WO1992012173A1

WO1992012173A1 - A method for constructing antigens

Info

Publication number: WO1992012173A1
Application number: PCT/US1992/000016
Authority: WO
Inventors: Ira H. Pastan; Michael M. Gottesman; Edward Bruggemann; Vijay K. Chaudhary
Original assignee: The United States Of America, As Represented By The Secretary, U.S. Department Of Commerce
Priority date: 1991-01-04
Filing date: 1992-01-02
Publication date: 1992-07-23
Also published as: AU1239892A

Abstract

The subject invention relates to a method of constructing antigens that are fusions between Pseudomonas exotoxin (PE) and specific regions of human P-glycoprotein. Furthermore, the invention also relates to the antigens themselves and to the uses of such antigens. For example, the antigens may be utilized in order to elicit large quantities of antibody formed during an immune response.

Description

A METHOD FOR CONSTRUCTING ANTIGENS BACKGROUND OF THE INVENTION Technical Field The subject invention relates to a method of constructing antigens that are fusions between

Pseudomonas exotoxin (PE) and specific regions of human P-glycoprotein. Furthermore, the invention also relates to the antigens themselves and to the uses of such antigens. For example, the antigens may be utilized in order to elicit large quantities of antibody formed during an immune response.

BACKGROUND INFORMATION P-glycoprotein is a 170 Kd plasma membrane protein that confers multidrug resistance to mammalian cells (Bradley et al., Biochiπi. Biophvs. Acta. 948:87-128

(1988), Endicott et al., Ann. Re . Biochem. 58:137-71 (1989) and Gottesman et al., J. Biol. Chem. 263: 12163- 66 (1988)). This human protein, encoded by the MDR1 gene, comprises 1280 amino acid residues and approximately 30 Kd of carbohydrate (Chen et al., Cell 47:381-89 (1986)). The amino acid sequence suggests that it has 12 transmembrane and two large cytoplasmic regions that may each bind one molecule of ATP. P- glycoprotein protects cells from a wide variety of cytotoxic drugs, probably by hydrolyzing ATP to transport these drugs out of the cell.

The objective of the present invention was to make a comprehensive panel of specific antibodies against P- glycoprotein. A comprehensive panel of antibodies is a set of individual antibodies, each of which recognizes only a specific region of P-glycoprotein; taken together these antibodies would cover the entire protein. These antibodies would recognize specific regions of P-glycoprotein, and they would not cross react with any other regions of P-glycoprotein. Such antibodies are used to elucidate the structure and function of individual domains of a protein. Several different methods for raising antibodies against specific regions of a eucaryotic protein were considered.

The first requirement for raising a panel of antibodies against specific regions of P-glycoprotein is a source of antigen. That is, it is necessary to make, or otherwise obtain, large quantities of specific fragments of the protein. One way to do this would be to start with eucaryotic cells in tissue culture that produce large quantities of P-glycoprotein. P- glycoprotein would be purified from these cells, then digested into fragments, using either chemical or enzymatic methods. Once the fragments obtained from this approach were separated and purified, they could be used as antigens for immunizations. Antisera produced this way would contain antibodies that specifically recognize the region of P-glycoprotein from which the fragments were produced.

This approach is not practical in the case of P- glycoprotein. First of all, it is extremely expensive and time consuming to grow up the large number of cells required. Secondly, even if the cells could grow up, current methods for the purification of P-glycoprotein are inefficient, and it is not clear that enough protein would remain after purification to continue (Ha ada et al., J. Biol. Che . 263:1454-58 (1988)). Chemical or enzymatic digestions, and subsequent separation and purification of the protein fragments, would result in further reductions of yield. Finally, there is no insurance that the fragments obtained from all possible methods of digestion would be from the desired regions of P-glycoprotein, and identifying the fragments could be a difficult project in itself. In fact, it is highly unlikely that a complete set of fragments of P-glycoprotein could be obtained by this approach, and the set of antibodies obtained would be similarly limited.

