G. Wong et al., "Human GM-CSF: Molecular Cloning of the Complementary DNA and Purification of the Natural and Recombinant Proteins," Science, 228, pp. 810-15 (1985)]. Animal cells transfected with this cDNA sequence synthesize glycosylated GM-CSF on the order of lμg/ml [Wiesbart et al., "Human Granulocyte-Macrophage Colony-Stimu¬ lating Factor is a Neutrophil Activator," Nature, 314, pp. 361-63 (1985); Wong, supra] . Therefore, these animal cells have not been able to produce human GM-CSF in sufficient quantities and with the necessary purity for biological and clinical use.

Accordingly, processes enabling the produc¬ tion of biologically active human GM-CSFs of high purity and in clinically useful amounts are needed.

SUMMARY OF THE INVENTION The present invention solves the problems referred to above by providing DNA sequences that code for human GM-CSF-like polypeptides and by expressing those sequences in high yields in appro¬ priate microbial hosts to produce efficiently and economically large quantities of polypeptides dis¬ playing a granulocyte and macrophage colony-stimu¬ lating activity. According to this invention, it is possible to modify the amino (5') terminal end of a DNA sequence coding for hGM-CSF and thereby to pro- duce hGM-CSF-like polypeptides in high yields in microbial hosts. By virtue of this invention, it is for the first time possible to obtain polypeptides displaying GM-CSF activity in quantities large enough for clinical use. Furthermore, the present invention pro¬ vides unglycosylated hGM-CSF-like polypeptides unexpectedly having a specific activity of at least

Q

1 x 10 Units/mg. This specific activity is sub¬ stantially higher than the specific activity of native hGM-CSF or of hGM-CSF produced in yeast or animal cells in culture. I !

-4-

The polypep ides of this invention may be used either as produced or after further deriya- tization or modification, against bacterial, fungal or parasitic infections, in the regeneration of leukocytes after irradiation or chemotherapy, as well as in methods and compositions for the treat¬ ment of leukemia. These compounds may also be used to reduce the likelihood of opportunistic infections in immunologically compromised individuals, such as those suffering from AIDS. In this case GM-CSF-like polypeptides are administered to the AIDS patients to increase their white blood cell count so as to prevent opportunistic infections, thus lengthening the life and reducing the expense of treating the AIDS patient.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 depicts the nucleic acid and de¬ duced amino acid sequence for an hGM-CSF DNA as iso¬ lated from the U937 cell line. The cleavage site for the signal peptide is indicated by an arrow.

The nucleotide and amino acid differences from Mo-cell derived cDNA are indicated above and below the U937 equivalent. Nucleotide 356 is either T or C in the Mo-cell derived cDNA. Figure 2 depicts the nucleic acid sequence for a hGM-CSF cDNA as isolated from the 5637 cell line. The cleavage site for the signal peptide is inciated by an arrow.

Figure 3 depicts a restriction endonuclease map of mouse GM-CSF cDNA, wherein both the coding region, and the nick-translated probe which was used to screen the U937 hGM-CSF library are indicated.

Figure 4 depicts the construction of the E.coli expression vector pPLmuGM-CSF (p210*) for high level hGM-CSF production in microbial cells in accordance with this invention. Figure 5a and 5b depict schematic repre¬ sentation of the E.coli expression vectors (p210* and pCI857, respectively) used to produce hGM-CSF in bacterial cells through a two-plasmid system and their construction through intermediate plasmids. The synthetic region used to replace the 5' coding sequences of hGM-CSF is indicated by a hatched region.

Figure 6 depicts the nucleotide sequence and deduced amino acid sequence of the synthetic linker used for E.coli expression (Ncol-Hgal) .

Figure 7 depicts the construction of the E.coli single plasmid expression vector p241-48.

Figure 8 depicts the yeast alpha mating factor fusion to the hGM-CSF coding region. Figures 9-10 are a schematic representation of the construction of the yeast expression vector p528/l for hGM-CSF production.

DETAILED DESCRIPTION OF THE INVENTION

In order that the invention herein described may be more fully understood, the following detailed description is set forth.

In the description the following terms are employed:

Nucleotide—A monomeric unit of DNA or RNA consisting of a sugar moiety (pentose), a phosphate, and a nitrogenous heterocyclic base. The base is linked to the sugar moiety via the glycosidic carbon (I* carbon of the pentose) and that combination of base and sugar is called a nucleoside. The base characterizes the nucleotide. The four DNA bases are adenine ("A"), guanine ("G"), cytosine ("C"), and thy ine ("T"). The four RNA bases are A, G, C, and uracil ("U").

DNA Sequence—A linear array of nucleotides connected one to the other by phosphodiester bonds between the 3' and 5' carbons of adjacent pentoses. < i

-6- I

Codon—A DNA sequence of three nucleotides

(a triplet) which encodes through mRNA an amino acid, a translation start signal or a translation termina- tion signal. For example, the nucleotide triplets TTA, TTG, CTT, CTC, CTA and CTG encode for the amino acid leucine ("Leu"), TAG, TAA and TGA are transla¬ tion stop signals and ATG is a translation start signal.

Reading Frame—The grouping of codons during the translation of mRNA into amino acid sequences. During translation the proper reading frame must be maintained. For example, the DNA sequence GCTGGTTGTAAG may be expressed in three reading frames or phases, each of which affords a different amino acid sequence:

GCT GGT TGT AAG—Ala-Gly-Cys-Lys

G CTG GTT GTA AG—Leu-Val-Val

GC TGG TTG TAA G—Trp-Leu-(STOP)

Polypeptide—A linear array of amino acids connected one to the other by peptide bonds between the α-amino and carboxy groups of adjacent amino acids.

Genome— he entire DNA of a cell or a virus. It includes inter alia the structural gene coding for the polypeptides of the substance, as well as operator, promoter and ribosome binding and interac¬ tion sequences, including sequences such as the Shine- Dalgamo sequences.

Gene—A DNA sequence which encodes through its template or messenger RNA ("mRNA") a sequence of amino acids characteristic of a specific polypeptide.

Transcription—The process of producing mRNA from a gene or DNA sequence.

Translation—The process of producing-a polypeptide from mRNA. Expressio — he process undergone by a gene or DNA sequence to produce a polypeptide. It is a combination of transcription and translation. Plasmid—A nonchromosomal double-stranded DNA sequence comprising an intact "replicon" such that the plasmid is replicated in a host cell. When the plasmid is placed within a unicellular organism, the characteristics of that organism may be changed or transformed as a result of the DNA of the plasmid. For example, a plasmid carrying the gene for tetra- cycline resistance (TET ) transforms a cell previously sensitive to tetracycline into one which is resistant to it. A cell transformed by a plasmid is called a "transformant". Phage or Bacteriophage—Bacterial virus, many of which consist of DNA sequences, encapsidated in a protein envelope or coat ("capsid").

Cloning Vehicle—A plasmid, phage DNA, cosmid or other DNA sequence which is able to repli- cate in a host cell, characterized by one or a small number of endonuclease recognition sites at which such DNA sequences may be cut in a determinable fashion without attendant loss of an essential bio¬ logical function of the DNA, e.g., replication, pro- duction of coat proteins or loss of promoter or binding sites, and which contains a marker suitable for use in the identification of transformed cells, e.g., tetracycline resistance or ampicillin resist¬ ance. A cloning vehicle is often called a vector. Cloning—The process of obtaining a popu¬ lation of organisms or DNA sequences derived from one such organism or sequence by asexual reproduction.

Recombinant DNA Molecule or Hybrid DNA—A molecule consisting of segments of DNA from different genomes which have been joined end-to-end outside of living cells and able to be maintained in living cells. i -8-

Expression Control Sequence—A sequence of nucleotides that controls and regulates expression of genes when operatively linked to those genes. They include the lac system, the β-lactamase system, the trp system, the tac and trc systems, the major operator and promoter regions of phage λ, the con¬ trol region of fd coat protein, the early and late promoters of SV40, promoters derived from polyoma virus and adenovirus, etallothionine promoters, the promoter for 3-phosphoglycerate kinase or other gly- colytic enzymes, the promoters of acid phosphatase, e.g., Pho5, the promoters of the yeast c^*-mating fac¬ tors, and other sequences known to control the ex¬ pression of genes of prokaryotic or eukaryotic cells and their viruses or combinations thereof.

GM-CSF-Like Polypeptide—A polypeptide displaying a biological activity of a GM-CSF. This polypeptide may include amino acids in addition to those of a mature GM-CSF or it may not include .all of the amino aqids of mature GM-CSF. Finally, it may include an N-terminal methionine.

The present invention relates to hGM-CSF- like polypeptides having a specific activity of at

Q least 1 x 10 units/mg, and to processes for the pro- duction of those polypeptides. In addition, this invention relates to the production of large amounts of hGM-CSF-like polypeptides in microbial cells. The polypeptides of this invention are clinically useful as described previously; they also permit the produc- tion of both polyclonal and monoclonal antisera to human GM-CSF. THE EXPRESSION SYSTEMS OF THIS INVENTION

A wide variety of host/expression vehicle combinations may be employed in producing the GM-CSF-like polypeptides this invention in high^' yields. Furthermore, a wide variety of host/expres¬ sion vehicle combinations may be employed to produce the high specific activity hGM-CSF-like polypeptides of this invention. For those hosts that glycosylate the produced hGM-CSF, a deglycosylation step is, of course, required to produce the high specific acti¬ vity hGM-CSF of the present invention.

