CA1302321C - Preparation of polypeptides - Google Patents

Abstract

ABSTRACT OF THE DISCLOSURE
A polypeptide comprising the A and B polypeptide chains of human insulin is prepared. A microorganism transformed by an expression transfer vector comprising a deoxynucleotide sequence coding for human proinsulin is incubated under conditions suitable for expression of the sequence coding for human proinsulin, and the human proinsulin is purified from the lysate or culture medium of the microorganism.

Description

130~:321 PREPAR~TION OF POLYPEPTIDES
The present invention relates to the preparation of a polypeptide comprising the A and B polypeptide chains of human insulin.
This application is a division of copending Canadian patent application Serial No. 360,565 filed September 11, 1980. The parent application describes DNA
transfer vectors comprising a deoxynucleotide sequence coding for human proinsulin and a deoxynucleotide sequence coding for human preproinsulin, microorganisms transformed by the transfer vectors and processes for the preparation of the DNA transfer vectors.
Insulin is a hormone produced primarily by the B cells of the pancreas. At the present time, the use of this hormone in the treatment of diabetis is well-known. Although slaughterhouses provide beef and pig pancreases as insulin sources, a shortage of this hormone is developing as the number of diabetics increases worldwide. Moreover, some diabetics develop an allergic reaction to beef and pig 20 insulin, with deleterious effects. The ability to produce human insulin in quantities sufficient to satisfy world needs is therefore highly desirable. The present invention provides genes, which are insertable into microorganisms, which are useful in the production of human insulin.
Insulin consists of two polypeptide chains, known as the A and B chains, linked together by disulfide bridges.
The A chain consists of 21 amino acids and the B chain consists of 30 amino acids. These chains are not synthesized independently in vivo but are derived from an ,. ~

130Z3Z~

immediate precursor, termed proinsulin: Proinsulin is a single polypeptide chain that contains a peptide, termed the C-peptide, which connects the A and B chains. See Steiner, D. F. et al., Science 157, 697 (1967). This C-peptide is excised during the packaging of insulin into the secretory granules of pancreatic B cells prior to secretion. See Tager, H. S. et al., Ann.Rev.Biochem. 43, 509 (1974). The current view of the function of the C-peptide is that it functions only in forming the three dimensional structure of the molecule. The amino acid sequence for human proinsulin, determined by conventional techniques, is given in Table l. In this table the B chain is amino acids 1-30, the C-peptide is amino acids 31-65 and the A
chain is amino acids 66-86.
Table 1 NH2-Phe-Val-Asn-Gln-His-Leu-Cys-Gly-Ser-His-Leu-Val-Glu-Ala-Leu-Tyr-Leu-Val-Cys-Gly-Glu-Arg-Gly-Phe-Phe-Try-Thr-Pro-Lys-Thr-Arg-Arg-Glu-Ala-Glu-Asp-Leu-Gln-Val-Gly-Gln-Val-Glu-Leu-~ly-Gly-Gly-Pro-Gly-Ala-Gly-Ser-Leu Gln-Pro-Leu-Ala-Leu-Glu-Gly-Ser-Leu-~ln-Lys-Arg-Gly-Ile-Val-Glu-Gln-Cys-Cys-Thr-Ser-Ile-Cys-Ser-Leu-Tyr-Gln-Leu-~lu-Asn-Try-Cys-Asn Chemical synthesis of this sequence of 86 amino acids though feasible is difficult using conventional techniques.
In the pancreatic B cells, the initial translation product is not proinsulin itself, but a pre-proinsulin that contains more than 20 additional amino acids on the amino terminus of proinsulin. See Cahn, S. J. et al., Proc.Nat.Acad.Sci. USA 73, 1964 (1976) and Lomedico, P. T.
_ ~30Z32~

et al., Nucl. Acid Res. 3, 381 (1976). The additional amino =
acid sequence is termed the signal peptide. In human pre-proinsulin (see Figure 2), the signal peptide has twenty-four amino acids and the sequence is. NH2-Met-Ala-Leu-Trp-Met-Arg-Leu-Leu-Pro-Leu-Leu-Ala-Leu-Leu-Ala-Leu-Trp-Gly-Pro-Asp-Pro-Ala-Ala-Ala. The twenty-four amino acid sequence is thought to be a specific signal for the vectorial trans-port of the s~nthesized polypeptide into the endoplasmic reticulum of the B cell, and is cleaved away from proinsulin during this phase. See Blobel, G. et al., J.Cell.Biol. 67, 835 (1975).
Several instances of signal peptides are known for eucaryotic proteins to be transported acros~ membrane barriers. A specific cleavage enzyme has been observed in a cell-free system which hydrolyzes the peptide bond between the signal peptide and the active protein concomitant with passage through a cell membrane. (See, Blobel, G. et al Proc.Nat.Acad.Sci USA 75, 361 (1978)).
=
Recent advances in biochemistry and in recombinant DNA
technology have made it possible to achieve the synthesis of specific proteins under controlled conditions independent of the higher organism from which they are normally isolated.
Such biochemical synthetic methods employ enzymes and sub-cellular components of the protein synthesizing machinery of living cells, either ln vitro, in cell-free systems, or in vivo, in microorganisms. In either case, the key element is provision of a deoxyribonucleic acid (DNA~ of specific sequen~.e which contains the infor~ation necessary to specify the desired amino acid sequence. Such a specific DNA is ~30232~