Another approach to producing useful antigens would be to make synthetic peptides (Dyson et al., Ann. Rev. Biophys. Biophys. Chem. 12:307-24 (1988) & Walter, J. Immunol. Methods 88:1419-61 (1986)). The cDNA for P-glycoprotein has been sequenced, so the amino acid sequence of the protein is known from deduction (Chen et al., Cell 47:381-89 (1986)). Peptides that correspond to specific regions of P-glycoprotein may be synthesized automatically. Synthetic peptides are always coupled to carrier proteins before injection into animals for the production of antibodies. The carrier proteins and the methods for coupling are widely used, standard, and do not present any special problems (Harlow et al., Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. (1988)). Large yields that are usually more than adequate for immunizations are typically obtained from automatic peptide synthesis. There is no need for substantial purification of the peptides, and there are few limitations on the sequences that can be synthesized. Because the peptide sequence is synthesized, antigens that correspond precisely to the desired regions of P-glycoprotein may be obtained.

There are disadvantages to this approach, however. The length of the synthesized peptide is limited by practical and technical considerations. Synthetic peptides usually do not exceed about 20 to 25 amino acid residues in length. Although immunizations with small peptides would result in antibodies that were specific for small, defined regions of P-glycoprotein, such small peptides frequently do not contain any good epitopes and they are not immunogenic. Furthermore, small peptides usually lack the complex secondary and tertiary structure of native proteins. These higher order structures are often important elements of good epitopes (Dyson et al., Ann. Rev. Biophys. Biophys. Chem. 12:307-24 (1988) & Walter, J. Immunol. Methods 88:1419-61 (1986)). Thus, many peptides must be synthesized for every region of a protein against which antibodies are desired, because many, if not most of these peptides will not illicit a good antibody response. Finally, the present inventors have attempted to use the synthetic peptide approach in the past to raise antibodies against P-glycoprotein. The majority of these peptides did not illicit any positive antisera, and the few that did resulted in antibodies of only low titer or low affinity (Bruggemann et al., J. Biol. Chem. 264:15483-88 (1989) & Richert et al., Biochemistry 27:7606-13 (1988)).

The present inventors have also produced fragments of P-glycoprotein directly in E. coli (Tanaka et al., Biochem. Biophys. Res. Comm. 166:180-86 (1990)). In this approach, fragments of the MDR cDNA are cloned into a bacterial expression vector. The resulting plasmid is transformed into the appropriate strain of E. coli, and the culture is induced to produce the protein fragments. After a substantial amount of the protein has been isolated or purified, it may be used to immunize animals. In the two cases in which this approach was used, good antibodies against P- glycoprotein were obtained.

Unfortunately, when it comes to making a comprehensive panel of antibodies against a eucaryotic protein, this approach has many disadvantages. For example, expression of eucaryotic proteins in bacteria is often poor. Often only small amounts of protein are produced. In addition, eucaryotic proteins are often subject to degradation by bacterial proteases. Even if the initial amount of protein expression is high, proteolytic degradation can result in a large reduction of yield. If initial expression is low, proteolytic degradation may make this approach impossible. To overcome these problems, different expression vectors and host strains are available, but finding the correct ones is largely a process of trial and error (Sambrook et al., Molecular Cloning: A Laboratory Manual. Second Edition, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. (1989)). Additionally, each eucaryotic protein fragment that is produced by this method is in essence a different protein. Consequently, expression and purification of each eucaryotic protein fragment presents a new and different challenge. Success with one does not ensure success with any other. Any attempt to systematically produce a comprehensive set of specific antigens will require substantial time and effort, much more than might be apparent from what is apparently a systematic approach. Finally, the presence or absence of restriction sites in the cDNA of the eucaryotic protein of interest, and the sites available in the expression vector itself, can limit the flexibility of this approach. With traditional cloning techniques, it may simply not be possible to clone precisely the desired regions of the eucaryotic protein. Recent PCR techniques, however, may partially, or completely, overcome this limitation.

