NO875114L - PROCEDURE FOR PREPARING A PREPARING POLYPEPTID CONTAINING A SPECIFIC POLYPEPTID. - Google Patents

Description

The present invention relates to a method for producing a precursor polypeptide containing a specific polypeptide, comprising expression of a DNA sequence in a recombinant microbial cloning vector.

Genetic information is encoded on double-stranded deoxyribonucleic acid ("DNA" or "genes") according to the order in which the DNA coding strand presents the characteristic bases in its repeating nucleotide components. "Expression" of the encoded information to form polypeptides involves a two-part process. Under the direction of certain control regions ("regulons") in the gene, RNA polymerase can be caused to move along the coding chain to form mRNA (ribonucleic acid) in a process called "transcription". In a subsequent "translation" step, the cell's ribosomes together with transfer RNA (tRNA) convert said mRNA "messages" into polypeptides. Included in the information that mRNA transcribes from DNA are signals for the start and termination of ribosomal translation as well as the identity and sequence of the amino acids that make up the polypeptide. The DNA coding chain comprises long sequences of nucleic triplets termed "codons" because the characteristic bases in the nucleotides in each triplet or codon encode specific pieces of information. For example, 3 nucleotides read as ATG (adenine-thymine-guanine) result in an mRNA signal interpreted as "start translation", while termination codons TAG, TAA and TGA are interpreted as "stop translation". Between the start and stop codons lie the so-called structural genes whose codons define the amino acid sequence that is finally translated. This definition proceeds according to the well-established "genetic code"

(eg, J.D. Watson, Molecular Biology of the Gene, W.A. Benjamin Inc., N.Y., 3rd ed. 1976) which describes the codons for the various amino acids. The genetic code is degenerate in the sense that different codons can give the same amino acid, but precisely since for each amino acid there is one or more codons for it and no other. Thus, e.g. all codons TTT, TTC, TTA and TTG read as such,

code for serine and no other amino acid. During transcription, the correct reading phase or reading frame must be maintained. One can e.g. see what happens when the transcribing RNA polymerase reads different bases at the beginning of a codon (underlined) in the sequence ...GCTGGTTGTAAG...:

The polypeptide that is finally produced will then depend entirely on the order of the structural gene with regard to regulons.

A clearer understanding of the process of genetic expression will emerge when certain components of genes are defined: Operon - A gene comprising structural gene(s) for polypeptide expression bg the control region ("regulon") that regulates this expression.

Promoter - A gene in the regulon to which RNA polymerase must bind for initiation of transcription.

Operator - A gene to which a repressor protein can bind and thus prevent RNA polymerase from binding to the adjacent promoter.

Inducer - A substance that deactivates the repressor protein thereby releasing the operator and allowing RNA polymerase to bind to the promoter and begin transcription.

Catabolite activator protein ("CAP") binding site - A gene that binds cyclic adenosine monophosphate ("cAMP") - mediated. CAP, also usually required for initiation of transcription. In special cases, the CAP binding seat may be unnecessary. A promoter mutation in the lactose operon in phage X-plac UV5 eliminates e.g. the necessity of CAMP ig CAP for expression, J. Beckwith et al., J. Mol. Biol. 69, ISS-160 (1972).

Promoter-operator-svstem - In the present context, this means an operable control region in an operon, with or without regard to whether it is included in a CAP binding site or is able to code for repressor protein expression.

For use in the discussion of recombinant DNA below, the following definitions are given: Cloning vector - Non-chromosomal double-stranded DNA comprising an intact "replicon" such that the vector is replicated when placed in a unicellular organism ("microbe") by a "transformation "-process. An organism thus transformed is termed a "transformant".

Plasmid - For present purposes, this is a cloning vector derived from viruses or bacteria, the latter being "bacterial plasmids".

Complementarity - A property given by the base sequences in single-stranded DNA and which enables the formation of double-stranded DNA through hydrogen bonding between complementary bases on the respective chains. Adenln (A) is a complement to thymine (T), while guanine (G) is a complement to cytosine (C).

Advances in biochemistry in recent years have led to the construction of "recombinant" cloning vectors in which e.g. plasmids are processed to contain exogenous DNA. In certain cases, the recombinant may contain "heterologous" DNA, i.e., DNA encoding polypeptides not normally produced by the organism susceptible to transformation by the recombinant vector. Thus, plasmids are cleaved to provide linear DNA having ligatable termini. These are linked to an exogenous gene having ligable termini to provide a biologically functional portion with an intact replicon and a desired phenotypic trait. The recombinant part is introduced into a microorganism by transformation and transformants are isolated and cloned with the aim of obtaining large populations that can express the new genetic information. Methods and means for creating recombinant cloning vectors and transforming organisms with these are widely reported in the literature. See e.g. H.L. Heynecker et al., Nature 263. 746-752 (1976); Cohen et al., Proe. night Acad. Pollock. USA 69, 2110 (1972)_;,. iMd.^, 70, .1293^ (1973);

al., Proe. Nat. Acad. Pollock. U.S.A. 71, 1743 (1974); Novick, Bacterological Rev., 33, 210 (1969); Hershfield et al., Proe. Soc. Nat'l. Acad. Pollock. U.S.A. 71, 3455 (1974) and Jackon et al., ibid. 69, 2904 (1972). A generalized discussion of the subject can be found in S. Cohen, Scientific American 233, 24 (1975).

A number of different techniques are available for DNA recombination according to which adjacent ends of separate DNA fragments are adapted in some way to facilitate ligation. The latter term refers to the formation of phosphodiester bonds between adjacent nucleotides, most often with the aid of the enzyme T4 DNA ligase. Thus blunt ends can be ligated directly. Alternatively, fragments containing complementary single chains at their adjacent ends are favored by hydrogen bonding, setting up the respective ends for subsequent ligation. Such single chains, termed cohesive termini, can be formed by addition of nucleotides to blunt ends using terminal transferase and sometimes simply by breaking down a chain at a blunt end with an enzyme such as X-exonuclease. Here too, and most commonly, one can resort to restriction endonucleases which cleave phosphodiester bonds in and around unique sequences in nucleotides of a length of approx. 4-6 base pairs. Many restriction endonucleases and their recognition sites are known, the so-called EcoRI endonuclease being the most widely used. Restriction endonucleases that cleave double-stranded DNA at rotationally symmetric "palindromes" leave cohesive termini. Thus, a plasmid or other cloning vector can be cleaved leaving termini each comprising half of the restriction endonuclease recognition site. A cleavage product of exogenous DNA obtained with the same restriction endonuclease will have ends complementary to those of the plasmid termini. As described below, alternative synthetic DNA comprising cohesive termini can be provided for insertion into the cleaved

vector. To counteract the reunification of the vector's cohesive

*^ermi<;>n;i-^ DNA -.mentioned termini can be degraded with alkaline phosphatase to provide molecular selection for closures including the exogenous fragment. Incorporation of a fragment that has the correct orientation in relation to other aspects of the vector can be improved when the fragment supplements the vector DNA extracted or removed with two different restriction endonucleases, and when it itself includes termini that respectively make up half of the recognition sequence of the different endonucleases.

Despite extensive work in recent years in recombinant DNA research, few results amenable to immediate and practical application have been obtained. This has been found to be particularly the case in the case of failed attempts to express polypeptides and the like encoded by "synthetic DNA", either by construction nucleotide by nucleotide in the conventional manner or obtained by reverse transcription from isolated in RNA ( complementary or "cDNA"). In the present context, what appears to represent the first expression of a functional polypeptide product from a synthetic gene is described together with related developments that may find wide application. The product referred to is somatostatin (US patent 3,904,594), an inhibitor of growth hormone, insulin and glucagon secretion, the effects of which suggest its use in the treatment of acromegaly, acute pancreatitis and insulin-dependent diabetes. See R. Guillemin et al., Annual Rev. With. 27, 379 (1976). The somatostatin model clearly shows the applicability of the new developments covered by the invention in several and useful areas, which will be apparent from the accompanying drawings and from what is stated below.