herein termed a gene. The coding relationship whereby a deoxynucleotide sequence is used to specify the amino acid sequence of a protein is described briefly, infra, and operates according to a fundamental set of principles that obtain throughout the whole of the known realm of living organisms.
A cloned gene may be used to specify the amino acid sequence of proteins synthesized by in vitro systems. DNA-directed protein synthesizing systems are well-known in the lo art, see, e.g., Zubay, G., Ann.Rev.Genetics 7, 267 (lg73~.
In addition, single-stranded DNA can be induced to act as messenger RNA in vitro, resulting in high fidelity trans-lation of the DNA sequence (Salas, J. et al., J Biol.Chem.
243, 1012 (1968). Other techniques well known in the art may be used in combination with the above procedures to enhance yields.
Developments in recombinant DNA technology have made it possible to isolate specific genes or portions thereof from higher organisms, such as man and other mammals, and to transfer the genes or fragments to a microorganism, such as bacteria or yea~t. The transferred gene is replicated and propagated as the transformed mircoorganism replicates. As a result, the transformed microorganism may become endowed with the capacity to make whatever protein the gene or fragment encodes, whether it be an enzyme, a hormone, an antigen or an antibody, or a portion thereof. The micro-organism passes on this capability to its progeny, so that in effect, the transfer has resulted in a new strain, having the described capability. See, for example, Ullrich, A. et al., Science 196, 1313 (1977), and Seeburg, P.H., et al., Nature 270, 486 (1977). A basic fact underlying the application of ehis technology for practical purposes is that DNA of all living organisms, from microbes to man, is chemically similar, being ~omposed of the same four nucleo-tides. The significant differences lie in the sequences of these nucleotides in the polymeric DNA molecule. The nucleotide sequences are used to specify the amino acid sequences of proteins that comprise the organism. Although most of the proteins of different organisms differ from each other, the coding relationship between nucleotide sequence and amino acid sequence is fundamentally the same for all organisms. For example, the same nucleotide sequence which codes for the amino acid sequence of HG~
in human pituitary cells, wlll, when transferred to a micro-organism, be recognized as coding for the same amino acid sequence.
Abbrevia~ions used herein are given in Table 2.
Table 2 DNA - deoxyribonucleic acid A Adenine RNA - ribonucleic acid T - Thymine cDNA - complementary DNA G - Guanine (enzymatically synthesized C - Cytosine from an mRNA sequence) U - Uracil 25 mRNA - messenger RNA ATP - adenosine triphosphate dATP - deoxyadenosine triphos- TTP - Thymidine triphosphate phate EDTA - Ethylenediaminetetra-dGTP - deoxyguanosine triphos- acetic acid phate 30 dCTP - deoxycytidine triphos-phate The coding relationships between nucleotide sequence in DNA and amino acid ~equence in protein are collectively known as the genetic code, shown in Table 3.

~302321 Table 3 Genetic Code Phenylalar.ine(Phe) TTK Histidine(His) CAK
Leucine(~eu) XTY Glutamine(Gln) CAJ
5 Isoleucine(Ile) ATM Asparagine(Asn) AAK
MethioninetMet) ATG Lysine(Lys) AAJ
Valine(Val) GTL Aspartic acid(Asp) GAK
Serine(Ser) QRS Glutamic acid(Glu) GAJ
Proline(Pro) CCL Cysteine(Cys) TGK
10 Threonine(Thr) ACL Tryptophan(Try) TG&
Alanine(Ala) GCL Arginine(Arg) ~JGZ
Tyrosine(Tyr) TAK Glycine(Gly) G&L
Termination signal TAJ
Termination signal TGA
Key: Each 3-letter deoxynucleotide triplet corresponds to a trinucleotide of mRNA, having a 5'-end on the left and a 3'-end on the right. All DNA sequences given herein are those of the strand whose sequence corresponds to the mRNA seq-uence, with thymine substituted for uracil. The letters stand for the purine or pyrimidine bases forming the deoxy-nucleotide sequence.
A - adenine J = A or G
G - guanine K = T or C
C = cytosine L = A, T, C or G
25 T - thymine M = A, C or T
X = T or C if Y is A or G
X - C if Y is C or T
Y = A, G, C or T if X is C
Y = A or G if X is T
W - C or A if Z is A or G
W - C if Z is C or T
Z = A, G, C or T if W is C
Z - A or G if W is A
QR - TC if S is A, G, C or T
QR - AG if S is T or C
S - A, G, C or T if QR is TC
S - T or C if QR is AG
An important feature of the code, for present purposes, is the fact that each amino acid is specified by a trinuc-leotide sequence, also known as a nucleotide triplet. Thephosphodiester bonds joining adjacent triplets are chemically indistinguishable from all other internucleotide bonds in 13023;~1 DNA. Therefore the nucleotide sequence cannot be read to code for a unique amino acid ~equence without additional information to determine the reading frame, which is the term used to denote the grouping of triplets used by the cell in decoding the genetic message.
Many recombinant DNA techniques employ two classes of compounds, transfer vectors and restric~ion enzymes, to be discussed in turn. A transfer vector is a DNA molecule which contains, inter alia, genetic information which insures its own replication when transferred to a host microorganism strain. Examples of transfer vectors commonly used in bacterial genetics are plasmids and the DNA of certain bacteriophages. Although plasmids have been used as the transfer vectors for the work described herein, it will be understood that other types of transfer vectors may be employed. Plasmid is the term applied to any autono-mously replicating DNA unit which might be found in a microbial cell, other than the genome of the host cell itself. A plasmid is not genetically linked to t~e chromo-some of the ho~t cell. Plasmid DNA's exist as double-stranded ring structures generally on the order of a few million daltons molecular weight, although some are greater than 108 daltons in molecular weight. They usually represent only a small percent of the total DNA Qf the ^ell. Transfer vector DNA is usually separable from host cell DNA by virtue of the great difference in size between them. Transfer vectors carry genetic information enabling them to replicate within the host cell, in most cases independently of the rate of host cell division. Some plasmids have the property 130232~

that their replication rate can be controlled by the inves-tigator by varia~ions in the growth conditions. By appropriate t~niques, the plasmid ~N~ rmg may be opened~ a fra~t of heter-olcgous DNA inserted, and the ring reclosed, forming an enlarged molecule comprising the inserted DNA segment.
Bacteriophage DNA may carry a segment of heterologous DNA inserted in place of certain non-essential phage genes.
Either way, the transfer vector serves as a carrier or vector for an inserted fragment of heterologous DNA.
Transfer is accomplished by a process known as trans-formation. During transformation, bacterial cells mixed with plasmid DNA incorporate entire plasmid molecules into the cells. Although the mechanics of the process remain obscure, it is possible to maximize the proportion of bacte_ial cel~s capable of taking up plasmid DNA and hence of being transformed, by certain empirically determined treatments. Once a cell has incorporated a plasmid, the latter is replicated within the cell and the plasmid replicas are distributed to the daughter cells when the cell divides. Any genetic information contained in the nucleotide se~uence of the plasmid DNA can, in principle, be expressed in the host cell. Typically, a transformed host cell is recognized by its acquisition of ~raits carried on the plasmid, such as resistance to certain antibiotics. Different plasmids are recognizable by the different capabilities or combination of capabilities which they confer upon the host cell containing them. Any given plasmid may be made in quantity by growing a pure culture of cells containing the plasmid and isolating the plasmid 1;~0232~

