CA2093643C - Peptide and protein fusions to thioredoxin and thioredoxin-like molecules - Google Patents

Abstract

This invention provides a fusion molecule comprising a DNA sequence encoding a thioredoxin-like protein fused to the DNA sequence encoding a selected heterologous peptide or protein. The peptide or protein may be fused to the amino terminus of the thioredoxin-like molecule, the carboxyl terminus of the thioredoxin-like molecule, or within the thioredoxin-like molecule, for example at the active-site loop of said molecule. Expression of this fusion molecule under the control of a regulatory sequence capable of directing its expression in a desired host cell, produces high levels of stable and soluble fusion protein. The fusion pro-tein, located in the bacterial cytoplasm, may be selectively released from the cell by osmotic shock or freeze/thaw procedures. It may be optionally cleaved to liberate the soluble, correctly folded heterologous protein from the thioredoxin-like portion.

Description

,~~~~~~J

PEPTIDE AND PROTEIN FUSIONS TO THIOREDOXIN
AND THIOREDOXIN-LIKE MOLECULES
The invention relates generally to the production of fusion proteins in prokaryotic and eukaryotic cells. More specifically, the invention relates to the expression in host cells of recombinant fusion sequences comprising thioredoxin or thioredoxin-like sequences fused to sequences for selected heterologous peptides or proteins, and the use of such fusion molecules to increase the production, activity, stability or solubility of recombinant proteins and peptides.
Backcrround of the Invention Many peptides and proteins can be produced via recombinant means in a variety of expression systems, e.g., various strains of bacterial, fungal, mammalian or insect cells. However, when bacteria are used as host cells for heterologous gene expression, several problems frequently occur.
For example, heterologous genes encoding small peptides are often poorly expressed in bacteria. Because of their size, most small peptides are unable to adopt stable, soluble conformations and are subject to intracellular degradation by proteases and peptidases present in the host cell. Those small peptides which do manage to accumulate when directly expressed in E. coli or other bacterial hosts are usually found in the insoluble or "inclusion body" fraction, an occurrence which renders them almost useless for screening purposes in biological or biochemical assays.
Moreover, even if small peptides are not produced in inclusion bodies, the production of small. peptides by recombinant means as candidates for new drugs or enzyme inhibitors encounters further problems. Even small linear peptides can adopt an .~.

enormous number of potential structures due to their degrees of conformational freedom. Thus a small peptide can have the 'desired' amino-acid sequence and yet have very low activity in an assay because the 'active' peptide conformation is only one of the many alternative structures adopted in free solution. This presents another difficulty encountered in producing small heterologous peptides recombinantly for effective research and therapeutic use.
Inclusion body formation is also frequently observed when the genes for heterologous proteins are expressed in bacterial cells. These inclusion bodies usually require further manipulations in order to solubilize and refold the heterologous protein, with conditions determined empirically and with uncertainty in each case.
If these additional procedures are not successful, little to no protein retaining bioactivity can be recovered from the host cells. Moreover, these additional processes are often technically difficult and prohibitively expensive for practical production of recombinant proteins for therapeutic, diagnostic or other research uses.
To overcome these problems, the art has employed certain peptides or proteins as fusion "partners" with a desired heterologous peptide or protein to enable the recombinant expression and/or secretion of small peptides or larger proteins as fusion proteins in bacterial expression systems. Among such fusion partners are included lacZ and trpE fusion proteins, maltose-binding protein fusions, and glutathione-S-transferase fusion proteins [See, generally, Current Protocols in Molecular Biology, Vol. 2, suppl. 10, publ. John Wiley and Sons, New York, NY, pp. 16.4.1-16.8.1 (1990); and Smith et al,.Gene 67:31-40 (1988)]. U. S. Patent 4,801,536 describes the fusion of a bacterial flagellin protein to a desired protein to enable the production of a heterologous gene in a bacterial cell and its secretion into the culture medium as a fusion protein. PCT
Patent Publication W091/11454 discloses fusion proteins using WO 92/13955 PC1'/US92/00944 biotinylated renin as the fusion partner. The renin is immobilized on a purification column to facilitate separation and cleavage.
However, often fusions of desired peptides or proteins to other proteins (i.e., as fusion partners) at the amino- or carboxyl- termini of these fusion partner proteins have other potential disadvantages. Experience in E. coli has shown that a crucial factor in obtaining high levels of gene expression is the efficiency of translational initiation. Translational initiation to in E. cola is vary sensitive to the nucleotide sequence surrounding the initiating methionine codon of the desired heterologous peptide or protein sequence, although the rules governing this phenomenon are not clear. For this reason, fusions of sequences at the amino-terminus of many fusion partner proteins affects expression levels in an unpredictable manner.
In addition there are numerous amino- and carboxy-peptidases in E.E. coli which degrade amino- or carboxyl-terminal peptide extensions to fusion partner proteins so that a number of the known fusion partners have a low success rate for producing 2o stable fusion proteins.
The purification of proteins produced by recombinant expression systems is often a serious challenge. There is a continuing requirement for new and easier methods to produce homogeneous preparations of recombinant proteins, and yet a number of the fusion partners currently used in the art possess no inherent properties that would facilitate the purification process. Therefore, in the art of recombinant expression systems, there remains a need for new compositions and processes for the production and purification of stable, soluble peptides and proteins for use in research, diagnostic and therapeutic applications.
Summary of the Invention In one aspect, the invention provides a fusion sequence comprising a thioredoxin-like protein sequence fused to a ;y5 selected heterologous peptide or protein. The peptide or protein may be fused to the amino terminus of the thioredoxin-like sequence, the carboxyl terminus of the thioredoxin-like sequence, or within the thioredoxin-like sequence (e. g., within the active-s site loop of thioredoxin). The fusion sequence according to this invention may optionally contain a linker peptide between the thioredoxin-like sequence and the selected peptide or protein.
This linker provides, where needed, a selected cleavage site or a stretch of amino acids capable of preventing steric hindrance between the thioredoxin-like molecule-and the selected peptide or protein.
As another aspect, the invention provides a DNA molecule encoding the fusion sequence defined above in association with, and under the control of, an expression control sequence capable of directing the expression of the fusion protein in a desired host cell.
Still a further aspect of the invention is a host cell transformed with, or having integrated into its genome, a DNA
sequence comprising a thioredoxin-like DNA sequence fused to the DNA sequence of a selected heterologous peptide or protein. This fusion sequence is desirably under the control of an expression control sequence capable of directing the expression of a fusion protein in the cell.
As yet another aspect, there is provided a novel method for increasing the expression of soluble recombinant proteins. The method includes culturing under suitable conditions the above-described host cell to produce the fusion protein.
In one embodiment of this method, if the resultinq fusion protein is cytoplasmic, the cell can be lysed by conventional means to obtain the soluble fusion protein. More preferably in the case of cytoplasmic fusion proteins, the method includes releasing the fusion protein from the host cell by applying osmotic shock or freeze/thaw treatments to the cell. In this case the fusion protein is selectively released from the interior of the cell via the zones of adhesion that exist between the ~~~J~~~
inner and outer membranes of E. cola. The fusion protein is then purified by conventional means.
In another embodiment of the method, if a secretory leader is employed in the fusion protein construct, the fusion protein 5 can be recovered from a periplasmic extract or from the cell culture medium.
An additional step in both of these methods is cleavage of the desired protein from the thioredoxin-like protein by conventional means.
Other aspects and advantages of the present invention will be apparent upon consideration of the following detailed description of preferred embodiments thereof.
Summary of the Drawings Fig. 1 illustrates the DNA sequence of the expression plasmid pALTRXA/EK/ILl7s Pro-581 and the amino acid sequence for the fusion protein therein, described in Example 1.
Fig. 2 illustrates the DNA sequence and amino acid sequence of the macrophage inhibitory protein-1a (MIP-1ec) protein used in the construction of a thioredoxin fusion protein described in Example 3.
Fig. 3 illustrates the DNA sequence and amino acid sequence of the bone morphogenetic protein-2 (BMP-2) protein used in the construction of a thioredoxin fusion protein described in Example 4.
Fig. 4 is a schematic drawing illustrating the insertion of an enterokinase cleavage site into the active-site loop of E.
coli thioredoxin (trxA) described in Example 5.
Fig. 5 is a schematic drawing illustrating random peptide insertions into the active-site loop of E. coli thioredoxin (trxA) described in Example 5.
Fig. 6 illustrates the DNA sequence and amino acid sequence of the human interleukin-6 (IL-6) protein used in the construction of a thioredoxin fusion protein described in Example 6.