The selection of an appropriate host is controlled by a number of factors recognized by the art. These include, for example, compatibility with the chosen vector, toxicity of proteins encoded by the hybrid plasmid, ease of recovery of the desired protein, expression characteristics, bio-safety and cost. A balance of these factors must be struck with the understanding that not all host vector com- binations may be equally effective for the expression of the particular recombinant DNA molecules of this invention.

Useful expression vectors include, for example, vectors consisting of segments of chromo- somal, non-chromosomal and synthetic DNA sequences, such as various known derivatives of SV40, known bacterial plasmids, e.g., plasmids from E.coli including col El, pCRl, pBR322, pMB9 and their derivatives, wider host range plasmids, e.g., RP4, phage DNAs, e.g., the numerous derivatives of phage λ, e.g., NM 989, and other DNA phages, e.g., M13 and Filamenteous single stranded DNA phages, yeast plasmids such as the 2μ plasmid or derivatives thereof, and vectors derived from combinations of plasmids and phage DNAs, such as plasmids which have been modified to employ phage DNA or other expression control sequences. In addition, any of a wide variety of expression control sequences — sequences that con¬ trol the expression of a DNA sequence when opera- tively linked to it — may be used in these vectors to express the DNA sequence of this invention. Such useful expression control sequences, include, for example, the early and late promoters of SV40, the lac system, the trp system, the TAC or TRC system, the major operator and promoter regions of phage λ, the control regions of fd coat protein, the promoter for 3-phosphoglycerate kinase or other glycolytic enzymes, the promoters of acid phosphatase, e.g., Pho5, the promoters of the yeast c^*-mating factors, and other sequences known to control the expression of genes of prokaryotic or eukaryotic cells or their viruses, and various combinations thereof.

The preferred expression vectors and con¬ trol sequences include the P.. promoter, the promoter of the yeast α-mating factor, and the yeast actin promoter.

A wide variety of host cells are also useful in producing the hGM-CSF-like polypeptides of this invention. These hosts may include well known eukaryotic and prokaryotic hosts, such as strains of E.coli, Pseudomonas, Bacillus, Streptomyces, fungi such as yeasts, and animal cells, such as CHO cells, African green monkey cells, such as COS1, COS7, BSC1, BSC40, and BMT10, and human cells and plant cells in tissue culture. Useful hosts may include strains of E.coli, such as E.coli W3110I^, E.coli JA221, E.coli C600, E.coli ED8767, E.coli DHl, E.coli LE392, E.coli HBlOl, E.coli X1776, E.coli X2282, E.coli MRCI, and strains of Pseudomonas, Bacillus, and Streptomyces, yeasts and other fungi, plant cells in culture or other hosts. The preferred hosts of this invention include E.coli K₁₂, E.coli strains SG927 [ATCC 39627], SG928 [ATCC 39628], SG935 [ATCC 39623] and SG936 [ATCC 39624] as well as its derivative, A89 [DSM 3869]., and yeast- strain Saccharo yces cerevisiae BJ1991.

THE hGM-CSF-LIKE POLYPEPTIDES OF THIS INVENTION

It should be understood that the hGM-CSF-like polypeptides (prepared in accordance with this inven- tion in those hosts) may include polypeptides in the form of fused proteins (e.g., linked to a prokaryotic, eukaryotic or combination N-terminal segment to direct excretion, improve stability, improve purification or improve possible cleavage of the N-terminal seg- ment), in the form of a precursor of GM-CSF-like polypeptides (e.g., starting with all or parts of a GM-CSF-like polypeptide signal sequence or other eukaryotic or prokaryotic signal sequences), in the form of a mature GM-CSF-like polypeptide, or in the form of a met-GM-CSF-like polypeptide.

One particularly useful form of a polypep¬ tide in accordance with this invention, or at least a precursor thereof, is a mature GM-CSF-like poly¬ peptide with an easily cleaved amino acid or series of amino .acids attached to the amino teiπαinus. Such construction allows synthesis of the protein in an appropriate host, where a translation start signal that may not be present in the desired GM-CSF is needed, and then cleavage in vivo or in vitro of the extra amino acids to produce the desired GM-CSF-like polypeptides. Such methods exist in the art.

The polypeptides of the invention also include hGM-CSF-like polypeptides that are coded for on expression by DNA sequences characterized by dif- ferent codons for some or all of the codons of the present DNA sequences. These substituted codons may code for amino acids identical to those coded for by the codons replaced but result in higher yield of the polypeptide. Alternatively, the replacement of one or a combination of codons leading to amino acid replacement or to a longer or shorter GM-CSF-like polypeptide may alter its properties in a useful way (e.g., increase the stability, increase the solu¬ bility or increase the therapeutic activity) . It should be understood that these polypeptides are also part of this invention. Finally, the most preferred hGM-CSFs of this invention are those characterized by a specific activity of at least 1 x 10 Units/mg. These poly¬ peptides may be produced directly in hosts that do not glycosylate their polypeptides, e.g., bacterial hosts such as E.coli. They may also be produced by deglycosylating polypeptides produced in hosts that glycosylate their polypeptides, e.g., yeasts and animal cells. Such deglycosylation may be accom¬ plished either in vitro or in vivo using conven- tional methods and compositions well known in the art. It may also be accomplished by inhibiting glycosylation during protein production using con¬ ventional agents and methods.

The GM-CSF-like polypeptides of the pre- sent invention may be purified by a variety of con¬ vention steps and strategies. These methods are know in the art.

COMPOSITIONS AND METHODS OF USING THE hGM-CSF-LIKE POLYPEPTIDES OF THIS INVENTION

While the GM-CSF-like polypeptides of this invention may be a^'dministered in compositions and methods of treatment in the form in which they are produced, it should also be understood that they may be formulated using known methods to prepare pharma¬ ceutically useful compositions. Such compositions also will preferably include conventional pharma- ceutically acceptable carriers and may include other medicinal agents, carriers, adjuvants, excipients, etc., e.g., human serum albumin or plasma prepara¬ tions. See, e.g., Remington's Pharmaceutical Sciences (E. W. Martin). The resulting formulations will contain an amount of hGM-CSF-like polypeptides effective in the recipient to stimulate the colony formation of granulocytes and macrophages. Admini¬ stration of these polypeptides, or pharmaceutically acceptable derivatives thereof, may be via any of the conventional accepted modes of administration of GM-CSF. These include parenteral, subcutaneous, or intravenous administration.

The GM-CSF-like polypeptides of this invention are particularly useful in compositions and methods for increasing the white blood cell count of immunologically-compromised patients so as to reduce the risk of infection in those patients. For example, the compositions and methods of this invention are useful in the therapy of AIDS patients- to prevent the occurrence of opportunistic infections which often shorten the life of the AIDS patient and add to the expense of their treatmen .^*

The dosage and dose rate will depend on a variety of factors for example, whether the treatment is given to a cancer patient after radiation therapy or to an AIDS victim to prevent opportunistic infec¬ tion. However, the dosage will likely be between 1 and 10 μg per day or between 10 and 100 μg per week, if the patient is to be treated steadily over a long period of time.

In order that this invention may be better understood, the following examples are set forth. These examples are for purposes of illustration only and are not to be construed as limiting the scope of the invention. EXAMPLE 1

CONSTRUCTION AND SCREENING OF A HUMAN cDNA LIBRARY

We describe below the construction of a human cDNA library from poly (A) mRNA isolated from human acrophage cell line U937 in λgtlO and the amplification of that library in E.coli C600 hfl cells.