DNA ~herefrom.
Restriction endonucleases are hydrolytic enzymes capable of catalyzing site-specific cleavage of DNA molecules. The locus of restriction endonuclease action is determinPd by the existence of a specific nucleotide sequence. Such a sequence is termed the recognition site for the restriction endonuclease. Restriction endonucleases from a variety of sources have been isolated and characterized in terms of the nucleotide sequence of their recognition sites. Some restriction endonucleases hydrolyze the phosphodiester bonds on both strands at the same point, producing blunt ends. Others catalyze hydrolysis of bonds separated by a few nucleotides from each other, producing free single stranded regions at each end of the cleaved molecule. Such single stranded ends are self-complementary, hence cohesive, and may be used to rejoin ~he hydrolyzed DNA. Since any DNA susceptible of cleavage by such an enzyme must contain the same recognition site, the same cohesive ends will be produced, so that it is possible to join heterologous sequences of DNA which have been treated with a restriction endonuclease to other sequences similarly treated. See Roberts, R.J., Crit.Rev.Biochem. 4, 123 (1976). Restriction -sites are relatively rare, however the general utility of restriction endonucleases has been greatly amplified by the chemical synthesis of double stranded oligonucleotides bearing the restriction site sequence. Therefore virtually any segment of DNA can be coupled to any other segment simply by attaching the appropriate restriction oligo-nucleotide to the ends of the molecule, and subjecting the product to the hydrolytic action of the appropriate restriction endonuclease, thereby producing the requisite cohesive ends. See Heyneker, H.L., et al., Nature 263, 748 (1976) and Scheller, R.H., et al., Science 196, 177 (1977). An important feature of the distribution of restriction endonuclease recognition sites is the fact that they are randomly distributed with respect to reading frame. Consequently, cleavage by restriction endonuclease may occur between adjacent codons or it may occur within a codon.
More general methods of DNA cleavage or for end sequence modification are available. A variety of non-specific endonucleases may be used to cleave DNA randomly, as discussed infra. End sequences may be modified by creation of oligonucleotide tails of dA on one end and dT
at the other, or of dG and dC, to create sites for joining without the need for specific linker sequences.
The term "expression" is used in reco~nition of the fact that an organism seldom if ever makes use of all its genetically endowed capabilities at any given time. Even in relatively simple organisms such as bacteria, many proteins which the cell is capable of synthesizing are not synthesized, although they may be synthesized under appro-priate environmental conditions. When the protein product, coded by a gi~en gene, is synthesized by the organism, the gene is said to be expressed. If ~he protein product is not made, the gene is not expressed. Normally, the expression of genes in E. coli is regulated as described generally, infra, in such manner that proteins whose function is not ~30Z321 useful in a given enviror~ent are not synthesized and meta-bolic energy is conserved.
The means by which gene expression is controlled in E. coli is well understood, as the result of ex~ensive studies over the past twenty years. See, generally, Hayes, W., The Genetics of Bacteria And Their Viruses, 2d edition, John Wiley & Sons, Inc., New Yor~ (1968), and Watson, J.D., The Molecular BiologY of the Gene, 3d edition, Benjamin, Menlo Park, California (1976). These studies have revealed that several genes, usually those coding for proteins carrying out related functions in the cell, are found clustered together in continuous sequence.
The cluster is called an operon. All genes in the operon are transcribed in the same direction, beginning with the codons coding for the N-terminal amino acid of the first protein in the sequence and continuing through to the C-terminal end of the last protein in the operon. At the beginning of the operon, proximal to the N-terminal amino acid codon, there exists a region of the DNA, termed the control region, which includes a variety of controlling elements including the operator, promoter and sequences for the ribosomal binding sites. The function of these sites is to permit the expression of those genes under their control to be respon~ive to the needs of the organism.
For example, those genes coding for enzymes required exclu-sively for utilization of lactose are normally not appreciably expressed unless lactose or an analog thereof is actually present in the medium. The control region functions that must be present for expression to occur are the initiation 13023Zl of transcription and the initiation of translation. Expression of the first gene in the sequence is initiated by the initiation of transcription and translation at the position coding for the N-terminal amino acid of the first protein of the operon. The expression of each gene downstream from that point is also initiated in turn, at least until a termination signal or another operon is encountered with its own control region, keyed to respond to a different set of environmental cues. While there are many variations in detail on this general scheme, the important fact is that, to be expressed in a procaryote such as E. coli, a gene must be properly located with respect to a control region having initiator of transcription and initiator of translation functions.
It has been demonstrated that genes not normally part of a given operon can be inserted within the operon and controlled by it. The classic demonstration was made by Jacob, F., et al., J.Mo .Biol. 13, 704 (1965). In that experiment, genes coding for enzymes involved in a purine biosynthesis pathway were transferred to a region controlled by the lactose operon. The expression of the purine biosynthetic enzyme was then observed to be repressed in the absence of lactose or a lactose analog, and was rendered unresponsive to the environmental cues normally regulating its expression.
In addition to the operator region regulating the initiation of transcription of genes downstream from it, there are known to exist codons which ful~ction as stop signals, indicating the C-terminal end of a given protein.