WO 92/13955 PCT/US92/0t1944 Fig. 7 illustrates the DNA sequence and amino acid sequence of the M-CSF protein used in the construction of a thioredoxin fusion protein described in Example 7.
Detailed Description of the Invention This invention permits the production of large amounts of heterologous peptides or proteins in a stable, soluble form in certain host cells that normally express limited amounts of such peptides or proteins. It enables release of the fusion protein from the production calls without the necessity of lysing the cells, thereby streamlining the purification process. Also, by using a small peptide insert in an internal region of the thioredoxin-like sequence (e.g. the active site loop of thioredoxin) the invention provides a ready cleavage site, accessible on the surface of the molecule. The fusion proteins of this invention also permit the desired peptide or protein to achieve its desired conformation.
According to the present invention, the DNA sequence encoding a heterologous peptide or protein selected for 2o expression in a recombinant system is fused to a thioredoxin-like DNA sequence for expression in the host cell. A thioredoxin-like DNA sequence is defined herein as a DNA sequence encoding a protein or fragment of a protein characterized by an amino acid sequence having at least 18% homology with the amino acid sequence of E. cola thioredoxin over an amino acid sequence length of 80 amino acids. Alternatively, a thioredoxin DNA
sequence is defined as a DNA sequence encoding a protein or fragment of a protein characterized by a crystalline structure substantially similar to that of human or E. coli thioredoxin.
The DNA sequence of glutaredoxin is one such sequence. The amino acid sequence of E. coli thioredoxin is described in H. Eklund et al, EMBO J. 3:1443-1449 (1984). The three-dimensional structure of E. coli -thioredoxin is depicted in Fig. 2 of A. Holmgren, J.
Biol. Chem. 264:13963-13966 (1989). Fig. 1 below nucleotides 2242-2568 contains a DNA sequence encoding the E. coli WO 92/13955 ~ ~ PCf/US92/00944 thioredoxin protein [Lim et al, J. Bacteriol. 163:311-316 (1985)].
As the primary example of a thioredoxin-like protein useful in this invention, E. coli thioredoxin has the following characteristics. E. cola thioredoxin is a small protein, only 11.7 kD, and can be expressed to high levels (>10~, corresponding to a concentration of 15 uM if calls are lysed at 10 A550~m1).
The small size and capacity for high expression of the protein contributes to a high intracellular concentration. E. coli ~thioredoxin is further characterized by a very stable, tight structure which can minimize the effects on overall structural stability caused by fusion to the desired peptide or proteins.
The three dimensional structure of E. cola thioredoxin is known. It contains several surface loops, including a unique active site loop between residues Cys33 and Cys36 which protrudes from the body of the protein. This active site loop is an identifiable, accessible surface loop region and is not involved 2o in any interactions with the rest of the protein that contribute to overall structural stability. It is therefore a good candidate as a site !or peptide insertions. Both the amino- and carboxyl-termini of E. cola thioredoxin are on the surface of the protein, and are readily accessible for fusions.
E-c"q~ thioredoxin is also stable to proteases. Thus, E.
coli thioredoxin may be desirable for use in E. coli expression systems, because as an E. cola protein it is characterized by ' stability to E. cola proteases. F. coli thioredoxin is also stable to heat up to 80'C and to low pH. Other thioredoxin-like proteins encoded by thioredoxin-like DNA sequences useful in this invention may share the homologous amino acid sequences, and similar physical and structural characteristics. Thus, DNA
sequences encoding other thioredoxin-like proteins may be used in place of E. cola thioredoxin according to this invention. For example, the DNA sequence encoding other species' thioredoxin, D.

e.g., human thioredoxin, may be employed in the compositions and methods of this invention. Both the primary sequence and computer-predicted secondary structures of human and E. coli thioredoxins are very similar. Human thioredoxin also carries the same active site loop as is found in the E, cola protein.
Insertions into the human thioredoxin active site loop and on the amino and carboxyl termini may be as well tolerated as-those in E. cola thioredoxin.
Other thioredvxin-like sequences which may be employed in this invention include all or portions of the proteins . glutaredoxin and various species' homologs thereof [A. Holmgren, cited above]. Although E. cola glutaredoxin and E. cola thioredoxin share less than 20~ amino acid homology, the two proteins do have conformational and functional similarities .
[Eklund et al, EMBO J. 3:1443-1449 (1984)].
All or a portion of the DNA sequence encoding protein disulfide isomerase (PDI) and various species' homologs thereof [J. E. Edman at al, Nature 317:267-270 (1985)] may also be employed as a thioredoxin-like DNA sequence, since a repeated domain of PDI shares >18~ homology with E. coli thioredoxin.
Similarly the DNA sequence encoding phosphoinositide-specific phospholipase C (PI-PhC), fragments thereof and various species' homologs thereof [C. F. Bennett et al, Nature 334:268-270 (1988)] may also be employed in the present invention as~a thioredoxin-like sequence based on the amino acid sequence homology with E. cola thioredoxin. All ar a portion of the DNA
sequence encoding an endoplasmic reticulum protein, such as Erp72, or various species homologs thereaf are also included as thioredoxin-like DNA sequences for the purposes of this invention -[R. A. Mazzarella et al, J. Biol. Chem. 265:1094-1101 (1990)]
based on amino acid sequence homology. Another thioredoxin-like sequence is a DNA sequence which encodes all or a portion of an D' WO 92/13955 ~ ~ PCT/US92/00944 adult T-cell leukemia-derived factor (ADF) or other species homologs thereof [N. Wakasugi et al, Prac. Natl. Aced. Sci UsA
x:8282-8286 (1990)] based on amino acid sequence homology to E.
coli thioredoxin.
It is expected from the definition of thioredoxin-like DNA
sequence used above that other sequences not specifically identified above, or perhaps not yet identified or published, may be useful as thioredoxin-like sequences ;based on their amino acid sequence similarities to E. coli thioredoxin and characteristic crystalline structural similarities to Es cola thioredoxin and the other thioredoxin-like proteins. Based on the above description, one of skill in the art should be able to select and identify, or, if desired, modify, a thioredoxin-like DNA sequence for use in this invention without resort to undue experimentation. For example, simple point mutations made to portions of native thioredoxin or native thioredoxin-like sequences which do not effect the structure of the resulting molecule are alternative thiorsdoxin-like sequences, as are allelic variants of native thiorsdoxin or native thioredoxin-like sequences.
DNA sequences which hybridize to the sequence for E. coli thioredoxin or its structural homologs under either stringent or relaxed.hybridization also encode thioredoxin-like. proteins for use in this invention. Stringent hybridization is defined herein ' as hybridization at 4XSSC at 65~C, followed by a washing in O.1XSSC at 65~C for an hour. Alternatively stringent hybridization is defined as hybridization in 50% formamide, 4XSSC
at 42'C. Non-stringent,hybridization is defined herein as hybridizing in at are 4XSSC at 50~C, or hybridization with 30-40%
formamide at 42'C. The use of all such thioredoxin-like sequences are believed to be encompassed in this invention.
Construction of a fusion sequence of the present invention, sDl ~~~J
which comprises the DNA sequence of a selected peptide or protein and the DNA sequence of a thioredoxin-like sequence, employs conventional genetic engineering techniques [see, Sambrook et al, Molecular Cloning. A Laboratory Mamual., Cold Spring Harbor 5 Laboratory, Cold Spring Harbor, New York (1989)]. Fusion sequences may be prepared in a number of different ways. For example, the selected heterologous protein may be fused to the amino terminus of the thioredoxin-like malecule. Alternatively, the selected protein sequence may be fused to the carboxyl 10 terminus of the thioredoxin-like molecule. Small peptide sequences could also be fused to either of the above-mentioned positions of the thioredoxin-like sequence to produce them in a structurally unconstrained manner.
This fusion of a desired heterologous peptide or protein to the thioredoxin-like protein increases the stability of the peptide or protein. At either the amino or carboxyl terminus, the desired heterologous peptide~or protein is fused in such a manner that the fusion does not destabilize the native structure of either protein. Additionally, fusion to the soluble thioredoxin-like protein improves the solubility of the selected heterologous peptide or protein.
It may be preferred for a variety of reasons that peptides be fused within the active site loop of t:he thioredoxin-like molecule. The face of thioredoxin surrounding the active site loop has evolved, in keeping with the protein's major function as a nonspecific protein disulfide oxido-reductase, to be able to interact with a wide variety of protein surfaces. The active site loop region is found between segments of strong secondary structure and offers many advantages for peptide fusions. A
small peptide inserted into the active-site loop of a thioredoxin-like protein is present in a region of the protein which is not involved in maintaining tertiary structure.
Therefore the structure of such a fusion protein should be stable. Previous work has shown that E. cola thioredoxin can be cleaved into two fragments at a position close to the active site ~~~'~~643 loop, and yet the tertiary interactions stabilizing the protein remain.
The active site loop of E. coli thioredoxin has the sequence NH2...Cys33-Gly-Pro-Cys36...COOH. Fusing a selected peptide with a thioredoxin-like protein in the active loop portion of the protein constrains the peptide at both ends, reducing the degrees of conformational freedom of the peptide, and consequently reducing the number of alternative structures taken by the peptide. The inserted peptide is bound at each end by cysteine l0 residues, which may form a disulfide linkage to each other as they do in native thioredoxin and further limit the conformational freedom of the inserted peptide.
Moreover, this invention places the peptide on the surface of the thioredoxin-like protein. Thus the invention provides a distinct advantage for use of the peptides in screening for bioactive peptide conformations and other assays by presenting peptides inserted in the active site loop in this structural context.
Additionally the fusion of a peptide into the loop protects 2o it from the actions of . o amino- and carboxyl-peptidases.
Further a restriction endonuclease cleavage site RsrII already exists in the portion of the E. cola thioredoxin DNA sequence encoding the loop region at precisely the correct position for a peptide fusion [see Figure 4]. RsrII recognizes the DNA sequence CGG(A/T)CCG leaving a three nucleotide lang 5~-protruding sticky end. DNA bearing the complementary sticky ends will therefore insert at this site in just one orientation.
A fusion sequence of a thioredoxin-like sequence and a desired protein or peptide sequence according to this invention may optionally contain a linker peptide inserted between the thioredoxin-like sequence and the selected heterologous peptide or protein. This linker sequence may encode, if desired, a polypeptide which is selectably cleavable or digestible by conventional chemical or enzymatic methods. For example, the selected cleavage site may be an enzymatic cleavage site.