A. EXTRACTION OF RNA FROM HUMAN U937 CELLS

We induced human macrophage U937 cells in culture with dexa ethasone (10^" M) and phorbol ester

_7 (10 M) and resuspended the pellets containing 1.2 x

9 10 cells in 48 ml lysis buffer (0.2 M Tris-HCl (pH 8.0), O.lM LiCl, 25 mM EDTA, 1% SDS) plus 5 mM vanadyl complex (Bethesda Research Labs) by vortexing. We lysed the cells by addition of 24 ml phenol and vor- texed for 5 min. We added 24 ml chloroform to the lysis mixture which was then shaken for 10 min. We separated the organic and aqueous phases by centri- fugation in a clinical centrifuge at room temperature for 10 min. We reextracted the aqueous phase two times with phenol:chloroform (1:1), then two times with chloroform only. We next ethanol-precipitated the nucleic acids in 0.3 M sodium acetate at -20°C overnight and pelleted the nucleic acid at 14k rpm in a Sorvall RC2B centrifuge (SS34 rotor) at 4°C for 20 min. We resuspended the pellets in 5 ml of 0.3 M sodium acetate buffer, and ethanol-precipitated the nucleic acid again as described above. We resuspended the final pellet in 300 μl H 0 and stored it at -20°C. Our RNA preparation was then enriched for poly(A) RNA by passage over an oligo(dT)-cellulose column (PL Biochem) . -15-

B. CONSTRUCTION OF A U937 cDNA-λqtlO LIBRARY

1. cDNA Synthesis

We synthesized cDNA from 20 μg poly (A) mRNA isolated as described above. We diluted the poly (A) mRNA to 500 μg/ml in H₂0, heated it to 65°C for 3 min, quick cooled it in a dry ice-propanol bath and then thawed it. The RNA was then added to a reaction mixture composed of 0.1 M Tris-HCl (pH 8.3) at 42°C, 0.01 M MgCl₂, 0.01 M DTT, 1 mM dCTP, 1 mM dGTP, 1 mM dTTP, 0.5 mM dATP and 100 μCi α-³²P-ATP (3000 Ci/mmole, Amersham or New England Nuclear), 20 μg oligo (^dτ) ₂_i8 (**^{?L B}i°^cnem)/ 0.03 M β-mercapto- ethanol, 5 mM Vanadyl Ribonucleoside Complex (Bethesda Research Labs), 169 U AMV Reverse Transcriptase

'(Seikagaku America). Final volume of the reaction mixture was 200 μl. We incubated this mixture for 2 min at room temperature and 6 h at 44°C. We termi¬ nated the reaction by addition of 1/10 vol 0.5 M Na₂EDTA (pH 8.0).

We adjusted the reaction mixture to 0.15 M NaOH and incubated the mixture at 37°C for 12 h fol¬ lowed by neutralization with 1/10 vol 1 M Tris-HCl (pH 8.0) and HCl. This was extracted with phenol: chloroform saturated TE buffer (10 mM Tris-HCl (pH 7.0) and 1 mM Na₂EDTA) . The aqueous phase was chromatographed through a 5 ml sterile plastic pipet containing a 7 x 29 cm bed of Sephadex G150 in 0.01 M (pH 7.4), 0.4 M NaCl, 0.01 M Na₂EDTA, 0.05% SDS. We pooled the front peak minus tail and precipitated the cDNA with 2.5 vol 95% ethanol at -20°C. This reaction yielded 1 μg of single-stranded cDNA.

2. Double Strand Synthesis

We resuspended the single-stranded cDNA in 200 μl (final vol) 0.1 M Hepes (pH 6.9), 0.01 M MgCl₂, 0.0025 M DTT, 0.07 M KC1, 1 mM dXTPs and 75 U Klenow fragment DNA polymerase 1 (Boehringer-Mannheim) and incubated the reaction mixture at 14°C for 21 h. Reaction was terminated by addition of Na₂EDTA (pH 8.0) to 0.0125 M, the mixture extracted with .. phenol:chloroform, as in the first cDNA step, and the aqueous phase chromatographed on a G150 column in 0.01 M Tris-HCl (pH 7.4), 0.1 M NaCl, 0.01 M Na₂EDTA, 0.05% SDS. We again pooled the radioactive peak minus the tail and ethanol-precipitated the DNA.

We then incubated the DNA obtained with 42 U reverse transcriptase in 50 μl 0.1 M Tris-HCl (pH 8.3), 0.01 M MgCl₂, 0.01 M DTT, 0.1 M KCl, 1 mM dXTPS, 0.03 M β-mercaptoethanol for 1 h at 37°C to complete double-strand synthesis. The reaction was terminated and processed as described above.

We cleaved the hairpin loop formed during double strand synthesis as follows: We redissolved the pellet in 50 μl 0.03 M sodium acetate (pH 4.5), 0.3 M NaCl, 0.003 M ZnCl₂ and treated it with 100

U S, nuclease (Sigma) for 30 min at room temperature. Reaction was terminated by addition of EDTA and pro¬ cessed as described above. The yield after S., treat¬ ment was 900 ng dsDNA. To assure blunt ends following S- nuclease digestion, we treated the DNA with Klenow in 0.01 M Tris-HCl (pH 7.4), 0.01 M gCl₂, 1 mM DTT, 0.05 M NaCl, 80 μM dXTP and 12.5 U Klenow in 60 μl for 90 min at 14°C, extracted with 50:50 phenol:chloroform, and chromatographed the DNA on a G50 spin column (1 ml syringe) in 0.01 M Tris-HCl (pH 7.4), 0.1 M NaCl, 0.01 M EDTA, 0.05% SDS.

We next methylated the dsDNA in order to avoid fragmentation under subsequent EcoRI digestion. We treated the DNA with EcoRI methylase in 30 μl final vol 0.1 M Tris-HCl (pH 8.0), 0.01 M Na₂EDTA, 24 μg BSA, 0.005 M DTT, 30 μM S-adenosylmethionine and 5 U EcoRI Methylase for 20 min at 37°C. The reaction mixture was heated to 70°C for 10 min, cooled, extracted with 50:50 phenol: chloroform and chromatographed on a G50 spin column as described above. We ligated our methylated ds cDNA to phos- phorylated EcoRI linkers (New England Biolabs) using the following conditions: 0.05 M Tris-HCl (pH 7.8), 0.01 M MgCl₂, 0.02 M DTT, 1 mM ATP, 50 μg/ml BSA, 0.5 μg linker, 300 U T4 DNA ligase in 7.5 μl final volume for 32 h at 14°C. We adjusted the reaction mixture to 0.1 M Tris-HCl (pH 7.5), 0.05 M NaCl, 5 mM MgCl₂, 100 μg/ml BSA and added 125 U EcoRI (New England Biolabs), incubated the mixture for 2 h at 37°C, extracted with 50:50 phenol: chloroform and chromatographed the DNA on a G50 spin column as described earlier.

We redissolved the cDNA in 100 μl 0.01 M Tris-HCl (pH 7.5), 0.1 M NaCl, 1 mM EDTA and chromato¬ graphed it on a 1 x 50 cm Biogel A50 (BIORAD) column which had been extensively washed in the same buffer (to remove ligation inhibitors). Aliquots of various fractions were run on a 1% agarose gel in TBE buffer (0.089 M Tris-HCl, 0.089 M boric acid and 2.5 mM Na₂EDTA), dried and exposed at -70°C overnight. We pooled all fractions that were larger than 500 base pairs and ethanol-precipitated the DNA in those frac¬ tions for cloning into an EcoRI-cut. λgtlO cloning vector. The size fractionation column yielded 126 ng of cDNA, average size approximately 1500 bp.

3. Library Construction

We incubated 5 μg EcoRI-cut λgtlO with 20 ng cDNA and T4 DNA ligase buffer at 42°C for 15 min to anneal cos sites, followed by centrifugation for 5 sec in an Eppendorf centrifuge and addition of ATP to 1 mM and 2400 U T4 DNA ligase (New- England Biolabs) in a final vol of 50 μl. [See Huynh, Young and Davis, "Constructing And Screening cDNA Libraries in λgtlO I

-18-

And λgtll", in DNA Cloning: A Practical Approach (D. Glover, ed.), IRL Press (Oxford 1984).] The ligation mixture was incubated at 14°C overnight. We packaged the λgtlO cDNA ligation mixture into phage particles using an Amersham packaging mix

[Amersha packaging protocol] and diluted with 0.5 ml SM buffer (100 mM NaCl, 10 mM MgS0₄, 50 mM Tris-HCl (pH 7.5) and 0.01% gelatin).

We next infected E.coli C600 hfl cells with these phage particles to form a cDNA library of 1 x 10 7 i.ndependent recombmants [see T. Maniatis et al.,

Molecular Cloning, p. 235 (Cold Spring Harbor, 1982)].

For plating and amplification of the library,

1 ml of cells plus 250 μl packaging mix was incubated at room temperature for 15 min, diluted to 50 ml in

LB plus MgS0₄ top agarose at 50°C and plated on LB Mg Nunc plates. This represented a plaque density of

2 x 10 plate. The plates were incubated at 37°C for approximately 8 h until plaques were nearly touching. We flooded the plates with 50 ml of cold

SM buffer (0.01 M Tris-HCl (pH 7.5), 0.01 M MgCl₂, 0.1 mM Na₂EDTA) and eluted on a gyro-rotary shaker overnight at 4°C. We pooled the eluants into 250 ml ' bottles and spun at 6k for 10 min in a Sorvall GSA rotor. We treated the supernatants with an equal volume of cold 20% PEG 4000-2 M NaCl in ice for 3 h and pelleted the phages by centrifugation at 4k for 30 min in an H4000 rotor in an RC-3B Sorvall centri¬ fuge. The phage pellets were thoroughly drained, resuspended in 60 ml SM, and spun at 10,000 rpm in a SS34 rotor to remove debris. The supernatants were adjusted to 3.5 M CsCl by addition of 7 g CsCl to 10 ml supernatant. We obtained phage bands by cen¬ trifugation in a 70.1 Beckman rotor at 50,000 rpm for 18 h at 15°C. We pooled the phage bands and stored them at 4°C for library stock. The titer obtained was 2.2 x 10 PFU/ml. ! -19-

4. Screening Of The Human GM-CSF cDNA Library With Murine GM-CSF cDNA

We screened the above-described human cDNA library with a fragment of murine GM-CSF cDNA sequence. The basis for our approach was that murine GM-CSF has some amino acid similarity to human GM-CSF [N. M. Gough, supra, p. 2; R. H. Weisbart, supra, p. 3]. Accordingly, we postulated that mouse GM-CSF cDNA might cross hybridize to human GM-CSF cDNA to an extent sufficient to allow selection of a human GM-CSF related cDNA from our human cDNA library.