13~Z32~

See Table 3. Such codons are known as termination signals and also as nonsense codons, since they do not normally code for any amino acid. Deletion of a termination signal between structural genes of an operon creates a fused gene s which could result in the synthesis of a chimeric protein consisting of two amino acid sequences coded by ad~acent genes, joined by a peptide bond. That such chimeric proteins are synthesiæed when genes are fused was demon-strated by Benzer, S., and Champe, S.P., Proc.Nat.Acad.Sci 10 USA 48, 114 (1962).
Once a given gene has been isolated, purified and inserted in a transfer vector, the over-all result of which is termed the cloning of the gene, its availability in substantial quantity is assured. The cloned gene is transferred to a suitable microorganism, wherein the gene replicates as the microorganism proliferates and from which the gene may be reisolated by conventional means.
Thus is provided a continuously renewable source of the gene for further manipulations, modifications and transfers to other vectors or other loci within the same vector.
Expression is obtained by transferring the cloned gene, in proper orientation and reading frame, into a control region such that read-through from the procaryotic gene results in synthesis of a chimeric protein comprising the 25 amino acid sequence coded by the cloned gene. A variety of specific protein cleavage techniques may be used to cleave the chimeric protein at a desired point so as to release the desired amino acid sequence, which may then be purified by conven~ional means. Techniques for con-structing an expression transfer vector having the cloned gene in proper juxtaposition with a control region are de~cribed in Polisky, B., et al., Proc.Nat.Acad.Sci USA 73, 3900 (1976); Itakura, K., et al., Science 198, 105~ (1977);
Villa-Komaroff, L., et al., Proc.Nat.Acad.Sci USA 75, 3727 (1978); Mercereau-Puijalon, O., et al., Nature 275, 505 (1978); Chang, A.C.Y., et al., Nature 275 (1978), and in our co ~ d~g Canadian patent application Serial No. 333,646 filed August 13, 1979.
In summary, the process whereby a mammalian pro~ein, such as human pre-proin~ulin or proinsulin, is produced with the aid of recombinant DNA technology first requires the cloning of the mammalian gene. Once cloned, the gene may be produced in quantity, further modified by chemical or enzymic means and transferred to an expression plasmid.
The cloned gene is also-useful for isolating related genes, or, where a fragment is cloned, for isolating the entire gene,, by using the cloned gene as a hybridlzation probe.
Further, the cloned gene is useful in pro~ing by hybridization, the identity or homology of independent isolates of the same or related genes. Because of the nature of the genetic code, the cloned gene, when translated in the proper reading frame, will direct the production only of the a~no acid seq~ce for which it codes and no o~er.
Some wor~ has been performed on the isolation and purifica~ion of rat proinsulin. Ullrich, A. et al., supra, and Villa-Komaroff, L. et al., supra dcscribe the isolation and purification of the rat proinsulin ~30232 gene and a method for transferring this gene to and replicating this gene in a microorganism. Ullrich et al. recovered several recombinant plasmids which contained the coding sequence for proinsulin, the 3' untranslated region and a part of the prepeptide.
Expression of the rat DNA eontaining the insulin coding sequence was diselosed m our eopendmg Canad~n applieation SerJal No.
333,646. Villa-Komaroff et al recovered one recombinant plasmid whieh eontained the eoding sequenee for amino acids 4-86 of proinsulin. This proinsulin sequence was separated from amino acids 24-182 of penicillinase, (~ -lactamase) by the coding sequenee for six glyeines. This penieillinase-proinsulin eoding sequenee was expressed to produee a fused protein. These artieles deseribe some of the basic procedures utilized in reeombinant DNA teehnology. However, they do not deseribe the isolation and purification of the human pre-proinsulin gene or human proinsulin gene.
A different approach to obtain human insulin has been taken by Crea, R. et al., Proc.Nat.Acad.Sci USA 75, 5765 (1978). This approach is to chemically synthesize coding sequences for (l) the A chain and (2) the B chain of human insulin, using codons favored by E. coli. These two se~uences can then be inserted into pla mids which can be expresA~ed to produce the A and B chains. Human insulin could then be generated by formation of the correct disulfide bonds berween the two protein chains.
The cloned gene for human pre-proinsulin is useful in a variety of ways. Transposition to an expression transfer vector will permit the synthesis of pre-proinsulin by a host microorganism transformed with the vectsr carrying the cloned gene. Growth of the transformed host will result in synthesis of pre-proinsulin as part-of a chimeric protein.
If the procaryotic portion of the fusion protein is the signal portion of an excre~ed or otherwise compartmentalized host protein, excretion or compartmentization can occur greatly enhancing the stabili~y and ease of purification of the pre-proinsulin fusion protein. Additionally, where the procaryotic portion is short, excretion from the io procaryotic host may be facilitated by the prepeptide itself, if the pre-sequence functions in the procaryotic host as it doeq in the eucaryotic cell. The pre-proinsulin gene may also be used to obtain the proinsulin gene using techniques as described below.
The cloned pre-proinsulin gene can be used in a variety of techniques for the production of pre-proinsulin.
Pre-proinsulin itself is useful because it can be converted to proinsulin by known enzymatic and chemical techniques.
For example, the prepeptide can be removed by a soluble enzymatic preparation, as described by Blobel, G. et al., supra, specific for remo~al of signal peptides. The cloned proinsulin gene can be used in a variet~ of techniques for the production of proinsulin. The proinRulin, produced from either gene, itself is useful becau~e it can be con-verted to insulin by known enzymatic and chemical techniques.See Kemmber, W., et al., J.Biol.Chem. 242, 6786 (1971).
In accordance with the present invention, there is provided a process for preparing a polypeptide comprising the A and B polypeptide chains of human insulin which comprises incubating a microorganism transformed by an expression transfer vector comprising a DNA sequence encoding human proinsulin, the proinsulin being of the formula B-C-A, under conditions suitable for expression of the DNA sequence encoding human proinsulin, purifying human proinsulin from the lysate or culture medium of the microorganism, and converting the proinsulin to insulin comprising the A and B chains.
The DNA sequence encoding human proinsulin comprises a DNA
sequence encoding a B chain of the sequence Phe-Val-Asn-Gln-His-Leu-Cys-Gly-Ser-His-Leu-Val-Glu-Ala-Leu Tyr-Leu-Val-Cys-Gly-Glu-Arg-Gly-Phe-Phe-Tyr-Thr-Pro-Lys-Thr and a DNA sequence encoding an A chain of the sequence Gly-Ile-Val-Glu-Gln-Cys-Cys-Thr-Ser-Ile-Cys-Ser-Leu-Tyr-Gln-Leu-Glu-Asn-Tyr-Cys-Asn.
The DNA sequence encoding human proinsulin comprises a plus strand having: (a) a DNA sequence (NaNbNC)j having a 5' and 3' end encoding a C peptide, wherein Na7 Nb and Nc may be A, T, G or C and j is any integer from O to 100 such that NaNbNC is not TGA, TAA or TAG, (b) a DNA sequence encoding the B chain of human insulin attached to the 5' end of the (NaNbNC)j; and (c) a DNA
sequence encoding the A chain of human insulin attached to the 3' end of the (NaNbNC)j.
The present invention and the invention described in the parent application provide the essential genetic elements needed for the production of human insulin by techniques adaptable to industrial processes. The naturally occurring structural gene has been cloned. Its expression in an appropriate host cell type, yields a protein product i302321 which is convertible to insulin by known methods. The present invention is fundamentally based upon the proinsulin molecule and takes advantage of the fact that the C-peptide region of the proinsulin molecule permits a spontaneous folding such that the A and B chains are properly juxt-aposed. In such configuration, the correct pairing of sulfhydryl groups is assured and the formation of the disulfide cross l.inks, as found in the active insulin molecule, are readily formed. Excision of the C-peptide is carried out by the combined use of trypsin, or an enzyme having similar substrate specificity, and carboxy-peptidase B or cathepsin-B to remove a C-terminal arginine on the B strand, as described in the prior art.
The entire structural coding portion of the human 15 insulin gene has been cloned as described herein, including a portion of the 5' untranslated region, all of the prepeptide sequence, the entire coding sequence for the B, C, and A
peptides, and all of the 3' untranslated region. The choice of host cell will effect what portions of the cloned gene is used. The entire preproinsulin coding sequence will be useful in transforming certain ~pecies of higher eucaryotic cells, since the presequence acts as a signal peptide to promote excretion of the peptide from the cell. In addition, those eucaryotic cell lines capable of responding to the signal peptide will catalyze the specific 130;~32~

