NO845186L - VECTOR SYSTEM FOR INTRODUCING HETEROLOGICAL DNA IN EYKARYOTIC CELLS. - Google Patents

Description

The present invention relates to the production of proteins of interest using eukaryotic cells. In particular, it relates to the achievement of high yields of secreted, commercially exploitable proteins from cultured eukaryotic cells transformed with genes encoding the protein and for an optional marker, and more particularly transformation vectors with the genes encoding the protein linked to complementary DNA promoting the expression of the protein.

The following definitions are given for ease of understanding

of the invention. To the extent that the definitions deviate from meanings that are common within the technique, the definitions below apply.

Amplification means the process by which the cells produce

repetitions of genes by, in its chromosomal DNA.

Cotransformation means the process of transforming

a cell with more than one exogenous gene foreign to the cell, one of which gives the cell a selectable phenotype.

Downstream means the direction towards the 3' end of a nucleotide

sequence.

An "enhancer" is a nucleotide sequence that can enhance the transcription of genes regardless of the gene's identity, the sequence's position in relation to the gene or the sequence's orientation.

A gene is a deoxyribonucleotide sequence that codes for a specific, "mature" protein. For the present purposes, a gene does not include untranslated flanking regions such as RNA transcription initiation signals, polyadenylation sites, promoters or enhancers.

A selection gene is a gene that gives a phenotype to the cells

which expresses the gene as a detectable protein.

A selection agent is a condition or substance that

enables one to detect the expression of a selection gene.

Phenotype means the observable characteristics of a cell

expressed by the cellular genotype.

A product gene is a gene that codes for a protein product

with desirable characteristics such as diagnostic or therapeutic applicability.

Genotype means the genetic information contained in one

cell, as opposed to its expression which is observed as the phenotype.

Ligation is the process of forming a phosphodiester bond between the 5' and 3<1-> ends of two DNA chains. This can be achieved using several well-known enzymatic techniques including butt-end ligation using T4 DNA ligase.

Orientation refers to the order of nucleotides in a DNA sequence. An inverted orientation of a DNA sequence is one in which the 5' to 3' order of the sequence is reversed relative to another sequence when compared relative to a reference point in the DNA from which the sequence was obtained. Such reference points may include the direction of transcription of other particular DNA sequences in the parent DNA or the origin of replication in replicable vectors containing the sequence.

Transcription means the synthesis of RNA from a DNA template.

Transformation means changing a cell's genotype by means of the cellular uptake of exogenous DNA. Transformation can be transient or stable and can be detected in some cases by a change in cell phenotype. Transformed cells are called transformants. Pre-transformation cells are referred to as parental cells.

Translation means the synthesis of a polypeptide from messenger RNA.

The introduction of DNA into eukaryotic cells is generally a well-known procedure and can be carried out using a number of standard methods. These include the use of protoplast fusion, DNA microinjection, chromosome transfection, lytic and non-lytic viral vectors (eg, Mulligan et al., "Nature" (London) 277:108-114 [1979], cell-cell -fusion (Fournier et al., "Proe. Nat. Acad. Sei." 74:319-323 [1977]), lipid structures (US Patent No. 4,394,448) and cellular endocytosis of DNA precipitates (Bachetti et al., "Proe. Nat. Acad. Sei. 74:1590-1594 [1977].

The cells can take up the DNA temporarily or the DNA can be taken up to give a stable cell line depending on the conditions. One way to achieve a stable cell line is by taking up a selection gene where the DNA contains the desired protein gene and keeping the cell line under selective pressure. In all cases, it is desirable to find ways to enhance or increase the expression of the protein from the transformed cells.

The present invention provides means for enhancing the expression of heterologous protein in eukaryotic cells.

In accordance with the present invention, a method for enhancing the expression of heterologous protein in eukaryotic cells comprises introducing into a eukaryotic cell DNA encoding a heterologous, non-viral protein and additional DNA, and culturing the cell to express the protein. The invention also provides transformation vectors comprising complementary DNA for use in eukaryotic cells to express a heterologous non-viral protein. In a preferred embodiment, a gene expressing an amplifiable, selectable phenotype is introduced into the eukaryotic cell to facilitate the increase of the introduced DNA encoding the desired heterologous protein, and the additional DNA. Fig. 1 (a) illustrates the structure of plasmid pAdD26SVpA(3).

Fig. 1(b) illustrates the structure of plasmid pCVSVL.

Fig. 1(c) illustrates the structure of plasmid pD20.

Fig. 1(d) illustrates the structure of plasmid pD17.

Fig. 1(e) illustrates the structure of plasmid pD61.

Fig. 2(a) illustrates the structure of plasmid pIFD-6.

Fig. 2 (b) illustrates the structure of plasmid pL58.

Fig. 2 (c) illustrates the structure of plasmid pQ2.

Fig. 2(d) illustrates the structure of plasmid pQ3.

Fig. 3 is a schematic drawing illustrating the production of plasmid pTPL from plasmid pAdD26SVpA(3). Fig. 4 is a schematic drawing continuing from fig. 3 and illustrates the preparation of plasmid p91023 from plasmid pTPL. Fig. 5 is a schematic drawing continuing from fig. 4 and illustrates the preparation of plasmid p91023(B) from p91023. Fig. 6 is a schematic drawing illustrating p91023-ATIII. Fig. 7 is the nucleotide sequence of human tissue plasminogen activator (tPA) and part of its non-coding flanking regions. Fig. 8a is an illustration of four overlapping cDNA clones encoding the entire tPA protein sequence. Figures 8b-8d are a schematic representation of a suitable method for obtaining cDNA for human tPA in replicative form. Figures 9a-9b are schematic representations of a method for producing tPA transformation vectors.

Fig. 10 is a schematic method for producing

a connected vector.

Fig. 11 is a schematic method for producing a cotransformation vector containing a translation activator.

Fig. 12a is a schematic drawing of a transformation vector containing a eukaryotic enhancer. Fig. 12b is a schematic drawing of a transformation vector containing a region receptive to an activator acting transcriptionally on the other side. Fig. 13 is the nucleotide sequence and the deduced amino acid sequence of the EPO clone lamda-HEP0FL13. Fig. 14 is a schematic illustration of the transformation vector pRK1-4.

In accordance with the present invention, extra DNA and DNA encoding a desired protein (i.e. a product gene) are introduced into eukaryotic cells. The extra DNA enhances the expression of the desired protein. For example, the additional DNA may increase transcriptional or translational efficiency or have other enhancing effects on the expression of the desired protein.

In accordance with the present invention, additional DNA is part of the vector system for transformation. It can be introduced into the cell before, simultaneously with or after the introduction of DNA that codes for the desired heterologous protein. Extra DNA is DNA that improves the stability of the product synthesized by the transformant cell line, increases amplification of heterologous gene introduced into the cell line, or increases transcription or translation. Extra DNA contains functions that are recognized by the parent cell. As such, it is more than just a carrier or filler DNA as it has hitherto been used in transformations.

One type of extra-DNA includes DNA that codes for translational activators. Translational activator genes produce protein or short, untranslated RNA products that react with the messenger RNA for the product to increase translation efficiency. An example is the adenovirus DNA which encodes the virus-associated (VA) RNA (Thimmappaya et al., "Cell" 31:543-551 [1982]). This DNA produces two species of small untranslated RNAs (VA 1 and VA 2). The RNA products of VA DNA are thought to associate with the tripartite leader sequence of the adenovirus major late promoter in a manner that is currently unclear, to enhance translation from mRNA containing the leader sequence. VA RNA also increases the translational capacity of other early adenovirus mRNAs (Svensson and Akusjaru, 1984, Mol. Cell Bio. 4, 736-742). VA 1 or VA 2 DNA is well known. It may, together with its promoter, be directly ligated into a linked vector (preferably upstream of the product gene promoter or downstream of the polyadenylation site) or it may be transfected in the unlinked state either to the product or selection gene.

The increased translation by VA RNA is very large when the tripartite leader structure of late adenovirus mRNAs is present in the mRNA.

Another type of extra DNA comprises genomic eukaryotic DNA from the parent cell. This DNA is thought to include origins of replication or sequences that promote DNA stability, but the mechanisms responsible for the beneficial effects of such DNA are unknown. This DNA can be obtained by random shearing or endonuclease cutting of the genome DNA of the cell line. The extra-genomic DNA should comprise fragments with a size varying from approx. 50 to approx. 5000 base pairs. These fragments are then ligated at available restriction sites upstream of the transformation vector promoter (and preferably upstream of any enhancer) or downstream of the polyadenylation site. These vectors are then suitable for transformation after which the transformants are sorted with regard to one or more desirable properties, e.g. product synthesis stability after serial cultivation of the transformant and/or high product yields.

The product yield from transformants can be improved by transforming the cells with a type of extra DNA called "trans-acting" transcriptional activators. Such DNA codes for proteins or protein derivatives that stimulate transcription. One type of transcription activator is referred to as immortalizing genes as they provide unlimited, successive progeneration; cells containing such genes do not die after a certain number of divisions. These activators do not need to be ligated to the same DNA strand as the product gene to increase transcription rates, and in this respect they differ from enhancers, Imperiale et al., "Cell" 35:127-236 [1983], Green et al. al., "Cell" 35:137-148 [1983]. Several trans-acting transcription activators that can be used in the present invention are known per se and have been cloned. The most thoroughly investigated examples include the human C-myc, SV40 large T antigen, polyoma large T antigen and adenovirus ElA genes.

The product genes that can be used are essentially unlimited. Genes that code for substances that are proteins or that can be produced by protein-based reactions, such as enzyme conversions, are suitable. Genes for proteins that can adversely affect host cells by synthesizing toxins or hydrolyzing host proteins, e.g. some enzymes from prokaryotic or lower eukaryotic sources may be used with such qualifications as providing anti-toxins in the culture medium or by selecting for the lower levels of expression that would otherwise be optimal. Genes for proteins or enzymes with action found in cells of higher animals such as mammals or vertebrates. The genes of interest for most therapeutic proteins will be of this type.

Examples of product genes are genes that code for coagulation or fibrinolysis proteins such as antihemophilic factor, tissue plasminogen activator, urokinase and the coagulation factors II, VII, IX, X or XIII, blood proteins such as fibronectin or albumin, protease inhibitors such as antithrombin III, α-l-antitrypsin and 2 macroglobulin, hormones or regulatory proteins including lymphokines such as erythropoietin and other T-cell active substances, growth hormones and platelet-derived growth factor, oncogene products, cell surface antigens, immune proteins such such as IgG, IgE and IgM as well as complement, and other proteins that are or will be of commercial interest.

Transformation vectors

Vectors used in transformation according to the present invention usually contain extra DNA and a product gene. In addition, other elements such as enhancers, promoters, introns, polyadenylation sites and 3<1->non-coding regions will usually be present in transformation vectors as will be described below. The extra DNA can be transformed using a special vector. When a selection gene is used, it can be linked to a product gene in a vector or be present in a separate vector that is transformed together with the vector containing the product gene.

Selection genes fall into three categories: demonstrably increased selection genes, dominant selection genes and demonstrably increased dominant selection genes.

Demonstrably increased selection genes are those where increase can be demonstrated by exposing host cells to changes in the selection agent. Detectably amplified genes that are not dominantly acting usually require a parental cell line that is genotypically deficient with respect to the selection gene. Examples include the genes for aspargine synthase; aspartate transcarbamylase; (Kemp et al.,

"Cell" 9:541 [1976], adenylate deaminase (DeBatisse et al., "Mol and Cell Biol." 2 (11) :1346-1353 [1982]), adenosine deaminase (Yeung et al., J.B.C. 258 15185 ( 1983), mouse dihydrofolate reductase (DHFR) and, with a defective promoter, mouse thymidine kinase (TK).

Dominant selection genes are those that are expressed in transformants regardless of the genotype of the parent cell. Most dominant selection genes are not demonstrably increased because the phenotype is so highly efficient at processing the selection agent that it is difficult to distinguish between cell lines that have or have not supplemented the gene. Examples of dominant selection genes of this type include the genes for prokaryotic enzymes such as xanthine-guanine-phosphoriboxyltransferase (Mulligan et al., "Proe. Nat. Acad. Sei." 78[4]:2072-2076 [1981]) and aminoglycoside- 3'-phosphotransferase (Colbere-Garapin et al., "J.Mol.Biol.", 150:1-14 [1981]).

Some dominant selection genes are also demonstrably increased. Suitable examples include the mutant DHFR gene described by Haber et al., "Somatic Cell Genet." 4:499-508 [1982], cell surface markers such as HLA antigens and genes encoding enzymes such as specific esterases which produce fluorescent or colored products from fluorogenic or chromogenic substances and which are known in the art.