As noted above, the present invention relates to fusion proteins. Such proteins are constructed by joining the individual genes for two different proteins and transforming the resulting gene into bacteria. One gene is a bacterial gene that codes for a (usually well-known) bacterial protein. PE is the bacterial gene used in the present invention. The other gene codes for the protein or protein fragment of interest. In the present invention, this is the eucaryotic protein P-glycoprotein. When the gene is expressed, the result is a fusion protein in which the two original proteins are covalently joined through a peptide (amide) bond. After an animal has been immunized with the fusion protein antigen, antibody response may be directed against both the bacterial protein and the eucaryotic protein or protein fragment attached to it. The antisera obtained by this method thus contains many antibodies in addition to antibodies against the desired protein, but these will usually not interfere with the use of the antisera.

Fusion proteins offer several advantages to the scientist who desires to raise antibodies against a eucaryotic protein. (1) If the bacterial protein, to which the eucaryotic protein of interest is fused, is expressed in large amounts, then there is an excellent chance that the fusion protein will also be expressed in large amounts. (2) If the chosen bacterial protein is not quickly degraded by the host bacteria, then there is a high probability that the fusion protein may be recovered intact. (3) The presence of the bacterial protein as part of the fusion protein immediately suggests methods of isolation and purification based on the properties of the bacterial protein. There is a good possibility that the same isolation and purification procedures may be used for the fusion proteins. (4) The size of the eucaryotic protein fragments fused to the bacterial protein is not limited. Protein fragments that are as small as synthetic peptides are possible, as well as fragments that are considerably larger.

These considerations suggest that large amounts of antigen suitable for immunizations may be obtained with this approach. Several systems for producing fusion protein are in common use. The most widely used is β- galactosidase (Carroll et al., "Production and purification of polyclonal antibodies to the foreign segment of β-galactosidase fusion proteins" in Glover, D.M. (ed.), DNA Cloning: A Practical Approach, Vol. Ill, IRL Press, Oxford (1987); Nielsen et al., J^. Immunol. Methods 111:1-9 (1988), & RUther et al. , EMBO J. 2:1791-94 (1983)). Others are TrpE (Dieckmann et al., J. Biol. Chem. 260:1513-20 (1985) & Shyjan et al. , Biochemistry 28:4531-35 (1989)) and protein A from Staphylococcus aureus (Nilsson et al., EMBO J. 4:1075- 80 (1985); Lόwendlar et al., EMBO J. 5:2393-98 (1986); Stahl et al., J. Immunol. Meth. 124:43-42 (1989) & Sjόlander et al., Infect. Immunol. 58:854-59 (1990)). In these other methods, the starting point is a plasmid that contains the gene for the bacterial protein. This gene is usually altered to contain convenient restriction sites in either the 5' or 3' end of the coding sequence, into which the eucaryotic cDNA may be inserted. The gene for the protein or protein fragment of interest must be cloned into these sites. In the method of the present invention, the starting plasmid contains one site at the 3' end of the coding sequence. PCR is used to amplify the gene or gene fragment and to introduce the appropriate restriction sites for cloning. The use of PCR to amplify fragments of the eucaryotic cDNA for cloning makes this technique extremely flexible and powerful. The precise region of interest may be amplified by use of the appropriate PCR primers. Furthermore, any restriction site may be introduced into the fragment of amplified eucaryotic cDNA by designing the site into the same PCR primers.

The fusion proteins of the present invention were created with the bacterial protein Pseudomonas exotoxin (PE) because it offers two substantial advantages over the other fusion protein systems. Firstly, the PE expression vectors produce large amounts of protein in E. coli upon induction with IPTG. It was determined that when MDRl sequences were cloned into the 3' end of the PE gene, the amount of protein produced upon induction does not decrease. The expression vectors consistently produce large amounts of fusion protein that is easily isolated in a form suitable for immunizations. The amount of fusion protein obtained for immunizations from the other systems is less consistent.

The second advantage of using PE arises from the properties of PE itself. PE is a secreted toxin that is normally produced by a pathogenic strain of bacteria. In animals, PE is highly toxic and immunogenic. Thus, it was thought that PE would enhance the immunogenicity of the P-glycoprotein fragments attached to it and increase the chances of obtaining positive antisera against P-glycoprotein.

Protein A may also have the same property (Lόwenadler et al., EMBO J. 5:2393-98 (1986); Terry et al., Arch.