removal of the presequence during transport of the protein across the cell membrane. In addition, such cells may be capable of further processing the expression product by excising the C-pr~tein after catalyzing formation of the proper disulfide bonds.
In procaryotic host cells, use of the proinsulin coding sequence is presently preferred. The step of converting proinsulin to insulin is carried out in detail using techniques known in the art. Two types of expression are already known, both involve inserting the proinsulin coding sequence in a region of transfer vector DNA whose expression was controlled by a procaryotic promoter. In some cases, a portion of the procaryotic structural gene and the promoter control is interposed such that the expression product is a fusion protein having its N-terminal portion composed of the N-terminal part of the procaryotic protein and its C-terminal portion is the cloned sequence, in this instance, proinsulin. Expression of proinsulin as a fusion protein has t~e advantage that the fusion protein may be more stable in the host cell than proinsulin itself. In addition, where the procaryotic protein is one that is normally excreted from the host cell, the fusion protein may also be excreted, making it easier to purify the expression product. The use of fusion proteins has a disadvantage that they must be specifically cleaved to yield the desired product. In the case of proinsulin, techniques exist which take advantage of the sequence of the amino acid sequence of the protein, to specifically cleave a fusion pro~ein, as described in Example 2.

~.3~2321 The cloned coding sequence can also be expressed direc~ly in the procaryotic cell by inserting the sequence directly adjacent ~o a promoter. The advantage in this instance is that specific cleavage is unnecessary The 5 principle disadvantage is that further purification is necessary.
The expression of mammalian genes inserted in E.
coli has now been obtained by insertion near the lac, trp and ~-lactamase promoters. The lac promoter is useful because its genetics are well characterized. There are two posssible insertion sites, providing long and short procaryotic leaders for ~he fusion protein. A large variety of genetic variants is available, having ~arious levels of endogenous repressor, being temperature inducible and the like. The trp promoter has the advantage of providing high levels of expression when induced. Expression plasmids having insertion sites in the trp promoter are available for all three reading frames. The beta lactamase promoter provides what amounts to a procaryotic signal protein, since the lactamase is normally e~creted. The B-lactamase fusion product is also excreted, or found in the periplasmic space.
As a result, fewer purification steps will be required to achieve pure proinsulin.
The present invention takes advantage of the function of the C-peptide of proinsulin, namely to facili~ate the spontan-eous folding of proinsulin to bEing the A and B chains toge~her in the correct configuration such that the sulfhydryl groups are properly paired and the correct disulfide cross links are formed, as found in insulin isolated from nature. Comparison ~30Z321 of the C-peptide sequences are not highly conserved during evolution. Therefore, many substitutions of amino acids sequence are possible in the C~peptide. In fact, the function of the C-peptide may simply be a matter of providing an amino acid loop of proper length to allow the A and B chains chains to fold together in the proper configuration.
Consequently, almost any coding sequence could be inserted in place of the C-peptide sequence of human proinsulin without substantially altering its primary function. There are, however, certain advantages favoring the use of the natural C-peptide; for example, removal of the peptide from the insulin preparations need not be complete, since the C-peptide is a natural component of insulin preparations.
Furthermore, it may hold some advantage in conferring the proF~r confuguration on the insulin or in removal by enzymes.
The present invention further demonstrates the universality of the genetic code. In particular, codons favored by mammalian cells are correctly translated in E. coli. Substitute codons, coding for the same amino acid, could be substituted in the sequence without affecting the sequence or function of the expressed protein. Therefore, the present invention is intended to encompass all synthetic coding variance of the basic coding sequence actually cloned herein, insofar as such variants code for human preproinsulin and human proinsulin.
The invention is illustrated further by the following Examples:
Example 1 Cloned human pre-proinsulin gene. The human insulin gene was cloned from RNA isolated from human insulinoma using ~3~23;~1 the procedure described by Ullrich, et al., supra for isolating RNA from islet cells. The RNA obtained as a pellet after centrifugation in 5.7 M CsCl containing 100 mM EDTA, was used without further fractionation as the t~mplate for c~N~ synthesis 5 using reverse transcriptase and Sl nuclease, as described by Ullrich, et al., supra. Approximately 40 ng of double stranded cDNA was prepared from 130 ug of unfractionated RNA.
The unfractionated cDNA was treated with terminal trans-ferase in the presence of ~ to generate oligo-C tails on the 3'-termini of the cDNA molecules. Similarly, the plasmid transfer vector pBR322 was cleaved with the restriction endonuclease Pst I and treated with terminal transferase in the presence of dGT~ to generate oligo-dG tails on the 3'-termini of the linear plasmid DNA. ~he plasmid DNA thus 15 treated cannot ~orm a circular DNA since the single-s~randed ends are not complementary. However, the oligo-dC ends of treated cDNA are complementary with the oligo-dG ends of the plasmid and can, therefore, be used to form circuLar plasmids having cDNA inserted at the Pst I site. Furthermore, the insertion regenerates both ends of the Pst I recognition sequences, at both ends of the insert, thereby providing for excision of the inserted sequences. (See Villa-Komeroff, et al, supra.) The foregoing tailing procedure was used to generate 25 pBR-322 plasmids having a hetrogeneous population of cDNA
inserts at the Pst I site. Such plasmids would be ampicillin-sensitive, since the Pst I site is in the ~ -lactamase gene, but remain tetracycline resistant. E. coli stain X1776 was transformed with the plasmid DNA containing inserts.