p c~

Examples of enzymatic cleavage sites include sites for cleavage by a proteolytic enzyme, such as enterokinase, Factor Xa, trypsin, collagenase, and thrombin. Alternatively, the cleavage site in the linker may be a site capable of being cleaved upon exposure to a selected chemical, e.g., cyanogen bromide, hydroxylamine, or low pH.
Cleavage at the selected cleavage site enables separation of the heterologous protein or peptide from the thioredoxin fusion protein to yield the mature heterologous peptide or protein. The mature peptide or protein may then be obtained in purified form, free from any polypeptide fragment of. the thioredoxin-like protein to which it was previously linked. The cleavage site, if inserted into a linker useful in the fusion sequences of this invention, does not limit this invention. Any desired cleavage site, of which many are known in the art, may be used for this purpose.
The optional linker sequence of a fusion sequence of the present invention may serve a purpose other than the provision of a cleavage site. The linker may also be a simple amino acid sequence of a sufficient length to prevent any steric hindrance between the thioredoxin-like molecule and the selected heterologous peptide or protein.
Whether or not such a linker sequence is necessary will depend upon the structural characteristics of the selected heterologous peptide or protein and whether or not the resulting fusion protein is useful without cleavage. For example, where the thioredoxin-like sequence is a human sequence, the fusion protein may itself be useful as a therapeutic without cleavage of the selected protein or peptide therefrom. Alternatively, where the mature protein sequence may be naturally cleaved, no linker may be needed.
In one embodiment therefore, the fusion sequence of this invention contains a thioredoxin-like sequence fused directly at its amino or carboxyl terminal end to the sequence of the selected peptide or protein. The resulting fusion protein is thus a soluble cytoplasmic fusion protein. In another embodiment, the fusion sequence further comprises a linker sequence interposed between the thioredoxin-like sequence and the selected peptide or protein sequence. This fusion protein is also produced as a soluble cytoplasmic protein. Similarly, where the selected peptide sequence is inserted into the active site loop region or elsewhere within the thioredoxin-like sequence, a cytoplasmic fusion protein is produced.
The cytoplasmic fusion protein can be purified by conventional means. Preferably, as a novel aspect of the present invention, several thioredoxin fusion proteins of this invention may be purified by exploiting an unusual property of thioredoxin.
The cytoplasm of . 1 is effectively isolated from the external medium by a cell envelope comprising two membranes, inner and outer, separated from each other by a periplasmic space within which lies a rigid peptidoglycan cell wall. The peptidoglycan wall contributes both shape and strength to the cell. At certain locations in the cell envelope there are ~~gaps~~
(called variously Bayer patches, Bayer functions or adhesion sites) in the peptidoglycan wall where the inner and outer membranes appear to meet and perhaps fuse together. See, M. E.
Bayer, J. Bacteriol. 93:1104-1112 (1967) and J~ Gen. Microbiol 5:395-404 (1968). Most of the cellular thioredoxin lies loosely associated with the inner surface of the membrane at these adhesion sites and can be quantitatively expelled from the cell through.these adhesion sites by a sudden osmotic shock or by a simple freeze/thaw procedure. See C. A. Lunn and V. P. Pigiet, J. Biol. Chem 257:11424-11430 (1982) and in ~~Thioredoxin and Glutaredoxin Systems~ Structure and Function~165-176 (1986) ed.
A. Holmgren et al., Raven Press, New York. To a lesser extent some EF-Tu (elongation factor-Tu) can be expelled in the same way [Jacobson et al, Biochemistry 15:2297-2302 (1976)], but, with the exception of the periplasmic contents, the vast majority of E.
coli proteins cannot be released by these treatments.
Although there have been reports of the release by osmotic shock of a limited number of heterologous proteins produced in the cytoplasm of E. coli [Denefle et al, Gene 85:499-510 (1989);
Joseph-Liauzun et al, Gene 86:291-295 (1990): Rosenwasser et al, J. Biol. Chem. 265:13066-13073 (1990)], the ability to be so released is a rare and desirable property not shared by the majority of heterologous proteins. Fusion of a heterologous protein to thioredoxin as described by the present invention not only enhances its expression, solubility and stability ae described above, but may also provide for its release from the cell by osmotic shock or freeze/thaw treatments, greatly simplifying its purification. The thioredoxin portion of the fusion protein in some cases, e.g., with MIP, directs the fusion protein towards the adhesion sites, from where it can be released to the exterior by these treatments.
In another embodiment the present invention may employ another component, that is, a secretory leader sequence, among which many are known in the art, e.g. leader sequences of phoA, MBP, ~-lactamase, operatively linked in frame to the fusion protein of this invention to enable the expression and secretion of the mature fusion protein into the bacterial periplasmic space or culture medium. This leader sequence-may be fused to the amino terminus of the thioredoxin-like molecule when the selected peptide or protein sequence is fused to the carboxyl terminus or to an internal site within the thioredoxin-like sequence. An optional linker could also be present when the peptide or protein is fused at the carboxyl terminus. It is expected that this fusion sequence construct when expressed in an appropriate host cell would be expressed as a secreted fusion protein rather than a cytoplasmic fusion protein. However stability, solubility and high expression should characterize fusion proteins produced using any of these alternative embodiments.
This invention is not limited to any specific type of heterologous peptide or protein. A wide variety of heterologous genes or gene fragments are useful in forming the fusion sequences of the present invention. While the compositions and 2~J~i~43 methods of this invention are most useful for peptides or proteins which are not expressed, expressed in inclusion bodies, or expressed in very small amounts in bacterial and yeast hosts, the heterologous peptides or proteins can include any peptide or 5 protein useful for human or veterinary therapy, diagnostic or research applications in any expression system. For example, hormones, cytokines, growth or inhibitory factors, enzymes, modified or wholly synthetic proteins or peptides can be produced according to this invention in bacterial, yeast, mammalian or 10 other eukaryotic calls and expression systems suitable therefor.
In the examples below illustrating this invention, the proteins expressed by this invention include IL-11, MIP-lec, IL-6, M-CSF, a bone inductive factor called HMP-2, and a variety of small peptides of random sequence. These proteins include 15 examples of proteins which, when expressed without a thioredoxin fusion partner, are unstable in E. cola or are found in inclusion bodies.
A variety of DNA molecules incorporating the above-described fusion sequences may be constructed for expressing the heterologous peptide or protein according to this invention. At a minimum a desirable DNA sequence according to this invention comprises a fusion sequence described above, in association with, and under the control of, an expression control sequence capable of directing the expression of the fusion protein in a desired host cell. For example, where the host cell is an E. cola strain,.the DNA molecule desirably contains a promoter which functions in E. coli, a ribosome binding site, and optionally, a selectable marker gene and an origin of replication if the DNA
molecule is extra chromosomal. Numerous bacterial expression vectors containing these components are known in the art for bacterial expression, and can easily be constructed by standard molecular biology techniques. Similarly known yeast and mammalian cell vectors and vector components may be utilized where the host cell is a yeast cell or a mammalian cell.
The DNA molecules containing the fusion sequences may be c...~'. ~:~