Demonstrably increased, dominant selection genes are preferred here. It should be understood that a dominant selection gene can in some cases be converted into a detectably increased gene by means of appropriate mutations in the gene.

It is preferred that the selection agent is one that prevents cell growth in the absence of the selection gene. In this way, revertant cells in long-term culture that delete or no longer express the selection gene (and presumably also the product gene) will not overgrow the population. However, it would be desirable in the commercial production of product proteins for therapeutic purposes to avoid the use of cell toxins and thereby simplify the purification steps for the product. A desirable selection gene will thus be one that makes it possible for transformants to use a nutrient that is critical for growth and which they would otherwise be unable to use. The TK gene described above is an example.

It should also be remembered that a selection gene can also be the product gene. For example, it may be that one wishes to harvest the products of selection genes for use as therapeutic or diagnostic agents. Product genes can become capable of acting as selection genes if the transformant's environment can be modified to make it possible for the product to give the transformants some selection advantage. For example, a product gene may produce an enzyme that will act on an otherwise unusable, critical substrate in the cell medium to release the critical substrate, e.g. an obligatory nutrient.

The selection and product genes used here may be accompanied by all or part of their wild-type, untranslated flanking sequences, but usually such sequences will not be included as further described below. In the case of a linked vector, particularly advantageous results with regard to product yield are obtained by placing the product gene upstream of and linked close to the translated regions of the selection gene. This usually means that most, if not all, of the 3<1->untranslated region found downstream of the wild-type product gene and the 5'-untranslated regions found upstream of the wild-type selection gene are excised using appropriate restriction endonucleases, or by smearing via oligonucleotide priming with M13 as described elsewhere. The result is that both genes are ligated directly without introducing a promoter between the product and selection gene.

The genes, as used here, will usually contain the translated sequences from the wild type which encode prepro polypeptides, e.g. control sequences for secretion, which may be desirable for the structure, stability and/or secretion of the mature product. If, however, translated wild-type driver sequences do not work correctly in the transformants, then such prepro sequences can be omitted (before assembly in a vector and after the cotransformation of the host cell) or they can be replaced with other prepro sequences that work correctly in the vector.

It is usually not important that the vectors or genes

found in the vectors is interrupted with introns as long as the messenger RNA transcripts from the vectors are correctly ligated to the transformants so that you get messenger RNA that is translated into the mature protein, plus any desired leader sequences. This would be the case with introns found in the genes of a particular higher eukaryotic cell to be processed by other higher eukaryotic cells. As most of the genes used here are reverse transcripts of cDNA, no introns will be present and the possibility of erroneous post-transcriptional splicing will be significantly reduced.

In the case of linked vectors, the coding chains of the selection and product genes are preferably linked together by directly ligating the stop codon of the product gene to the start codon of the selection gene. Optionally, the genes are ligated across an oligo-deoxyribonucleotide bridge that is rich in guanine- and cytosine-deoxyribonucleotides (G and C, respectively). The bridge should be free of termination and start codons, as well as palindromes to reduce the likelihood of forming RNA "hairpin loops".

The terms "product gene" and "selection gene" are not intended to mean that only one of each gene or a single copy of each gene is intended for use in the vectors. First, a particular selection phenotype may require the synthesis of more than one distinct protein. In this case, e.g. the selection gene for each protein be present in a vector that is not covalently linked either to the product gene or to any other selection gene, or each selection gene may be ligated to the product gene as described above or in US Patent No. 4,399,216. It is not preferred to use more than one selection gene when each gene would give a selection phenotype for the same selection agent independently of another selection gene.

It may be desirable to transform with a vector or vectors containing several separate product genes. Separate product genes that code for proteins with a beneficial effect on each other are of particular interest. For example, one can simultaneously express a protein that stabilizes another product protein or that is part of a multi-protein system in its biologically active form.

In addition, one or both of the selection and product genes may be repeated in the vector or vectors, i.e. present in multiple, usually two, copies. In such cases, each of the repeated genes will preferably contain the entire RNA sequence for processing and transcriptional control that is present in vectors containing only a single copy of the genes.

The vectors mentioned here may also include enhancers. Enhancers are functionally distinct from promoters, but appear to operate together with the promoters. Their function at the cellular level is not well understood, but their unique property is the ability to activate or enhance transcription without being position- or orientation-dependent. Promoters must be upstream of the gene while enhancers can be present upstream or 5'

from the promoter, within the gene such as the intron, or downstream of the gene between the gene and a polyadenylation site or 3' from the polyadenylation site. Inverted promoters are not functional, but inverted enhancers are. Enhancers are cis-acting, i.e.

that they have an effect on promoters only if they are present on the same DNA molecule. For a general discussion of enhancers, see Khoury et al., "Cell" 33:313-314 [1983].

Preferred enhancers are obtained from animal viruses such as simian virus 40, polyomavirus, bovine papillomavirus, retrovirus or adenovirus. Ideally, the enhancer should be from a virus to which the host cell is susceptible, i.e. which usually infect cells of the host type. Viral enhancers can be easily obtained from commonly available viruses. The enhancer regions of several viruses, e.g. Rous sarcoma virus and simian virus 40 are well known. See Luciew et al., "Cell" 33:705-716 (1983). It will be routine chemistry to excise these regions on the basis of published restriction maps for the virus in question and, if necessary, to modify the binding sites to enable or facilitate the insertion of the enhancer into the product gene or, if desired, the selection gene vector . See e.g. Kaufman et al., Mol Cell Biol. 2,

pp. 1304-19 (1982). Optionally, the enhancer can be synthesized from sequence data; the size of viral enhancers (usually less than about 150 bp) is sufficiently small for this to be carried out in practice.

The enhancer can be repeated within the vector, either in a

double shaped (as in the case of the SV40 virus enhancer as found in nature) or spaced outwards in the vector on

the places mentioned above. Preferably, the enhancer is in the same orientation relative to the product and/or selection gene as it was relative to the genes under its influence in the wild-type source. The enhancer is preferably located upstream of all promoters present in the vector. A number of different enhancers can be used and the enhancer should not be ligated to either the product or the selection gene to be useful in transformation. In an unlinked vector system, enhancers are preferably present on the vector containing the product gene. This will increase the probability that the product and selection genes will be physically bound in the cotransformants. Approximately 1 to 100 times more transformants can be obtained in this way with a given amount of transforming DNA depending on the promoter used for transcription of the selection gene.

The enhancer need not contain any of the flanking sequences found in its wild-type environment, e.g. viral origins of replication or related promoter elements such as "TATA boxes", "cap" sites or transcription primer sequences. However, the absence of random restriction enzyme attack sites that enable the removal of these sequences may make it more convenient to include all or part of the sequences. It may also be more convenient to use a DNA fragment containing both an enhancer and the promoter which is usually under the wild-type biological control of the enhancer.

Enhancer and enhancer-promoter regions can also be selected from eukaryotic cells instead of viruses. These are preferably enhancers associated with genes that produce protein in large, significant quantities in the cells of origin. It has been found that they will simultaneously give high product yields from genes other than those normally controlled in the wild-type environment. The host cell to be transformed is preferably a cell line of the same somatic or bacterial origin as the cell from which the enhancer is obtained. For example, the activating or enhancing region for the immunoglobulin gene (immunoglobulin enhancer) in the J^-C^ intron can be introduced either upstream or downstream of the product gene, and the structure is introduced together with a selection gene into a myeloma cell line. See Gillies et al., "Cell" 33:717-728 (1983) for further details regarding this enhancer. The immunoglobulin enhancer is preferably used on vectors intended to be introduced into myeloma cells to produce product protein.

Both the product gene and the selection gene are ligated to promoters to come under the transcriptional control of a promoter, except in the vector-linked embodiment described above. The promoter may be the wild-type promoter in question

gene, or the gene may be ligated to a promoter from another gene in the transformed cell line, from another eukaryotic cell line or from eukaryotic viruses. Obviously, the promoter should be recognized by the host to be transformed (in the absence of random recombination with promoters in the transformed cells), but otherwise the choice of a promoter is not believed to be critical. Particularly desired promoters are linked to 5' untranslated leader sequences and activated transcriptionally or translationally by means of exogenous agents or conditions, e.g. extragen products (transcripts and polypeptides), heavy metal ions, heat shock or viral infection. A preferred promoter for product gene expression is the major late promoter of adenoviruses with the tripartite leader sequence.

The vector system need not contain an enhancer if a promoter is chosen which is enhancer-independent, e.g. the mouse α-globin promoter. However, vectors with an enhancer and an enhancer-dependent strong promoter, such as the major adenovirus late promoter, are preferably used over such enhancer-independent strong promoters. A strong promoter is one that results in the same or more transcripts than the SV40 early promoter under controlled conditions.

Another element that should preferably be present in the vector construction is a site for polyadenylation. This is a DNA sequence located downstream of the translated regions of a gene, where adenine ribonucleotides are added to form a poly-adenylate tail at the 3' end of the messenger RNA. Polyadenylation is important for stabilizing the messenger RNA against degradation in the cell, an event that reduces the level of messenger RNA and consequently the level of product protein.

Eukaryotic polyadenylation sites are well known. A corresponding sequence is found among eukaryotic genes: the hexanucleotide 5'-AAUAAA-3' is located 11-30 nucleotides from the point in the RNA where polyadenylation starts. DNA sequences containing polyadenylation sites can be obtained from viruses according to published reports. For example, polyadenylation sequences can be obtained from genes for beta-globin in mice, genes for simian virus 40 late or early regions, etc. Viral polyadenylation sites are preferred. As these sequences are known, they can be synthesized in vitro and ligated to the vectors in the usual way.

A polyadenylation site should be located downstream

for the product gene in vectors that are linked or unlinked. In vectors that are unlinked, it is optionally ligated downstream of the selection gene. The sequence that separates the polyadenylation

the site from the stop codon for translation is preferably an untranslated DNA region such as e.g. a non-promoter driven eukaryotic gene. The oligonucleotide preferably has a considerable extent, in the order of magnitude up to approx. 1000 bases, from the stop codon to the polyadenylation site. This 3'-untranslated nucleotide sequence usually results in an increase in product yields. The vector can be terminated from approx. 30 bp downstream of the corresponding sequence, but it preferentially retains the 3' sequence found downstream of the polyadenylation site in its wild-type environment. These sequences usually range from approx. 200 to 600 base pairs downstream of the polyadenylation site.

It has been found that the presence of introns in the untranslated, transcribed portion of the vector can increase product yields. Such introns can be obtained from sources other than both the host cells and the gene sources. For example, a 3'-spliced site from an immunoglobulin gene introduced into the untranslated region of the essential late transcript of adenoviruses instead of part of the normal intron of this transcript can result in an increase in product yields.

Cells transfected with transcriptional activators, translational activators, and strong promoters or promoter-enhancer combinations offer the greatest potential for increased product synthesis. You will get a spreading effect from the cells that have been transformed with transcription and translation activators.

The vectors are preferably supercoiled, double-stranded, ring-shaped constructs. This is the form in which the vectors are obtained by the usual prokaryotic cloning procedures by which they are produced. However, the vectors can be straight-chained, i.e. covalently cleaved at one point, which belongs to other steps such as ligation to extra-genomic DNA.

The following table includes representative examples of suitable vector systems for transformation.

Vector synthesis

Linked vectors with short or no oligonucleotide sequences between the product and selection gene can be produced in several ways. In the absence of a restriction site adjacent to, or immediately downstream of, the product gene's translational stop codon, the product gene can be cut at a restriction site upstream of the wild-type translational stop codon and the codon then ligated to the selection gene fragment using synthetic linkers. Another way of producing such a vector is by the use of a bridge-forming nucleotide probe to remove unwanted sequences that lie between the product and selection gene. If the amino acid at the carboxylic acid end of the product protein is not necessary for biological activity, then a synthetic stop codon can optionally be ligated to a suitable 3' end. Use of a gene fragment here is not a preferred embodiment as it is usually desired to synthesize product protein that is as identical to that found in nature as possible.

The 5<1->untranslated regions adjacent to the selection gene in its wild-type environment can be removed in the same way as described above for the product gene, and this can be carried out simultaneously with the removal from the product gene as shown in example 4. however, it is not necessary that the protein expressed by the selection gene or a derivative thereof is identical to the wild type, e.g. more active against a growth-limiting substrate or more resistant to a toxin than the wild-type protein. In linked vectors, the selection gene is preferably under the control of the promoter of the product gene. This will help to stabilize the cell's genotype and expression of the product protein as cells that eliminate the product gene will not survive the selection agent in the absence of recombination of the transformants. The opposite has been found to be the case with unlinked vectors, although here the same promoter can be used for both the selection and product gene.