The accompanying drawings illustrate a field in which preferred embodiments of the invention find application, i.e. expression of the hormone somatostatin by bacterial transformants containing recombinant plasmids. Figure 1. Schematic overview of the procedure: The gene for somatostatin, produced by chemical DNA synthesis, is fused with the E. coli3-galactosidase gene on the plasmid pBR322. After transformation in E. coli, the recombinant plasmid directs the synthesis of a precursor protein which can be specifically cleaved in vitro at methionine residues by means of cyanogen bromide to obtain active mammalian polypeptide hormone. A, T, C and G denote the characteristic bases (adenine, thymine, cytosine and guanine respectively) in the deoxyribonucleotides of the coding chain of the somatostatin gene. Figure 2. Schematic structure of a synthetic gene whose coding chain (ie the "upper" chain) comprises codons for the arainoic acid sequence in somatostatin (given). Figure 3. Schematic illustration of the preferred method for construction of nucleotide trimers used in the construction of synthetic genes. In the convention The conventional way used to depict nucleotides in Figure 3 is 5'-OH on the left and 3'-OH on the right, e.g. Figure 4. Flow chart for the construction of a recombinant plasmid (eg, pSOM11-3) capable of expressing a somastostatin ("SOM")-containing protein, starting with a parent plasmid pBR322. On figure 4 is the pmjrfnt-l-tge jnp^élt-^ indicated 1 "darton ("d" ) f"~Åpr~ and Tcr bétlegnef hen respectively genes for ampicillin and tetracycline resistance, while Tc<s>denotes tetracycline sensitivity resulting from removal of part of the Tc<r>gene. The relative positions of various restriction endonuclease-specific cleavage sites on the plasmids are indicated (eg EcoRI, Baml, etc.). Figures 5A and 5B. The nucleotide sequences of key parts in two plasmids are indicated along with the direction of an mRNA transcription, which always proceeds from the 5' end of the coding strand. Restriction endonuclease substrate sites are as shown. Each sequence shown contains both the control elements of the lac (lactose) operon and codons for expression of the amino acid sequence I somatostatin (italics). The amino acid sequence numbers of 3-galactosidase ("p<->gal") are given in parentheses. Figures 6-8. As more particularly described in the experimental section below, these indicate the results of comparative radioimmunoassay experiments demonstrating the somatostatin activity of product expressed by the recombinant plasmids. Figure 9. Schematic structure of synthetic genes if

coding chains comprise codons for the amino acid sequences of the A and B chains of human insulin.

Figure 10. Flow chart for the construction of a recombinant

plasmid which can express the B chain in human insulin.

1. Production of genes encoding heterologous polypeptide

DNA encoding any polypeptide of known amino acid sequence can be prepared by selecting codons according to the genetic code. To facilitate purification, etc., oligodeoxyribonucleotide fragments of e.g. from approx. 11 to approx. 16 nucleotides prepared separately and then assembled in the desired sequence. Thus, first and second series of oligodeoxyribonucleotide fragments of appropriate size are produced. The first, when joined in the correct sequence, provides a DNA coding chain for polypeptide expression (see, eg, Figure 2, fragments A, B, C and D). The second series, when similarly joined in the correct sequence, gives a chain complementary to the coding chain (eg, Figure 2, fragments E, F, G and B). The fragments in the respective chains preferably overlap each other so that complementarity promotes their self-assembly through hydrogen bonding of the cohesive termini in fragment blocks. After assembly, the structural gene is completed by ligation in the conventional manner.

The degeneracy in the genetic code allows substantial freedom in the choice of codons for any given amino acid sequence. For the present purpose, however, the codon choice was advantageously guided by three considerations. First, codons and fragments were chosen, and fragment composition arranged, so as to avoid unreasonable complementarity of the fragments, one with another, with the exception of fragments adjacent to each other in the intended gene. Second, sequences rich in AT base pairs (eg, about five or more) are avoided, especially when preceded by a sequence rich in GC base pairs, to avoid premature termination of transcription. Third, at least the majority of the codons selected are those preferred in the expression of microbial genomes (see, eg, W. Fiers et al., Nature 260, 500 (1976)). For present purposes, the following are defined as codons "preferred for expression of microbial genomes":

In the case of somatostatin, the amino acid (codon ) ratios in the structural gene are most preferred: gly (GGT); cys (TGT); light (AAG); trp (TGG); ala (GCT, GCG); asn (AAT, AAC); phe (TTC, TTT); thr (ACT, ACG); and see (TCC,

TCG).

When the structural gene 1 a desired polypeptide is to be introduced into a cloning vector for expression as such, the gene is preceded by a "start" codon (eg ATG) and immediately followed by one or more termination or stop codons (see Fig . 2). However, as described below, the amino acid sequence of a particular polypeptide may be expressed with additional protein preceding and/or following it. If the intended use of the polypeptide requires cleavage of the additional protein, appropriate cleavage sites are encoded for the adjacent junction polypeptide-additional protein codon. In Figure 1 as an example, the expression product is thus a precursor protein comprising both somatostatin and the majority of the ε-galactosidase polypeptide. Here, the ATG is not required to code for the translation start, because ribosomal construction of the additional e-gal protein reads through the structural somatostatin gene. Incorporation of the ATG signal, however, codes for the production of methionine, an amino acid that is specifically cleaved by cyanogen bromide, providing a facile method for converting the precursor protein to the desired polypeptide.

Figure 2 also exemplifies a further feature preferred in heterologous DNA for recombinant use, ie the provision of cohesive termini, preferably comprising one of the two strands of a restriction endonuclease recognition site. For the reasons mentioned above, the termini are preferably constructed to create respectively different recognition sites by recombination.

While the developments described herein have been demonstrated to be successful with the somatostatin model, it will be understood that heterologous DNA encoding virtually any known amino acid sequence may be used, mutatis mutandis. The techniques discussed above and in the following are thus applicable, mutatis mutandis, in the production of poly(amino) acids such as polyleucine and polyalanine; enzymes; serum proteins; analgesic polypeptides such as endorphins, which modulate pain thresholds, etc. Most preferably, the polypeptides produced as such will be mammalian hormones or intermediates thereof. Among such hormones can be mentioned e.g. somatostatin, human insulin, human and bovine growth hormone, leutinizing hormone, ACTH, pancreatic polypeptide, etc. Intermediates include e.g. human preproinsulin, human proinsulin, the A and B chains of human insulin, etc. In addition to DNA produced in vitro, the heterologous DNA may comprise cDNA resulting from reverse transcription from mRNA, see e.g. Ullrich et al., Science 196. 1313 (1977).

2. Recombinants that encode the expression of precursor protein

In the process illustrated schematically in Figure 1, expression yields a precursor protein comprising both a polypeptide encoded by a specific heterologous structural gene (somatostatin) and additional protein (including part of the B-galactosidase enzyme). A selective cleavage site adjacent to the somatostatin amino acid sequence allows subsequent separation of the desired polypeptide from excess protein. The case illustrated is representative of a large class of methods made available through the techniques described herein.

Most commonly, cleavage will be effected outside the replicative environment in the plasmid or other vector such as e.g. after harvesting the microbial culture. In this way, temporary conjugation of small polypeptides with excess protein can preserve the former against e.g. in vivo degradation by endogenous enzymes. At the same time, the additional protein will usually deprive the desired polypeptide of bioactivity until extracellular cleavage, with the effect that the biosafety of the method is increased. In special cases, it may of course be desirable to effect cleavage inside the cell. E.g. cloning vectors can be provided with DNA encoding enzymes that convert insulin precursors to the active form, by operation in tandem with other DNA encoding the expression of the precursor form.

In a preferred case, the particularly desired polypeptide lacks internal cleavage sites corresponding to those used to disperse excess protein, although it will be understood that when this condition is not satisfied, competing reactions will still give the desired product, albeit in lower yield. When the desired product is methionine-free, cyanogen bromide cleavage at methionine adjacent to the desired sequence has proven very effective. Likewise, arginine- and lysine-free products can be split enzymatically with e.g. trypsin or chymotrypsin at arg-arg, lys-lys or similar cleavage sites adjacent to the desired sequence. In the case where splitting leaves e.g. unwanted arginine attached to the desired product, this can be removed by carboxypeptidase degradation. When trypsin is used to cleave at arg-arg, lysine sites in the desired polypeptide can first be protected as with maleic or citraconic anhydrides. The cleavage techniques discussed here as examples are only representative of the many variations that a person skilled in the art will be aware of.