further modified to contain different codons to optimize expression in the selected host cell, as is known in the art.
These DNA molecules may additionally contain multiple copies of the thioredoxin-like DNA sequence, with the heterologoue protein fused to only one of the DNA sequences, or with the heterologous protein fused to all copies of the thioredoxin-like sequence. It may also be possible to integrate a thioredoxin-like/heterologous peptide or protein-encoding fusion sequence into the chromosome of a selected host to either replace or to duplicate a native thioredoxin-like sequence.
Host cells suitable for the present invention are preferably bacterial cells. For example, the various strains of E. cola (e. g., HB101, W3110 and strains used in the following examples) are well-known as host cells in the field of biotechnology. E.
coli strain GI724, used in the following examples, has been deposited with a United States microorganism depository as described in detail below. Various strains of~B. subtilis, Pseudomonas, and other bacteria may also be employed in this method.
Many strains of yeast and other eukaryotic cells known to those skilled in the art may also be useful as host cells for expression of the polypeptides of the present invention.
Similarly known mammalian cells may also be employed in the expression of these fusion proteins.
To produce the fusion protein of this invention, the host cell is.either transformed with, or has integrated into its genome, a DNA molecule comprising a thioredoxin-like DNA sequence fused to the DNA sequence of a selected heterologous peptide or protein, desirably under the control of an expression control sequence capable of directing the expression of a fusion protein.
The host cell is then cultured under known conditions suitable for fusion protein production. If the fusion protein accumulates in the cytoplasm of the cell it may be released by conventional bacterial cell lysis techniques and purified by conventional procedures including selective precipitations, solubilizations and column chromatographic methods. If a secretory leader is incorporated into the fusion molecule substantial purification is achieved when the fusion protein is secreted into the periplasmic space or the growth medium.
Alternatively, for cytoplasmic thiaredoxin fusion proteins, a selective release from the cell may be achieved by osmotic shock or freeze/thaw procedures. Although final purification is still required for most purposes, the initial purity of fusion proteins in preparations resulting from these procedures is superior to that obtained in conventional whole cell lysates, reducing the number of subsequent purification steps required to attain homogeneity. In a typical osmotic shock procedure, the packed cells containing the fusion protein are resuspended on ice in a buffer containing EDTA and having a high osmolarity, usually due to the inclusion of a solute, such as 20% w/v sucrose, in the buffer which cannot readily cross the cytoplasmic membrane.
During a brief incubation on ice the cells plasmolyze as water leaves the cytoplasm down the osmotic gradient. The cells are then switched into a buffer of low osmolarity, and during the osmotic re-equilibration both the contents of the periplasm and proteins localized at the Bayer patches are released to the exterior. A simple centrifugation following this release removes the majority of bacterial cell-derived contaminants from the fusion protein preparation. Alternatively, in a freeze/thaw procedure the packed cells containing the fusion protein are first resuspended in a buffer containing EDTA and are then frozen. Fusion protein release is subsequently achieved by allowing the frozen cell suspension to thaw. The majority of contaminants can be removed as described above by a centrifugation step. The fusion protein is further purified by well-known conventional methods.
These treatments typically release at least 30% of the fusion proteins without lysing the cell cultures. The success of these procedures in releasing significant amounts of a wide variety of thioredoxin fusion proteins is surprising, since such is techniques are not generally successful with a wide range of proteins. The ability of these fusion proteins to be substantially purified by such treatments, which are significantly simpler and less expensive than the purification methods required by other fusion protein systems, may provide the fusion proteins of the invention with a significant advantage over other systems which are used to produce proteins in E. coli.
The resulting fusion protein is stable and soluble, often with the heterologous peptide or protein retaining its bioactivity. The heterologous peptide or protein may optionally be separated from the thioredoxin-like protein by cleavage, as discussed above.
In the specific and illustrative embodiments of the compositions and methods of this invention, the E. cola thioredoxin (trxA) gene has been cloned and placed in an E. coli expression system. An expression plasmid pALtrxA-781 was constructed. This plasmid containing modified IL-il fused to the thioredoxin sequence and called pALtrxA/EK/ILlls Pro-581 is described below in Example 1 and in Fig. 1. A modified version of this plasmid containing a different ribosome binding site was employed in the other examples and is specifically described in Example 3. Other conventional vectors may be employed in this invention. The invention is not limited to the plasmids described in these examples.
Plasmid pALtrxA-781 (without the modified IL-11) directs the accumulation of >10% of the total cell protein as thioredoxin in E. coli host strain GI724. Examples 2 through 6 describe the use of this plasmid to form and express thioredoxin fusion proteins with BMP-2, IL-6 and MIP-lac, which are polypeptides.
As an example of the expression of small peptides inserted into the active-site loop, a derivative of pALtrxA-781 has been constructed in which a 13 amino-acid linker peptide sequence containing a cleavage site for the specific protease enterokinase [Leipnieks and Light, J. Biol. Chem. 254:1077-1083 (1979)) has been fused into the active site loop of thioredoxin. This 2~~J~~~

plasmid (pALtrxA-EK) directs the accumulation of >10% of the total cell protein as the fusion protein. The fusion protein is all soluble, indicating that it has probably adopted a 'native' tertiary structure. It is equally as stable as wild type thioredoxin to prolonged incubations at 80'C, suggesting that the strong tertiary structure of thioredoxin has not been compromised by the insertion into the active site loop. The fusion protein is specifically cleaved by enterokinase, whereas thioredoxin is not, indicating that the peptide inserted into the active site loop is present on the surface of the fusion protein.
As described in more detail in Example 5 below, fusions of small peptides were made into the active site loop of thioredoxin. The inserted peptides were 14 residues long and were of totally random composition to test the ability of the system to deal with hydrophobic, hydrophilic and neutral sequences.
The methods and compositions of this invention permit the production of proteins and peptides useful in research, diagnostic and therapeutic fields. The production of fusion proteins according to this invention has a number of advantages.
As one example, the production of a selected protein by the present invention as a carboxyl-terminal fusion to E. coli thioredoxin, or another thioredoxin-like protein, enables avoidance of translation initiation problems often encountered in the production of eukaryotic proteins in E. coli. Additionally the initiator methionine usually remaining on the amino-terminus of the heterologous protein is not present and does not have to be removed when the heterologous protein is made as a carboxyl terminal thioredoxin fusion.
The production of fusion proteins according to this invention reliably improves solubility of desired heterologous proteins and enhances their stability to proteases in the expression system. This invention also enables high level expression of certain desirable therapeutic proteins, e.g., IL-11, which are otherwise produced at low levels in bacterial host ~~~~ P
y cells.
This invention may also confer heat stability to the fusion protein, especially if the heterologous protein itself is heat stable. Because thioredoxin, and presumably all thioredoxin-like 5 proteins, are heat stable up to 80'C, the present invention may enable the use of a simple heat treatment as an initial effective purification step for some thioredoxin fusion proteins.
In addition to providing high levels of the selected heterologous proteins or peptides upon cleavage from the fusion to protein for therapeutic or other uses, the fusion proteins or fusion peptides of the present invention may themselves be useful as therapeutics. Further the thioredoxin-like fusion proteins may provide a vehicle for the delivery of bioactive peptides. As one example, human thioredoxin would not be antigenic in humans, 15 and therefore a fusion protein of the present invention with human thioredoxin may be useful as a vehicle for delivering to humans the biologically active peptide to which it is fused.
Because human thioredoxin is an intracellular protein, human thioredoxin fusion proteins may be produced in an E. cola 20 intracellular expression system. Thus this invention also provides a method for delivering biologically active peptides or proteins to a patient in the form of a fusion protein with an acceptable thioredoxin-like protein.
The present invention also provides methods and reagents for screening libraries of random peptides for their potential enzyme inhibitory, hormone/growth factor agonist and hormone/growth factor antagonist activity. Also provided are methods and reagents for the mapping of known protein sequences for regions of potential interest, including receptor binding sites, substrate binding sites, phosphorylation/modification sites, protease cleavage sites, and epitopes.
Bacterial colonies expressing thioredoxin-like/random peptide fusion proteins may be screened using radiolabelled proteins such as hormones or growth factors as probes. Positives arising from this type of screen would identify mimics of ~~~J~~~~