Five hundred twenty-five tetracycline resistant trans-formants were obtained. These were replica-plated and screened for insulin sequences by in situ colony hybridization, essentially as described by Gruenstein and Hogness, Proc.
Nat. Acad. Sci. USA 72, 3961 (1975). The previously cloned rat preproinsulin cDNA (Ullrich, A., et al, supra) was used as a hybridization probe, after labeling the rat cDNA by nick translation using DNA polymerase I. Since the amino acid sequences of rat and human insulins are quite similar (insulin, 92Z homology; proinsulin, 83~ homology), it was anticipated that the cloned rat cDNA would cross-hybridize with human insulin sequences under conditions of reduced stringency known in the art. Autoradiography revealed that two out o the 525 colonies hybridized with the cloned rat preproinsulin-I probe. Both colonies were ampicillin-sensitive. The plasmids isolated from the colonies were approxi~ately 250 base pairs (bp) and 500 bp longer than pBR322 itself. The larger plasmid was designated pcHI-l.
The nucleotide sequence of the inserted cDNA fragment of pcHI-l was determined by the method of Maxam and Gilbert, Proc. Nat. Acad. Sci. USA 74, 560 (1977). A schematic diagram of the insert lS sho~n in Fi~ure 1. The entire coding region for human pre-proinsulin was contained in the insert, together with a portion of the 5'-untranslated region, all of the 3'-untranslated region and part of the 3'-polydA
region. The restriction sites used to cleave the insert for sequencing are also shown in ~. 1, together with the direction of sequencing in each fragment and the regions of overlap.

The mRNA sequence derived from the cDNA sequence, and the amino acid sequence deduced from one of the reading frames is shown in ~ 2. The primary structure of human proinsulin determined in this manner agrees precisely with 5 that obtained by previous amino acid sequencing experiments (Dayhoff, M.O., Atlas of Protein Sequence and Structure, 5, Supp. 2, pp. 127-130 (1976) and Suppl. 3, pp. 150-151 (1978). The amino acids of the proinsulin coding sequence are numbered from 1 to 86; those of the pre-sequence are 10 numbered from -24 to -1. Fi~. 2 further shows the location of certain restriction sites useful for the construction of the proinsulin gene from the pcHI-l insert. It will be understood that the cDNA sequence of the coding strand of the pcHI-l insert, is the same as that shown in ~. 2 15 except that Thymine (T) is substituted for Uracil (U).
Large quantities of pcHI-l and other plasmids are prepared by transforming E. coli HB-101 therewith. The HB-101 strain therefore serves as a convenient host for maintaining and replicating plasmids having cloned inserts 20 as described herein.
Examp1e 2 Construction of proinsulin transfer vector. The insert cDNA of pcHI-l contains the entire coding sequence for human pre-proinsulin. For certain applications, such 25as transfer to another species of eucaryotic cell, the normal processing and removal of the pre-sequence and C-peptide, and the attainment of a correctly folded con-figuration yield~ng active insulin may be expected. In other circumstances, transfer to a procaryotic cell such as a bacterium, may be expected to yield the unprocessed protein.
In the latter situation, construction of a coding sequence for proinsulin will be advantageous, slnce proinsulin is readily converted to active insulin in vitro, using techniques well known in the art. See Kemmler, W., et al., J. Biol. Chem.
242, 6786 (1971). Three alternative methods for construction of a proinsulin coding sequence are disclosed herein. A
plasmid transfer vector comprising pBR322 with an inserted proinsulin coding sequence is designated pcHP-l.
A. A chemically synthesized coding sequence for the human insulin B chain has been described by Crea, R., et al., Proc. Nat. Acad. Sci. USA, 75, 5765 (1978). The synthetic nucleotide sequence differs from the naturally occuring sequence disclosed herein because the synthetic sequence was designed to exploit codon assignments more favored by a procaryotic host, such as E. coli. Also, a triplet coding for methionine was incorporafed just prior to amino acid 1 (phe).
Fortuitously, however, the two sequences are identical in the region of amino acids 13-14, which region contains the only Alu I site common to both sequences. The locations of the Alu I sites on the natural sequence are shown in F g. 2.
Synthetic proinsulin DNA is treated with Alu I endonuclease to yield two fragments of 43 bp and about 56 bp, respectively.
Similarly, the cDNA insert of pcHI-l preferably obtained by Hha I cleavage as described in Example 2B, is cleaved by partial hydrolysis catalyzed by Alu I endonuclease to form fragments of about 75, 90, 110, 135, 165, 200, 245, 275, 375, and 455 bp, respectively;these are fractionated by gel electro-phoresis to obtain the 375 bp fragment, the result from a 130232~