Cleavable protein can be expressed adjacent to either the C- or N-termini of a specific polypeptide or even within the polypeptide itself, as is the case for the included sequence that separates proinsulin and insulin. Again, the vector used may code for expression of protein comprising repeated sequences of the desired polypeptide, each separated by selective cleavage sites. Most preferably, however, codons for excess protein will be translated before the structural gene in the desired product, as in the case illustrated in the figures. In each case, care must be taken to maintain the correct reading frame in relation to the regulation.

3. Expression of immunogens

The ability to express both a specific polypeptide and excess protein represents useful tools for the production of immunogenic substances. Polypeptide "haptens" (ie, substances containing determinants specifically bound by antibodies and the like, but usually too small to elicit an immune response) can be expressed as conjugates with additional protein of sufficient size to confer immunogenicity. The 3-gal-somatostatin conjugate exemplified here is actually of immunogenic size and can be expected to produce antibodies that bind the somatostatin hapten. Proteins comprising more than 100 amino acids, most commonly more than 200, exhibit an immunogenic character.

Conjugates prepared in the above manner can be used to produce antibodies that are useful in radioimmunoassays or other assays for the hapten, and alternatively in the preparation of vaccines. In the following, an example of the latter application will be described. Cyanogen bromide or other cleavage products of viral envelope protein will give oligopeptides that bind to antibody to the protein itself. Given the amino acid sequence of such an oligopeptide-hapten, heterologous DNA can therefore be expressed as a conjugate with additional protein that confers immunogenicity. The use of such conjugates as vaccines can be expected to reduce the side effects that accompany the use of the coat protein itself to effect immunity.

4. The control elements

Figure 1 shows a process where a transformant organism expresses polypeptide product from heterologous DNA brought under the control of a regulon which is "homologous" to the organism in its untransformed state. Thus, lactose-dependent E. coli chromosomal DNA comprises a lactose or "lac" operon which mediates lactose degradation by, among other things, triggering the enzyme p-galactosidase. In the particular case illustrated, the lac control elements are obtained from a bacteriophage, X-plac 5, which is infectious for E. coli. The phage's lac operon is in turn derived by transduction from the same bacterial species, hence the "homology". Homologous regulons suitable for use in the described process can alternatively be derived from plasmid DNA which is native to the organism.

The simplicity and efficiency of the lac promoter-operator system recommend its use in the systems described herein, as does its ability to be induced by IPTG (isopropylthio-B-D-galactoside). Other operons or parts thereof can of course also be used, e.g. X promoter operator, arabinose operon (phi 80 data) or choline E1, galactose, alkaline phosphatase or tryptophan operons. Promoter operators derived from the latter (ie the "tryp operon") would be expected to give 100% repression before induction (with indole acrylic acid) and harvest.

5. General plasmid construction

The details of the procedure illustrated schematically in Figure 4 appear from the experimental section below. However, at this point it may be useful to briefly describe various of the techniques used in the construction of the recombinant plasmid in the preferred embodiment.

The cloning and expression of the synthetic somatostatin gene uses two plasmids. Each plasmid has an EcoRI substrate site at a different region in the structural ε-galactosidase gene (see Figures 4 and 5). The introduction of the synthetic somatostatin DNA fragment into the EcoRI sites of these plasmids places the expression of the genetic information in this fragment under the control of the lac operon controlling elements. After the introduction of the somatostatin fragment into these plasmids, translation should result in a somatostatin polypeptide preceded by either 10 amino acids (pSOM1) or by virtually the entire p<->galactosidase subunit structure (pSOM11-3).

The plasmid construction procedure starts with plasmid pBR322, a well-characterized cloning vector. Introduction of the lac elements into this plasmid was achieved by introducing a Hael II restriction endonuclease fragment (203 nucleotides) containing the lac promoter, CAP binding site, operator, ribosome binding site and the first 7 amino acid codons of the structural β-galactosidase gene. The HaelII fragment was derived from β-plac5 DNA. The EcoRI-digested pBR322 plasmid that had its termini repaired with T4 DNA polymerase and deoxyribonucleotide triphosphates was blunt-ligated to the HaelII fragment to form EcoRI termini at the insertion points. Joining of these Haelll and repaired EcoRI termini develops EcoRI restriction sites (see Figures 4 and 5) at each terminus. Transformants of E. coli RR1 with this DNA were selected for resistance to tetracycline (Tc) and ampicillin (Ap) on 5-bromo-4-chloro-indolylgalactoside (X-gal) medium. On this indicator medium, colonies essential for the synthesis of ε-galactosidase, by virtue of the increased number of titrating repressors for lac operators, are identified by their blue color. Two orientations of the HaeIII fragment are possible, but these were masked by the asymmetric location of the Hha restriction site in the fragment. Plasmid pBHIO was further modified to eliminate the EcoRI endonuclease site distal to the lac operator (pBH20).

The eight chemically synthesized oligodeoxyribonucleotides (figure 2) were labeled at the 5'-terminus with [<32>P]-^-ATP using polynucleotide kinase and joined with T4 DNA ligase. Through hydrogen bonding between the overlapping fragments, the somatostatin gene self-assembles and eventually polymerizes into large molecules due to the cohesive restriction site termini. The ligated products were treated with EcoRI and BamHI restriction endonucleases to develop the somatostatin gene as shown in Figure 2.

The synthetic somatostatin gene fragment with EcoRI and BamHI termini was ligated into the pBH20 plasmid, pretreated with EcoRI and BamHI restriction nucleases and alkaline phosphatase. The treatment with alkaline phosphatase provides a molecular selection for plasmids carrying the inserted fragment. Ampicillin-resistant transformants obtained with this ligated DNA were analyzed for tetracycline sensitivity and several were examined for the introduction of an EcoRI-BamHI fragment of the appropriate size.

Both chains of the EcoRI-BamHI fragments in plasmids from two clones were analyzed by nucleotide sequence analysis starting from the BamHI and EcoRI sites. The sequence analysis was extended into the lac controlling elements; The lac fragment sequence was intact and in one case, pSOM1, the nucleotide sequence of both chains was independently determined and each gave the sequence shown in Figure 5A.

The EcoRI-Pst fragment in the pSOM1 plasmid, with the lac controlling element, was removed and replaced with the EcoRI-Pst fragment in pBR322 to make the plasmid pSOM11. The EcoRI fragment of plac5, which carries the lac operon control region and most of the structural 3-galactosidase gene, was inserted into the EcoRI site of pSOM11. Two orientations of the EcoRI lac fragment in X-plac5 were expected. One of these orientations would maintain the correct reading frame in the somatostatin gene, the other would not. Analysis of independently isolated clones for somatostatin activity then identified clones containing the correctly oriented gene of which the clone designated pSOM11-3 was one.

6. The microorganism

Various unicellular microorganisms have been proposed as candidates for transformation, such as bacteria, fungi and algae. That is the unicellular organisms that can be grown in cultures or that can be fermented. Bacteria are mostly the most appropriate organisms to work with. Bacteria susceptible to transformation include members of the Enterobacteriaceae, such as strains of Escherichia coli and Salmonella; Bacillaceae such as Bacillus subtilis; Pneumococcus, Streptococcus and Haemophilus influenzae.

The particular organism chosen for the somatostatin work that will be discussed below was E. coli strain TT1, genotype: Pro_Leu_Thi~RB~MBrec A<+>Str<r>Lac y~. E. coli RR1 is derived from E. coli HB101 (H.W. Boyer et al., J. Mol. Biol. (1969) 41, 459-472) by crossing it with E. coli K12 strain KL16 as the Efr donor. See J.H. Miller, Experiments in Molecular Genetics (Cold Spring Harbor, New York, 1972). Cultures of both E. coli RR1 and E. coli RR1 (pBR322) have been deposited with the American Type Culture Collection without restriction on access, ATCC Nos. 31343 and 31344, respectively.

The somatostatin-producing organism has also been deposited (ATCC No. 31447).

In the case of human insulin, the A and B chain genes were cloned in E. coli K12 strain 294 (end A, thi~, hsr", hsmk<+>), ATCC No. 31446, and the organism used in the expression of A -chain (E. coli K12 strain 294 [pIAl], ATCC no. 31448). The B-chain in human insulin was first expressed in a derivative of BH101, i.e. E. coli K12 strain D1210 a lac<+>(i°o <+>z<t>y<+>), and the B gene-containing organism has likewise been deposited (ATCC No. 31449). Alternatively, the B gene can be introduced into and expressed from the organism mentioned first, i.e. strain 294 .