receptor binding sites and may lead to the design of compounds with therapeutic uses. Bacterial colonies expressing thioredoxin-like random peptide fusion proteins may also be screened using antibodies raised against native, active hormones or growth factors. Positives arising from this type of screen could be mimics of surface epitopes present on the original antigen. Where such surface epitopes are responsible for receptor binding, the 'positive' fusion proteins would have biological activity.
Additionally, the thioredoxin-like fusion proteins or fusion peptides of this invention may also be employed to develop monoclonal and polyclonal antibodies, or recombinant antibodies or chimeric antibodies, generated by known methods for diagnostic, purification or therapeutic use. Studies of thioredoxin-like molecules indicate a possible B cell/T cell growth factor activity [N. Wakasuki et al, cited above], which may enhance immune response. The fusion proteins or peptides of the present invention may be employed as antigens to elicit desirable antibodies, which themselves may be further manipulated by known techniques into monoclonal or recombinant antibodies.
Alternatively, antibodies elicited to thioredoxin-like sequences may also be useful in the purification of many different thioredoxin fusion proteins.
The following examples illustrate embodiments of the present invention, but are not intended to limit the scope of the disclosure.
~.~iPLE 1 - THIOREDOXIN-IL-11 TOrI~ MOLE~C'L1r F
A thioredoxin-like fusion molecule of the present invention was prepared using E. coli thioredoxin as the thioredoxin-like sequence and recombinant IL-11 as the selected heterologous protein. The DNA and amino acid sequence of IL-11 has been published. See Paul et al, Proc. Natl. Acad. Sci.U.S.A.
87:7512-7516 (1990) and PCT Patent publication W091/0749, published May 30, 1991. IL-11 DNA can be obtained by cloning 4~~
SJ
(~ J

based on its published sequence. The E. coli thioredoxin (trxA) gene was cloned based on its published sequence and employed to construct various related E. cola expression plasmids using standard DNA manipulation techniques, described extensively by Sambrook, Fritsch and Maniatis, Molecular Cloninq. A Laboratorv anus , 2nd edition, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY (1989).
A first expression plasmid pALTRxa-781 was constructed containing the E. coli trxA gene without fusion to another sequence. This plasmid further contained sequences which are described in detail below for the related IL-11 fusion plasmid.
This first plasmid, which directs the accumulation of >10% of the total cell protein as thioredoxin in an E. coli host strain GI724, was further manipulated as described below for the construction of a trxA/IL-11 fusion sequence.
The entire sequence of the related plasmid expression vector, pALtrxA/ER/ILl7e Pro-581, is illustrated in Fig. 1 and contains the following principal features:
Nucleotides 1-2060 contain DNA sequences originating from the plasmid pUC-18 [Norrander et al, Gene 26: 101-106 (1983)]
including sequences containing the gene for ~-lactamase-which confers resistance to the antibiotic ampicillin in host E. cola strains, and a colEl-derived origin of replication. Nucleotides 2061-2221 contain DNA sequences for the major leftward promoter (pL) of bacteriophage 1 [Sanger et al, J. Mol. Biol. 162:729-773 (1982)],_ including three operator sequences, OL1, OL2 and OL3.
The operators are the binding sites for.tcI repressor protein, intracellular levels of which control the amount of transcription initiation from pL. Nucleotides 2222-2241 contain a strong ribosome binding sequence derived from that of gene 10 of bacteriophage T7 [Dunn and Studier J. Mol. Biol. 166:477-535 (1983)].
Nucleotides 2242-2568 contain a DNA sequence encoding the E.
coli thioredoxin protein [Lim et al, J. Bacteriol. 163:311-316 (1985)]. There is no translation termination codon at the end of the thioredoxin coding sequence in this plasmid.
Nucleotides 2569-2583 contain DNA sequence encoding the amino acid sequence for a short, hydrophilic, flexible spacer peptide "--GSGSG--". Nucleotides 2584-2598 provide DNA sequence encoding the amino acid sequence for the cleavage recognition site of enterokinase (EC 3.4.4.8), "--DDDDK--" [Maroux et al, J.
$iol. Chem. 246:5031-5039 (1971)].
Nucleotides 2599-3132 contain DNA sequence encoding the amino acid sequence of a modified form o;f mature human IL-11 [Paul et.al, Proc. Natl. Acad. SCi. USA 87:7512-7516 (1990)], deleted for the N-terminal prolyl-residue normally found in the natural protein. The sequence includes a translation termination codon at the 3'-end of the IL-11 sequence.
Nucleotides 3133-3159 provide a "Linker" DNA sequence containing restriction endonuclease sites. Nucleotides 3160-3232 provide a transcription termination sequence based on that of the . co i asuA gene [Takagi et al, ~lucl. Acids Res. 13:2063-2074 (1985)]. Nucleotides 3233-3632 are DNA sequences derived from pUC-18.
As described in Example 2 below, when cultured under the appropriate conditions in a suitable . o i host strain, this plasmid vector can direct the production of high levels (approximately 10% of the total cellular protein) of a thioredoxin-IL-11 fusion protein. By contrast, when not fused to thioredoxin, IL-11 accumulated to only 0..2% of the total cellular protein when expressed in an analogous host/vector system.