single site cleavage in the codon for amino acid number 14.
The synthetic gene cleavage fragments and the 375 bp single site cleavage fragment of the cDNA insert are joined by blunt-end ligation. Correct joining of the 43 bp synthetic fragment with the 375 bp cDNA fragment is maximized by providing that the 375 bp cDNA is present in molar excess.
The possibilitie~ for incorrect joining are also reduced by the fact that the synthetic fragments have single-stranded protruding ends that are not complementary with these of the 375 bp cDNA fragment.
The joined molecule, a composite of the synthetic sequence coding for methionine followed by amino acids 1-13, and the naturally occuring sequence coding for amino acids 14-86 of proinsulin, constitutes a coding sequence for proinsulin. The proinsulin coding sequence may be inserted in any chosen expression plasmid by either filling or excising the single-stranded ends and then attaching the appropriate linker oligonucleotides.
Expression yields a fusion protein which may be cleaved at the methionine residue by treatment with cyanogen bromide to yield proinsulin. Proinsulin is converted to insulin by the method of Kemmler, et al., supra.
B. The cDNA insert o pcHI-l has a Hha I site in the sequence coding for amino acids -14 to -13, as shown in Fi~. 2. In addition, the transfer vector pBR322 has a Hha I site just 22 bp from the Pst I site at the 5'-end of the insert. It is therefore possible to reisolate a sequence including all of the proinsulin coding sequence and a 22 bp region of pBR322 DNA, by treating pcHI-l with Hha I endonuclease. This procedure is preferred to reisolating the insert wi~h Pst I endonuclease, since the insert contains two internal Pst I sites and the yield of intact insert DNA by P5 t I endonuclease treatment is low.
The & a I isolated sequence is also perfectly suitable for use in the procedures of Examples 2A and 2C herein.
Treatment of either isolate with Hha I endonuclease results in clevage of the plus strand between amino acids -14 and -13 of the pre-sequence. (The plus strand is defined as the strand whose nucleotide sequence corresponds to the mRNA sequence. The minus strand is the strand whose sequence is complementary to the mRNA sequence.) rhe remaining presequence may be specifically removed by exploiting the 3' to 5' exonuclease activity of T4 DNA polymerase, which acts on the minus strand at the pre-sequence coding endJ and on the plus strand at the opposite end. The exonucleolytic reaction may be stopped at a defined nucleotide by putting the same nucleotide, in triphosphate form, in the reaction mixture. The exonucleoytic action is then thwarted by the polymerase activity of the enzyme which continually replaces the specific nucleotide, as described by Englund, P.T. in J.
Biol. Chem. 246, 3269 ~1971) and in J. Mol. Biol. 66, 209 (1972).
The remaining pre-sequence may be digested specifically to the N-terminal phenylalanine codon of amino acid number 1 by three cycles of T4 polymerase digestion. Cycle l is carried out in the presence of TTP which will terminate digestion opposite the A of the glycine codon in amino acid poSitiOQ -7. Cycle 2 is carrie~ out in the presence of dCTP, which terminates digestion opposite the G of ~3~232~.

codon position -5 (Asp). Cycie 3 is caxried out in the presence of dATP which terminates digestion opposite the T which is the first nucleotide of position 1 of proinsulin.
After each cycle of T4 polymerase digestion, the trip~oshate of the just-completed cycle must be removed and the triphosphate for the forthcoming cycle introdùced.
Minicolumns of Sephadex G~75 1/ are employed to separate triphosphates from the reaction mixture. The columns may be equilibrated with a suitable buffer and samples collected 10 in a buffer containing the triphosphate of the succeeding cycle. In this way, enzyme comigrating with the DNA will not digest the DNA beyond the next selected stopping point.
To prevent such digestion in the lower part of the column after resolution of the preceding triphosphate, the chromatography 15 is carried out in the cold (4C) and elution is hastened by centrifugation. Alternatively, the enzyme is heat inactivated at 650C before the chromatography step. The succeeding cycle is then i~itiated by addition of fresh, active enzyme.
As a result of three T4 polymerase digestion cycles, the minus 20 strand of the DNA is digested completely and specifically to the Llrst codon of ~he proinsulin coding sequence. The sequence of the plus strand at the opposite end of the molecule which also has a 3' end, is such that only a few nucleotides are removed by the foregoing cycles of digestion.
The plus strand at the pre-sequence end is then specifically digested with Sl exonuclease, which acts on single-stranded ends to yield a blunt-ended molecule. Care must be taken to prevent partial digestion by Sl nuclease beyond the beginning of 1/ Trademar~, Pharmacia, Inc., Uppsala, Sweden.

~3~23;~.

the first codon. Such partial digestion may occur because of helix "breathing", a partial and transitory unpairing of DNA strands. Breathlng occurs throughout the molecule but most frequently in A-T rich regions and at the ends of DNA
5 molecules. A transitory unpairing at the A-T rich codon number 1 could permit Sl nucleolytic action beyond the desired stopping point. Such a result is prevented by carrying out Sl digestion under conditions of maximum helix integrity; low temperature (room temperature or less) 10 and high salt which is normally employed in Sl reaction buffer.
Samples of DNA obtained at successive stages of Sl nuclease digestion are cloned into a suitable transfer vector according to procedures known in the ar~. Sequence analysis of the smallest Alu I fragment of such clones is 15 used to screen for pro-insulin coding clones having the entire pre-sequence removed.
An attractive method for cloning the proinsulin-coding cDNA involves the incorporation of oligo-A tails, using terminal transferase, on the 3' ends of the cDNA. Oligo-T
20 tails are generated on the 3'-end at a suitable site on an expression transfer vector, such as the EcoRl site in the -galactosidase gene on plasmid pTS~ llrich, A~, et al) See also our Canadian patent application Serial No. 333,646.
Insertion of the proinsulin-coding cDNA by the foregoing method yields a correctly oriented insert in phase as to reading frame with the ~-galactosidase gene of the plasmid with a 1/6 probability. Expression of the oligo-A tails results in the incorporation of-lysine residues just prior to the beginning of the proinsulin 13~Z321 sequence. Mild trypsin digestion of the fusion protein yields proinsulin, which is converted to active insulin as previously described. Alternatively, the fusion protein is treated with a combination of ~rypsin and carboxypeptidase B (or s cathepsin B) to yield active insulin fro~ the fusion protein in a single reaction.
C. A proinsulin coding sequence is constructed by selective cleavage at an internal site in the proinsulin coding region, followed by ligation of a chemically synthesized sequence coding for that part of the proinsulin coding region removed by the previous cleavage. The plas~id pcHI-l is used as a source of the proinsulin coding region, which is selectively excised by treatment with Pst I endo-nuclease or preferably by treatment wieh Hha I endonuclease, 15 as described in Example 2B.
Either fragment, after isolation is treated with alkaline phosphatase to ~emove the 5' terminal phosphate g~oups, then cleaved by treatment with a restriction endonuclease having a unique cleavage point in 20 the proinsulin coding sequence. Preferably the restriction site is located near one of the ends of the proinsulin coding sequence. The Alu I site in region of amino acids 13-14 provides a convenient cleavage point (see Fig. 2).
The Hha I fragment of pcHI-1 is partially cleaved with Alu I endonuclease to generate two fragments of approx-imately 76 bp and approximately 375 bp, respectively. The Alu I fragments are fractionated by gel electrophoresis, as described in Example 2A, and the 375 bp fragment is recovered.