Virol. 104:63-75 (1989) & Sjόlander et al., Infect. Immunol. 58:854-59 (1990)), but β-galactosidase and

TrpE do not. With this method, the present inventors successfully obtained good antibodies against most of the regions of P-glycoprotein that were attempted.

Because this approach is easy and reliable, it may be generally useful to raise antibodies against other eucaryotic proteins that are difficult to prepare in large amounts.

All U.S. patents and publications referred to herein are hereby incorporated by reference. SUMMARY OF THE INVENTION

The present invention relates to a method for constructing antigens which are fusions between

Pseudomonas exotoxin (PE) and specific regions of human

P-glycoprotein. The antigens can be used to immunize an animal, thereby eliciting the production of large quantities of the respective antibody. Thus, the method may be utilized to produce large quantities of antibodies which are difficult to obtain using other methods. In particular, the present invention relates to a method of producing a fusion protein comprising the steps of:

(i) cloning at least one DNA sequence which encodes a protein or a portion thereof, into a vector, said vector comprising a gene which encodes an inactivated form of a bacterial toxin; (ii) transforming a host cell with the resulting vector of step (i) , thereby allowing for expression of said fusion protein.

The DNA sequence which encodes said protein or portion thereof may be, for example, a human drug resistance gene. For instance, the DNA sequence which encodes the protein or portion thereof may be MDR 1. The protein itself may be, for example, human P- glycoprotein. Several parent vectors may be used, for example, pVC4D.

The gene which is present in the vector comprises a DNA sequence which encodes, for example, an inactivated form of Pseudomonas exotoxin. More specifically, the gene may comprise, for example, a DNA sequence having a codon deletion for one amino acid of Pseudomonas exotoxin (PE) . The gene having this codon deletion is referred to as PEΔ553. The amino acid which corresponds to the deleted codon is Glu553.

The fusion protein produced according to the above method is referred to as PE-p-glycoprotein (PEPG) . Another aspect of the invention relates to a recombinant DNA molecule comprising:

(i) a DNA segment which encodes a protein or a portion thereof; (ii) a vector for introducing said DNA segment into host cells, said vector comprising a gene which encodes an inactivated form of a bacterial toxin.

The vector utilized may be, for example, pVC4D, and the protein may be, for instance, the human P- glycoprotein. The gene which is present in the vector comprises a DNA sequence which has a codon deletion for one amino acid of Pseudomonas exotoxin. In such a case, the gene is referred to as PEΔ553. The amino acid which corresponds to the deleted codon is Glu553.

The invention further relates to a host cell stably transformed with the recombinant DNA molecule, referred to above, in a manner allowing expression of a fusion protein encoded by said recombinant DNA molecule. The host cell is a procaryotic cell such as, for example, an E. coli cell. Additionally, the present invention also relates to a reco binantly produced fusion protein consisting of a toxin protein encoded by a parent vector and a eucaryotic protein or fragment thereof. The toxin protein may be, for example, Pseudomonas exotoxin (PE) protein, and said PE protein lacks the amino acid

Gly553. The eucaryotic protein or portion thereof may be, for example, human P-glycoprotein.

The invention also relates to an antibody or antibodies raised against the fusion protein described above. In particular, the antibody is raised against a recombinantly produced fusion protein, said protein consisting of a eucaryotic protein or fragment thereof and a Pseudomonas exotoxin (PE) protein, said PE protein lacking the amino acid Gly553. The antibody may be polyclonal or monoclonal.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 represents the expression vector pVC4D. Base pair 1 is the first T in the EcoRl site GAATTC. The size of the plasmid is 4368 bp. Figure 2 represents a model of P-glycoprotein in the plasma membrane. The regions of P-glycoprotein that were fused to PE are indicated in black. The nucleotide binding sites are circled, and the possible sites for glycosylation are sown as wiggly lines on the first extracellular loop.