EXPERIMENTAL PART

I. SOMATOSTATIN

1. Construction of somatostatin gene fragments Eight oligodeoxyribonucleotides respectively marked A to E in Figure 2 were first constructed, mainly according to the modified triester method of K. Itakura et al., J. Am. Chem. Soc. 97, 7327 (1975). In the case of fragments C, E and H, however, an improved technique was used in which fully protected trimers are first prepared as building blocks for the construction of longer oligodeoxyribonucleotides. The improved technique is schematically shown in Figure 3 where B is thymine, N-benzoylated adenine, N-benzoylated cytosine or N-isobutylated guanine. Briefly, and with reference to Figure 3, the coupling reaction between an excess of I (2 mmol), and II (1 mmol) went almost to completion 1 within 60 minutes using a powerful coupling reagent, 2,4,6-triisopropylbenzenesulfonyltetrazolide (TPSTe, 4 mmol; 2). After removal of the 5'-protecting group with 2# benzene-sulfonic acid solution, the 5'-hydroxyl dimer V could be separated from an excess of 3'-phosphodiester monomer IV by simple solvent extraction with aqueous NaHCOQ solution in CHCl 3 . The fully protected trimer block was prepared successively from the 5'-hydroxyl dimer V, I (2 mmol), and TPSTe (4 mmol) and isolated by chromatography on silica gel, as in B.T. Hunt et al., Chem. and Ind. 1967, 1868 (1967). The yields of trimers prepared according to the improved technique are shown in Table II.

The eight oligodeoxyribonucleotides, after removal of all protecting groups, were purified by high pressure liquid chromatography on "Permaphase AAX" (R.A. Henry et al., J. Chrom. Sei. II. 358 (1973)). The purity of each oligomer was checked by homochromatography on thin-layer DEAE cellulose and also by gel electrophoresis in a 20% acrylamide plate after labeling the oligomers with (-γ-<32>P)-ATP in the presence of polynucleotide kinase. A labeled major product was obtained from each DNA fragment.

2. Ligation and acrylamide gel detection of somatostatin DNA

The 5' OH termini of the chemically synthesized fragments A to H were phosphorylated separately with T4 polynucleotide kinase. (<32p>)-■y-ATP was used in the phosphorylation so that the reaction products could be monitored audioradiographically, although it will be understood that unlabeled ATP could also be used where autoradiography could do without. Just before the clinase reaction, 25 µCi of (^-<32>P)ATP (about 1500 Ci/mmol) (Maxam and Gilbert., Proe. Nat. Acad. Sei. U.S.A., 74, 1507 (1977)) was evaporated to dryness in 0.5 ml Eppendorf tube. 5 pg fragment was incubated with 2 units of T4 DNA kinase (hydroxylapatite fraction, 2500 units/ml; 27), in 70 mM Tris-HCl, pH 7.6, 10 mM MgCl2, 5 mM dithiothreitol in a total volume of 150 µl in 20 minutes at 37°C. To ensure maximal phosphorylation of the fragments for ligation purposes, 10 µl of a mixture consisting of 70 mM

Tris-HCl, pH 7.6, 10 mM MgCl 2 , 5 mM dltlotreltol, 0.5 mM ATP and 2 units of DNA kinase added and incubation continued for an additional 20 minutes at 1°C. The fragments (250 ng/µl) were stored at -20°C without further processing. Kinase-treated fragments of A, B, E and F (1.25 µg each) were ligated in a total volume of 50 µl in 20 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 , 10 mM dithiothreitol, 0.5 mM ATP and 2 units of T4 DNA ligase (hydroxylapatite fraction, 400 units/l; 27), for 16 h at 4°C. Fragments C, D, G and H were ligated under similar conditions, samples of 2 µl were removed for analysis by electrophoresis on a 10% polyacrylamide gel followed by autoradiography (H.L. Heyneker et al., Nature 263, 748 (1976)) where unreacted DNA fragments are represented by rapidly migrating material and where the monomeric form of the ligated fragments migrates with broraphenol blue (BPB) dye. Some dimerization also occurs due to the cohesive ends of the ligated fragments A, B, E, and F, and the ligated fragments C, D, G, and H. These dimers represent the slowest migrating material and can be cleaved by restriction endonuclease EcoRI and BamHI, respectively.

The two half molecules (ligated A+B+E+F and ligated C + D + G + H) were joined by a further ligation step performed in a final volume of 150 µl at 4°C for 16 hours. 1 µl was removed for analysis. The reaction mixture was heated for 15 min at 65°C to inactivate the T4 DNA ligase. The heat treatment does not affect the migration pattern of the DNA mixture. Sufficient restriction endonuclease BamHI was added to the reaction mixture to cleave the multimeric forms of the somatostatin DNA for 30 minutes at 37°C. After addition of NaCl to 100 mM, the DNA was digested with EcoRI endonuclease. The restriction endonuclease digests were terminated by phenol-chloroform extraction of the DNA. The somatostatin DNA fragment was purified from unreacted and partially ligated DNA fragments by preparative electrophoresis on a 10% polyacrylamide gel. The band containing the somatostatin DNA fragment was excised from the gel and the DNA was eluted by cutting the gel into small pieces and extracting the DNA with elution buffer (0.5 M ammonium acetate, 10 mM MgCl2. 0.1 mM EDTA, 0, 1* SDS) overnight at 65* C. The DNA was precipitated with 2 volumes of ethanol, centrifuged, redissolved in 200 µl 10 mM Tris-HCl, pH 7.6 and dialyzed against the same buffer resulting in a somatostatin DNA concentration of 4 pg/ml.

3. Construction of recombinant plasmids

Figure 4 schematically shows the way in which recombinant plasmids comprising the somatostatin gene were constructed, and can be referred to in connection with the following special mention.

A. The parental plasmid pBR322

The plasmid chosen for experimental somatostatin cloning was pBR322, a small (molecular weight approx. 2.6 megadaltons) plasmid containing resistance genes to the antibiotics ampicillin (Ap) and tetracycline (Tc). As indicated in Figure 4, the ampicillin resistance gene contains a cleavage site for the restriction endonuclease PstI, the tetracycline resistance gene contains a similar site for the restriction endonuclease BamHI, and an EcoRI site is located between the Ap<r> and TC<r> genes. Plasmid pBR322 is derived from pBR313, a 5.8 megadalton Ap<r>Tc<r>Col<lmni->plasmid (R.L. Rodriquez et al., ICN-UCLA Symposia on Molecular and Cellular Biology, 471-77 (1976), R.L. Rodriquez et al., Construction and Characterization of Cloning Vehicles, in Molecular Mechanisms in the Control of Gene Expression, pp. 471-77, Academic Press, Inc. (1976). Plasmid pBR322 is characterized and the method of its preparation fully described in F. Bolivar et al., "Construction and Characterization of New Cloning Vehicles II. A Multipurpose Cloning System", Gene (November 1977).