130Z32~

A nucleotide sequence coding for the first 13 amino acids of proinsulin'with a 5'-terminal G (on the plus strand), to complete the codon for alanine at position 14, is synthesized by the phosphotriester method, Itakura, K.1 et al. J. Biol. em. 250, 4592 (1975) and Itakura, K., et al., J. AM. Chem. Soc. 97, 7327 (1975). The plus strand of the synthetic DNA has the sequence 5'-TTTGTGAACCAACACCTG
TGCGGCTCACACCTGGTGGAAG-3', corresponding to the natural sequence. However, other sequences coding for the same 10 amino acids may be synthesized. In general the sequence is s'-TTKGTLAAKCAJCAKXTYTGK&GLQRSCAKXTYGTLCAJG-3'. The resulting sequence is blunt-end ligated wieh the approximately 375bp fragment of the Hha I fragment of pcHI-l. Since the latter has a 5'-phosphate only at the end to be joined, ~he two 15 fragments wili be joined in the correct order. The synethetic fragment is correctly joined to the larger fragment in approximately 50Z of the reactions.
The ligase-treated DNA is then cloned into a suieable expression plasmid, either by oligo-A tailing, as 20 described in ~xample 2B, or by attachment o~ 'linkers and insertion into expression plasmids of known reading frames.
In the case of oligo-A tailed inserts, expression of pro-insulin is observed in about 1/12 of the clones. In the ca~e of direct insertion where the reading frame is known 25 to be correct,'the frequency of expression clones is about 50%.
Example 3 Expression of human preproinsulin and proinsulin.The cloned inserts coding for preproinsulin (Example 1) or ~ 302321 proinsulin (Example 2) are inserted in an expression transfer vector. When ~he insertion occurs in the correct orientation with respect to initiation of translation at the insertion site, and the insert is in reading frame phase 5 with the pro~otor and ribosome binding site, the protein product of the cloned gene is synthesized by actively metabolizing host cells transformed by the transfer vector.
The protein product is a fusion protein if the expression transfer vector contains a portion of a procaryotic gene 10 between the promoter and the insertion sit~ owever, the insertion may be made im~ediately adjacent to a promoter site. In such cases, the protein coded by the insert is synthesized directly ~oth techniques present advantages and dis-advantages. Fusion proteins have the advantage that they 15 tend to stabilize the foreign protein coded by the inserted gene. Also, desirable functional properties such as excretion from the host cell are con~erred by fusion with certain host proteins such as~ -lactamase. On the other hand, purification of the insert coded sequence is co~plicated by the general 20 desirability of specifically removing the host portion of the fusion protein. Such removal is accomplished by known techniques as described in Examples 2A and 2B. Direct synthesis of the desired protein obviates the need for specific cleavage but generally precludes the possibility 25 of excretion from the host cell.
Expression plasmids have been developed wherein expression is controlled by the lac promoter (Itakura, et al., Scie~ce 198, 1056 (1977), Ullrich, A., et al., Excer~ Medica, ~1979); by the ~ promoter (Martial, et al., Science 205, 602 (1979); and by the ~-lactamase promotoer, as described in our copending Canadian patent application Serial No.
353,090 filed March 30, 1980.
Expression is detected by measurement of a product capable of binding immunochemically with anti-insulin antibody, or anti-proinsulin antibody. Radioimmunoassay, in which the antibody is radioactively labeled and antigen-antibody pairs are precipitated by a preparation of heat-killed Staphylcccoccus aureus C is employed. (See Morgan and Lazarow, Diabetes 12, 115 (1963) and Kessler, S.W., J. Immunol., 115, 1617 (1975).
Radioimmune screening, as described by Erlich, H.A., et al., Cell 10, 681 (1978) or by Broome, S. and Gilbert, W., Proc.
Nat. Acad. Sci. USA, 75, 2746(1978), is used for detecting expression in bacterial colonies.
Fusion proteins indicative of expression are detected by comparing molecular weights of the host protein contributing the N-tenminal part of the fusion protein, in host cells transformed by expression plasmids with and without an insert.
A preferred variant is to employ the minicell-producing E.
coli strain P678-54 as host. Radioactively labeled amino acids are incorporated into minicell proteins, comparing strains transformed with expression transfer vectors with and without the inserted proinsulin coding sequence. The proteins are fractionated by SDS-acrylamide gelelectrophoresis and the protein positions detected by autoradiography.
Expression of proinsulin is indicated by the presence of a labeled protein band found only in minicells transformed by 34 13 02 32~

the proinsulin expression plasmid. The position of the electrophoretic band provides a measure of the molecular weight of the expressed protein, and is consistent with the known length of the inserted gene and of the N-terminal procaryotic portion.
Removal of the procaryotic portion and conversion of proinsulin to insulin in vitro are carried out by known procedures, as described in detail supra.

Claims

The embodiments of the invention in which an exclusive property or privilege is claimed are defined as follows:
1. A process for preparing a polypeptide comprising the A and B polypeptide chains of human insulin, which comprises incubating a microorganism transformed by an expression transfer vector comprising a DNA sequence encoding human proinsulin, said proinsulin being of the formula B-C-A wherein A and B represent the A and B chains respectively of human insulin and C is encoded by a DNA sequence (NaNbNc)j having a 5' and 3' end, wherein Na, Nb and Nc may be A, T, G or C and j is any integer from 0 to 100 provided that NaNbNc is not TGA, TAA or TAG, and wherein said DNA sequence encoding the B chain of human insulin is attached to the 5' end of (NaNbNn)j; and wherein said DNA sequence encoding the A chain of human insulin attached to the 3' end of (NaNbNc)j under conditions suitable for expression of said DNA sequence encoding human proinsulin, purifying human proinsulin from the lysate or culture medium of said microorganism, and converting said proinsulin to insulin comprising said A
and B chains.
2. The process of claim 1 wherein the DNA sequence encoding human proinsulin comprises a DNA sequence encoding a B chain of the sequence:

and a DNA sequence encoding an A chain of the sequence 3. The process of claim 1, wherein said proinsulin is converted to insulin by oxidizing sulfhydryl groups on the A and B peptides of said human proinsulin prepared to form disulfide crosslinks between said A and B
peptides, and then excising the C peptide by an enzyme-catalyzed hydrolysis specific for the bonds joining the C peptide to the A and B peptides.
4. The process of claim 1 wherein the DNA sequence encoding human proinsulin comprises a plus strand having:
(a) a DNA sequence encoding the naturally occurring C peptide;
(b) a DNA sequence encoding the B chain of human insulin attached to the 5' end of the C-peptide encoding sequence; and (c) a DNA sequence encoding the A chain of human insulin attached to the 3' end of C-peptide encoding sequence.