Otherwise, the vectors described herein can be synthesized using techniques well known to those skilled in the art. The components of the vectors such as product or selection genes, enhancers, promoters and extra DNA such as "trans-acting" transcription* activators, translational activators and the like can be obtained from natural sources or can be synthesized as described above. If the components are found in DNA that is available in large quantities, e.g. constituents such as viral groups, or if they can be synthesized, e.g. polyadenylation sites, then, with the appropriate use of restriction enzymes, essentially large quantities of vector can be obtained by simply culturing the parent organism, cleaving its DNA with an appropriate endonuclease, separating the DNA fragments, identifying the DNA containing the elements of interest, and recovering these by using well-known techniques. Typically, a transformation vector can be assembled in small quantities and then ligated into a suitable autonomously replicating synthesis vector such as a prokaryotic plasmid or a phage. ;The plasmid pBR322 can be used in most cases. See Kaufman; et al., (1982) above. Larger transformation vectors may require high-capacity synthesis vectors such as cosmids (a vector where DNA from a phage is packaged in the phage's capsid but replicates as a plasmid when transfected into a susceptible host). ;The synthesis vectors are used to clone the ligated transformase ion vectors in the usual way, e.g. by transfection of a susceptible prokaryotic organism, replication of the synthesis vector to high copy number and recovery of the synthesis vector by cell lysis and separation of the synthesis vector from cell debris by methods well known to those skilled in the art. The resulting yield of synthetic vector can be transfected directly into eukaryotic cells, or the transformation vector can be isolated from the synthesis vector by appropriate endonuclease cleavage, separation by molecular weight, and recovery of the transformation vector. Isolation of the transformation vector is not necessary as long as the remainder of the synthesis vector does not adversely affect eukaryotic gene amplification, transcription or translation. The preferred synthesis vector is here e.g. a mutant of E. coli plasmid pBR322 in which sequences harmful to eukaryotic cells have been removed. See Lusky et al., Nature 293:79-81 (1981). Use of this mutant prevents any need to remove the remaining plasmid before transformation. ;Transformation;The cells to be transformed can be any eukaryotic cell, including yeast protoplast, but is preferably a non-fungal cell. Primary explants (including relatively undifferentiated cells such as stem cells), and immortalized and/or transformed cell lines are suitable. Candidate cells need not be genotypically deficient with respect to the selection gene as long as the selection gene, when used, is dominantly acting. ;In certain settings, there is no need for a stably transformed cell line. In such cases, a suitable transformation vector may depend on a temporary introduction of DNA into mammalian cells (Mellon, P., Parker, Y. Gluzman, T. Maniatis 19 81 Cell 22 279-288). In order to isolate the desired transformants, it is not required that all the cells in the population stably contain exogenous genes that express the desired protein product. It is possible to temporarily introduce exogenous genes into a subpopulation of cells so that the subpopulation expresses the desired product over a period of several days. As a selectable marker is not required in the transformation vector for the DNA transfection and expression system according to the present invention, the exogenous DNA can be lost after culturing the cells over a period of 1-2 weeks. 2-3 days after transfection of suitable mammalian cells, however, it is found that the desired products are synthesized and can be detected. Thus, a host-vector system for cloning a gene for a mammalian protein can be based on the development of CV-1 monkey cell lines transformed with an SV40 DNA molecule lacking the origin of replication (Gluzman, Y., Cell 23 175- 182, ; 1981) . SV40 is a small DNA tumor virus that replicates lytically in monkey cells. In the absence of DNA replication, SV40 will transform monkey cells. The transformed CV-1 monkey cells containing defective SV40 DNA, designated COS (CV-1, "origin defective", SV40), do not contain a complete copy of the SV40 genome, but produce high levels of T antigen and allow replication of SV40 DNA. They also efficiently support the replication of SV40 by deletions in the early region and by bacterial plasmids containing the SV40 origin of replication (Myers, R.M. & Tjian, R. 1980 PNAS 1J_ 6491-6495). Accordingly, this system provides a means to amplify transfected exogenous DNA (including the protein gene and extra DNA) via SV40-mediated DNA replication to increase the amount of mRNA and protein expressed from the exogenous DNA. In other embodiments, the cells are preferably stably immortalized mammalian cell lines. Cell lines known to stably integrate selection genes into their chromosomal DNA are preferred, e.g. Chinese hamster ovary (CHO) cell lines. Also HeLa cells, COS monkey cells, melanoma cell lines such as Bowes cell line, L cells from mice, fibroblast cells from mice, NIH3T3 cells from mice and the like. ;Transformation with unlinked DNA can be carried out consecutively or simultaneously. Methods are known to facilitate cellular uptake of DNA. Microinjection of the vector into the cell nucleus will give the highest transformation yields, but exposure of parental cells to the DNA in the form of a calcium phosphate precipitate is usually more convenient. Generally, the frequency of transformants expressing a product is greater upon transformation with a molar excess of product gene relative to selection gene, preferably in the order of 10:1 or more. ;The selection transformants are then sorted by ligation of the product gene in their chromosomes, or by expression of the product itself. The former method can be performed by using "Southern blot" analysis, the latter by means of standard immunological or enzymatic tests. Once the transformants have been identified, steps can be taken to further enhance expression of the product gene by subcloning in the presence of a selection agent in constant or increasing amounts. In general, this involves taking the transformant cell population and (a) selecting one or more cells from the cell population that express the product in an advantageous manner compared to other cells in the population, (b) growing the selected cell or cells into a subsequent cell population under conditions designed to select for a change in the expression of the phenotype and (c) further selecting one or more cells from the subsequent cell population that express the product in a preferred manner compared to the other cells in the subsequent population. Step (b) is preferably carried out with several of the clones from step (a). This method is further described in US patent application no. 565,627 filed on 27 December 1983. The following examples are given to further illustrate the invention and are not intended to imply any limitation of the scope of the claims. In the examples, C° is used. ;Example 1;Cell culture;The SV40-transformed COS monkey cells (clone M6) were provided by M. Horowitz (Horowitz et al., J. Mol. App. ;Genet. 2, pp. 147-9 (1983). Adenovirus 2 was provided by P. Sharp. Infections with adenovirus were performed by absorbing 20 pfu/cell for 90 min at 37° C. After rinsing, the cells were incubated for 18 h at 37° C. At this time, it was observed a noticeable cytopathic effect. ;DNA transfections were performed as described by Sompayrac;&Danna, Proe. Nati. Acad. Sei., 78, pp. 7575-8 (1981), with the addition of "chlorquine" treatment (Luthman and Magnusson , Nuc. Acids Res. 11, pp. 1295-1308 (1983). Cells were rinsed with serum-free media and incubated at 37°C for 12 hours in Dulbecco's modified Eagle's medium containing DEAE-dextran (molecular weight 5000,000, Pharmacia, 250 mg/ ml) with 0.1M tris, pH 7.3 and 2 µg/ml plasmid DNA. After incubation, the cells were rinsed with serum-free media; and treated for 2 hours at 37°C with 0.1M chlorquine in media containing serum . Cells were then supplied with media and incubated for the indicated time periods. ;Plasmid construction;In order to analyze adenovirus-specific influencing factors that act on the adenovirus MLP, a series of DHFR cDNA genes has been constructed and these are drawn in figures l(a)-l(e). pAdD26SVpA(3) contains a mouse DHFR cDNA under the control of the adenovirus MLP that includes the first leader sequence and the 5' splice site from the adenovirus late mRNA. This 5' splice site can be properly spliced to a 3<1> splice site introduced from a mouse immunoglobulin gene (Kaufman & Sharp, Mol. Cell Biol., 2, pp. 1304-19 (1982)). An SV40 early polyadenylation signal is present at the 3' end of the DHFR coding region. The cDNA gene is cloned into a -BR3 22 derivative (pSVOd) lacking sequences detrimental to replication in mammalian cells (Lusky, et al., Nature (London), 293, pp. 79-81 (1981)), and which contains the SV40 origin of replication (Mellon et al.). The plasmids pAdD26SVpA(3) (Fig. 1(a)) and pCVSVL (Fig. 1(b)) have been described (Kaufman & Sharp, supra). pCVSVL2 is identical to pCVSVL except that it contains a duplicate of the SV40 Avall D fragment introduced by addition of XhoI linkers and insertion into the XhoI site (15.83 m.u.) upstream of the adenovirus MLP. The AvaIID fragments are both oriented so that the SV40 sen promoter is in the same orientation as the Ad 2 MLP. ;pCVSVL2 is derived from pAdD26SVpA(3) but lacks the SV40 origin from pSVod and contains the SV40 enhancer and origin of replication upstream of the adenovirus MLP. pD20 (Fig. 1(c)) and pD17 (Fig. 1(d)) are identical to pAd26SVpA(3) ; (Fig. 1(a)) except for the insertion of a 138 bp DNA sequence encoding the second and 2/3 of the third leader sequence of adenovirus, derived from a cDNA clone for fiber protein of adenovirus (Zain et al., Cell, 16, pp. 851-61 (1979)). pD20 contains the 13 8 bp region in the correct orientation introduced into the PvuII site which is 8 bp 5' from the 5<1-> splice site. pD20 thus contains the first 170 bp from the spliced late mRNA encoding the complete first, second and 2/ 3 of the third leader sequence, and further contains 8 bp from the first late leader sequence to the 5<1-> splice site. pD17 (Fig. 1(d)) contains the 138 bp fragment in the opposite orientation. pD61 (Fig. 1 (e)) is identical to ;pCVSVL2 except that it contains the second and third leader sequences present in pD20 (Fig. 1 (c)). ;To obtain pDl7 and pD20, pJAW43 (Zain et al ., supra) column t with Xhol, treated with Klenow fragment of DNA polymerase 1 and then cleaved with PvuII. The 13 8 bp fragment was isolated and ligated into pAdD26SVpA(3) which had previously been cleaved with PvuII. DNA was transferred into E. coli HB101 and isolated using a radiolabeled 138 bp XhoI-PvuII fragment from pJAW43. DNA was prepared from positively hybridizing clones and examined for orientation by digestion with PvuII and HindIII followed by acrylamide gel electrophoresis. pD20 contains the 138 bp fragment in the same orientation and pD17 contains the fragment in the opposite orientation with respect to the direction of transcription of adenovirus MLP. pD61 was constructed by inserting the EcoRl-XhoI partial 1.1 kB fragment from pCVSVL into the large fragment from pD20 that had previously been digested completely with EcoRl and XhoI. The resulting plasmid pD61 contains the SV40 enhancer and the adenovirus promoter with a tripartite leader sequence. ;These recombinants have been used to investigate adenovirus factors that affect expression from these cDNA genes. The experimental procedure involved DNA transfection and adenovirus superinfection of COS monkey cells. 36 h after DNA transfection, the transfected cells were infected with adenovirus and incubated for 18 h. The cells were then labeled for 1 hour with 35S methionine and analyzed for DHFR synthesis by immunoprecipitation and polyacrylamide gel electrophoresis. The results indicate that both AdD26SVpA(3) and pCVSVL2 can control the synthesis of DHFR compared to artificially transfected samples. In artificially transfected COS cells, it is possible to detect monkey ^DHFR migrating just above mouse DHFR. ;When DHFR expression from pCVSVL2 and pCVSVL was compared in COS cells and upon transformation of CHO DHFR cells, no differences were found between the two plasmids. In addition, superinfection with adenovirus has little effect on the synthesis of DHFR from both pAdD26SVpA(3) and pCVSVL2. However, expression from pCVSVL2 is twofold greater than from pAdD26SVpA(3). This is due to the introduction of the SV40 enhancer upstream of the adenovirus MLP in pCVSVL2. Plasmids containing the ;138 bp leader insert from the spliced adenovirus tripartite leader sequence (pD20, pD17 and pD61) were also found to synthesize DHFR. In addition, both pD20 and pD61 respond to adenovirus infection by a 3-10-fold increase in DHFR synthesis in different experiments. In contrast, expression from pD17, which contains the 13 8 bp tripartite leader segment in the opposite orientation, was not affected by adenovirus superinfection. These results suggest that there is a sequence within the tripartite second and third director segments that responds in an orientation-dependent manner to adenovirus infection. To determine whether the mRNA level is affected by adenovirus infection, 3' Sl mapping has been performed as described (Kaufman et al., Mol. Cell Biol. 2, 1304-1319 (1982)). A 3' end-labeled DNA probe was prepared and hybridized with all RNA isolated from transfected and adenovirus-superinfected COS monkey cells. For the plasmids pCVSVL2, pD20 and pD61, a single 550 base pair fragment corresponding to an mRNA polyadenylated in the SV40 polyadenylation site is observed. pD20 lacking the SV40 enhancer exhibits approximately 2-fold less DHFR