5. Preparation of Murine GM-CSF probe

We obtained our murine GM-CSF cDNA probe for screening our human cDNA library, as follows: We generated a cDNA library from mRNA isolated from EL-4 mouse thymoma cells [ATCC TIB 39] induced with phorbol-12-myristate-13-acetate (PMA). We then reverse transcribed mRNA into cDNA using the Okayama- Berg protocol [H. Okayama and P. Berg, "High-Effi- ciency Cloning of Full-Length cDNA," Mol. Cell Biol. , 2, pp. 161-70 (1982)]. We inserted the cDNA into the plasmid vector pHG327, a modified pKCR vector [K. O'Hare et al., "Transformation of Mouse Fibro- blasts to Methotrexate Resistance by a Recombinant Plasmid Expressing a Prokaryotic Dihydrofolate

Reductase," Proc. Natl. Acad. Sci. USA, 78, pp. 15, 27-31 (1981)], which has a unique SstI site for cDNA cloning and two flanking BamHI sites to allow con¬ venient excision of the inserted cDNA sequence The resultant library (a gift of W. Boll and C. Weissmann) consisted of approximately 2 X 10 individual cDNA molecules. We screened this mouse cDNA library by hybridization with two oligomer DNA probes synthesized on the basis of the published cDNA sequence for mouse lung cell derived GM-CSF [N. M. Gough et al., "Mole¬ cular Cloning of a cDNA encoding a Murine Haemato¬ poietic Growth Regulator Granulocyte-Macrophage Colony I

-20-

Stimulating Factor", Nature, 309, pp. 763-67 (1984)] using the solid-phase phosphotriester method [H. Ito et al., "Solid Phase Synthesis of Polynucleotides: VI. Further Studies on Polystyrene Copolymers .For the Solid Support," Nucl. Acids Res., 10, pp. 1755-69 (1982).

These probes had the following sequences:

— GMCSF-2 probe CCAACTCCGGAAACGGACTG

— GMCSF-1 probe CTTAAAACCTTTCTGACTG 0.02% of the 2 x 10 cDNA molecules scored positive.

We determined the sequence of the longest positive insert using conventional methods. For example, either the chain termination method [F. Sanger et al., Proc. Natl. Acad. Sci. USA, 74, pp. 5463-67 (1977); J. Messing and J. Vieira, "A New Pair of M13 Vectors For Selecting Either DNA Strand of Double-Digest Restriction Fragments," Gene, 19, pp. 269-76 (1982)] or by chemical degradation of the DNA chain [A. M. Maxam and W. Gilbert, Methods in Enzymology, 65, pp. 499-560 (1980)] can be used. We found that the DNA sequence was similar to that of GM-CSF from mouse lung cells. Over the coding region there were only three base changes (at positions 186, 213, and 482). However, only the base change at position 482 generates an amino acid substitution, yielding valine instead of glycine. These changes represent one of the two variants reported for different isolates of mGM-CSF [(N. M. Gough et al., "Structure and Expression of the mRNA for Murine Granulocyte-Macrophage Colony Stimulating Factor", EMBO J., 4, pp. 645-53 (1985)].

6. Screening of the Human cDNA library

We screened our human cDNA library for human GM-CSF sequences, using a labeled probe which consisted of a 365 base pair Ncil-Hinfl fragment of our murine GM-CSF (Figure 3). The Ncil-Hinf fragment was purified on a 5% polyacrylamide gel and radio- labled by nick-translation. For screening, we used the plaque hybridization technique of Benton and Davis [W. D. Benton and R. W. Davis, "Screening lambda gt recombinant clones by hybridization to a single plaque in situ", Science, 196, p. 180 (1977)].

To prepare our library for screening with this DNA probe, we prepared an overnight culture of C600 hfl cells in L broth and 0.2% maltose and pelleted and resuspended the cells in an equal volume of SM buffer. We then pre-adsorbed 0.9 ml of cells with 2 x 10 phage particles at room temperature for 15 min. We diluted the suspension to 50 ml in LB plus 10 mM MgS0₄ and 0.7% agarose at 55°C and plated it on LB Mg Nunc plates. We screened 10 such plates. We incubated the plates at 37°C for approximately 8 h until plaques were nearly touching. We then chilled the plates at 4°C for 1 h to allow the agarose to harden. We transferred the λgtlO phage particles from the plaque library plates to nitrocellulose. We placed the filters onto the plates con¬ taining the recombinant plaques for varying times from 30 seconds to 5 minutes depending on the number of different filters used, and then lifted and incubated the filters with the phage-containing side up on LB + 10 mM MgS0₄ plates at 37°C for 5 h.

These filters were then lysed by placing them onto a pool of 0.5 N NaOH for 5 min, then neu¬ tralized on 1 M Tris-HCl (pH 7.0), submerged into 1 M Tris-HCl (pH 7.0) and then heated at 80°C for 2 hours.

We hybridized the filters to the radiolabled 365 base pair Ncil-Hinfl fragment cDNA probe in [6 x SSC] 0.2% polyvinyl-pyrrolidone (M.W. 40,000), 0.2% ficoll (M.W. 40,000), 0.2% bovine serum albumin, 0.05 M Tris-HCl (pH 7.5), 1 M sodium chloride, 0.1% sodium pyrophosphate, 0.1% SDS, 10% dextran sulfate (M.W. 500,000) and denatured salmon sperm DNA (__ 100 μg/ml) at 65°C. They were then washed in a I -22- similar buffer (2 x SSC, 0.1% SDS) at 65°C. We detected hybridizing cDNA sequences by autoradiog- raphy. By means of this technique, we screened 1 x 10 phage and we picked 20 positive plaques. .. These positive cDNAs were further purified and characterized. The largest cloned cDNA was 768 nucleotides in length. We sequenced this cDNA insert using conventional methods. Its sequence is depicted in Figure 1. The longest open reading frame found in this cDNA (nucleotides 7-438) encodes a protein of 144 amino acids which is identical over the coding region to that reported for Mo-cell derived hGM-CSF with the exception of a single base (297). The first 17 residues comprise a series of hydrophobic amino acids consistent with their role as putative a signal sequence for the remaining protein. The cleavage site between serine (the terminal amino acid of the putative signal sequence) and alanine (the first amino acid of the coding sequence) as depicted in Figure 1, is also in agreement with the reported first amino acid of the mature protein (alanine). The A to G substitution at position 297 results in a codon for- an isoleucine instead of the reported methionine. We also constructed a second cDNA library in λgtlO from poly(A) RNA of the human bladder car¬ cinoma cell line 5637 (ATCC HTB9) substantially as described previously. We screened this library, of 10 recombinant phage for hGM-CSF sequences according to the plaque hybridization technique described above using a 240 nucleotide fragment (Pstl-Apal) from the coding region of the hGM-CSF cDNA of Figure 1 as a probe. Positive signals were found at a frequency of one in 500 plaques (0.02%). The largest of these cDNAs was 911 nucleotides in length. We sequenced this DNA insert using conventional methods. Its sequence is depicted in Figure 2. The coding sequence of the 5637-derived cDNA is identical to that reported for the Mo-deriyed cDNA. The 5637-derived cDNA con¬ tains additional non-coding sequences at both the 5' and 3' ends not found in either the U937 or Mo-cell derived cDNA's.

EXAMPLE 2

EXPRESSION OF hGM-CSF

A. In E.coli

Referring now to Figures 4, 5a and 5b, we have shown therein a schematic outline of one embodi- ent of a process for preparing a recombinant DNA molecule (pPLmuGM-CSF) characterized in that it pro¬ duces hGM-CSF in high yield. It has a U937-derived DNA sequence coding for human GM-CSF, which had been modified at 5' end of its coding region to minimize any potential disadvantageous RNA secondary structure, fused to a DNA sequence derived from mu and carrying a Shine Dalgarno sequence from mu, the combined DNA sequence being operatively-linked to a PL promoter dervied from bacteriophage λ. To construct expression vector pPLmuGM-CSF, we first prepared a synthetic oligonucleotide DNA sequence or linker to replace the coding sequence between the first alanine codon and a unique Hga I site at nucleotide 120 of our hGM-CSF cDNA (Figure 1). This Ncol-Hgal linker (Figure 6) was constructed to reduce the potential of mRNA secondary structure which might make the ribosomal binding site inacces¬ sible. In this linker we substituted adenosine (A) for the naturally occurring nucleotides wherever the degeneracy of the genetic code allowed the retention of the same amino acid sequence. [G. N. Buell et al., "Optimizing the Expression in E.Coli of a Synthetic Gene Encoding Somatomedin-C (IGF-I)," Nucl. Acids Res. , 13, pp. 1923-38 (1985)]. These base substi- tutions resulted in a change in the potential free energy due to mRNA secondary structure of 15 kcal (ΔG = -28.3 to ΔG = -13.3 kcal) increasing the free energy beyond that predicted to form stable stem and loop structures. We also used our linker to add. an ATG start codon directly in front of the N-terminal alanine of our coding sequence.