Figure 3 represents the expression and purification of fusion proteins. A: Before the culture was induced with IPTG, a small portion of the cells were removed for analysis. A similar portion was removed after induction. The cells were lysed at 100°C in Laem li sample buffer for 5 minutes and run on an 8% SDS polyacrylamide gel (Laemmli, U.K., Nature 227:680- 85 (1970)). 3 μg of protein was applied to each lane. Cells contained plasmid pEB37 which encodes PEPG7; pEB17, which encodes PEPG8; pVC4D, which encodes PEΔ553; or no plasmid. PEΔ553 runs at about 67 Kd, and the fusion proteins PEP67 and PEPG8 run at about 74 Kd. B: Inclusion body samples were heated at 100°C in Laemmli sample buffer for 5 minutes and run on a 10% SDS polyacrylamide gel. 3 μg of protein was applied to each lane. Figure 4 shows an antiserum assay. Preimmune serum (P) and antiserum (A) from each rabbit were used to immunoprecipitate P-glycoprotein from KB-C1 cells that had been labelled with ³H-azidopine. The results were run on a 7% gel. PEPG7A and PEPG8A fusion proteins were injected into the rabbits. For each lane 10⁶ KB-C1 cells were labelled with 1 μCi of ³H-azidopine and immunoprecipitated with 5 μl of serum. Rabbit antiserum 4007, which is used here as a positive control has been previously described (Tanaka et al. , Biochem. Biophvs. Res. Comm. 166:180-86 (1990)). DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method of expressing antigens which are fusions between Pseudomonas exotoxin and specific regions of human P- glycoprotein. These antigens or fusion proteins can be used for immunization in order to produce antibodies against P-glycoprotein. The same method can be utilized to raise antibodies against other eukaryotic proteins. Vector pVC45Df(+)T was initially chosen in order to carry out the present method (Jinno et al. , J. Biol. Chem. 264:15953-59 (1989)). This vector, which was constructed by the present inventors, contains the gene for mature PE from which the codon for Glu553 has been deleted. The Glu553 residue is essential for the catalytic activity of PE, but it apparently affects no other function of the protein (Carroll et al. , J. Biol. Chem. 262:8707-11 (1987); Douglas et al., J. Bacteriol. 169:4967-71 (1987) & Lukac et al., Infect. Immun. 56:3095-98 (1988)). It is important to inactivate the ADP-ribosylation activity of PE because active PE would kill the animal when the final fusion protein is injected. The receptor binding and translocation functions of PE are not affected by this deletion. The purpose was to start with a protein that was as similar as possible to native PE, but would not kill the immunized animal. It was believed that in this way PE was most likely to act as its own adjuvant. The protein produced by this gene is called PEΔ553. In addition to the codon deletion mentioned above, there are other ways in which PE can be rendered inactive. For example, additions to the carboxyl end of PE, or deletions of two or more amino acid residues, often inactivates PE (Chaudhary et al., Proc. Natl. Acad. Sci. 87:308-12 (1990)). In addition, mutations at positions 276 and 279 of PE, in which the arginine residues are changed to glycine or other amino acids, inactivates PE (Jinno et al., J. Biol. Chem. 264:15953- 59 (1989)). Also, deletion of part or all of domain I of PE causes inactivation of the molecule (Hwang et al., Cell 48:129-136 (1987)). The pVC45Df(+)T vector contains, in addition to the PEΔ553 gene described above, the sequence coding for the OmpA single sequence 5' to the PEΔ553 gene. This short peptide sequence will cause the entire protein to be translocated to the periplasm. It was believed that this feature would make it easier to purify the final fusion protein. Two MDR sequences were cloned into this vector. The expressed proteins comprised the OmpA signal sequence at the amino terminus which would be removed upon translocation, followed by PEΔ553, followed by the P-glycoprotein fragment. Unfortunately, the fusion protein thus constructed was degraded by proteolytic enzymes, either after translocation to the periplasm or during the translocation event. Consequently, the present inventors decided to use a vector that did not contain the signal sequence.

Rather than delete the signal sequence from pVC45Df(+)T, pVC4 was chosen which contains the gene for the full-length, mature PE gene (Chaudhary et al., Proc. Natl. Acad. Sci. 85:2939-43 (1988)). No signal sequence precedes the PE gene. The gene is preceded by an ATG initiation codon. To move the Glu553 deletion from pVC45Df(+)T into pVC4, the following procedure was used:

Both pVC4 and pVC45Df(+)T were digested with Xhol and EcoRL. The 417 base pair fragment from PVC45Df(+)T, which includes the Glu553 deletion, was gel purified and then ligated to the 3.9 Kbp fragment from pVC4, which was also gel purified. The final expression vector, pVC4D, contains the initiating ATG codon followed by the coding sequence for the mature PEΔ553 gene. The protein it encodes, PEΔ553, is 613 amino acid residues long.