specific mRNA compared to plasmids containing the SV40 enhancer. After superinfection with adenovirus, there is no change in the level of DHFR mRNA from either cDNA gene. This indicates that the observed increase in DHFR synthesis after adenovirus superinfection is not due to an increase in DHFR mRNA level. The increased DHFR synthesis after adenovirus superinfection is due to increased translational efficiency. ;Example 2;To analyze the effect of VA-RNA on translation, a new set of plasmids expressing human gamma interferon has been constructed (Figs. 2(a)-2(d)). A cDNA encoding human gamma interferon has been cloned from mRNA isolated from human peripheral blood lymphocytes using oligo dG-oligo dC tailing methods (Maniatis et al.) and was provided by Dr. S. Clark (Genetics Institute ). The Pstl clone encoding gamma interferon was inserted into the Pstl site in a derivative of pCVSVL2 that has previously deleted the Pstl site 3' to the SV40 polyadenylation signal. The resulting plasmid p(gamma-IF) D-6 (Fig. 2(a)) contains the gamma interferon coding region 5' from the DHFR cDNA and 3' from the 3' splice site for mRNA biogenesis. To obtain the corresponding plasmid with the tripartite leader sequence from adenovirus, both p(gamma-IF) D-6 and pD61 were digested with SalI and BglII. Sali cleaves each plasmid once in the tetracycline resistance gene of pBR322, and Bglll cleaves each plasmid once 5' in the 3' splice site from the immunoglobulin gene. The gamma-IF-containing fragment from p(gamma-IF) D-6 was ligated to the tripartite styrene-containing fragment pD61 and the DNA then used to transform E. coli HB101 to tetracycline resistance. pL58 (Fig. 2 (b)) is a plasmid containing the tripartite leader sequence of adenovirus, but is otherwise identical to p(gamma-IF) D-6. The adenovirus VA genes were isolated from a Hind III Ad 2B fragment and inserted into pL58. The Hind III B fragment of adenovirus (17.1mu-31.7mu) had previously been cloned into pBR322 such that the Hind III site in 31.7mu was adjacent to the EcoR1 site of pBR322. This plasmid was digested with HpaI (28.0mu) and EcoRl linkers were used. After digestion with EcoRl, the 1.4 KB band was isolated and cloned into the EcoRl site of pL58. pQ2 (Fig. 2 (c)) contains the VA genes present in the vector so that they are transcribed in the opposite direction as adenovirus MLP. pQ3 (Fig. 2(d)) is identical to pQ2 with the exception that the VA genes are present in the opposite orientation. ;Gamma interferon;Gamma interferon assays were performed by measuring protection against the cytopathic effect of vesicular stomatitis virus or encephalomyocarditis virus on CCL54 cells (ATCC No. CCL54, transfer 24) (Stewart, W.E. II, The Interferon System, Springer, Berlin, 1979). Gamma-interferon units are expressed with reference to the NIH α-interferon standard. Chloramphenicol acetyl transferase was measured in cell extracts as described by Corman et al., Mol. Cell. Biol. 2^1044-1051; (1982). ;The expression of gamma interferon has provided a sensitive test for quantifying effects of VA-RNA. p-gamma-IF-6 and pL58 are identical to pD20 and pD61 except that the gamma interferon cDNA has been inserted in the correct orientation in a Pstl site between the 3<1> splice site and the DHFR cDNA one. Finally, PQ2 and PQ3 are identical to pL5 8 except that the adenovirus VA genes (adenovirus map units 28.02 and 31.00) have been inserted into an EcoR1 site in both orientations. ;Transformation in COS cells;Plasmids pQ2, pQ3, P158 and p-gamma-IF-6 were all transfected in COS cells. (Table IA). Thirty-six hours later, a sample of the adapted media was taken for analysis, and one of two duplicate plates was infected with adenovirus. Twenty hours after infection, samples were re-analyzed for gamma-IF activity. p-(gamma-IF) D-6 exhibits low activity and only a slight increase after adenovirus superinfection. pL58 exhibits several times more activity than p6 and exhibits a twofold increase in gamma-IF activity after infection with adenovirus. In contrast, both pQ2 and pQ3 had significantly higher levels of gamma-IF activity (50-70 times greater than p(gamma-IF) D-6 in the absence of adenovirus superinfection. After adenovirus superinfection there was a reduction in gamma- IF Activity To determine whether VA RNA mediates the increased expression of gamma interferon, plasmids pQ2 and p(gamma-IF) D-6 have been mixed with either pSVod or PVASVod (a derivative of pSVOd containing adenovirus VA -genes), in a ratio of 1:1 and transfected into COS cells. 48 hours after transfection, samples were taken for gamma-IF analysis. p(gamma-IF) D-6 exhibits a 2-3-fold increase in activity when cotransfected with pVASVOd. In contrast, pL58 shows more than a 10-fold increase after cotransfection with PVASVOd. Finally, pQ2, which contains the VA genes, has high gamma-IF activity when cotransfected with pVASVOd. These results indicate that the presence of VA -genes in trans can facilitate expression from mRNAs containing the majority of the tripartite the leader sequences from adenovirus late mRNAs. The reduction in gamma-IF activity from pQ3-superinfected samples shown in Table 1 may be due to toxic effects due to the adenovirus infection after cellular metabolism or from competition between adenovirus late mRNAs and transcripts from the modulated cDNA gene. The results from the activity tests on gamma interferon have also been worked together with both immunoprecipitation and methods for protein staining (data not shown). The results indicate that pQ2 expresses gamma interferon at ;1 µg/ml 72 hours after transfection. ;These results confirm that VA-RNA is responsible for the increased expression after adenovirus superinfection. The increased translation in the presence of adenovirus VA-RNA also increases the expression of chloramphenicol acetyltransferase from the early SV40 promoter. Cotransfection of a plasmid containing the bacterial gene for chloramphenicol acetyltransferase (CAT) under the control of the early SV40 promoter with plasmids containing the adenovirus VA genes results in 5-10-fold higher levels of CAT activity compared to cotransfection with plasmids lacking VA- the genes. The increased expression is due to increased translation as RNA spot analysis indicates that there is no increase in CAT mRNA after cotransfection with the VA genes. Expression of a plasmid containing the bacterial gene for xanthine-guanine-phos foribosyltransferase under the control of the early SV40 promoter also increases after transfection with plasmids containing the VA genes. In the same way, the expression of other heterologous proteins such as e.g. tissue plasminogen activator, is significantly increased by using extra DNA such as VA genes. A similar increase in translation efficiency has been observed by introducing pL58 and pQ3 into monkey CV1 cells by cotransformation with a plasmid encoding the large T antigen of SV40. The increased translational efficiency is therefore not limited to COS cells. In addition, the translation of both internal and secreted (DHFR and gamma-interferon) proteins can be stimulated by VA-RNA. Cells transfected with the cDNA recombinant for interferon gamma, pQ2, produce interferon at an amount of 1 µg/10^ cells 72 hours after transfection. ;This is a very high expression level when one assumes a transfection efficiency of 5-10%. Finally, the increased translation efficiency is not unique to DHFR and gamma interferon as both human interleukin 2 and human proinsulin have been efficiently expressed in similar vectors. ; Text to Table 1A;γ-interferon activity in transfected and adenovirus-infected COS cells;A. COS cells were transfected with the indicated plasmid DNAs. Where indicated, cells were then either artificially infected or infected with adenovirus as described above. Samples were taken for γ-interferon analysis at the time of the adenovirus infection (36 hours after transfection) and 18 hours after infection (56 hours after transfection). The activities were determined and expressed as units/ml/10 cells as described above. B. COS cells were transfected with the indicated plasmid DNAs (2 µg/ml with an equal amount of either pSVOd or pVASVOd) and samples were taken for γ-interferon assay 60 hours after transfection. ;Example 3;Construction of vector p91023( B);The transformation vector pAdD26SVpA(3) was described by Kaufman et al., Mol. Cell Biol. 2 (11):1304-1319 (1982). It has the structure illustrated in fig. 3. Briefly, this plasmid contains a mouse dihydrofolate reductase (DHFR) cDNA gene that is under the transcriptional control of the major late promoter from adenovirus 2 (Ad2). In adenovirus DNA, a 5<1> splice site is included and a 3' splice site, derived from an immunoglobulin gene, is present between the major late promoter from Ad2 and the DHFR coding sequence. The early SV40 polyadenylation site is present downstream of the DHFR coding sequence. The prokaryote-derived portion of pAdD26SVpA(3) is from pSVOd (P. Mellon, V. Parker, Y. Gluzman and T. Maniatis, 1981, Cell 27:279-288) and does not contain the pBR322 sequences known to inhibit replication in mammalian cells (M. Lusky and M. Botchan, 1981, Nature (London) 293:79-81). ;pAdD26SVpA(3) was transformed into plasmid PTPL as illustrated in fig. 3. pAdD26SVpA(3) was converted into plasmid pAdD26SVpA(3)(d) by removing one of the two Pstl sites in pAdD26SVpA(3). This was done by a partial digestion with Pstl (using a lack of enzyme activity so that a subpopulation of plasmids made straight-chain and in which only one Pstl site is cleaved can be obtained), then treatment with Klenow, ligation for recirculating the plasmid, transforming E. coli and sorting to remove the Pstl site located 3' from the SV40 polyadenylation sequence. ;The tripartite director from adenovirus and virus-associated genes (VA genes) was inserted into pAdD26SVpA(3)(d) as illustrated in fig. 3. First, pAdD26SVpA(3)(d) was digested with PvuII to produce a straight-chain molecule opened in the 3' part of the first of the three elements comprising the tripartite director. Then pJAW 43 (Zain et al., 1979, Cell 16 851) was digested with Xho 1\, treated with Klenow, digested with PvuII and the 138 bp fragment containing the second director and part of the third director was isolated by electrophoresis on an acrylamide gel (6% in tris-borate buffer; Maniatis et al., [1982] supra). The 138 base pair fragment was then ligated into the PvuII-cleaved pAdD26SVpA(3)(d). The ligation product was used to transform E. coli to tetracycline resistance and colonies were sorted using the method of Grunstein-Hogness using a 32p labeled probe hybridizing to the 140 base pair fragment. DNA was prepared from positively hybridized colonies to test whether the reconstructed PvuII site was 5' or 3' to the inserted 138 base pair DNA specific for the second and third late adenovirus leaders. In the correct orientation, the PvuII site is on the 5' side of the 138 base pair insert. This plasmid is designated pTPL in fig. 3. The Ava II D fragment from SV40 containing the SV40 enhancer sequence was obtained by cleaving SV40 DNA with Ava II, ligating Xho-1 linkers to the fragments, cleaving with Xho-1 to open the Xho-1 site and isolating the fourth largest (D) fragment by gel electrophoresis. Introduction of insertion of the linked D fragment provided simple direct replication of the SV40 enhancer. ;This was the result of proportioning the amount of pAdD26SVpA(3) to the Xho-1 linked D fragment in the ligation. The orientation of the SV40 D fragment in pCVSVL2-TPL was such that the late SV4 0 promoter was in the same orientation as the major late adenovirus promoter. To introduce the adenovirus virus-associated (VA) genes into pCVSVL2-TPL, a plasmid pBR322 containing the adenovirus type 2 Hind III B fragment was first constructed. Adenovirus type 2 DNA was digested with Hind III and the B fragment was isolated after gel electrophoresis. This fragment was then inserted into pBR322 which had previously been digested with Hind III. After transformation of E. coli to ampicillin resistance, the recombinants were sorted for insertion of the Hind III B fragment and the orientation after insertion was determined by restriction enzyme digestion. pBR322 - Ad Hind III B contains the adenovirus type 2 Hind III B fragment in the orientation shown in fig. 4. As illustrated in fig. 4, the VA genes were easily obtained from plasmid pBR322-Ad Hind III B by cleavage with Hpa I, addition of EcoRl linkers and cleavage with EcoRl, and recovery of the 1.4 kb fragment. The fragment with EcoRl protruding ends was then ligated into the EcoRl site of pTPL (which had previously been cleaved with EcoRl). After transf ojma^jon__av_Ej. coli HBlOp. and selection for tetracycline resistance, colonies were sorted by filter hybridization to a DNA probe specific for the VA genes. DNA was prepared from positively hybridized clones and characterized by restriction endonuclease digestion. The product plasmid was designated p91023. ;The 2 EcoRl sites in p91023 were removed. p91023 was cut completely with EcoRl, yielding two DNA fragments: one of ca. ;7kb containing Ad-MLP,- TPL,-DHFR, the SV40pA site, the SV40 Avall fragment and the pBR322 Tet gene, and origin of replication (minus the venom region) and, secondly, a 1.3Kb fragment containing VA- the genes. The ends of both fragments are blunted using the Klenow fragment from Poll, and then both fragments, i.e. 1.3Kb and 7Kb, again ligated together. A plasmid p91023(A) containing the VA genes and corresponding to p91023 except for the deletion of the 2 EcoR1 sites was identified by Grunstein-Hogness sorting with the VA gene fragment, and by conventional restriction site analysis. ;Next, the single Pstl site in p91023(A) was removed and replaced with an EcoRl site (Fig. 5). p91023(A) was cut completely with Pstl and then treated with Klenow fragment of Poll to produce blunt ends. EcoRl linkers were ligated to the aligned Pstl site of p91023(A). The straight-chain p91023(A) together with EcoRl linkers attached to the aligned Pstl site was separated from unligated