Referring now to Figure 5a, we depict therein the construction of vector p210* through various intermediates. To carry out the above de- scribed DNA sequence modification we subcloned the

U937-derived hGM-CSF cDNA of Figure 1 into pUC-8 [J. Viera and J. R. Messing, Gene, 19, pp. 259-268 (1982)]. The resulting vector, which we designated pUC8 GM-CSF was cut with Hgal and HindiII. We ligated the result- ing small fragment together with our synthetic Ncol-Hgal linker. We then cut our pPLmu vector [E. Remaut et al., Gene, 15, pp. 81-93 (1981); G. Gray et al., Gene, 32, in press] with Ncol and Hindlll and inserted our fragment therein. We then removed the 3' non-coding sequences between the Bal I site and Sma I site. We designated this vector pPLmu:hGM-CSF (p210*). We similarly prepared a second vector which contained the 5637-derived hGM-CSF of Figure 2. This vector was designated (p210*-5637). We tested the ability of plasmid p210* to direct the synthesis of hGM-CSF in an in vitro transcription-translation system. For comparison, we assayed the same expression vector carrying the native cDNA sequence deleted to the alanine codon by Bal 31 digestion (p208 Bal31). pPLMU:hGM-CSF (p210*) yielded significantly more hGM-CSF than p208 Bal31, proving that our synthetic linker and its inhibition of mRNA secondary structure markedly improved the production of hGM-CSF-like polypeptide. 1. E.coli Host Selection and Fermentation

We introduced the expression plasmid p210* into E.coli strain C600 which also carried a second plasmid encoding the thermosensitive respressor

(CI_g57) of the λ PL promoter [E. Remaut et al. Gene, 22, pp. 103-13 (1983)] (Figure 5b). We could have also used strains in which the repressor is part of the chromosome (see description, infra). Transcrip- tion was regulated by growing the cells at 28°C, a permissive temperature for repressor activity, or at 42°C, non-permissive for repressor activity. Using this E.coli C600 host strain, we were unable to produce sufficient quantities of hGM-CSF to be visible on SDS-PAGE analysis.

Subsequently we tested E.coli strain SG936 [lac (am), trp (am), pho (am), sup C (ts), rpsl, mal (am), htpR (am), tsx:TN10, Ion R9] [ATCC 39624] which is an htpR Ion mutant. This mutant is deficient -in its production of Ion protease. As a result, this strain, as well as Ion mutant strains SG935 [ATCC 39623], SG927 [ATCC 39627], and SG928 [ATCC 39628], exhibits a reduced capacity to degrade foreign pro¬ teins upon their accumulation within the cell or at high temperatures. [S. A. Goff et al., "Heat Shock Regulatory Gene LtpR Influences Rates of Protein Degradation and Expression of the Ion gene in Escherichia coli," Proc. Natl. Acad. Sci. USA, 81, pp. 6647-6651 (1984).] We transformed E.coli strain SG936 with plasmid 210*, using standard induction protocol. We examined the ability of this strain to produce hGM-CSF under controlled fermentation conditions: we grew the transformed E.coli SG936 at 28°C on L-Broth with 50 mg/ml Ampicillin and 200 mg/ml Kanamycin (10%) to an optical density (650) of approximately 30. After ^• i

-26-

I heat shock at 42°C, and the addition of extra medium, we grew the cells for three hours and harvested them.

We analyzed total cell aliquots of the fermentation mixture for hGM-CSF production by -SDS- ^■ 5 polyacrylamide gel electrophoresis at various intervals: inoculation, 4 hours after inoculation, 1 hour after induction at high temperature (42°C) and 3 hours after induction. A strong protein band appeared 1 hour after induction. The induced protein

10 (with an apparent molecular weight of 14,500) accorded well with the expected 14,574 molecular weight of our expected recombinant hGM-CSF protein. We per¬ formed a densitometer scan of a portion of the stained gel which represented the hGM-CSF production of an

15 aliquot taken after harvesting. This scan indicated that 8-9% of the total protein was hGM-CSF.

We also examined other host strains for their ability to produce and accumulate hGM-CSF. ^ e compared wild type E.coli strains C and B with mutant

20 E.coli strain SG936, because the mutant strain grows poorly. We found only B to be positive for hGM-CSF. This accumulation of hGM-CSF in E.coli B accords with the known low levels of Ion protease in that E.coli strain. Thus, our high yields are improved

25 in Ion protease mutants which do not break down the recombinant protein once it is produced.

2. Single Plasmid Vector

In a more preferred embodiment, we con¬ structed a single plasmid vector for the production

30 of human GM-CSF. This single plasmid was more advan¬ tageous than the above-described two-plasmid system for synthesis of hGM-CSF in E.coli for several reasons. First, the two separate plasmids required two different antibiotic selections for their main-

35 tenance in .the host cell. One plasmid (210*) employed ampicillinase as its resistance marker. Unfortunately, the a picillin required in its growth medium is an undesirable element in fermentation for a human pharmaceutical product. In addition, coordinate growth of the two plasmids was not certain. Thus, the quantity of repressor encoded by one plasmid

(pcI857) could not be certain to match the number of promoter copies present on the second plasmid (210*). For these and other reasons we constructed a single-plasmid system which would contain all the elements previously found on the two plasmids. Those elements included the leftward transcriptional pro¬ moter of phase Lambda (PL) followed by: the ribosome binding site for the ner-1 gene of phage Mu, the altered sequence encoding the mature human GM-CSF protein (preceded by a methionine codon to initiate translation), the transcriptional terminator from phage T4 gene 32, the tetracycline resistance gene from plasmid pBR322, the .origin of replication and deletion mutant from plasmid pAT153, phage Lambda cl gene mutant 857 encoding the thermolabile repressor of PL. We combined these elements from previously described plasmids as shown in Figure 7.

In order to conduct this single plasmid vector we conducted a three-fragment ligation: the first fragment, which contained the cI857 thermolabite repressor, was isolated from plasmid 153-PL-T4-hTNFCA3 cts [DSM 3460] after subjecting it to digestion by EcoRI and Sail; the second frag¬ ment was isolated from plsmid T4 CA5 Dra 6 Hinlll; after subjecting it to digestion by Sail and HindiII and isolating the smaller fragment; the final fragment which contained the hGM-CSF cDNA, was isolated from p 210* after EcoRI and HindiII digestion. Single plasmid vector p 241-8 resulted. Another significant element in our construc¬ tion of the single plasmid system was the introduction of a transcriptional terminator. Our data indicate I

-28- that this fragment is useful for stabilizing the single plasmid vector.

The bacterial host strain (SG936) in the high expression system described above was also..

I altered to allow use of the new single plasmid vector. We removed the tetracycline resistance already present in SG936. Selection of the bacteria on fusaric acid for sensitivity to tetracycline yielded a new strain: A89. When the single plasmid vector 241-8 was introduced into strain A89 and production of hGM-CSF was initiated by heat induction, hGM-CSF comprised 8-10% of the total cell protein.

3. Purification of E.coli Produced hGM-CSF To purify the hGM-CSF we had produced, we lysed the recombinant-protein-bearing E.coli cells using the French press. We centrifuged and then washed the pellet, which included the dense inclusion bodies containing recombinant-hGM-CSF, with a solu- bilizing buffer of 0.75 guanidinium hydrochloride, 1% tween 40, 50mM EDTA, O.lM tris HC1 (pH 7.5) to remove - soluble contaminants. We extracted the recombinant- hGM-CSF from the pellet using a solution buffer of 5mM KH₂ P0₄ and 6M urea. We further purified our hGM-CSF from E.coli by gel filtration and traced the eluent of the G-100 sephadex column by optical density. To do this, we applied the washed pellet to the column. The sample was chromatographed and we monitored the eluent at O.D. 280. Three peaks were observed. We then analysed our sample containing the recombinant hGM-CSF produced in E.coli by SDS-PAGE, stained by coomassie- blue to visualize the results, and found that the last peak yielded a single band of the expected molecular weight.

We also analyzed this last peak using reverse phase HPLC analysis,, column: Brownlee RP 30, 2.1 x 200 mm C8. This further confirmed that the third G-100 sephadex column peak showed essentially a single protein peak. The other two peaks produced by the gel filtration chromatography represente contaminants and a di er molecule, respectively.

4. Biological Activity Assays

We assayed samples of lysed E.coli cells which carried the hGM-CSF expression plasmid described above, and were induced for recombinant protein pro- duction for biological activity using the bone marrow clonal assay and the CML assay.