A thioredoxin-IL-11 fusion protein was produced according to the following protocol using the plasmid constructed as described in Example 1. pALtrxA/EK/ILl7d Pro-581 was transformed into the E. coli host strain GI724 (F-, lacIq, lacPLB, ampC::~,cI+) by the procedure of Dagert and Ehrlich~, Gene 6:23 (1979). The untransformed host strain E. coli GI724 was deposited with the American Type Culture Collection, 12301 Parklawn Drive, cs~
4 ~ .J
C.~ ~ 2 4 Rockville, Maryland on January 31, 1991 under ATCC No. 55151 for patent purposes pursuant to applicable laws and regulations.
Transformants were selected on 1.5% w/v agar plates containing IMC medium, which is composed of M9 medium [Miller, "Experiments in Molecular Genetics", Cold Spring Harbor Laboratory, New York (1972)] supplemented with 0.5% w/v glucose, 0.2% w/v casamino acids and 100 ~ g/ml ampicillin.
GI724 contains a copy of the wild-type.lcI repressor gene stably integrated into the chromosome at the amnC locus, where it has been placed under the transcriptional control of Salmonella typhimurium ~p promoter/operator sequences. In GI724, 1 cI
protein is made only during growth in tryptophan-free media, such as minimal media or a minimal medium supplemented with casamino acids such as IMC, described above. Addition of tryptophan to a culture of GI724 will repress the ~p promoter and turn off synthesis of ~,cl, gradually causing the induction of transcription from pL promoters if they are present in the cell.
GI724 transformed with pALtrxA/ER/IL1~ Pro-581 was grown at 37'C to an A550 of 0.5 in IMC medium. Tryptophan was added to a final concentration of 100 ~ g/ml and the culture incubated for a further 4 hours. During this time thioredoxin-IL-11 fusion protein accumulated to approximately 10% of the total cell protein.
All of the fusion protein was found to be in the soluble cellular i~raction, and was purified as follows. Cells were lysed in a french pressure cell at 20,000 psi in 50 mM HEPES pH 8.0, 1 mM phenylmethylsulfonyl fluoride. The lysate was clarified by centrifugation at 15,000 x g for 30 minutes and the supernatant loaded onto a QAE-Toyopearl column. The flow-through fractions were discarded and the fusion protein eluted with 50 mM HEPES pH
8.0, 100 mM NaCl. The eluate was adjusted to 2M NaCl and loaded onto a column of phenyl-Toyopearl. The flow-through fractions were again discarded and the fusion protein eluted with 50 mM
HEPES pH 8.0, 0.5 M NaCl.
The fusion protein was then dialyzed against 25 mM HEPES pH

~~~J~~J
8.0 and was >80% pure at this stage. By T1165 bioassay [Paul et al, cited above] the purified thioredoxin-IL-11 protein exhibited an activity of 8x105U/mg. Thie value agrees closely on a molar basis with the activity of 2x106U/mg found for COS cell-derived 5 IL-11 purified to homogeneity and measured for activity in the same assay. One milligram of the fusion protein was then cleaved at 37'C for 20 hours with 1000 units of bovine enterokinase [Leipnieks and Light, ~. Biol. Chem. 25:1677-1683 (1979)] in 1 ml lOmM Tris-C1 (pH 8.0)/lOmM CaCl2. IL-11 was recovered from 10 the reaction products by passing them over a QAE-Toyopearl column in 25 mM HEPES pH 8.0, where homogeneous IL-11 was found in the flow-through fractions. Uncleaved fusion protein, thioredoxin and enterokinase remained bound on the column.
The homogeneous IL-11 prepared in this manner had a 15 bioactivity in the T1165 assay of 2.5x106 U/mg. Its physical and chemical properties ware determined as follows:
(1) Molecular Weiq~
The molecular weight of the IL-11 was found to be about 21 20 kD as measured by 10% SDS-PAGE under reducing conditions (tricine system) in accordance with the methods of Schagger, et al., Anal $iochem. 166:368-379.(1987). The compound ran as a single band.
(2) ~ndotoxin Content 25 The endotoxin content of the IL-11 was found to be less than 0.1 nanogram per milligram IL-11 in the LAL (Limulus amebocyte lysate, Pyrotel, available from Associates of Cape Cod, Inc., Woods Hole, Massachusetts, U.S.A.) assay, conducted in accordance with the manufacturer s instructions.
(3) Isoelectric Point The theoretical isoelectric point of IL-11 is pH 11.70. As measured by polyacrylamide gel isoelectric focusing using an LKB
Ampholine PAGplate with a pH range from 3.5 to 9.5, the IL-11 ran at greater than 9.5. An exact measurement could not be taken ~\~J
r ~;asJ

because IL-11 is too basic a protein for the reliable gels available.
(4) Fluorescence Absorption Spectrum Fluorescence absorption spectrum of the IL-11, as measured on a 0.1% aqueous solution in a 1 cm quartz cell showed an emission maximum at 335-337 nm.
(5) W Absorbtion W absorption of the IL-11 on a 0.1% aqueous solution in a 1 . cm quartz cell showed an absorbance maximum at 278-280 nm.
(6) Amino Acid Composition The theoretical amino acid composition for IL-11, based on its amino acid sequence is as follow:

Amino Acid Number ole Ala 20 11.3 Asp Acid 11 6.22 Cysteine 0 Glu 3 1.70 Phe 1 0.57 Gly 14 7.91 His 4 2.26 Ile 2 1.13 Lys 3 1.70 Leu 41 23.16 Met 2 1.13 Asn 1 0.57 Pro 21 11.86 Gln 7 3.96 Arg 18 10.17 Ser 11 6.22 Thr 9 5.09 Val 5 2.83 Trp 3 1.70 Tyr 1 0.57 A sample of homogenous IL-11 was subjected to vapor phase hydrolysis as follows:
6 N HCl and 2 N Phenol reagent were added to hydrolysis vessel in which tubes containing 45 ~1 of 1:10 diluted (w/H20) ~~~~~~-~3 IL-11, concentrated to dryness are inserted. Samples were sealed under vacuum and hydrolyzed for 36 hours at 110°C. After the hydrolysis, samples were dried and resuspended in 500 ~l Na-S
sample dilution buffer. Amino acid analysis was performed on a Beclanan 7300 automated amino acid analyzer. A cation exchange column was used for separation of amino acids following post column derivatization with ninhydrin. Primary amino acids were detected at 570 nm and secondary amino acids were detected at 440 nm. Eight point calibration curves were constructed for each of l0 the amino acids.
Because certain amino acids are typically not recovered, results for only 5 amino acids are given below. Since the hydrolysis was done without desalting the protein, 100% recovery was achieved for most of the amino acids.
The relative recovery of each individual amino acid residue per molecule of recombinant IL-11 was determined by normalizing GLX ~ 10 (the predicted number of glutamine and glutamic acid residue in IL-11 based on cDNA sequence). The value obtained for the recovery of GLX in picomoles was divided by 10 to obtain the GLX quotient. Dividing the value obtained for the recovery in picomoles of each amino acid by the GLX quotient for that sample gives a number that represents the relative recovery of each amino acid in the sample, normalized to the quantitative recovery of GLX residues. The correlation coefficient comparing the expected versus the average number of residues of each amino acid observed is greater than 0.985, indicating that the number of residues observed for each amino acid is in good agreement with that predicted sequence.
0 Amino 1 No. of Residues 2 No. of Residues 3 Correlation Acids Calculated Expected Coefficient 1 Asp 12.78 12 2 Glu 10.00 10 3 Gly 12.80 14 0.9852 Arg 16.10 18 5 Pro 18.40 2 :L

(7) Amino Terminus Sequencing IL-11 (buffered in 95% acetonitrile TFA) was sequenced using an ABI 471A protein sequences (ABI, Inc.) in accordance with the manufacturer's instructions. Amino terminus sequencing confirmed that the thioredoxin fusion protein produced IL-11 contained the correct IL-11 amino acid sequence and only one amino terminus observed.

(8) Peptide Mapping to The IL-11 was cleaved with Endoproteinase Asp-N (Hoehringer Mannheim) (1:500 ratio of Asp-N to IL-11) in 10 mM Tris, pH 8, 1 M urea and 2 mM 4-aminobenzamidine dihydrochloride (PAHA), at 37°C for 4 hours. The sample was then run on HPLC on a C4 Vydac column using an A buffer of 50 mM NaHP04, pH 4.3, in dH20, a B
buffer of 100% isopropanol with a gradient at 1 ml/min from 100%A
to 25%A and 75%B (changing 1%/minute). The eluted peptide fragments were then sequenced using an AHI 471A protein sequences (AHI, Inc.) in accordance with the manufacturer's instructions.
Peptide mapped confirmed the IL-11 produced from the thioredoxin 2o fusion protein contained the proper IL-11 N-terminal and C-terminal sequences.

(9) Solubility IL-11 protein was tested for solubility in the substances below with the following results:
Water very soluble Ethyl Alcohol very soluble Acetone very soluble 1M sodium chloride very soluble 10% sucrose very soluble (10) Sugar Composition and Protein,LPolysaccharide Content in %
The absence of sugar moieties attached to the polypeptide backbone of the IL-11 protein is indicated by its amino acid sequence, which contains none of the typical sugar attachment sites.

WO 92/13955 ~ ~ PCT/US92/00944 209~6r~3 Human macrophage inflammatory protein sec (MIP-7a ) was expressed at high levels in E. coli as a thioredoxin fusion protein using an expression vector similar to pALtrxA/EK/
IL11~ Pro-581 described in Example 1 above but modified in the following manner to replace the ribosome binding site of bacteriophage T7 with that of 1 CII. In the plasmid of-Example l, nucleotides 2222 and 2241 were removed by conventional means.
Inserted in place of those nucleotides was a sequence of nucleotides formed by nucleotides 35566 to 35472 and 38137 to 38361 from bacteriophage lambda as described in Sanger et al (1982) cited above.
To express a thioredoxin-MIP-la fusion the DNA sequence in the thusly-modified pALtrxA/EK/ILl7e Pro-581 encoding human IL-11 (nucleotides 2599-3132) is replaced by the 213 nucleotide DNA sequence shown in Fig. 2 encoding full-length, mature human MIP-7e~ [Nakao et al, MQI. Cell. Biol 10:3646-3658 (1990)].
The host strain and expression protocol used for the production of thioredoxin-MIP-7s fusion protein are as described in Example 1. As was seen with the thioredoxin-IL-li fusion protein, all of the thioredoxin-MIP-1ac fusion protein was found in the soluble cellular fraction, representing up to 20% of the total protein.
Cells ware lysed as in Example 1 to give a protein concentration in the crude lysate of l0 mg/ml. This lysate was then heated at 80~C for 10 min to precipitate the majority of ' contaminating E, coli proteins and was clarified by centrifugation at 130,000 x g for 60 minutes. The pellet was discarded and the supernatant loaded onta a Mono Q column. The fusion protein eluted at approximately 0.5 M NaCl from this column and was >80% pure at this stage. After dialysis to remove salt the fusion protein could be cleaved by an enterokinase treatment as described in Example 1 to release MIP-lcc.