linkers and cleaved completely with EcoRl, then re-ligated. A plasmid p91023(B) was recovered and identified with a structure similar to p91023(A); but with an EcoRl seat placed in the former Pstl seat. Plasmid p91023(B) with a product gene inserted into the EcoR1 site has been deposited with and is available from the American Type Culture Collection, Rockville, Maryland under accession number ;ATCC 39754 as pCSF-1 in E. coli. ;Example 4;The applicability of the vectors containing a translational activator (adenovirus VA genes) as described in Examples 2 and 3 above has been demonstrated both in transiently transfected cells and stably transformed cells with a cDNA encoding human antithrombin III (ATIII) . ;Cloning of the human antithrombin III cDNA;A full-length cDNA encoding human antithrombin III was isolated from a human liver cDNA pool by hybridization to specific oligonucleotide probes derived from the amino acid sequence (T.E. Petersen, G. Dudek-Wojciechowska, L . Sottrup-Jensen and S. Magnusson (1979)) in Physiological Inhibitors of Coagulation and Fibrinolysis (D. Coilen, B. Wiman and M. Verstraete, eds., Elsevier, Amsterdam, pp. 43-54). A pool of human liver cDNA was prepared using standard procedures (Maniatis et al.). PolyA<+->;mRNA was isolated from human liver, treated with methylmercury and synthesis of the first cDNA chain was performed with reverse transcriptase. After base treatment, synthesis of the second strand was performed with Klenow fragment of DNA polymerase I and the double-stranded cDNA was treated sequentially with Sl nuclease, EcoRl methylase and Klenow fragment of DNA polymerase I to smooth the ends. EcoRl linkers were added, excess linkers were removed by EcoRl cleavage and passage through the CL4B column. The cDNA was ligated to gtlO, encapsulated in vitro and plated (C600 hfl, high-lysogenization frequency from Ron Davis, Stanford University). A total of 500,000 were examined with 2 sets of oligomers made from amino acids 243 to 248 (Met-Met-Tyr-Gln-Glu-Gly) and from amino acids 304 to 309 (Glu-Glu-Met-Met-Leu-Val). Based on the degeneracy of the amino acid code, one set of oligonucleotides comprised 8 different 17-mers and the other comprised 16 different 17-mers. ;Hybridization was performed in 5X SSC with 5 µg/ml denatured salmon sperm DNA, 0.5% SDS and 5X Denhart<1>'s solution. 38 positive samples that hybridized to both probes were isolated, plaque-purified and identified by rehybridization to the oligonucleotide probes. During these experiments, S.C. published Bock et al,, ;(1982) (Bock, S.C, Wion, K.L. , Vehar, G.A. and Lawn R.M. Nuc. Acids Res. 10, 8113-8125) the nucleotide sequence of human ATIII. ;In order to isolate full-length clones, a 17-mer oligonucleotide was therefore prepared that included the initiator ATG [T A T T C C A A TGTGATAGG] and hybridized to the 38 positive samples. ;Of 13 positive samples, one was prepared and designated lambda-ATIII-C3 (AT3-C3) which contained an addition of approximately 1.4 kb. ;Insertion of ATIII cDNA into expression vector;The expression vector used for ATIII expression was derived from pQ2 (Fig. 2 (c)). pQ2 was digested with PstI which removed the interferon cDNA and the 7.5 KB fragment isolated from an agarose gel. As the ATIII cDNA contained no internal EcoRI sites, it was possible to excise by EcoRI digestion and isolate the 1.4 KB fragment from lambda-AT3-C3 by agarose gel electrophoresis. The EcoRI-ATIII fragment was inserted into the vector into which the synthetic spacer had previously been added: Pstl<5>'gGCGAGCCTG<3>' EcoRl ;ACGTCCGCTCGGACTTAA;Addition of the synthetic spacer was achieved by phosphorylating only the 5' end of the 10- more than. This leaves a non-phosphorylated EcoRI end that depletes the background of transformants. After removal of excess spacers by low-melting agarose gel electrophoresis, the 7.5 KB fragment was extracted and ligated to the EcoRI-ATIII cDNA fragment. The ligation mixture was used to transform E. coli HB10L to tetracycline resistance and positive samples identified by sorting by the method of Grunstein-Hogness (1975 PNAS, ;72, 3961-3965) using a T4 end label (P.T. Englund, ;(1971), J.B.C., 246, 3269-3276) cDNA fragment encoding ATIII cDNA. Two clones obtained, p91023 ATIII-C3, contain ATIII cDNA in the correct orientation with respect to the adenovirus promoter and p91023 ATIII-E1 contain cDNA in the opposite orientation. Plasmid p91023 ATIII-C3 (pATIII-C3) (Fig. 6) has been deposited in a strain of E. coli-HBlOl at, and may be obtained from, the American Type Culture Collection, Rockville, Maryland under accession number ATCC 39941. p91023-ATIII -C3 and p91023-E1 have the pBR322 origin of replication and the tetracycline resistance gene for propagation in E. coli and the SV40 origin and 1.1 KB deletion of the pBR322 "poison" region for replication in COS monkey cells (Lusky, M. and M. Botchan, (1981), Nature, 293, 79:81). Elements for activating expression in cDNAs in COS monkey cells include the adenovirus major late promoter (AdMLP; the SV40 enhancer; (SV40-AVA-IID fragment); a cDNA copy of the adenovirus tripartite leader; a hybrid -intron comprising a 5' splice site from the second intron of the tripartite leader sequence and a 3' splice site from a mouse immunoglobulin gene (Kaufman, R.J. and Sharp P.A. (1982) Mol. Cell. Biol., 2, 1304-1319 ); the early polyadenylation signal region from SV40; and the adenovirus VA genes. The orientation of the 1.4 Kb cDNA encoding human antithrombin III is indicated. p91023-ATIII-C3 contains the ATIII cDNA in the correct orientation for transcription from Ad-MLP.;COS cell expression;p91023-ATIII-C3 and p91023-ATIII-E1 were introduced by means of DEAE-dextran-mediated transfection into COS monkey cells as described in Example 1. 24 hours after transfection, the cells were rinsed with serum-free Dulbecco's modified Eagle's medium and adjusted ed media were removed 24 hours later for radioimmunological analysis using rabbit anti-human antithrombin III (obtained from Robert Rosenberg, Massachusetts Institute of Technology). Alternatively, cells were labeled with ;35 ;S methionine (lmCi/ml) for 4 hours 48 hours after transfection. Appropriate media were removed and cell extracts prepared for immunoprecipitation and SDS-gel electrophoresis as described in ex. ;1. Test results for ATIII in transfected COS cells in adapted media are shown in Table 2. ; Chinese hamster ovary (CHO) cells lacking DHFR have been described (Urlaub and Chasin, PNAS, 77, 4216-4220, ;[1982]). Plasmids p91023-ATIII-C3, pSV2Neo (P. Southern and ;P. Berg, 1982, J. Mol. Appl. Genet., 1, 327-341) and pAdD26SVpA(3) ;(R.J. Kaufman and P.A. Sharp, Mol. Cell. Bio., (1982), 2, 1304-1319) were mixed (25 µg p91023-ATIII-C3, 2.5 µg pSV2Neo and 2.5 µg pAdD26SVpA3) and precipitated by addition of sodium acetate (pH 4.5) up to 0.3M and 2.5 mol ethanol. Precipitated DNA was allowed to air-dry, resuspended in 2X HEBSS (0.5 ml) (Chu and Sharp "Gene", 13:1970202 [1981]), mixed thoroughly with 0.25 M CaCl2 (0.5 ml); as described (Kaufman and Sharp, "J. mol. Biol.", 150:601-621 ; [1982]). The calcium phosphate-DNA precipitate was incubated for 30 min. at room temperature. Then, media from CHO DUKX-B1 cells previously subcultured at 5x10 3 cells/10 cm plate (24 hours before transfection) was removed and the DNA-calcium phosphate precipitate added to the monolayer. After 30 minutes of incubation at room temperature, 5 ml of a media (Flow Labs) with 10% fetal calf serum was added and the cells were incubated at 37°C for 4.5 hours. The media was then removed from the monolayer of cells, 2 ml of a-media containing 10% glycerol was added over 3 min. at room temperature (24°C) and then removed, and the cells were rinsed and fed with α media containing 10% fetal calf serum, 10 µg/ml of thymidine, adenosine, deoxyadenosine, penicillin and streptomycin respectively. 2 days later, the cells were subcultured 1:15 in α media with 10% dialyzed fetal calf serum, penicillin and streptomycin, but without the nucleosides and containing 1 mg/ml G418 (GIBCO). The cells were re-infused with the same selective media (which lacked nucleosides) after 4-5 days. Colonies appeared 10-12 days after subculturing in selective media. Colonies (approximately 20) from each plate were pooled and propagated in increasing concentrations of methotrexate except in the presence of G418 to further select pSV2Neo. However, the maintenance of pSV2Neo was not essential for the amplification of the ATIII cDNA gene. The first pools of transformants showed no ATIII activity whether assayed by the radioimmunoassay of adapted media or by 35S-methionine labeling and immunoprecipitation followed by SDS-polyacrylamide gel electrophoresis. However, after selection in media containing 0.02, 0.1 and 1.0 µM methotrexate, ATIII could be observed using both methods (Table 1). Then, cloned cell lines were obtained by dilution plating. It was also demonstrated that ATIII produced in COS and CHO cells was active in its ability to bind thrombin and that binding was accelerated in the presence of heparin. ;Example 5;cDNA encoding tissue plasminogen activator (human) ;The cDNA gene encoding human tPA was obtained by reverse transcription from mRNA isolated from Bowes melanoma cell line in accordance with conventional methods previously used with other cells that contains large amounts of the desired mRNA. ;A tPA protein was isolated from the Bowes melanoma cell line (available from Dr. Rifken, New York University). Amino acid sequence analysis showed that the protein found in adapted medium from Bowe's melanoma cells contained two different N-termini, one N-terminus containing three more amino acids than the other. These two N-termini are referred to as the glycine N-terminus (Gly- Ala - Arg - Ser - Tyr - Gin - Val - Ile - Cys - Arg - Asp - Glu - Lys - Thr - Gin - Met - Ile - Tyr - Gin - Gin - His) and the serine N-end (Ser - Tyr - Gin - Val - Ile - Cys - Arg - Asp - Glu - Lys - Thr - Gin - Met - Ile - Tyr - Gin - Gin - His) . ;Messenger RNA was isolated from Bowes melanoma cells and used as a template for the synthesis of cDNA as is commonly known in the art. The cDNA was copied to produce a double strand. It was introduced into tetracycline-resistant plasmid vectors by homopolymer end-joining or by synthetic linkers as is well known in the art. ;Collections of cloned plasmids where each clone containing a unique cDNA copy of an mRNA species present in the Bowes melanoma cells were prepared by transformation of E. coli with the vectors and selection of antibiotic-resistant ;E. coli cells. Plasmids in the cDNA collections containing cDNA encoding at least a portion of tPA were identified using the standard screening method of Grunstein and Hogness (Grunstein et al., 1975 PNAS 72, 3961). In this method, DNA is immobilized in a single-stranded form and examined for its ability to hybridize or bind to radioactively labeled probes. Such probes are short DNA oligonucleotides with a sequence corresponding to the amino acid sequence of a small part of the tPA protein. The amino acid sequence of amino acids 15-20 from the N-terminus of tPA was used (indicated in Fig. 7). As some of these amino acids are encoded by more than one trinucleotide codon, a mixture of 17-mer probes covering all possibilities was prepared following standard procedures. A cDNA collection was then screened using the method of Grunstein and Hogness until a clone was found in the collection that hybridized to the probe. Confirmation that the clone encoded a portion of tPA was obtained by the routine procedure known as partial dideoxy primer extension, which determines the sequence upstream of the primer (Wallace et al., "Nuc. Acids Res." 9, 3647-3656 [ 1981]). This showed that the clone contained codons corresponding to the N-terminal amino acid sequence of tPA. ;The cDNA present in the resulting clone was isolated from the plasmid, radiolabeled, and then used as a probe to identify other overlapping fragments of tPA cDNA found in the cDNA pool. This procedure was continued until fragments were identified which together (except for the overlapping duplicate sequences) encoded the "mature" tPA protein and parts of the 5'-3' untranslated regions. Two different collections were screened, whereby cDNA clones were obtained which together covered the entire coding sequence for tPA. The cDNA clones Sam1 and S20 were isolated from an "asymmetrically-linked" collection. This assembly was prepared following conventional procedures (Maniatis et al., Molecular Cloning - A Laboratory Manual, Cold Spring Harbor Laboratories, Cold Harbor, N.Y. (1982)) where cDNA from Bowes cell mRNA was ligated to EcoRl linkers in 3 ' end and SalI linkers at the 5' end, and the linked cDNA inserted into appropriate vectors. cDNA clone Ed1 was isolated from a GC-end-extended cDNA pool (Maniatis et al, supra). Fig. 8a shows that the cDNA clones Sam1, S20 and Ed1 comprise a partially overlapping set spanning the entire coding sequence for tPA. The technique that was used to bring together a single coding sequence from the three fragments is shown schematically in fig. 8b, 8c and 8d. Fig. 8b illustrates the construction of the plasmid used to receive the fragment, and for replication of the complete cDNA gene. The construction of plasmid YI-p5 is described by Botstein et al., "Gene" 8:17-24 [1979]. The yeast 2y plasmid is commercially available. The designation "+" indicates restriction enzyme sites in the plasmids. Treatment with restriction enzymes and/or the Klenow fragment of DNA polymerase I (hereafter sometimes referred to as "Klenow") is indicated by indications next to the arrows. These enzymes are commercially available. Conventional enzyme reaction conditions were used. Note that ligation of the cleavage products of certain restriction enzymes will yield a DNA sequence that is not hydrolyzable by any restriction enzyme. As an example, reference is made to the elimination of the HpaI site after construction of the Y0p4 plasmid. Figures 8c and 8d illustrate the construction of a plasmid containing the full-length