We first assayed biological activity on human bone marrow cells. Normal human bone marrow was fractionated in Ficoll by density and depleted of macrophages by absorption to plastic [B. Pike and

W. A. Robinson, J. Cell. Physiol, 76, pp. 77-84 (1970)]. We then grew the bone marrow cells (50,000 cells) in serum-free methylcellulose medium or agarose medium in the presence of 10% calf serum [J. F. Eliason and N. Odartchenko, "Colony Formation By Primitive Hemo¬ poietic Progenitor cells in Serum-Free Medium", Proc. Natl. Acad. Sci. USA, 82, pp. 775-79 (1985)]. We then compared the effect of a reference GM-CSF, a partially purified hGM-CSF supplied by Chugai, and our recombinant hGM-CSF, by growing the bone marrow cells in the presence of the respective CSFs at dif¬ ferent dilutions. For each, duplicate one milliliter aliquots of the cell suspension were plated in 35 mm diameter Petri dishes and incubated with the respec- tive CSFs at 37°C in a fully humidified atmosphere of 5% C0₂, 5% 0₂ and 90% N . After 7 days and after 14 days, we counted the colonies. We typed them as granulocyte, macrophage or mixed, based on the size of the cells and the colony morphology. In some cases we also stained with Geimsa to confirm clas¬ sification of colonies. This assay demonstrated a concentration dependent stimulation of colony formu¬ lation on human bone marrow cells with both the native and recombinant material. Accordingly, the recombinant hGM-CSF is active without glycosylation. We also performed a CML assay to quantify hGM-CSF activity. This assay measures 3H-thymidme uptake on granulocytes which have been purified from the peripheral blood of patients with chronic myeloge- nous leukemia (CML). The cell preparation contained 7% blood cells, which proliferated in response to hGM-CSF [J. D. Griffin et al., Blood, 63, pp. 904-11 (1984)]. The cells (10 /ml) were incubated for 48 hours with varying concentrations of GM-CSF and proliferation was measured by a 6 hour incorporation

3 . . of H-thymidme (lOμ Ci/Ml, 200 Ci/nmole). One unit

3 . . . per ml induces 50% of maximal H-thymidme incorporation of CML cells. Such uptake in the presence of GM-CSF is directly proportional to the incubation time and the number of cells. We assayed the -activity of E.coli produced hGM-CSF from: (1) an induced culture of SG936 cells carrying p 210* and pcI857; (2) an uninduced culture of SG936 cells carrying p 210* and pcI857; and (3) an induced culture of SG936 cells carrying pPLmu and pcI857. CML cells incubated with samples (2) and (3) incorporated only background levels of

3H-t pi.dm. e. CML cells incubated with sample (1) showed a dilution-dependent stimulation of

3 . . . H-thymidme incorporation, confirming the presence of active GM-CSF.

5. Preferred Method of Purification of E.coli Produced hGM-CSF

To reconstitute our host cells, we sus¬ pended E.coli cells (lOOg) in 5 volumes of buffer I (lOOmM sodium phosphate; 5mM benzamidine-HCl, 5mM EDTA; 0.5 mM phenylmethy1sulfonyl fluoride (solu- -31- bilized in ethanol, 10% of w/v); 25% sucrose; 0.17% (w/w of cells) lysozyme; pH=7.0). After the cells were sonicated for 2 minutes in an ice water bath, we incubated them at 22°C for one hour. The mixture was then cooled to 4°C and passed through a french pressure cell twice at 16,000 psi. We then centri- gued the crude homogenate at 10,400 xg (GSA rotor, 40 min) and discarded the supernatant. The pellet was sonicated for 90 seconds in buffer II (buffer I, except with 0.75 M guanidium-HCl; 1% (w/v) Tween 40, and without sucrose and lysozyme). We then centri- fuged as above for 30 minutes. The resulting pellet was dissolved in 85 ml of buffer III (lOOmM sodium phosphate; 3M gu-HCl; lOmM 2-mercaptoethanol; lmM EDTA; pH=7.0) . We centrifuged this solution (30 min) as described above and subjected the supernatant to ultracentrifugation for one hour at 40,000 rpm (Beck an Ti-45 rotor). The supernatant from the ultracentrifugation was applied to a column of Sephacryl S-200 (5.0 x 87 cm) precquilibrated with buffer III. The flow rate of the gel filtration chromotagraphy was maintained at 70 ml/h and we collected 17 ml fractions. We monitored the frac- tionation of proteins by A_28Q nm and SDS gel electro- phoresis (12.5% acrylamide, proteins visualized by Coomasie blue staining).

The fractions from the S-200 column that contained hGM-CSF of at least 80% purity were pooled and diluted to a protein concentration of 0.25 mg/ml (708 ml) with buffer IV (15mM sodium phosphate; 3M urea; pH=7.5). We dialyzed this material against the dilution buffer until the concentration of 2-mercap- toethanol was less than 25% of the protein concentra¬ tion. We also included phenylmethylsulfonyl fluoride (0.5 mM) during the first dialysis to avoid pro- teolysis. We monitored the oxidation of protein through an assay for free sulfhydryl groups using DTNB, while continuing dialysis against the same buffer. The half-life of the reduced material -under these conditions was found to be about 4 h; therefore, the oxidation was approximately 95% complete after 18 hours. We then dialyzed the material twice against five volumes of a buffer containing 30 mM sodium phospate (pH = 7.5). The resulting solution was centrifuged (30 min, 10,400 x g) and the supernatant concentrated to 230 ml.

We centrifuged the concentrate again, as above, and applied the supernatant to a column of Fast Flow Q (2.6 x 15 cm) which had been equilibrated with 30 mM phosphate buffer (pH 7.5). The column was washed with the same buffer until the absorbance (280 nm) of the effluent was 0. The column was then developed with a gradient of sodium phosphate (30 to 130 mM,- 600 x 600 ml). We analyzed 15 ml fractions containing hGM-CSF for the purest product using SDS gel electrophoresis. Those fractions were pooled and concentrated to 10 ml.

The concentrated material was applied to a column of Ultrogel ACA-54 (LKB, 2.6 x 90 cm) equili- brated with 30 mM sodium phosphate and 130 mM NaCl

(pH = 7.5). We pooled and stored 5ml fractions con¬ taining pure hGM-CSF at -80° C. Using this purifica¬ tion method we produced in 30 to 60 mg of pure pro¬ tein, indicating a 12.5% to 25% yield.

B. In Yeast

We have also constructed expression vectors for the production in high yield of human GM-CSF in yeast. Figures 9 and 10 show a schematic outline of one embodiment of a process for constructing expres- sion vector p528/l which, when used to transform appropriate yeast cells, expressed hGM-CSF in high yields.

We made use of the leader or signal peptide of the yeast alpha (α) mating factor to express. hGM-CSF in yeast. Two genes, denoted MFαl and MF 2 have been reported to encode the alpha mating factor of yeast [J. Kurjan and I. Herskowitz, "Structure of a yeast pheromone gene (MFc^*): a putative factor precursor contains four tandem copies of mature α-factor". Cell, 30., pp. 933-43 (1982); A. Singh et al., "Saccharomyces cerevisiae contains two dis¬ crete genes coding for the α-factor pheromone", Nucl. Ac. Res. , 11, pp. 4049-63 (1983)]. MFαl codes for a precursor of 165 amino acids, containing four copies of alpha factor. The alpha factor repeats are preceded by a secretion leader sequence of 83 amino acids. The junctions between the secretion leader and the first repeat, and between each of the repeats have the following structure:

(leader or repeat)-lys-arg-(glu/asp-ala)₂_₃-(repeat)

The optimal cleavage of the secretion leader from the heterologous protein portion of the fusion pre¬ cursor may be obtained when the first amino acid of the heterologous protein is placed behind the -lys-arg processing site (see Figure 8). Accordingly, we used this alpha mating factor signal sequence hGM-CSF fusion in our vectors.

1. Expression vectors

In constructing our expression vector, we used two different genes encoding hGM-CSF. Gene 1, which was present on plasmid p210*, is characterized by a U937-derived DNA sequence coding for hGM-CSF, which was modified at the 5'-end of its coding region by a synthetic oligonucleotide DNA sequence or linker (Figure 6). Gene 2, which was present on plasmid p208 corresponds to unmodified hGM-CSF coding sequence.

In order to place an Nco I site in a posi- tion preceding the first hGM-CSF codon in our recom¬ binant molecule, we digested p208 with Pst I, and then treated it with Bal 31. We then restricted the digested sequence with Bam HI and isolated the smaller heterologous fragment. (Pst I is located upstream, and Bam HI downstream, of the hGM-CSF coding sequence.) We next cut p210* with Nco I and filled in with dNTPs. (The Nco I site marks the start of our synthetic linker. ) We then restricted our sequence with Bam HI and isolated the large fragment. We created a group of recombinant plasmids by insert¬ ing the heterologous small fragments taken from p208 (containing a DNA sequence coding for hGM-CSF) into the larger fragment taken from p210* (containing- the remainder of the expression vector) (see Figure 9). In one selected plasmid, this iigatio reconstructed an Nco I site at the start of the hGM-CSF coding sequence.

We cut our two GM-CSF plasmids (p208 Bal 31 and p210*) with Nco I and treated with SI nuclease (see Figure 10). We next cut the plasmids with

Hind III. The smaller fragments, which contained the native hGM-CSF coding sequence (in the case of 208 Bal 31) or the synthetic linker as well as the remainder of the hGM-CSF DNA sequence (in the case of p210*) were isolated.