B. Construction of plasmid pBHIO

5 µg of plasmid pBR322 DNA was digested with 10 units of the restriction endonuclease EcoRI in 100 mM Tris-HCl, pH 7.6, 100 mM NaCl, 6 mM MgCl2 at 37°C for 130 minutes. The reaction was terminated by phenol-chloroform extraction; The DNA was then precipitated with 2V volumes of ethanol and resuspended in 150 µl of T4 DNA polymerase buffer (67 mM Tris-HCl, pH 8.8, 6.7 mM MgCl2»16.6 mM (NH^gSC^, 167 pg/ ml bovine serum albumin, 50 pM of each of the dNTPs, A. Panet et al., Biochem. 12, 5045

(1973). The reaction was started by adding 2 units of T4 DNA polymerase. After incubation for 30 minutes at 37°C, the reaction was terminated by a phenol-chloroform extraction of the DNA followed by precipitation with ethanol. 3 pg of X-plac<5>DNA (Shapiro et al., Nature, 224. 768 (1969)) was digested for one hour at 37°C with the restriction enzyme Haelll (3 units) in 6 mM Tris-HCl, pH 7 ,6, 6 mM MgCl2»6 mM p<->mercaptoethanol in a final volume of 20 µl. The reaction was stopped by heating for 10 minutes at 65°C. The pBR322-treated DNA was mixed with the HaelII-digested X-plac<5>DNA and end-ligated in a final volume of 30 µl with 1.2 units of T4 DNA- ligase (hydroxylapatite fraction; A. Panet et al., supra) in 20 mM Tris-HCl, pH 7.6, 10 mM gCl 2 , 10 mM dithiothreitol, 0.5 mM ATP 1 12 hours at 12-C. The ligated DNA- the mixture was dialyzed against 10 mM Tris-HCl, pH 7.6 and used for transformation of E. coli strain RR1. Transformants were selected for tetracycline and ampicillin resistance on minimal plates containing 40 pg/ml 5-bromo-4-chloro-indolylgalactoside ( X-gal) medium (J.H. Miller, Experiments in Molecular Genetics (Cold Spring Harbor, New York, 1972)). Colonies essential for the synthesis of ε-galactosidase were identified by their blue color. After examination of 45 independently isolated blue colonies three of them were found to contain plasmid DNA containing two EcoRI sites separated by about 200 base pairs.The position for an asymmetric be horizontal Hhal fragment 1,203 b.p. The Haelll lac control fragment (W. Gilbert et al., in Protein-Ligand Interactions, H. Sand and G. Blauer, Eds. (De Gruyter, Berlin (1975 ), pp. 193-210) enables determination of the orientation of The Haelll fragment, now an EcoRI fragment in these plasmids.Plasmid pBHIO was shown to contain the fragment in the desired orientation, i.e. the lac transcription enters the Tc<r> gene of the plasmid.

C. Construction of plasmid pBH20

Plasmid pBHIO was then modified to eliminate the EcoRI site distal to the lac operator. This was achieved by selective EcoRI endonuclease cleavage at the distal site involving partial protection by RNA polymerase of the second EcoRI site located between the Tc<r> and lac promoters, which is only approx. 40 base pairs apart. After binding of RNA polymerase, the DNA (5 µg) was digested with EcoRI (1 unit) in a final volume of 10 µl 1 for 10 minutes at 37°C. The reaction was stopped by heating at 65°C for 10 minutes. The EcoRI -cohesive termini were degraded with Sl-nuclease in 25 mM Na-acetate, pH 4.5, 300 mM NaCl, 1 mM ZnCl2 at 25°C for 5 min. The reaction mixture was stopped by addition of EDTA (10 mM final) and Tris- HCl, pH 8 (50 mM final). The DNA was phenol-chloroform extracted, ethanol precipitated and resuspended in 100 µl in T4 DNA ligation buffer. T4 DNA ligase (1 µl) was added and the mixture incubated at 12<* >C for 12 h. The ligated DNA was transformed into E. coli strain RR1, and Ap<r>Tc<r->transformants were selected on X-gal-antibiotic medium.Restriction enzyme analysis of DNA selected from 10 isolated blue colonies showed that these colonies carried plasmid DNA with an EcoRI site. Seven of these colonies had retained the EcoRI site located between the lac and Tc<r> promoters. The nucleotide sequence of the EcoRI site in the lac control region in one of these plasmids, pBH20, was confirmed. This plasmid was then used to clone the somatostatin gene.

D. Construction of plasmid pSOM1

20 µg of the plasmid pBH20 was completely digested with restriction endonucleases EcoRI and BamHI in a final volume of 50 µl. Bacterial alkaline phosphatase was added (0.1 unit of Worthington BAPF) and incubation was continued for 10 minutes at 65°C. The reactions were stopped by phenol-chloroform extraction and the DNA was precipitated with 2 volumes of ethanol, centrifuged and dissolved in 50 µl of 10 mM Tris-HCl, pH 7.6, 1 mM EDTA. The alkaline phosphatase treatment effectively prevents self-ligation of the EcoRI, BamHI-treated pBH20 DNA, but circular recombinant plasmids containing somatostatin DNA can still be formed by ligation. Since E. coli RR1 is transformed with very low efficiency by linear plasmid DNA, the majority of transformants will contain recombinant plasmids. 50 µl of somatostatin DNA, (4 pg/ml) was ligated with 25 µl of the BamHI, EcoRI, alkaline phosphatase-treated pBH20 DNA in a total volume of 50 µl containing 20 mM Tris-HCl, pH 7.6, 10 mM MgCl2. 10 mM dithiothreitol, 0.5 mM ATP, and 4 units of T4 DNA ligase at 22°C. After 10, 20, and 30 min, additional somatostatin DNA (40 ng) was added to the reaction mixture (the gradual addition of somatostatin DNA can favoring ligation to the plasmid over self-ligation).Ligation was continued for one hour followed by dialysis of the mixture against 10 mM Tris-HCl, pH 7.6. In a control experiment, BamHI, EcoRI, alkaline phosphatase-treated pBH20 DNA was ligated in the absence of somatostatin DNA under similar conditions. Both preparations were used without further treatment to transform E. coli RR1. The transformation experiments were carried out in a P3 apparatus (National Institutes of Health, USA, Recombinant DNA Research Guidelines, 1976). Transformants were selected on minimal medlum plates containing 20 pg/ml AP and 40 pg/ml X-gal. 10 transformants, all of which were sensitive to Tc, were isolated. For reference, these were designated pSOMl, pSOM2, etc. pSOMlO. In the control experiment, no transfo romants. Four of the ten transformants contained plasmids with both an EcoRI site and a BamHI site. The size of the small EcoRI, BamHI fragment in these recombinant plasmids was in all cases equal to the size of the in vitro produced somatostatin DNA. Base sequence analysis according to Maxam and Gilbert, Proe. Nat. Acad. Pollock. USA, 74, 560 (1977), showed that the plasmid pSOM1 had the desired somatostatin DNA fragment inserted.

DNA sequence analysis of the clone containing plasmid pSOM1 predicted that it would produce a peptide comprising somatostatin. However, no somatostatin radioimmunoactivity has been detected in extracts of cell pellets or culture supernatants, nor has the presence of somatostatin been detected when the growing culture is added directly to 70* formic acid and cyanogen bromide. E. coli RR1 extracts have been observed to degrade exogenous somatostatin very rapidly. The absence of somatostatin activity in clones containing plasmid pSOM1 could well result from intracellular degradation by endogenous proteolytic enzymes. Plasmid pSOM1 was therefore used to construct a plasmid encoding a precursor protein comprising somatostatin and of sufficient size to be expected to resist proteolytic degradation.