cDNA clone from the three cDNA fragments identified in the collections. In fig. 8c, the plasmid Y0p4 is digested with XbaI, the ends blunted with Klenow and the straight-stranded plasmid digested again with EcoRl. The tPA cDNA fragment containing the 5' end of the gene (including part of the untranslated 5' end - see the boxed SalI site in Fig. 7) was isolated from its plasmid by digestion with SalI, treatment with Klenow and cleaved with EcoRl. Straight chain Y0p4 and the tPA cDNA 5' fragment SamI were ligated to produce plasmid 1901 creating a SalI site after ligation between the XbaI and SalI sites. Plasmid 1901 was replicated in E. coli and screened for ampicillin resistance. Plasmid 1901 was then treated as shown in fig. 8b and ligated to the 3' fragment (Ed 1) of the tPa cDNA (liberated from its plasmid by sequential treatment with Bgl II, Klenow and EcoRl). The Bglll seat, indicated in fig. 7 with a frame, is destroyed by ligation to the ClaI cleavage product from plasmid 1901. The Pstl-Bglll and EcoRl-Pstl subfragments of the tPA cDNA are discarded without effect, in the first case due to the region being non-coding, in the second case because the region overlaps the third fragment, S20 (Fig. 8c), which supplements the discarded sequence. These steps yield the plasmid J118 which was replicated in E. coli, and it was screened for ampicillin resistance. Plasmid J118 was digested with EcoR1 and ligated to the central tPA cDNA fragment S20 as shown in Fig. 8c. Correct orientation of the S20 fragment in J205 was confirmed by asymmetric endonuclease cleavage as is generally known in the art. The tPA cDNA gene differs from the previously published sequence for human tPA by a number of nucleotides and an amino acid substitution as shown in fig. 7. ;Example 6;Construction of transformation vector pLDSG ;The starting plasmid in this example is known as pAdD26SVpA(3) ;(Kaufman et al., Mol. Cell. Biol. 2(11):1304-1319 [1982]). ;It has the structure described in fig. 9a. Briefly, this plasmid contains a mouse dihydrofolate reductase (DHFR) cDNA gene that is under transcriptional control of the major late promoter of adenovirus 2 (Ad2). A 5' splice site is included in the adenovirus DNA and a 3' splice site derived from an immunoglobulin gene is present between the major late Ad2 promoter and the DHFR coding sequence. The early SV40 polyadenylation site is present downstream of the DHFR coding sequence. The prokaryotic-derived portion of pAdD26SVp(A)3 is from pSVOd (P. Mellon, V. Parker, Y. Gluzman and T. Maniatis, 1981, Cell 27:279-288) and does not contain the pBR322 sequences known for to inhibit replication in mammalian cells (M. Lusky and M. Botchan, 1981, Nature (London) 293:79-81). ;pAdD26SVpA(3) is converted into plasmid pCVSVL2 at the first step shown in fig. 9a. The Ava II D fragment of SV40 was obtained by cleaving SV40 DNA with Ava II, ligation of Xho 1 linkers to the fragments, cleavage with Xho 1 to open the Xho-1 site, and isolation of the fourth largest (D) fragment by gel electroph oresis . Insertion of the linked D fragment as shown in the first step provided simple direct replication of the SV40 enhancer. This was the result of proportioning the amount of pAdD26SVpA(3) to Xho-1 linked D fragment in the ligation. The orientation of the SV4 0 D fragment in pCVSVL2 was such that the late SV4 0 promoter is in the same orientation as the major late adenovirus promoter. ;pCVSVL2 is converted into plasmid pB2L2 by a three-step procedure designed to make it compatible for joining with the J205 tPA gene from Example 5. First, one of the two Pstl sites in pCVSVL2 is removed by a partial cleavage with Pstl ;( using a lack of enzyme activity to obtain; a subpopulation of plasmids made straight chain, where only one Pstl site is cleaved), then treatment with Klenow, ligation to make the plasmid circular again, transformation of E. coli and sorting to remove the Pstl site located 3' to the SV40 polyadenylation sequence. Second, the plasmid was digested with BglII, treated with Klenow, ligated and transformed colonies of E.coli were sorted by destruction of the BglII site in the immunoglobulin intron (3<1->splice site in Fig. 3a). ;Third, the Pstl site was converted to a Bglll site by cleavage with Pstl, Klenow treatment, ligation to Bglll linkers, and cleavage with an excess (100 units/µg DNA) of Bglll. The resulting straight-stranded DNA was isolated by gel electrophoresis in a low-melting agarose gel (1.2%) in Tris-acetate buffer. This DNA was ligated in vitro at a concentration of 1 µg/ml at 24°C and used to transfect E. coli HB101 (see Maniatis et al., supra). Tetracycline-resistant colonies were grown and DNA prepared and analyzed for the presence of the Bglll site at the 5<1> end of the DHFR cDNA by digestion with Bglll and PvuII, and Pstl digestion. Large-scale DNA preparations were performed by "banding" DNA twice in CsCl. pJ205 from Example 5 was digested with Hind III and SalI, treated with Klenow and ligated to BamHI linkers. ;The resulting DNA was phenol-extracted, chloroform-extracted and ethanol-precipitated by adding sodium acetate ;(pH 4.5) up to 0.3M and 2.5 parts by volume of ethanol. DNA was recovered by centrifugation and the pellet dried. The pellet was resuspended and digested with BamHI (100 units/µg DNA) to cleave the BamHI linkers. The cleavage product was applied to an agarose gel and the 2.1 kb band identified. The band was recovered and DNA obtained by adding an equal volume of a buffer containing 10 mM tris-HCl (pH 7.4) and 1 mM EDTA, and heating for 15 min. at 68°C. The DNA was then phenol-extracted (2x), chloroform-extracted (2x) and ethanol precipitated by adding sodium acetate (pH 4.5) up to 0.3 M and 2.5 parts by volume of ethanol. As illustrated in fig. 9b, vector pB2L2 was digested with BglII, treated with bovine alkaline phosphatase, extracted with phenol (2x), chloroform (2x) and ethanol precipitated by addition of NaOAc (pH 4.5) up to 0.3 M and 2.5 volumes of ethanol. This precipitated DNA was ligated to the BamHI-linked 2.1 kb fragment from pJ205. The ligated DNA was used to transform E. coli HB101. Tetracycline-resistant colonies were screened with a 32p-labeled probe specific for tPA cDNA using the method of Grunstein-Hogness. The probe was prepared by cleavage of J2O5 with Hind III and SalI, and then labeled with T4 DNA polymerase using 32p-α-dCTP. DNA was prepared from positively hybridizing clones and screened for the presence of tPA cDNA by separate restriction enzyme digestions with Hind III, Pvu II and Sacl. These cleavages showed not only the presence of the tPA cDNA in the vector, but also the orientation of the cDNA relative to the promoter. pLDSG is a plasmid containing tPA in the correct orientation. ;Example 7;Co-transformation and amplification;The plasmids pLDSG and pAdD26SVpA(3) (Ex. 6) were mixed together (50 µg pLDSG and 0.5 µg pAdD26SVpA(3) and precipitated by addition of NaOAc (pH 4.5) until 0 .3 M and 2.5 volumes of ethanol. Precipitated DNA was allowed to air dry, resuspended in 2x HEBSS (0.5 ml) (Chu and Sharp "Gene", 13:197-202 [1981]) and mixed vigorously with 0 .25 M CaCl 2 (0.5 ml) as described (Kaufman and Sharp, "J. Mol. Biol.", 150:601-621 [1982]).The calcium-phosphate-DNA precipitate was allowed to stand for 30 min. at room temperature and added to CHO-DUKX-B1 cells (Chasin and Urlaub, P.N.A.S. 77:4216-4220; [1980]). The maintenance of the growth of these cells has been described (Kaufman and Sharp, "J. Mol. Biol." supra and Chasin and Urlaub, supra). ;DUKX-Bl cells are subcultured at 5x10^/10cm plate for 24 hours before transfection. The media is removed and the DNA-calcium-phosphate precipitate is added to the monolayer. After 30 minutes of incubation at room temperature, added 5 ml a-media (Flow) with 10% fetal calf serum and the cells were incubated at 37°C for 4.5 hours. The media was then removed from the monolayer of cells, 2 ml a-media (Flow) containing 10% glycerol was added during ;3 min. at room temperature (24°C) and then removed and the cells rinsed and fed with a-media containing 10% fetal calf serum, 10 µm/ml of thymidine, adenosine, deoxyadenosine, penicillin and streptomycin respectively. 2 days later, the cells were subcultured 1:15 in α media with 10% dialyzed fetal calf serum, penicillin and streptomycin, but lacking nucleosides. The cells were then re-fed with the same selective media (lacking nucleosides) after 4-5 days. ;The colonies appear 10-12 days after subculturing in selective media. Two methods have been followed for methotrexate (MTX) selection and amplification. In the first method shown in Table 3 below, single independent cloned transformants are isolated on the basis of uptake of exogenous DNA (selection gene) and then each clone is propagated under conditions to increase expression of the product gene, i.e. growth in increasing concentrations of methotrexate. In the second method, a mixture of several independent transformants is isolated on the basis of uptake of the exogenous DNA (selection gene) and propagated under conditions to increase expression of the product gene, i.e. growth in increasing concentrations of methotrexate. Individual clones are then isolated from the mass-selected population and analyzed for expression of the product gene. Those clones exhibiting the highest levels of product gene expression are recultured under conditions to further increase product expression (i.e., growth in increasing concentrations of methotrexate in the culture media). ;Results obtained by following procedure 1 are shown in table 4 below. Single clones growing in α-media without nucleosides were selected, propagated and analyzed for tPA activity. Activities are measured as CTA milliunits/cell/day, i.e. mU/cell/day, (see below). Clones exhibiting tPA activity were then selected for resistance to 0.0 2 µM MTX, 0.1 µM MTX and 0.5 µM MTX, respectively. ;Clone 4C1 did not synthesize increasing tPA activity under MTX selection. However, 4C1 was grown in the absence of MTX to generate subclones (H3B and B10A) that together amplify and express high levels of tPA when selected for MTX resistance. Without wishing to limit the present method to a particular hypothesis, this is believed to be due to the segregation of DHFR genes that are not linked to tPA genes in offspring from the original 4Cl transformant. Subclones containing a single DHFR gene linked to tPA genes amplify the DHFR and tPA genes together in these clones after MTX selection. ; Results obtained by following method 2 are shown in table 5. A mixture of approx. 300 of the colonies obtained from subculture after selection in media lacking nucleosides (the colonies from four selection plates). These transformants were selected for sequential resistance to 0.0 2 µM MTX and 0.0 5 µM MTX. Although the initially mixed transformants contained tPA activities of 0.03, 0.9 and 1.9 mU/cell/day after cultivation in media containing 0, 0.0 2 and 0.5 µM MTX, respectively, they expressed the individual clones in the population tPA in the amounts shown in Table 5. The individual clones varied 50-fold in tPA expression level from 0.2 mU/cell/day up to 1OmU/cell/day. This method is preferred for rapidly identifying clones that express large tPA activities. It is obvious that method 1 can be combined with method 2 to give even more fertile transformants. ; Measurement of tPA expression; There are sensitive tPA tests that measure the tPA-catalyzed conversion of plasminogen to plasmin in the presence of fibrin. Plasmin can be detected by the release of 125i-fibrin fragments from<125>i-fibrin which has been immobilized on a plastic plate (Strickland et al, J.Biol. Chem. 251:5694-5702 (1976)), or by the cleavage of a chromogenic substrate (Drapier et al., Biochemie 61:403-471 [1979]). A suitable chromogenic substrate designated S2251 is available from Kabi: Diagnostics, Inc., Greenwich, CT. These samples are well known to those skilled in the art and are referenced in Strickland et al., supra, and in Drapier et al., supra. tPA activity was measured by rinsing a plate of cells (4 x 10<5> cells/10 cm plate) with 5 ml of serum-free media and then adding 4 ml of serum-free media. Cells were incubated for 20 hours at 37°C and samples were taken from the adapted media for analysis. Measurements of activity are expressed as the number of milliunits (mU); per cell per day. Under these test conditions, the Bowes melanoma cell line yields 0.02 mU/cell/day. The specific activity of the human tPA is 100,000 units/mg. The rate of cleavage ;125 ;of the I-fibrin or S2251 substrate is directly proportional to the amount of tPA-catalyzed conversion of plasminogen to plasmin. ;Samples of adapted media from cells lacking tPA expression produce rather small background interference in these samples. These small background disturbances are due to proteases that are not fibrin-activated as the background in the chromogenic substrate sample does not change with the elimination of fibrin from the standard sample. In contrast, the activity of tPA-producing CHO cell lines exhibits fibrin activation very similar to that exhibited by Bowes melanoma tPA. Quantification of tPA activity is obtained by comparison with a standard curve utilizing urokinase (Leo Pharmaceuticals, Belgium). A unit (CTA, Committee on Thrombolytic Agents) activity is defined by comparison with the standard reference urokinase preparation of the WHO. (Johnson et al., Thromb. Diath. Haemorrh 21, p. 259 (1969)). ;The synthesis of tPA is measured by rinsing twice a monolayer of cells (2x10 /lOcm) with 5 ml of methionine-free media and adding 1 ml of methionine-free media containing lmCi 35g~methionine. Cells were incubated for 4 hours at 37°C and the adapted media were analyzed by immunoprecipitation with rabbit anti-human tPA using Staphylococcus aureus as immuno-absorbent. ;Results from gel electrophoresis of the total amount of labeled secreted protein from clone H3B indicate the only major difference between secreted protein from the original H3B subclone propagated in a medium without nucleosides and H3B cells selected after growth in 0.1 µM MTX is the presence of a strong band migrating at 67,000 daltons. This band is immunoprecipitated specifically with a rabbit anti-human tPA antibody and comigrates with tPA co-produced from the Bowes melanoma cell line. These results indicate that after selection of clone H3B for resistance to 0.1 µM MTX, the expression of human tPA is increased 10-fold. ;Example 8;Construction of vector with adjacent linkage ;This example illustrates the construction of a transformation vector where the stop codon for tPA is ligated directed against the start codon of DHFR. This involves the removal of tPA- ;3' and DHFR-5<1> untranslated regions from pLDSG constructed in ;ex. 6. This example is based on the use of DNA oligonucleotides to initiate DNA synthesis from M13 phage templates (Wallace et al., "Science", 209:1396 (1980); Zoller et al., Methods in Enzymology vol . 