We next constructed recombinant plasmid p216 using our fragments containing the respective hGM-CSF coding sequences and the larger Stul-Hindlll fragment of pMATA 21/51. That fragment contains the promoter and secretion leader of the MATαl gene.

To construct that fragment, we first isolated a DNA sequence encoding MFαl from a yeast genomic library -35- by using an oligonucleotide probe corresponding to amino acids 97 to 102 of MFαl precursor [(5') GTA CAT TGG TTG G/GCC G/A/TGG (3')] which we synthesized according to the published sequence of MFαl. [K. A. Nasmyth and S. I. Reed, "Isolation of genes by comple¬ mentation in yeast: molecular cloning of cell-cycle gene", Proc. Natl. Acad. Sci. USA, 77, 2119-23 (1980).]

We subcloned the 1.7 kb EcoRI fragment that we selected with this probe into pUC18. We mutagenized the resulting plasmid (p220/3) using primer mutagenesis [B. A. Oostra et al., "Transform¬ ing activity of polyoma virum middle-T antigen probed by site-directed mutagenesis", Nature, 304, pp. 456-59 (1983)] to introduce a StuI site at the posi¬ tion corresponding to the -lys-arg cleavage. We also inserted a 500 bp HindiII fragment carrying a synthetic SMC gene starting with a unique NcoI site [G. Buell et al., "Optimizing the expression in E.coli of a synthetic gene encoding somatomedin-C (IGF-1)", Nucl. Ac. Res., 13, pp. 1923-38 (1985)] into the HindiII site of p220/3. The result of this construction was a plasmid (p216) that had the secre¬ tion leader of MFαl ligated to the first codon (Ala) of the hGM-CSF coding sequence. We then transferred the MFαl/hGM-CSF gene fusion on an EcoRI-HindiII fragment to expression vector pl60/l, which carries origins of replication for E.coli and yeast (ori; 2 origin of replication), as well as selectable markers for both organisms (E.coli: β-lactamase; yeast: URA3) .

MFαl^*-hGM-CSF^' (5' )...AAA AGA GCA CCC ...(3* ) MFαl/208 Bal 31 lys arg ala pro (5' )...AAA AGAtGCA CCA ...(3^») MFαl/210*

KEX2 cleavage Recombinant plasmid p216 was cut with Eco RI and Hind III and we isolated the small fragment, which contained the MFαl/hGM-CSF fusion. This frag¬ ment was transferred to expression vector pl60/-I which carries an origin of replication for E.coli, the yeast URA 3 gene, the origin of replication of the 2μ circle, and the origin of replication of ARS 1 (autonomously replicating sequence, which allows replication in the yeast cell independently of the yeast chromosone), and the upstream region of PYK 1 (PUR) [D. T. Stinchomb et al., "Isolation and • Characterization Of A Yeast Chromosomal Replicator," Nature, 282, pp. 39-43 (1979)].

We designated the plasmid that resulted from using the DNA sequence of p208 Bal31 (i.e., the authentic cDNA sequence) as plasmid 528/1. We desig¬ nated the plasmid that resulted from using the DNA sequence of p210* (i.e., the 5' altered sequence) as plasmid 525/2. We also constructed a third expres- siόn plasmid, designated p545/l, in which an actin promoter replaced the MFαl promoter of pMATA 21/51 and a high-copy number vector (JDB207) was used as a base vector. [J. D. Beggs "Multiple-copy yeast vectors," Molecular Genetics in Yeast, Alfred Benzon Symposium 16, ed. D. Van Wettstein, J. Fries, M.

Kielland-Brandt & Stenderup, Munksgaard, Copenhagen (1981).] The low expression of the LEU 2 gene on this vector requires elevated copy numbers in trans- formants to allow growth on selective media [E. Erhart and E. P. Hollenberg, J. Bacteriol, 156, pp. 625-35 (1983)].

2. - Expression results

We transformed the hGM-CSF expression plas¬ mids into Saccharomyces cerevisiae strain BJ1991 (MATα ura3-52 leu2-3,112 trpl prbl-1122 pep4-3) (obtained from E. Jones, Carnegie-Mellon University). We grew the transformants in "SD" medium [F. Sherman et al., "Methods in yeast genetics", Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. (1981)], con¬ taining tryptophan and leucine (plasmids p525/2 and p528/l), or tryptophan and uracil (p545/l), to an optical density (600 mm) of 2. These cultures were used to inoculate "production medium", consisting of SD-medium containing 4% casamino acids and tryptophan (inoculum was 10% of final volume of production medium). At appropriate intervals during growth,

0.5 ml of the cultures were pelleted using a micro- fuge and then cell pellets were redissolved in 1/10 of the original culture volume by boiling 5 minutes in SDS sample buffer [U. K. Laemmli, "Cleavage of the Structural Proteins During the Assembly of the Head of Bacteriophage T4", Nature, 227, pp. 680-85 (1970)]. Produced proteins were then blotted to nitrocellulose and probed with antibodies against hGM-CSF (Western blotting technique) or .with concan- avalin A [J.C.S. Clegg, "Glycoprotein Detection in Nitrocellulose Transfers of Electrophoretically Separated Protein Mixtures Using Concanavalin A and Peroxidase: Application to Arenavirus and Flavovirus Proteins", Analytical Biochemistry, 127, pp. 389-94 (1982)].

Absolute expression levels realized were as follows (measured at OD₆₀₀ 10) in mg/litre: plasmid medium cells p525/2 10 5 p545/l 20 8 p528/l <0.1 <0.1 The fact that only the two constructions using the 5' modified hGM-CSF gene, but not the construction using the natural cDNA sequence expressed high levels of hGM-CSF indicates that the structure of the mRNA encoding the secretion precursor significantly in¬ fluences gene expression. Presumably, an unfavorable structure of the mRNA in p528/l severely interferes with its translation. This effect is different from the observed effects of mRNA secondary structure in the vicinity of the ATG start codon in bacterial expression [Buell et al., supra] because the altered DNA stretch in the mRNA encoded on plasmids p525/2 and p545/l is separated from the ATG start codon by 240 base-pairs.

As demonstrated from the above results, 70-80% of the total produced hGM-CSF was secreted into the culture medium. The secreted hGM-CSF occurred in 3 main forms:

(a) an unglycosylated form (14.5 kd) . This form comprised approximately 10% of the total secreted hGM-CSF. The size of this mole¬ cule corresponded exactly to non-glycosylated native hGM-CSF, indicating that the secretion leader had been correctly processed to generate ^**; an hGM-CSF beginning with Ala Pro. (b) a low-molecular-weight glycosylated form (18 kd). This form comprised up to 5% of the total secreted hGM-CSF. Native hGM-CSF is glycosylated; the hGM-CSF protein sequence contains two potential sites of N-linked glyco- sylation [Wong et al., Science, 228, pp. 810-15 (1985)]. The size of the 18 kd form was con¬ sistent with either the presence of two core- glycosyl side chains attached to each of both potential glycosylation sites, or the presence of an extended core glycosyl chain attached to only one glycosylation site.

(c) a high-molecular-weight glycosylated form (about 43 kd) . This form comprised 80-90% of the total secreted hGM-CSF. Treatment of the above-described glycosy¬ lated hGM-CSF forms with endoglycosidase H reduced their molecular weight to approximately 14.5 kd, indicating that all glycosyl side chains were N-linked to the protein backbone.

Replacement of the MFαl promoter by the actin promoter and use of a high-copy-vector (expres- sion plasmid p545/l) improved hGM-CSF expression approximately two-fold. However, while the amounts of the glycosylated forms in the culture medium increased, no increase of the unglycosylated form in the medium was observed. Instead, 2-3 fold more unglycosylated hGM-CSF was cell-associated indicating that solubility may limit the amount of hGM-CSF in the medium.

The supernatants from the cultures of yeast cells producing hGM-CSF were tested for biological activity. In the bone marrow clonal assay, described supra at pp. 23-24, yeast-secreted hGM-CSF stimulated colony-formation in a dose-dependent manner. Simi¬ larly, the CML assay, described supra at pp. 24-25, showed a dose dependent response to yeast produced hGM-CSF.

C. Comparison of Glycosylated And Deglycosylated hGM-CSF

Natural hGM-CSF is known to be a glycopro¬ tein with a molecular weight of about 22 kd [G.G. Wong et al., Science, 228, supra] . In the polypep¬ tide chain there are two asparagine residues at positions 27 and 37 which are potential sites for N-linked glycosylation (Asn-X-Thr/Ser) . Thus, it was heretofore believed that recombinantly produced hGM-CSF had to be produced in animal cells and gly¬ cosylated, in order to be active biologically. If the hGM-CSF was to be produced in bacterial cells, a second glycosylation step was thought to be necessary to confer activity. Contrary to this supposition, we have dis¬ covered that unglycosylated hGM-CSF has an unexpec¬ tedly higher specific activity than glycosylated I !