E. Construction of plasmids pSOMll and pSOMll-3

A plasmid was constructed in which the somatostatin gene could be located at the C-terminus of the ε-galactosidase gene, translation being kept in phase. The presence of an EcoRI site near the C-terminus of this gene and the available amino acid sequence of this protein (B. Polisky et al., Proe. Nat. Acad. Sei. USA 73, 3900 (1976 ), A.V. Fowler et al., Id. at 74, 1507 (1976), A-I- Bukhari et al., Nature New Biology 243, 238 (1973) and K. E. Langley, J. Biol. Chem., 250. 1587 (1975)) allowed introduction of EcoRI, BamHI -somatostatin gene in the EcoRI site while maintaining the correct reading frame. For construction of such a plasmid, pSOM1 DNA (50 µg) was digested with the restriction enzymes EcoRI and PstI in a final volume of 100 µl. A preparative 5* polyacrylamide gel was used to separate the large Pst-EcoRI fragment carrying the somatostatin gene from the small fragment carrying the lac control elements. The large band was excised from the gel and the DNA eluted by cutting the gel into small pieces and extracting the DNA at 65°C overnight. Similarly, plasmid pBR322 DNA (50 pg) was digested with PstI and EcoRI restriction endonucleases and the two resulting DNA fragments purified by preparative electrophoresis on a 5* polyacrylamide gel. The small Pstl-EcoRI fragment from pBR322 (1 pg) was ligated with the large Pstl-EcoRI DNA fragment (5 pg) from pSOM1 in a final volume of 50 µl with one unit of T4 DNA ligase at 12°C for 12 hours. The ligated mixture was used to transform E. coli KRI, and transformants were selected for ampicillin resistance on X-gal medium. As expected, almost all of the Ap<r->transformants (95*) gave white colonies (no lac operator) on X-gal indicator plates. The resulting plasmid, pSOMll, was used in the construction of plasmid pSOMll-3. A mixture of 5 pg pSOM11 DNA and 5 pg X-plac<5>DNA was digested with EcoRI (10 units for 30 minutes at 37°C). The restriction endonuclease digestion was terminated by phenol-chloroform extraction. The DNA was then ethanol precipitated and resuspended in T4 DNA ligase buffer (50 µl). T4 DNA ligase (one unit) was added to the mixture and incubated at 12°C for 12 hours. The ligated mixture was dialyzed against 10 mM Tris-HCl, pH 7.6 and used to transform E. coli strain RR1. Transformants were selected for Aβ on X-gal plates containing ampicillin and selected for basal β-galactosidase production. About 2* of the colonies were blue (pSOM11-1, 11-2, etc.). Restriction enzyme analysis of plasmid DNA obtained from these clones showed that all plasmids carried a new EcoRI fragment of approx. 4.4 megadaltons, carrying lac operon control sites and most of the β-galactosidase gene. Because two orientations of the EcoRI fragment are possible, the asymmetric location of a HindIII restriction site was used to determine which of these colonies carried this EcoRI fragment with lac transcription further into the somatostatin gene. HindIII-BamHI double digests indicated that only the clones carrying plasmids pSOMll-3, pSOMll-5, pSOMll-6 and pSOMll-7 contained the EcoRI fragment in this orientation. 4. Radioimmunoassay for somatostatin activity The standard radioimmunoassays (RIA) for somatostatin (A. Arimura et al., Proe. Soc. Exp. Biol. Med., 148, 784 (1975)) were modified by reducing the assay volume and using phosphate buffer. Tyr<11>somatostatin was iodinated using a chloramine-T method. (Id.) To assay for somatostatin, the sample, usually in 70* formic acid containing 5 mg/ml cyanogen bromide, was dried in a conical polypropylene tube (0.7 ml, Sarstedt) over moist KOH under vacuum. 20 µl of PBSA buffer (75 mM NaCl; 75 mM sodium phosphate, pH 7.2; 1 mg/ml bovine serum albumin; and 0.2 mg/ml sodium azide) was added, followed by 40 µl of a (<125>I ) somatostatin "cocktail" and 20 µl of a 1000-fold dilution PBSA of rabbit anti-somatostatin immune serum S39 (Vale et al., Metabolism, 25, 1491 (1976). Said (<125>I)-somatostatin "cocktail" contained per ml of PBSA buffer: 250 pg normal rabbit gamma globulin (Antibodies, Inc.), 1500 units of protease inhibitor ("Trasylol", Calblochem) and about 100,000 counts of (125 I )Tyr1:1-somatostatin After at least 16 hours at room temperature, 0.333 ml of goat anti-rabbit gamma globulin (Antibodies, Inc., P=0.03) in PBSA buffer was added to the test tubes.The mixture was incubated for 2 hours at 37°C, cooled to 5°C , then centrifuged at 10,000 x g for 5 minutes. The supernatant was removed and the pellet counted in a gamma counter. With the amount of antiserum used, 20* of the counts precipitated without any unlabeled competing somatostatin. The background with u final somatostatin (200 mg) was usually 3*. Half maximal competition was achieved with 10 pg somatostatin. Initial experiments with extracts of E. coli strain RR1 (the recipient strain) indicated that less than 10 pg of somatostatin could be readily detected in the presence of 16 pg or more of cyanogen bromide-treated bacterial protein. More than 2 µg of protein from formic acid-treated bacterial extracts interfered somewhat by increasing the background, but cyanogen bromide cleavage greatly reduced this interference. Reconstruction experiments showed that somatostatin is stable in cyanogen bromide-treated extracts.

A. Competition of bacterial extracts

Strains E. coli RR1 (pSOMll-5) and E. coli RR1 (pSOMll-4) were grown at 37"C to 5 x 10<8> cells/ml in Luria-bul jong. Then IPTG was added to 1 mM and growth continued for 2 h. 1 ml aliquots were centrifuged for a few seconds in an Eppendorf centrifuge and the pellets were suspended in 500 µl 70* formic acid containing 5 mg/ml cyanogen bromide. After about 24 h at room temperature, aliquots were diluted 10-fold in water and the volumes indicated in Figure 6A were analyzed in triplicate for somatostatin. In Figure 6A, "B/Bq" is the ratio of (<125>I)-somatostatin bound in the presence of sample to that bound in the absence of competing somatostatin. Each point is the average of three parallel tubes The protein content of the undiluted samples was determined to be 2.2 mg/ml for E. coli RR1 (pSOMll-5) and 1.5 mg/ml for E. coli RR1 (pSOM-4) .

B. Initial screening of pSOMll clones for somatostatin Cyanogen bromide-treated extracts of 11 clones (pSOMll-2, pSOMll-3, etc.) were prepared as described above for the case of Figure 6A. 30 µl of each extract was taken in three parallels for radioimmunoassay, the results of which appear in Figure 6B. The range of analysis points is indicated. The picogram somatostatin values were read from a standard curve obtained as part of the same experiment.

M M M

The radioimmunoassay results described so far can be summarized as follows. In contrast to the results from experiments with pSOM1, four clones (pSOM11-3, 11-5, 11-6 and 11-7) were found to have readily detectable somatostatin radioimmunoactivity, Figures 6A and 6B. Restriction fragment analysis showed that pSOMll-3, pSOMll-5, pSOMll-6 and pSOMll-7 had the desired orientation of the lac operon, while pSOMll-2 and 11-4 had the opposite orientation. There is thus a perfect correlation between the correct orientation of the lac operon and the production of somatostatin radioimmunoactivity.

C. Effects of IPTG induction and CNBr cleavage

on positive and negative clones

The construction of the somatostatin plasmid predicts that the synthesis of somatostatin will be under the control of the lac operon. The lac repressor gene is not included in the plasmid and the recipient strain (E. coli RR1) contains the chromosomal lac repressor wild-type gene which hare produces 10-20 repressor molecules per cell (15). The plasmid copy number (and therefore the number of lac operators) is around 20-30 per cell, so that complete repression is impossible. As shown in Table III below, the specific activity of somatostatin in E. coli RR1 (pSOM11-3) was increased by IPTG, an inducer of the lac operon. As expected, the level of induction was low, ranging from 2.4 to 7-fold. In experiment 7 (Table III), α-activity, which is a measure of the first 92 amino acids of α-galactosidase, was also induced by a factor of two. In several experiments, no somatostatin radioimmunoactivity can be detected before cyanogen bromide cleavage of the total cell protein. Since the antiserum used in the radioimmunoassay, S39, requires a free N-terminal alanine, no activity was expected before cyanogen bromide cleavage.

D. Gel filtration of cyanogen bromide-treated extracts

Formic acid and cyanogen-treated extracts of the positive clones (pSOM11-3, 11-5, 11-6 and 11-7) were collected (total volume 250 µl), dried and resuspended in 0.1 ml 50* acetic acid.

(%) leucine was added and the sample was applied to a 0.7 x 47 cm column of Sephadex G-50 in 50* acetic acid. 50 µl aliquots of the column fractions were analyzed for somatostatin. Collected negative clone extracts (11-2, 11-4 and 11-11) were processed

in an identical manner. The results appear from Figure 5C. On the same column, known somatostatin (Beckman Corp.) is eluted as indicated (SS). In this system, somatostatin is well separated from excluded large peptides and fully included small molecules. Only extracts of clones positive for somatostatin showed radioimmunoactivity in the column fractions and this activity eluted in the same position as chemically synthesized somatostatin.