100 :468-509) . See Fig. 10. ;pLDSG is digested with Bam HI and the 4.5 kb fragment is isolated by gel electrophoresis. This region contains most of the non-pBR/322 regions of pLDSG. Phage Ml3 mp8 digested with BamHl. Then the straight-chained phage DNA is ligated to the 4.5 kb fragment from pLDSG and the Ml3 construct is used to transform a bacterial host (JM103). Plaques are sorted using the method of Benton-Davis "Science" 196 [ 1980] using ^^ P-labeled probe that hybridizes to the tPA gene. Positively hybridizing clones are grown and DNA imprinted is prepared. Correct orientation of the tPA gene can be determined according to known techniques using restriction endonuclease digestion (BamHl , EcoRl and PvuII).; Phage DNA is isolated from a plaque that exhibits correct o orientation. This single-stranded DNA can then be used with a synthetic primer that "bridges" the region of the phage to be removed. The function of the primer is to hybridize to the DNA that is 5' and 3' of the removed region, thereby bringing the unwanted DNA out of position by loop formation and preventing it from acting as a template in the subsequent primer extension step . In fig. 10 are the 5'- and 3'-ends of the primer and the corresponding sites in M13-LDSG are designated B and A respectively. gene-TAA-ATG-the following 9 translated nucleotides at the amino end of DHFR-3<1>. The phage DNA and primer are mixed and the primer is extended by T4 DNA ligase and the Klenow fragment of DNA polymerase I as generally indicated in the publications of Wallace et al., and Zoller et al., supra. The product is used to transform JM103. Phage DNA imprinted from plaques hybridizing to ^^-labeled primer and confirmed by Sanger (Sanger et al., Proe. Nat. Acad. Sei. 74, pp. 5463-5467 (1977)) by dideoxynucleotide sequencing has had the unwanted region removed, is cleaved with BamHl and the tPA gene-containing fragment is isolated by gel electrophoresis. This fragment is isolated from the gel and ligated to the pLDSG 2.5 kb BamHI fragment obtained in the BamHI digestion described above. The ligation products are used to transform E. coli HB101, tetracycline-resistant clones are identified, plasmid DNA is prepared and characterized. A plasmid with the correct orientation of the 2.5 and 4.5 kb fragments is identified as pLDSGL. ;This plasmid can be used as a linked vector in example 7 instead of pLDSG and pAdD26SVpA(3). ;Example 9;Co-transformation with PBR deleted vectors;pLDSG is cleaved with BamHI and the 4.5 kb fragment is isolated as shown in fig. 10. Practically the entire pBR322 region is discarded together with the 2.5 kb fragment from the BamHI cleavage. pAdD26SVpA(3) is cleaved in the same way. The tPA and DHFR gene-containing fragments are used to transform CHO cells as shown in Ex. 7 for pLDSG and pAdD26SVpA(3). ;Example 10;Vector containing the tripartite adenovirus driver;and VA genes;The method according to this example is illustrated in;Fig. 11. To construct a vector containing the tripartite adenovirus driver, start with pB2L2 and cleave it with PvuII to produce a straight-chain molecule. Then pJAW 43 (Zain et al., 1979, Cell 16, 851) is digested with Xho 1, treated with Klenow, digested with PvuII and the 138 base pair fragment is isolated by electrophoresis on an acrylamide gel (6% in tris-borate buffer; Maniatis et al., [1982] supra). The 138 base pair fragment is then ligated to the PvuII-cleaved pB2L2. The ligation product is used to transform E. coli to tetracycline resistance and colonies are sorted using the Grunstein-Hogness method using a P-labeled probe that hybridizes to the 140 base pair fragment. DNA is prepared from positively hybridizing colonies to test whether the reconstructed PvuII site is 5' or 3' to the inserted 138 base pair DNA specific for the second and third late adenovirus leaders. In the correct orientation, the PvuII site will be on the 5' side of the 138 base pair insert. This plasmid is designated pB2L2-TPL. To introduce the adenovirus virus-associated (VA) gene into pB2L2-TPL, adenovirus type 2 DNA is first digested with HindIII. The E-fragment is isolated after gel electrophoresis according to known methods. This fragment contains the VA genes. This fragment is treated with Klenow, digested with HpaI, ligated to EcoRl linkers (Collaborative), digested with EcoRl and the 1.3 kb band is isolated from an agarose gel. This fragment is ligated into the EcoRl site of pB2L2-TPL (which has been previously digested with EcoRl). After transformation of E. coli HB101 and selection for tetracycline resistance, the colonies are sorted by filter hybridization to a DNA probe specific for the VA genes. DNA is prepared from positively hybridizing clones and characterized by restriction endonuclease digestion. The product plasmid is designated pB2L2-TPLVA. It is used with advantage instead of pB2L2; in e.g. 6 to produce modified pLDSG which is again used with pAdD26SVpA(3) in the method according to ex. 7. ;Example 11;Construction of Vector Containing an Immunoglobulin Enhancer ;The vector pSer has been described (Gillies et al., "Cell", 33:717-728 [1983]). It is a derivative of pSV2GPT in which the SV40 enhancer has been deleted (PvuII to Sph 1, base pairs 64-270) (see Mulligan et al., "Science" 209:1422-1427 [1980]). This DNA codes for resistance to mycophenolic acid when cells are grown in the presence of xanthine and hypoxanthine. A 1 kb fragment containing an immunoglobulin enhancer derived from the mouse immunoglobulin constant region gene has been inserted into the EcoRl site of pSer to derive pSerx2/3. ;This vector has a single BamH1 site. It is cleaved with BamHl, treated with alkaline phosphatase and ligated to a fragment of pLDSG prepared by BamHl digestion and isolation of the large (4.5 kb) DNA fragment as shown in Ex. 8. Ligated DNA is used to transform E. coli HB101 which is selected for ampicillin resistance, and the colonies are sorted by the method of Grunstein-Hogness for the presence of the tPA gene by hybridization to a 32p-mer]cet tPA probe. This probe can be prepared by cleavage of J205 with HindIII and SalI and labeling of DNA with T4 DNA polymerase in the presence of a<32>P-α-dCTP and cold dTTP, dATM and dGTP. Positively hybridizing clones are cultured and plasmid DNA is harvested. Fig. 12a illustrates this vector (p7B1). This DNA can be introduced into myeloma cells J558L as described by Gillies et al., supra, which is a modification of the method of Sandri-Goldin for protoplast fusion ("Mol. Cell. Bio." 1:743-752 (1981)). 36 hours after protoplast fusion, cells are cultured in media containing 5 mg/ml mycophenolic acid, 250 µg/ml xanthine and 15 mg/ml hypoxanthine to select for cells that have incorporated the plasmid DNA. One particular clone isolated in this way expressed tPA at a level of 0.05 mU/cell/day. Another suitable parental cell line is ATCC CRL1580 or another non-differentiating myeloma cell line. ;Example 12;Construction of transcriptionally activated vector;The trans-acting transcriptional activator EIA;is used in conjunction with a vector containing a promoter for early adenovirus region 2. This promoter, in contrast to the major late adenovirus promoter, is activated by E1A- the protein product. ;pCVSVL is described by Kaufman et al., "Mol. Cell Biol." 2(11:1304-1319 (1982). pCVSVL is cleaved with Bal 1, treated with Klenow, ligated to XhoI linkers (Collaborative), XhoI cleaved, EcoRl cleaved and the 4.2 kb fragment isolated by gel electrophoresis. ;Adenovirus 2 DNA is digested with EcoRl and the EcoRl F fragment (map unit 70.7-75.9) is recovered. The F fragment was digested with Xhol and the E2 promoter subfragment is isolated by gel electrophoresis. This subfragment has terminal EcoRl and Xhol cohesive ends and ligated appropriately to the terminal EcoRl and XhoI ends of the 4.2 kb fragment from pCVSVL obtained as described above. ;The resulting plasmid is used to transform bacteria. ;A clone that hybridizes to a labeled E2 probe is recovered as pE2-7. pE2-7- is a DHFR cDNA gene utilizing the adenovirus early region 2 promoter. The construction of pE2-7 was provided by Drs R. Kingston and P. Sharp. See Mol. Cell Biol. vol. 4 , pp. 1970-1977 (1984). ;pE2-7 is partially cleaved with Pstl, treated with Klenow, ligated to Bglll linkers and Bglll-s shared. This allows the appropriate insertion of the tPA gene from pJ205 (Fig. 9b) into pE2-7. A part of pJ205 carrying the tPA gene is excised from pJ205 as shown in fig. 9b and the cohesive BamH1 ends are ligated to the protruding BglII ends of pE2-7, yielding plasmid pE2-7PA shown in FIG. 12 (b) . pE2-7PA is a plasmid containing a tPA gene utilizing the adenovirus early region 2 promoter. The E1A gene and promoter are obtained as the KpnI fragment (0 to 5.8 map units) from adenovirus 2. ;pE2-7PA, pE2- 7 and the ElA gene vector are used in molar ratios of (20:1:4) to transform CHO cells as previously described in ex. 7. Transformants isolated as such produce 0.008m units/cell/day of tPA activity. Transformants obtained by omitting the ElA gene (pE2-7PA with pE2-7 in a 20:1 ratio) express only 0.0003 mU/cell/day. ;Example 13;Changes in cell morphology associated with elevated tPA expression ;Growth and selection under conditions to optimize for tPA synthesis result in cells with altered morphology that are unable to adhere to tissue culture plates. ;The mixture of 300 transformants grown in a medium without nucleosides (as described in Ex. 7, Table 4) producing low levels of tPA (0.01 mU/cell/day) has a uniform morphology characteristic of CHO cells. After selection as described in Table 4 for growth in MTX up to 0.05 µM, levels of tPA increase to 0.6 mU/cell/day and changes in cell morphology become apparent. These changes are even more striking in clones producing more than 1.2 mU/cell/day (eg clones 9, 12 and 20). These cells do not adhere to the tissue culture plate and are therefore difficult to select for higher levels of tPA expression. This is overcome by adding aprotinin (Sigma) (0.5% to 5% vol/vol) to the culture medium or by using fetal calf serum that is free of plasminogen. The result of these treatments is to alleviate the toxicity imposed on cells that produce high levels of tPA. Plasminogen-free serum can be obtained by passing the serum over a column of lysine-Sepharose 4B (Pharmacia) as described (Deutch & Mertz, Science (1970), p. 1095). ;Example 14;Expression of EPO in COS cells;By means of genetic engineering, a clone containing the erythropoietin (EPO) gene was produced from a human fetal liver collection. An EPO clone designated Iambda-HEP0FL13 has the nucleotide and amino acid sequence illustrated in FIG. 13. The clone lambda-HEPOFL13 was inserted into the p91023 (B) vector (Ex. 3) and was transfected into COS-1 cells using standard genetic engineering procedures well known to those skilled in the art. 8 µg of each purified plasmid DNA was then used to transfect 5 x 10 COS-1 cells using the DEAE-dextran method (Sompayrac et al., Proe. Nat. Acad. Sei., 78:7575- 7578 (1981) and Luthman et al., Nuc. Acids Res., 11:1295-1308 (1983)). After 12 hours, the cells were washed and treated with Chloroquine (0.1M) for 2 hours at 37°C, washed again and exposed to 10 ml medium containing 10% fetal calf serum for 24 hours. The medium was replaced with 4 ml of serum-free medium and it was harvested 48 hours later. Production of immunologically active EPO (a) was quantified using a radioimmunological assay as described by Sherwood and Goldwasser (Blood, 54:885-893 (1979)) and was found to be 300 ng/ml. The p91023(B) vector containing lambda-HEP0FL13 was also transfected into COS-1 cells without the chloroquine treatment and the cells were cultured as described above. In vitro biologically active EPO was measured using either a colony-forming assay with mouse fetal liver cells as a source of CFU-E or a<3>H-thymidine uptake assay using spleen cells from phenylhydrazine-injected mice, obtaining values of resp. 2 and 3 U/ml. In vivo biologically active EPO was also measured using either the method with hypoxic mice or starved rats, whereby values of resp. 1 (CFU-E) and 2 (3H-Thy) U/ml. ;Example 15;Expression of EPO in CHO cells- cells;Construction of vector RK1-4;Bam HI-PvuII fragment from plasmid pSV2DHFR (Subramani et al., Mol. Cell. Biol. 1:854-864 (1981) containing the promoter of the SV40 early region flanking the mouse dihydrofolate reductase (DHFR) gene, an SV40 enhancer, the small t-antigen intron and the SV40 polyadenylation sequence was isolated (fragment A). The remaining fragments were obtained from the vector p91023 ( A) (supra) as follows. p91023(A) was cleaved with Pst I in the only Pst I site near the adenovirus promoter to make the plasmid straight-chain and either ligated to synthetic Pst I to EcoRl converters and re-circularized ( by creating the sites Pst I : EcoRl - Pst I in the original Pst I site; 91023(B') or treated with the large fragment of DNA polymerase I to destroy the Pst I sites and ligate to a synthetic EcoRI linker , and re-ring-shaped (by producing an EcoRl site in the original Pst I site; 91023 (B). Both of the two resulting plasmids 91023(B) and 91023(B<*>) were digested with Xba and EcoRl to produce two fragments (A and B).