-40- hGM-CSF. Thus such unglycosylated hGM-CSFs are -an important part of this invention. As demonstrated in this application, these hGM-CSFs have a specific

Q activity of at least 1 x 10 Units/mg. Such deglycosylated hGM-CSFs may be pro¬ duced in several ways. For example, they may be produced in bacterial cells that do not glycosylate the proteins they produce. For example, when the polypeptide chain is produced by E.coli, it contains no attached carbohydrate and possesses a molecular weight of 14.5 kd.

These unglycosylated polypeptides may also be produced by deglycosylating yeast or animal cell produced proteins. For example, we isolated a high molecular weight fraction (MW 50-70 kd) of hGM-CSF, produced in yeast cells, as described above, using ConA-chromatography and gel filtration. We have also produced hGM-CSF in animal cells, by growing transfected Chinese Hamster Ovary (CHO) cells for three days in 10% fetal calf -serum containing medium. We used on CHO-cell clone which was derived from transfection with a vector that had the hGM-CSF gene isolated from U937 cells. The transcription was promoted by a SV40 early and Adenovirus major late promoter. Gene amplification was carried out with methotrexate selection. We isolated a high mole¬ cular weight fraction (MW 26-30 kd) by immuno- and lectin chromatography, followed by gel filtration. We then deglycosylated the isolated frac- tion with a mixture of endo-and exoglycosidases.

Deglycosylation of the high molecular weight frac¬ tions was established by immunoblotting. The digested fractions showed a reduction of the mole¬ cular weight close to that observed with the E.coli produced hGM-CSF (i.e., 14.5 kd) .

We determinated hGM-CSF concentration with a competition Radio Immuno-Assay (RIA), using as "I C tracer I labelled E.coli-derived hGM-CSF and anti- hGM-CSF fixed on Staph. aureus cells. The CML and BMC Marrow assays which we then conducted, are described above.

The results of our deglycosylation assays are summarized in the Table 1:

Table 1

Sample RIA CML Bone Marrow

A. Mammalian hGM-CSF before digestion 230 ng/ml 0.4xl0⁷ U/mg 0.2xl0⁸ U/mg after digestion 600 ng/ml 3.2xl0⁷ U/mg 2.3xl0⁸ U/mg

B. yeast hGM-CSF before digestion 0.05 mg/ml 1.2xl0⁷ U/mg 0.25xl0⁸ U/mg after digestion 0.20 mg/ml 9 .5xl0⁷ U/mg 3.8xl0⁸ U/mg C. E.coli hGM-CSF 2 x 10⁸ U/mg 1.8xl0⁹ U/mg

Our results demonstrate that the E . coli derived hGM-CSF produced entirely without carbohy¬ drate or deglycosylated yeast or animal cell derived hGM-CSF , are superior to the glycosylated native hGM-CSF for use in clinical applications .

Microorganisms and recombinant DNA mole¬ cules prepared by processes of this invention are exemplified by cultures deposited in the Deutsche Sammung von Mikroorganism, Grisebachstrasse 8 , D-3400 Gottingen, West Germany, on September 2 , 1985 and identified there as B84, B85 , B102 , and YE464 , and on October 4 , 1986 and identified there as Bill (p241-8 ) .

A. E . coli K₁₂ (p210* ) B. E.coli SG 936 (p210*) and (pcI857)

C. E.coli K_χ2 (p210*-5637) D. Yeast strain S.cerevisiae BJ1991 (p525/2).

E. E.coli A89 (p241)

These cultures were assigned accession numbers DSM 3473, DSM 3474, DSM 3475, DSM^* 3465, and DSM 3869, respectively.

While we have hereinbefore presented a number of embodiments of this invention, it is apparent that our basic construction can be altered to provide other embodiments which utilize the pro- cesses and compositions of this invention. Therefore, it will be appreciated that the scope of this inven¬ tion is to be defined by the claims appended hereto rather than by the specific embodiments which have been presented hereinbefore by way of example.

Claims

We claim:

1. A recombinant DNA molecule comprising a DNA sequence encoding a human granulocyte-macrophage colony stimulating factor (hGM-CSF)-like polypeptide, said DNA sequence encoding the polypeptide being characterized by a 5' terminal alteration and being operatively linked to an expression control sequence in the molecule, said alteration allowing the produc¬ tion of said polypeptide in higher yield than the native DNA sequence coding for hGM-CSF.

2. The recombinant DNA molecule according to claim 1, wherein the 5' alteration is selected from those of the formula: ATGGCACCAGCAAGAAGCCCGAGT CCGTCGACACAACCGTGGGAGCATGTGAATGCGATCCAGGAG, and GCA CCAGCAAGAAGCCCGAGTCCGTCGACACAACCGTGGGAGCATGTGAATGCGA TCCAGGAG.

3. The recombinant DNA molecule according to claim 1 or 2, wherein said expression control sequence is selected from the group consisting of the lac system, the β-lactamase system, the trp system, the tac system, the trc system, major operator and promoter regions of phage λ, the control region of fd coat protein, the promoter for 3-phosphoglycer- ate kinase or other glycolytic enzymes, the promoters of acid phosphates, the promoters of the yeast -mating factors, and other sequences which control the expression of genes of prokaryotic or eukaryotic microbial cells and their viruses and combinations thereof.

4. The recombinant DNA molecule according to claim 1, which is selected from the group consist¬ ing of p210*, p210*-5637, p241-8, p525/2, and p545/l. 'i l ⁱ

_44_

5. A microbial host transformed with at least one recombinant DNA molecule according to any of claims 1 to 4.

6. The microbial host according to claim 5, selected from the group of microbial hosts consisting of E.coli SG936, E.coli SG935, E.coli SG928, E.coli SG927, E.coli A89 and other E.coli strains which are Ion mutants and which are charac¬ terized by production of low levels of Ion protease.

7. The microbial host according to claim 5, being Saccharomyces cerevisiae (BJ1991).

8. A process for producing human granu¬ locyte-macrophage colony stimulating factor (hGM-CSF)- like polypeptides comprising the step of culturing a microbial host transformed with a recombinant DNA molecule according to any one of claims 1-4.

9. The process according to claim 8, wherein said hGM-CSF-like polypeptide is selected from polypeptides having the formulae: MetAlaProAlaArgSerProSerProSerThrGlnProTrpGluHisVal AsnAlalleGlnGluAlaArgArgLeuLeuAsnLeuSerArgAspThrAla AlaGluMetAsnThrValGluVallleSerGluMetPheAspLeuGlnGlu ProThrCysLeuGlnThrArgLeuGluLeuTyrLysGlnGlyLeuArgGly SerLeuThrLysLeuLysGlyProLeuThrMetlleMetAlaSerHisTyr LysGlnHisCysProProThrProGluThrSerCysAlaThrGlnllelle ThrPheGluSerPheLysGluAsnLeuLysAspPheLeuLeuValIlePro PheAspCysTrpGluProValGlnGlu, AlaProAlaArgSerProSer ProSerThrGlnProTrpGluHisValAsnAlal1eGlnGluAlaArgArg LeuLeuAsnLeuSerArgAspThrAlaAlaGluMetAsnThrValGluVal IleSerGluMetPheAspLeuGlnGluProThrCysLeuGlnThrArgLeu GluLeuTyrLysGlnGlyLeuArgGlySerLeuThrLysLeuLysGlyPro LeuThrMetlleMetAlaSerHisTyrLysGlnHisCysProProThrPro GluThrSerCysAlaThrGlnll lleThrPheGluSerPheLysGluAsn LeuLysAspPheLeuLeuVallleProPheAspCysTrpGluProValGln Glu and polypeptides coded for by DNA sequences charac- terized by a 5¹ terminal alteration of the DNA sequence coding for hGM-CSF, said alteration allowing the production of said polypeptide in higher yield than the native DNA sequence coding for hGM-CSF.

10. The process according to claim 8, wherein the microbial host is selected from the group consisting of E.coli SG936, E.coli SG935, E.coli SG928, E.coli SG927, E.coli A89 and other E.coli strains which are Ion mutants and which are characterized by their production of low levels of Ion protease.

11. The. rocess according to claim 8, wherein the microbial host is Saccharom ces cerevisiae (BJ1991).

12. A hGM-CSF-like polypeptide unacco - panied by glycosylation, said polypeptide having a

Q specific activity of at least 1 x 10 Units/mg.

13. The hGM-CSF-like polypeptide according to claim 12, produced by a method comprising the steps of culturing a bacterial host transformed with a DNA sequence coding for said polypeptide.

14. The hGM-CSF-like polypeptide according to claim 12, produced by a method comprising the steps of culturing a eukaryotic host transformed with a DNA sequence coding for said polypeptide and preventing the glycosylation of or deglycosylating the produced polypeptide !

-46- !

15. A pharmaceutical composition compris¬ ing a polypeptide produced according to the process of any of claims 8 through 11, is an amount effec¬ tive to stimulate granulocyte and macrophage formation and a pharmaceutically acceptable carrier.

16. A pharmaceutical composition compris¬ ing an hGM-CSF-like polypeptide as defined in any one of claims 12 through 14, in an amount effective to stimulate granulocyte and macrophage formation and a pharmaceutically acceptable carrier.

17. A method of reducing the likelihood of opportunistic infection comprising the step of treating an immunologically compromised human with a pharmaceutical composition as defined in claims 15 and 16.