At the 3 ' end of the PEΔ553 gene is a PpuMI site that is suitable for cloning. There is a second PpuMI site within the PEΔ553 gene, but this site is not recognized by the enzyme if the plasmid is prepared from E. coli strain HB101. Presumably, this second PpuMI site is somehow modified but the nature of the modification is not known. To clone MDRl sequences onto the 3' end of PEΔ553, pVC4D was digested with PpuMI (New England Biolabs) and EcoRI, and the 105 bp fragment was removed by gel purification. Any sequence can be cloned into this vector that contains a PpuMI end and a EcoRI end.

MDRl sequences were prepared by PCR from plasmid pMDR200XS, which was constructed by the present inventors, and contains the full-length human MDRl cDNA (Pastan et al., Proc. Natl. Acad. Sci. 85:4486-90 (1988) & Ueda et al., Proc. Natl. Acad. Sci. 84: 3004- 08 (1987)). The 5' primers all contained the sequence AGGACCTG, which introduced a PpuMI site, flanked on each side by five complete codons of the appropriate MDRl sequence. These 5' primers reconstruct all but the final codon of PEΔ553, Lys613. Thus, in the final PE-P-glycoprotein fusion (PEPG) , PEΔ553 is complete except for the final amino acid residue. This last amino acid residue is not known to be essential for any function of PE. The 3' primers contained the sequence TGATGAATTC which introduces two stop codons and an EcoRI site. This sequence was also flanked on each side by five complete codons of the appropriate MDRl sequence. It was necessary to include a stop codon because the stop codon for PEΔ553 in pVC4D was removed with the 105 bp PpuMI/EcoRI fragment described above. It is a common practice under these circumstances to include two stop codons, if possible, to ensure that translation actually stops. TGA was chosen as a stop codon because the EcoRI site would overlap it. The result is a lObp mismatch during PCR rather than a 12 bp mismatch. This mismatch was kept to the minimum length possible.

PCR was performed with reagents and equipment from Perkin Elmer-Cetus. After PCR, the amplified MDRl fragments were digested with PpuMI and EcoRI, ligated to pVC4D that had already been prepared by PpuMI/EcoRI digestion (described above) , transformed into E. coli strain HB101, and selected on ampicillin. If there was no internal Avail site in the amplified MDRl fragment, AvII (Bethesda Research Laboratories) was used in place of PpuMI to digest the amplified fragments. Avail was more reliable than PpuMI. The constructions were screened by sequencing the MDRl region with Sequenase (United States Biochemicals) to ensure that the correct reading frame was maintained and that no mutations had been introduced by the primers, PCR itself, or the cloning steps. (The recombinant plasmid (i.e., E_j_ coli strain B52) was deposited with the American Type Culture Collection in Rockville, Maryland. Its ATCC designation number is 68508.) The expression vector pVC4D contains a T7 promoter upstream from the PEΔ553 gene. The protein may be expressed by transforming the plasmid into E. coli strain BL21(ΛDE3) which contains the gene for T7 RNA polymerase under control of a lac promoter, and inducing the culture with IPTG. This system expresses proteins very strongly under conditions of induction which is important if substantial amounts of protein are desired. Without induction, very little protein is expressed. Overexpression, or even partial expression, of foreign proteins in E. coli frequently kills the cells. Thus, tight control over expression of foreign proteins is a useful feature.