bringing fragment A from p91023 (B) and fragment p91023 (B') together created two new plasmids containing either an EcoRI-Pst I site or a Pst I - EcoRl site in the original Pst I site. The plasmid containing a Pst I-EcoRI site where the Pst I site is closest to the major late adenovirus promoter was designated p90123(C).

The vector p91023(C) was digested completely with Xhol and

the resulting straight-stranded overhanging DNA was blunt-ended by an end-filling reaction with the E. coli large fragment of DNA polymerase I. To this DNA was ligated a 340 bp Hind III-EcoRI fragment containing the SV40 enhancer prepared as follows.

The Hind III-PvuII fragment from SV40 containing the SV40 origin of replication and the enhancer was inserted into the plasmid tt lac (Little et al., Mol. Biol. Med. 1:473-488 (1983)).

The vector tt lac was prepared by cleaving the tt lac DNA with BamHI, filling in the overhanging end with the large fragment of DNA polymerase I and cleaving the DNA with Hind III. The resulting plasmid (TrSVHPlac) regenerated the BamHI site by ligation to the blunt PvuII end. The EcoRl - Hind III fragment was prepared from SVHPlac and ligated to the EcoRl - Hind III fragment of PSVod (Mellon et al., supra) containing the plasmid origin of replication and it was selected for the resulting plasmid pSVHPod. The EcoRl - Hind III fragment from PSVHPod of 340 bp containing the SV40 origin/enhancer was then prepared and both ends ligated with the large fragment by DNA polymerase I. - The resulting plasmid (p91023(CO/Xho/butt plus EcoRI/Hind III/- blunt SV40 origin plus -enhancer) where the orientation of Hind III

- The EcoRI fragment was such that the BamHI site in the fragment closest to the VA gene was designated pES105. The plasmid pES105

was digested with BamHl and PvuII, and PvuII alone and the BamHl - PvuII fragment containing the major late adenovirus promoter (fragment B), and the PvuII fragment containing the plasmid in the resistance gene (tetracycline resistance), and other sequences (fragment C) were also isolated. Fragments A, B and C were ligated and the resulting plasmid shown in fig. 14 were isolated and designated RK1-4. Plasmid RK1-4 has been deposited at the American Type Culture Collection, Rockville, Maryland, where it is available under accession number ATCC 39940.

Expression of EPO

DNA from clone lambda HEPOFL13 was cleaved with EcoRl and the small RI fragment containing the EPO gene was subcloned into the EcoRl site of plasmid RKl-4. This DNA (RKFL13) was then used to transfect the DHFR-deficient CHO cells directly (without cleavage) and the selection and amplification was performed as indicated below. After growth for 2 days, cells were selected that incorporated at least one DHFR gene in α-medium without nucleosides and supplemented with 10% dialyzed fetal bovine serum. After cultivation for 2 weeks in selective media, colonies were removed from the original plates, pooled into groups of 10-100 colonies per well. mixture, re-plated and grown to confluence in α-medium without nucleosides. Supernatant media from the mixtures cultured before methotrexate selection were analyzed for EPO by RIA. Mixtures showing positive EPO production were immediately subcloned, reanalyzed and secondary positive cultures were then subjected to stepwise methotrexate selection (see below). For mixtures that were negative by RIA, methotrexate-resistant colonies obtained from the opposite cultures grown in the presence of methotrexate (0.2 µm) were again reanalyzed in mixtures for EPO by RIA. Those cultures that were positive were subcloned and grown in further increasing concentrations of methotrexate.

Stepwise methotrexate (MTX) selection was achieved by repeated cycles of culturing the cells in the presence of increasing concentrations of methotrexate and selecting for surviving cells. For each round, EPO was measured in the culture supernatant by RIA and by in vitro biological activity. The methotrexate levels used in each stepwise boost were 0.02 µM, 0.1 µM and 0.5 µM.

The RKFL13 DNA was also introduced into CHO cells by microinjection.

The levels of EPO expression obtained are shown below

in table 6.

The present invention has been described in detail including the preferred embodiments. However, it should be understood that those skilled in the art, after taking this description into account, can make modifications and improvements within the spirit and scope of the present invention.

Claims (28)

1. Vector system for use when introducing heterologous DNA into eukaryotic cells, characterized in that the vector system comprises a product gene and extra DNA. 2. Vector system according to claim 1, characterized in that the additional DNA comprises a gene that codes for a translation activator, a gene that codes for a trans-acting transcription activator, or a fragment of genomic DNA from the eukaryotic cell to be transformed. 3. Vector system according to claim 2, characterized in that the additional DNA comprises a gene that codes for a translation activator. 4. Vector system according to claim 1, characterized in that it further comprises an adenovirus^ important promoter which contains the tripartite controller and where the additional DNA comprises a gene which codes for a translation activator. 5. Vector system according to claim 4, characterized in that the gene that codes for the translation activator is an adenovirus VA gene. 6. Vector system according to claim 2, characterized in that the additional DNA comprises a gene that codes for a trans-acting transcription activator. 7. Vector system according to claim 6, characterized in that the gene is the EIA polyoma large T antigen, the SV40 large T antigen or the myc gene. 8. Vector system according to claim 1, characterized in that the product gene codes for human tissue plasminogen activator. 9. Plasmid p91023(B) containing a product gene. 10. Eukaryotic cell containing the plasmid according to claim 9. 11. Cell according to claim 10, characterized in that the cell is a CHO cell. 12. The plasmid pRK1-4 containing a product gene. 13. Eukaryotic cell containing the plasmid according to claim 12. 14. Cell according to claim 13, characterized in that the cell is a CHO cell. 15. Vector system according to claim 1, characterized in that it comprises an immunoglobulin enhancer. 16. Cell containing the vector system according to claim 15. 17. Cell according to claim 16, characterized in that the cell is a myeloma cell. 18. Vector system according to claim 1, characterized in that it comprises a bacterial plasmid containing an SV40 enhancer, an adenovirus important late promoter, an adenovirus tripartite director, an SV40 polyadenylation site and an adenovirus VA gene. 19. Vector system according to claim 18, characterized in that the bacterial plasmid further comprises an SV40 origin site for replication. 20. Vector system according to claim 18, characterized in that the bacterial plasmid further comprises a 3' and a 5' splice site so that an intron is formed. 21. Vector system according to claim 18, characterized in that the bacterial plasmid further comprises a DHFR 3' non-coding sequence located 3' in relation to the product gene. 22. Vector system according to claim 18, characterized in that the bacterial plasmid further comprises a DHFR transcription sequence which is capable of expressing DHFR. 23. Vector system according to claim 21, characterized in that the DHFR transcription sequence comprises an SV40 early promoter, an SV40 early polyadenylation site and a DHFR gene. 24. Method for producing high yields of a heterologous, non-viral protein in a eukaryotic host, characterized by (l) culturing in a culture medium eukaryotic cells containing (a) an expression vector system containing the gene for the heterologous the protein and a promoter operably linked for expression of the gene, and (b) a gene encoding a translational activator; and (2) recovering and separating the heterologous protein from the host cells and culture medium. 25. Method according to claim 24, characterized in that the promoter is the most important late promoter from adenovirus containing the tripartite controller, and that the gene that codes for a translational activator is an adenovirus VA gene. 26. Method according to claim 24 or 25, characterized in that the protein is a tissue plasminogen activator. 27. Protein, characterized in that it is produced by the method according to claim 26. 28. Protein exhibiting the biological properties of human tissue plasminogen activator, characterized in that it is produced by the steps of (A) growing a eukaryotic cell transformed with a DNA sequence as shown in FIG. 7, the DNA sequence being operatively linked to a promoter for expression of the protein, and (B) recovering and separating the expressed protein from the culture medium.