SUMMARY OF ACTIVITY INFORMATION

The data establishing the synthesis of a polypeptide containing the somatostatin amino acid sequence can be summarized as follows: (1) Somatostatin radioimmunoactivity is present in E. coli cells carrying the plasmid pSOMll-3, which contains a somatostatin of proven correct sequence and has the correct orientation of the lac-EcoRI DNA fragment. Cells with the related plasmid pSOM11-2, which has the same somatostatin gene but an opposite orientation of the lac-EcoRI fragment, produce no detectable somatostatin activity; (2) as predicted by the construction scheme, no detectable somatostatin radioimmunoactivity is observed until after cyanogen bromide treatment of the cell extract; (3) somatostatin activity is under the control of the lac operon which could be shown by the induction of IPTG, an inducer of the lac operon; (4) the somatostatin activity will be chromatographed with known somatostatin on Sephadex G-50; (5) The DNA sequence of the cloned somatostatin gene is correct. If translation is out of phase, a peptide will be formed that is different from somatostatin in each position. Radioimmunoactivity is detected, which indicates that a peptide is formed which is closely related to somatostatin, and translation must be in phase. Since translation takes place in phase, the genetic code will ensure that a peptide with the exact sequence of somatostatin is formed; (6) Finally, the above samples of E. coli RR1 (pSOMll-3) extract inhibit the release of growth hormone from rat pituitary cells, while samples of E. coli RR1 (pSOMll-2) prepared in parallel and with identical protein concentration , have no effect on growth hormone release.

STABILITY. YIELD AND PURIFICATION OF

SOMATOSTATIN

The strains containing the EcoRI-lac operon fragment (pSOMll-2, pSOMll-3, etc.) are segregated with respect to the plasmid phenotype. E.g., after approx. 15 generations was approx. half of the E. coli RR1 (pSOMll-3) culture was constitutive for p<->galactosidase, i.e. contained the lac operator, and of these, approx. half ampicillin resistant. Strains that were positive (pSOMll-3) and negative pSOMll-2) for somatostatin are unstable and therefore the growth bias is probably due to overproduction of large, but incomplete and inactive galactosidase. The yield of somatostatin has varied from 0.001 to 0.03* of the total cell protein (Table I) probably as a result of selection for cells in culture having plasmids with a deleted lac region. The highest yields of somatostatin have been from preparations where growth was initiated from a single amplicillin-resistant, constitutive colony. Even in 1 of these cases, 30* of the cells at harvest had omissions of the lac region. Storage in the frozen state, (lyophilization) and growth for harvest from a single such colony is therefore indicated for the described system. Dividends can be increased, e.g. by using bacterial strains that overproduce lac repressor so that expression of precursor protein is substantially completely suppressed before induction and harvest. Alternatively and as previously discussed, a tryptophan system or other operator-promoter system which is usually completely repressed can be used.

In the crude extract resulting from cell breakdown in e.g. an Eaton press, the β-galactosidase-somatostatin precursor protein is insoluble and is found in the first pellet from low-speed centrifugation. The activity can be solubilized in 70* formic acid, 6M guanididium hydrochloride, or 2* sodium dodecyl sulfate. Most preferably, however, the crude extract from the Eaton press is extracted with 8M urea and the residue digested with cyanogen bromide. In initial experiments, the somatostatin activity derived from E. coil strain RR1 (pSOM11-3) has been enriched about 100-fold by alcohol extraction of the cleavage product and chromatography on Sephadex G-50 in 50% acetic acid. When the product is again chromatographed on Sephadex G-50 and then subjected to high pressure liquid chromatography, substantially pure somatostatin can be obtained.

II. HUMAN INSULIN

The techniques described above were then applied to the production of human insulin. Thus, the genes for insulin B chain (104 base pairs) and for insulin A chain (77 base pairs) were constructed from the amino acid sequence of the human polypeptides, each with single-stranded cohesive termini for the EcoRI and BamHI restriction endonucleases and each constructed for separate introduction into pBR322 plasmids. The synthetic fragments, deca- to pentadeca-nucleotides, were synthesized by the block phosphotriester method using trinucleotides as building blocks and finally purified by high performance liquid chromatography (HPLC). The synthetic genes for human insulin A and B chain were then cloned separately into plasmid pBR322. The cloned synthetic genes were fused to an E. coli ε-galactosidase gene as before to provide efficient transcription, translation and a stable precursor protein. Insulin peptides were cleaved from 3-galactosldase precursor, detected by radioimmunoassay, and purified. Insulin radioimmunoassay activity was then developed by mixing the E. coli products.

1. Construction and synthesis of human insulin genes The genes constructed for human insulin are shown in figure 9. The genes for human insulin, B chain and A chain, were constructed from the amino acid sequence in the human polypeptides. The 5' ends of each gene have single-stranded cohesive termini for EcoRI and BamHI restriction endonucleases, for correct insertion of each gene into plasmid pBR322. A HindIII endonuclease recognition site was incorporated into the middle of the B-chain gene for the amino acid sequence Glu-Ala to allow amplification and verification of each half of the gene separately from the construction of the entire B-chain gene. The B-chain dg A-chain genes were engineered to be built from 29 different oligodeoxyribonucleotides, ranging from decamers to pentadecamers. Each arrow indicates the fragment synthesized by the improved phosphotriester method, Hl to H8 and Bl to B12 for the B-chain gene and Al to All for the A-chain gene. 2. Chemical synthesis of oligodeoxyribonucleotides Materials and methods for the synthesis of oligodeoxyribonucleotides were essentially those described in Itakura, K. et al.,

(1975) J. Biol. Chem. 250. 4592 and Itakura, K. et al., (1975) J. Amer. Chem. Soc. 97, 7327 with the exception of these modifications: a) The fully protected raononucleotides, 5'-0-dimeth-oxytrityl-3'-p-chlorophenyl-p-cyanoethyl phosphates, were synthesized from the nucleoside derivatives using the monofunctional phosphorylation radical p-chlorophenyl- ε-cyanoethyl phosphorochloridate (1.5 molar equivalent) in acetonitrile in the presence of 1-methylimidazole Van Boom, J.H. et al.,

(1975) Tetrahedron 31, 2953. The products were isolated in large

scale (100-300 g) by preparative liquid chromatography (Prep 500 LC, Waters Associates).

b) Using the solvent extraction method (Hirose, T. et al., (1978) Tetrahedron Letters, 2249) 32

bifunctional trimers synthesized (see Table IV) in 5-10 mmol scale, and 13 trimers, 3 tetramers and 4 dimers

as 3'-terminus blocks, 1 a 1 mmol scale. The homogeneity of the fully protected trimers was checked by thin-layer chromatography on silica gel in two methanol/chloroform solvent systems: Solvent a, 5* v/v and solvent b, 10* v/v (see

table IV). Starting from this library of compounds, 29 oligodeoxyribonucleotides with a defined sequence were synthesized, 18 for the B-chain gene and 11 for the A-chain gene.