~~~J
i,, Human Bone Morphogenetic Protein 2 (HMP-2) was expressed at high levels in E. cola as a thioredoxin fusion protein using the modified expression vector described in Example 3. The DNA
5 sequence encoding human IL-11 in the modified pALtrxA/EK/ILlle Pro-581 (nucleotides 2599-3132) is replaced by the 345 nucleotide DNA sequence shown in Fig. 3 encoding full-length, mature human BMP-2 [Wozney et al, Science 242:1528-1534 (1988)].
10 In this case the thioredoxin-BMP-2 fusion protein appeared in the insoluble cellular fraction when strain GI724 containing the expression vector was grown in medium containing tryptophan at 37'C. However, when the temperature of the growth medium was lowered to 20'C the fusion protein was found in the soluble 15 cellular fraction.
APL 5 - THIO DOXIN-S nTT pEpTIDE FUSION MOLECU7F_c Native ~ coli thioredoxin was expressed at high levels in E.E. coli using strain GI724 containing the same plasmid expression 20 vector described in Example 3 deleted for nucleotides 2569-3129, and employing the growth and induction protocol outlined in Example 1. Under these conditions thioredoxin accumulated to approximately 10% of the total protein, all of it in the soluble cellular fraction.
25 Fig. 4 illustrates insertion of 13 amino acid residues encoding an enterokinase cleavage site into the active site loop of thioredoxin, between residues G34 and P35 of the thioredoxin protein sequence. The fusion protein containing this internal enterokinase site was expressed at levels equivalent to native 30 thioredoxin, and was cleaved with an enterokinase treatment as outlined in Example 1 above. The fusion protein was found to be as stable as native thioredoxin to heat treatments, being resistant to a 10 minute incubation at 80°C as described in Example 4.
Below are listed twelve additional peptide insertions which ~~~J~~J

were also made into the active site loop of thioredoxin between G34 and P35. The sequences are each 14 amino acid residues in length and are random in composition. Each of the thioredoxin fusion proteins containing these random insertions were made at levels comparable to native thioredoxin. All of them were found in the soluble cellular fraction. These peptides include the following sequences:
Pro-Leu-Gln-Arg-Ile-Pro-Pro-Gln-Ala-Leu-Arg-Val-Glu-Gly, Pro-Arg-Asp-Cys-Val-Gln-Arg-Gly-Las-Ser-Leu-Ser-Leu-Gly, Pro-Met-Arg-His-Asp-Val-Arg-Cys-Val-Leu-His-Gly-Thr-Gly, Pro-Gly-Val-Arg-Leu-Pro-Ile-Cys-Tyr-Asp-Asp-Ile-Arg-Gly, Pro-Lys-Phe-Ser-Asp-Gly-Ala-Gln-Gly-Leu-Gly-Ala-Val-Gly, Pro-Pro-Ser-Leu-Val-Gln-Asp-Asp-Ser-Phe-Glu-Asp-Arg-Gly, Pro-Trp-Ile-Asn-Gly-Ala-Thr-Pro-Val-Lys-Ser-Ser-Ser-Gly, Pro-Ala-His-Arg-Phe-Arg-Gly-Gly-Ser-Pro-Ala-Ile-Phe-Gly, Pro-Ile-Met=Gly-Ala-Ser-His-Gly-Glu-Arg-Gly-Pro-Glu-Gly, Pro-Asp-Ser-Leu-Arg-Arg-Arg-Glu-Gly-Phe-Gly-Leu-Leu-Gly, Pro-Ser-Glu-Tyr-pro-Gly-Leu-Ala-Thr-Gly-His-His-Val-Gly, and Pro-Leu-Gly-Val-Leu-Gly-Ser-Ile-Trp-Leu-Glu-Arg-Gln-Gly.
The inserted sequences contained examples that were both hydrophobic and hydrophilic, and examples that contained cysteine residues. It appears that the active-site loop of thioredoxin can tolerate a wide variety of peptide insertions resulting in soluble fusion proteins. Standard procedures can be used to purify these loop ~~inserts".

Human.interleukin-6 (IL-6) was expressed at high levels in E. coli as a thioredoxin fusion protein using an expression vector similar to modified pALtrxA/EK/ILl7o Pro-581 described in Example 3 above. To express a thioredoxin-IL-6 fusion the DNA
sequence in modified pALtrxA/EK/ILl~n Pro-581 encoding human~IL-11 (nucleotides 2599-3132) is replaced by the 561 nucleotide DNA
sequence shown in Figure 6 encoding full-length, mature human IL-6 [Hirano et al, Nature 324:73-76 (1986)x. The host strain and ~~':a~

expression protocol used for the production of thioredoxin-IL-6 fusion protein are as described in Example 1.
When the fusion protein was synthesized at 37°C, approximately 50% of it was found in the "inclusion body" or insoluble fraction. However all of the thioredoxin-IL-6 fusion protein, representing up to 10% of the total cellular protein, was found in the soluble fraction when the temperature of synthesis was lowered to 25'C.

Human Macrophage Colony Stimulating Factor (M-CSF) was expressed at high levels in E. cola as a thioredoxin fusion protein using the modified expression vector similar to pALtrxA/EK/ILlle Pro-581 described in Example 3 above.
The DNA sequence encoding human IL-11 in modified pALtrxA/EK/ILils Pro-581 (nucleotides 2599-3135) is replaced by the 669 nucleotide DNA sequence shown in Fig. 7 encoding the first 223 amino acids of mature human M-CSF'~ [G. G. Wong et al, Science 235:1504-1508 (1987)]. The host strain and expression 2o protocol used for the production of thioredoxin-M-CSF fusion protein was as described in Example 2 above.
As was seen with the thioredoxin-IL-11 fusion protein, all of the thioredoxin-M-CSF fusion protein was. found in the soluble cellular fraction, representing up to 10% of the total protein.
EXAMPLE_8 - RELEASE OF FUSION PROTEIN VIA OSMOTIC SHOCK OR
FREEZE/THAW
To determine whether the fusions of heterologous proteins to thioredoxin according to this invention enable targeting to the host cell's adhesion sites and permit the release of the fusion proteins from the cell, the cells were exposed to simple osmotic shock and freeze/thaw procedures.
Cells overproducing wild-type E. cola thioredoxin, human thioredoxin, the E. coli thioredoxin-MIPIa fusion or the E. coli thioredoxin-IL-11 fusion were used in the following procedures.