The basic units used to construct polynucleotides were two types of trimer blocks, ie the bifunctional trimer blocks in Table IV and corresponding 3'-terminus trimers protected with an anisoyl group at the 3'-hydroxy. The bifunctional trimer was hydrolyzed to the corresponding 3'-phosphodiester component with a mixture of pyridine-triethylamine-water (3:1:1 v/v) and also to the corresponding 5'-hydroxyl component with 2* benzenesulfonic acid. The previously mentioned 3'-terminus block was treated with 2% benzenesulfonic acid to obtain the corresponding 5'-hydroxyl. However, the coupling reaction of an excess of the 3'-phosphodiester trimer (1.5 molar equivalent) with the 5'-hydroxyl component, obtained (1 molar equivalent) in the presence of 2,4,6-triisopropyl-benzenesulfonyltetrazolide (TPSTe, 3-4 equivalents) was almost completely within 3 hours. To remove the excess of the 3*-phosphodiester block reactant, the reaction mixture was passed through a short silica gel column mounted on a sintered glass filter. The column was washed, first with CHCl 3 to elute some byproducts and the coupling reagent, and then with CHCl 3 :MeOH (95:5 v/v) in which almost all of the fully protected polymer was eluted. Under these conditions, the added 3'-phosphodiester block reactant remained in the column. In the same way, block connections were repeated until the desired length was constructed. High performance liquid chromatography (HPLC) was used extensively during oligonucleotide synthesis for a) analysis of each trimer and tetramer block, b) analysis of the intermediate fragments (hexamers, nonamers and decamers), c) analysis of the final coupling reaction, and d) purification of the end products. HPLC was performed using a Spectra-Physlcs 3500B liquid chromatograph. After removal of all protecting groups with conc. NH 4 OH at 50°C (6 h) and 80* AcOH at room temperature (15 min), the compounds were analyzed on a Permaphase AAX (DuPont) (Van Boom, J. et al., (1977) J. Chromatography 131. 169 ) column (1 m x 2 mm), using a linear gradient of solvent B (0.05 M KH2PO4- 1.0 M KC1, pH 4.5) in solvent A (0.01 M KH2PO4, pH 4.5) . The gradient was formed by starting with buffer A and applying 3* buffer B per minute. The elution was carried out at 60<*>C with a flow rate of 2 ml per minute. The purification of the 29 final oligonucleotides was also performed on Permaphase AAX, under the same conditions as reported above. The desired peak was collected, desalted by dialysis and lyophilized. After labeling the 5'-termini with (-γ-<32>P)ATP using T4 polynucleotide kinase, the homogeneity of each oligonucleotide was checked by electrophoresis on a 20* polyacrylamide gel. 3. Assembly and cloning of the B-k. 1 ed gene and the A- k. 1 ed gene The gene for the B chain of insulin was engineered to have an EcoRI restriction site at the left end, a HindIII site in the middle and a BamHI site at the right end. This was done so that both halves, the left EcoRI-HindIII half (BE) and the right EindIII-BamEI half (BB), could be cloned separately in the appropriate cloning vector pBR322 and, after their sequences had been verified, joined to obtain of the complete B gene (Figure 10). The BB half was assembled by ligation from 10 oligodeoxyribonucleotides, labeled B1 to B10 in Figure 9, prepared by chemical phosphotriester synthesis. B1 and B10 were not phosphorylated and thereby unwanted polymerization of these fragments through their cohesive ends (EindIII and BamEI) was eliminated. After purification by preparative acrylamide gel electrophoresis and elution of the largest DNA band, the BB fragment was inserted into plasmid pBR322 which had been digested with Hlndlll and BamHI. About 50* of the ampicillin-resistant colonies obtained from the DNA were sensitive to tetracycline, indicating that a non-plasmid HindIII-BamHI fragment had been introduced. The small HindIII-BamHI fragments from four of these colonies (pBB101 to pBB104) were sequenced and found to be correct as constructed.

The BH fragment was prepared in a similar manner and introduced into pBR322 which had been digested with EcoRI and HindIII restriction endonucleases. Plasmids from three ampicillin-resistant, tetracycline-sensitive transformants (pBH1 to pBH3) were analyzed. The small EcoRI-HindIII fragments were found to have the expected nucleotide sequence.

The A-chain gene was assembled in three parts. The left four, the middle four and the right four oligonucleotides (see Figure 9) were ligated separately, then mixed and ligated (oligonucleotides A1 and A12 were unphosphorylated). The assembled A-chain gene was phosphorylated, purified by gel electrophoresis and cloned into pBR322 at the EcoRI-BaHI sites. The EcoRI-BaHI fragments from two ampicillin-resistant, tetracycline-sensitive clones (pA10, pA11) contained the desired A gene sequence.

4. Construction of plasmids for expression of Å and B insulin genes

Figure 10 illustrates the construction of lac-insulin B plasmid (pIB1). Plasmids pBH1 and pBB101 were digested with EcoRI and HindIII endonucleases. The small BH fragment in pBHl and the large fragment in pBBlOl (containing the BB fragment and most of pBR322) were purified by gel electrophoresis, mixed and ligated in the presence of EcoRI-cleaved X-plac 5. The EcoRI megadalton fragment in X-plac 5 contains the lac control region and the majority of the structural ε-galactosidase gene. The configuration of the restriction seats ensures correct connection of the BH to the BB. The Lac-EcoRI fragment can be introduced in two orientations; thus, half of the clones obtained after transformation should have the desired orientation. The orientation of ten ampicillin-resistant, ε-galactosidase constitutive clones was checked by restriction analysis. Five of these colonies contained the entire B gene sequence and the correct reading frame from the ε-galactosidase gene into the B chain gene. One, pIB1, was selected for subsequent experiments.

In a similar experiment, the 4.4 lac megadalton fragment from X-plac 5 was inserted into the pA1 plasmid at the EcoRI site to form pIA1. pIAl is identical to pIBl with the exception that the A gene fragment is inserted instead of the B gene fragment. DNA sequence analysis showed that the correct A and B chain gene sequences were retained in pIAl and pIBl, respectively.

5. Expression

The strains containing the insulin genes correctly attached to ε-galactosidase both produce large amounts of a protein the size of β-galactosidase. About 20* of the total cell protein was this ε-galactosidase-insulin A or B chain hybrid. The hyrid proteins are insoluble and were found in the first pellet obtained at low speed where they make up approx. 50* of the protein.

To detect the expression of the Insulin A and B chains, a radioimmunoassay (RIA) based on the reconstitution of complete insulin from the separate chains was used. The insulin reconstitution method of Katsoyannls et al.

(1967) Biochemistry 6, 2642-2655, adapted to a 27-µl assay volume, represents a very suitable assay. Easily detectable insulin activity is achieved after mixing and reconstitution of S-sulfonated derivatives in the insulin chains. The separate S-sulfonated chains in insulin do not react significantly after reduction and oxidation with the anti-insulin antibody used.

To use the reconstitution assay, the B-galactosidase A or B chain hybrid gene was partially purified, cleaved with cyanogen bromide, and S-sulfonated derivatives were formed.

Evidence that correct expression has been achieved from chemically synthesized genes for human insulin can be summarized as follows: a) Radioimmunoactivity has been detected for both chains b) The DNA sequences obtained after cloning and plasmid construction have been directly verified to be correct such as constructed. Since radioimmunoactivity is obtained, translation must be in phase. Therefore, the genetic code ensures that peptides with the sequences of human insulin are produced. c) The E. coli products behave after cyanogen bromide cleavage as insulin chains in three different chromatographic systems that separate according to different principles (gel filtration, clone exchange and reverse phase HPLC). d) The E. coli-produced A-chain has been purified on a small scale by HPLC and has the correct amino acid composition.

Claims (14)

1. Method for producing a precursor polypeptide containing a specific polypeptide, comprising expression of a DNA sequence in a recombinant microbial cloning vector, characterized in that a heterologous DNA sequence is used as the DNA sequence which codes for a first amino acid sequence corresponding to the specific polypeptide in reading frame with a DNA sequence encoding another amino acid sequence so that the expression gives said precursor polypeptide having the amino acid sequence of the specific polypeptide and the second amino acid sequence, there being a selective cleavage site adjacent to said first amino acid sequence. 2. Method according to claim 1, characterized in that the second amino acid sequence is redundant, and that the polypeptide is cleaved after expression at said site to obtain the specific polypeptide. 3. Method according to claim 1 or 2, characterized in that the second amino acid sequence includes one or more amino acid sequences that correspond to that in the first amino acid sequence, and that all the amino acid sequences in the first amino acid sequence in the polypeptide are separated from each other by a selective cleavage site. 4. Method according to claim 2 or 3, characterized in that the cleavage is effected in a system which is exogenous to the replicating environment in the cloning vector. 5. Method according to claim 4, characterized in that the first amino acid sequence does not contain any similar selective cleavage site. 6. Method according to claim 5, characterized in that the first amino acid sequence lacks arginine and lysine, and that said selective cleavage site is cleaved with trypsin. 7. Method according to claim 5, characterized in that the first amino acid sequence lacks arginine, the selective cleavage site is cleaved with trypsin, and any lysine in the first amino acid sequence is first protected against cleavage. 8. Method According to claim 5, characterized in that the first amino acid sequence is methionine-free and that cleavage is effected with cyanogen bromide. 9. Method according to any one of the preceding claims, characterized in that the vector is a bacterial plasmid. 10. Method according to any one of the preceding claims, characterized in that the heterologous DNA sequence is followed, in the reading frame, by one or more termination codons. 11. A method according to any one of the preceding claims, characterized in that the second amino acid sequence precedes the first amino acid sequence. 12. Method according to any one of the preceding claims, characterized in that the first amino acid sequence is a mammalian hormone or an intermediate thereof. 13. Method according to claim 12, characterized in that the polypeptide lacks the hormone's bioactivity. 14. Method according to claim 12 or 13, characterized in that said hormone is somatostatin, insulin, proinsulin, the A-chain in human insulin or the B-chain in human insulin.