~~33~43 For an osmotic shock treatment, cells were resuspended at 2 A550/ml in 20 mM Tris-C1 pH 8.0/2.5 mM EDTA/20% w/v sucrose and kept cold on ice for 10 minutes. The cells were then pelleted by centrifugation (12,000 xg, 30 seconds) and gently resuspended in the same buffer as above but with sucrose omitted. After an additional 10 minute period on ice, to allow for the osmotic release of proteins, cells were re-pelleted by centrifugation (12,000 xg, 2 minutes) and the supernatant ("shockate") examined for its protein content. Wild-type E. coli thioredoxin and human to thioredoxin were quantitatively released, giving "shockate"
preparations which were >80% pure thioredoxin. More significantly >80% of the thioredoxin-MIP7a and >50% of the thioredoxin-IL-11 fusion proteins were released by this osmotic treatment.
A simple freeze/thaw procedure produced similar results, releasing thioredoxin fusion proteins selectively, while leaving most of the other cellular proteins inside the cell. A typical freeze/thaw procedure entails resuspending cells at 2 A550/ml in mM Tris-C1 pH 8.0/2.5 mM EDTA and quickly freezing the 2o suspension in dry ice or liquid nitrogen. The frozen suspension is then allowed to slowly thaw before spinning out the cells (12,000 xg, 2 minutes) and examining the supernatant for protein.
Although the resultant "shockate" may_require additional purification, the initial "shockate" is characterized by the absence of nucleic acid contaminants. Compared to an initial lysate,,the purity of the "shockate" is significantly better, and does not require the difficult removal of DNA from bacterial lysates.
Thus, this release step can be substituted for the lysis step of Example 2. The supernatant obtained after centrifugation is then further purified in tc:e manner disclosed in that Example.
Numerous modifications and variations of the present invention are included in the above-identified specification and are expected to be obvious to one of skill in the art. Such modifications and alterations to the compositions and processes WO 92/13955 PCf/US92/00944 ~k~t"~ ~) of the present invention are believed to be encompassed in the scope of the claims appended hereto.

Claims (41)

WHAT IS CLAIMED IS:

1. A DNA sequence encoding a fusion protein for expression in a prokaryotic expression system, said sequence comprising DNA encoding a thioredoxin-like protein fused to a DNA sequence encoding a non-thioredoxin-like protein heterologous to said prokaryotic expression system.

2. A DNA sequence of claim 1 wherein said DNA encoding said thioredoxin-like protein comprises the amino terminus of said fusion protein.

3. A DNA sequence of claim 1 wherein said DNA encoding said thioredoxin-like protein comprises the carboxyl terminus of said fusion protein.

4. A DNA sequence of claim 1, 2 or 3 wherein said DNA encoding said thioredoxin-like protein is selected from the group consisting of E. coli thioredoxin and human thioredoxin.

5. A DNA sequence of claim 1, 2 or 3 wherein said DNA encoding said non-thioredoxin-like protein is selected from the group consisting of IL-11, IL-6, Macrophage Inhibitory Protein 1.alpha. and Bone Morphogenic Protein 2.

6. A DNA sequence of claim 1, 2 or 3 additionally comprising a linker DNA sequence fused between said DNA encoding said thioredoxin-like protein and said DNA encoding said non-thioredoxin-like protein.

7. A plasmid DNA molecule comprising a DNA sequence of claims 1-6, said sequence being under the control of an expression control sequence capable of directing the expression of a fusion protein in a selected host cell.

8. An E. coli host cell transformed with, or having integrated into the genome thereof, a plasmid of claim 7.

9. A method of making a non-thioredoxin-like protein comprising (a) culturing in a culture medium a host cell of claim 8;
(b) recovering the fusion protein produced thereby from said culture medium;
(c) cleaving said non-thioredoxin-like protein from said fusion protein and (d) isolating said non-thioredoxin-like protein.

10. The method of claim 9 wherein the thioredoxin-like protein is thioredoxin.

11. A fusion DNA comprising a first DNA encoding a thioredoxin-like protein for expression in a prokaryotic expression system fused to a second DNA encoding a non-thioredoxin-like protein heterologous to said prokaryotic expression system, said fusion DNA capable of encoding a fusion protein.

12. The fusion DNA of claim 11 wherein said first DNA
encodes the amino terminus of said fusion protein.

13. The fusion DNA of claim 11 wherein said first DNA
encodes the carboxyl terminus of said fusion protein.

14. The fusion DNA of claim 11, 12 or 13 wherein said first DNA encodes a thioredoxin-like protein selected from the group consisting of human thioredoxin, glutaredoxin, and the thioredoxin-like domains of protein disulfide isomerase, form-1 phosphoinositide-specific phospholipase C and ERp72.

15. The fusion DNA of claim 14 wherein said first DNA
encodes E. coli thioredoxin.

16. The fusion DNA of claim 11 wherein said second DNA
encodes a protein selected from the group consisting of: IL-11, IL-6, Macrophage Inhibitory Protein 1.alpha. or Bone Morphogenic Protein 2.

17. The fusion DNA of claim 11 further comprising a third DNA encoding a linker peptide fused between said first DNA and said second DNA.

18. The fusion DNA of claim 17 wherein said third DNA
encodes a cleavage site.

19. The fusion DNA of claim 17 wherein said linker further provides for preventing steric hindrance between said thioredoxin-like protein and said heterologous protein.

20. The fusion DNA of claim 17, wherein said third DNA
sequence encodes the amino acid sequence G-S-G-S-G-D-D-D-D-K
(nucleotides 2569-2601 of Figure 1).

21. The fusion DNA of claim 17, comprising thioredoxin-G-S-G-S-G-D-D-D-D-K-des-Pro-IL-11 of Figure 1.

22. A fusion DNA comprising thioredoxin G-S-G-S-G-D-D-D-D-K-des-Pro-IL-11 of Figure 1.

23. A protein encoded by a DNA of claim 22.

24. A fusion DNA comprising a first DNA encoding a thioredoxin-like protein fused to a second DNA encoding a linker peptide.

25. The fusion DNA of claim 24 wherein said first DNA
encodes E. coli thioredoxin.

26. The fusion DNA of claim 24 wherein said first DNA
encodes a protein selected from the group consisting of human thioredoxin, glutaredoxin, and the thioredoxin-like domains of protein disulfide isomerase, form-1 phosphoinositide-specific phospholipase C and ERp72.

27. The fusion DNA of claim 24 wherein said linker peptide encodes a cleavage site.

28. The fusion DNA of claim 11 comprising pALtrxA/EK/ILII.DELTA.
Pro-581 having the nucleotide sequence as depicted in Figure 1.

29. A plasmid DNA encoding a fusion protein for expression in a prokaryotic expression system comprising a first DNA
encoding a thioredoxin-like protein fused to a second DNA
encoding a non-thioredoxin-like peptide or a non-thioredoxin-like protein heterologous to said prokaryotic expression system, said plasmid DNA under the control of an expression control sequence, and comprising a promoter, a ribosome binding site, an origin of replication and an optional selectable marker, said control sequence capable of directing the expression of said fusion protein in a selected host cell.

30. A plasmid DNA comprising a DNA sequence of any one of claims 11 to 22 and 24 to 28, said DNA under the control of a suitable expression control sequence capable of directing the expression of said fusion protein in a selected host cell.

31. A host cell transformed with, or having integrated into the genome thereof, a DNA according to claims 11 to 22 and 24 to 28, said DNA under the control of an expression control sequence capable of directing the expression of a cytoplasmic fusion protein.

32. An E. coli host cell transformed with, or having integrated into the genome thereof, a plasmid according to claims 29 and 30.

33. A fusion protein expressed in a prokaryotic expression system comprising a thioredoxin-like protein fused to a non-thioredoxin-like peptide or a non-thioredoxin-like protein heterologous to said prokaryotic expression system.

34. The fusion protein of claim 33 wherein said thioredoxin-like protein is E. coli thioredoxin.

35. The fusion protein of claim 34, which is a thioredoxin/IL-11 fusion protein.

36. A fusion protein comprising a thioredoxin-like protein fused to a linker peptide providing a cleavage site.

37. The fusion protein of claim 36, wherein said linker peptide further provides for preventing steric hindrance.

38. The fusion protein of claim 37, wherein said linker peptide is -G-S-G-S-G-D-D-D-D-K (amino acids 110-119 of Figure 1).

39. A method of making a non-thioredoxin-like protein comprising:
(a) culturing in a culture medium under suitable conditions a host cell of claim 31;

(b) recovering said fusion protein produced thereby from said culture medium;
(c) cleaving said non-thioredoxin-like protein from said fusion protein; and (d) isolating said non-thioredoxin-like protein.

40. The method according to claim 39 wherein said recovering step comprises treating said transformed and cultured host cell by osmotic shock to release said fusion protein from said host cell.

41. The method according to claim 39 wherein said recovering step comprises treating said transformed and cultured host cell by freezing and thawing to release said fusion protein from said host cell.