The plasmids were transformed into E. coli strain BL21(ADE3), a 500 ml culture was grown up, and expression of the fusion proteins was induced with IPTG (Rosenburg et al., Gene 75:323-27 & Studier et al., J. Mol. Biol. 189:113-30 (1986)). Under these conditions the cells produce large amounts of fusion protein, or PEΔ553 itself, in the form of inclusion bodies. Inclusion bodies may be easily isolated, and the protein that comprises them recovered in nearly pure form (Marston, "The purification of eucaryotic polypepetides expressed in Escherichia coli," in Glover, D.M. (ed.), DNA Cloning: A Practical Approach, Vol III., IRL Press, Oxford (1987)). To isolate the inclusion bodies, the induced cells were harvested and treated with Tris, EDTA, and lysozyme to digest the bacterial outer membrane. The cells were then lysed with deoxycholate. DNase was added at this step to digest the bacterial DNA. The suspension was centrifuged, and the pellet was washed several times, once with Triton X-100. The purpose of these washes was to remove various uncharacterized proteins from the inclusion body pellet. At this point, the pellet, which consists of isolated inclusion bodies and is approximately 95% pure, was resuspended in phosphate- buffered saline (PBS) .

The yield from a 500 ml culture is about 50 mg of protein. There are some contaminating bands in the inclusion body preparations that contain fusion proteins that are not present in the control with PEΔ553. It is believed that these contaminants are degradation products from the fusion protein. These contaminants represent only a small fraction of the total protein, and they do not interfere with the immunization of animals. The inclusion body suspension in PBS may be stored frozen at -20°C. The inclusion bodies may be denatured by adding SDS to 2%, with or without β-mercaptoethanol to 5%. Incubation at room temperature for 5 to 10 minutes is usually sufficient for denaturation, but the samples may also be heated to 100^CC for 3 to 5 minutes.

Inclusion body preparations were injected into rabbits either directly as a particulate suspension of inclusion bodies, or after they were solubilized by denaturation. To immunize the rabbits, the fusion proteins were diluted first in PBS and then mixed with one volume of either complete Freund's adjuvant (CFA) or incomplete Freund's adjuvant (IFA) . For each immunization or boost, a total of 1 g protein was used. Injections were done at multiple subcutaneous sites according to the following schedule: Day 1, immunize with CFA; Days 22 and 43, boost with IFA; Day 50 bleed and test serum. Female white New Zealand rabbits were used to raise antiserum.

Sera were screened by attempting to immunoprecipitate P-glycoprotein that had been photoaffi .ni.ty labeled wi.th 3H-azidopme (Bruggemann et al., J. Biol. Chem. 264:15483-88 (1989)). This screening assay was chosen because it is relatively easy and quick. The present inventors were also particularly interested in obtaining antibodies that would be useful for immunoprecipitation. The positive antisera were assayed by Western blot against plasma membrane proteins prepared from multidrug resistant cells. Most gave positive results with P-glycoprotein, although the reactions were generally weaker than in the im unoprecipitations. In the Western blots, background bands were observed, some of which were also present in the preimmune sera.

Inclusion bodies are generally believed to be good antigens (Harlow et al., Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. (1988)), better results were obtained with the denatured, solubilized fusion proteins. Although inclusion bodies frequently gave positive antisera, soluble fusion protein often gave stronger antisera. In no case have were better antisera obtained with rabbits immunized with particulate inclusion bodies.

When the antigens were designed corresponding regions from the two halves of P-glycoprotein were chosen, but highly homologous sequences were avoided because of the desire to obtain antisera that would distinguish between the amino and carboxy halves of P-glycoprotein. The positive antisera were tested by immunoprecipitation against proteolytic fragments of P- glycoprotein, and none have been found that cross react with the corresponding regions in the opposite half of the protein.

Several modifications and other uses of the present method are possible. New restriction sites could be added to the PpuMI site at the 3' end of the PEΔ553 gene. New restriction sites could also be introduced at the EcoRI site, or the 105 base pair PpuMI/EcoRI fragment could be removed from pVC4D and replaced with a linker that introduces a new site, or with a linker that includes a multiple cloning region. These modifications might be advantageous if the cDNA fragment to be cloned into pVC4D contains either a PpuMI site or a EcoRI site. Modifications such as these might introduce some extra amino acid residues into the fusion protein that belong neither to PE nor to the eukaryotic protein or protein fragment. Additional amino acid residues would probably not affect expression of the fusion protein or the production of antibodies against the eukaryotic protein.

No limitations to the cDNA sequences that can be cloned into the PEΔ553 gene or protein sequences that can be fused to PEΔ553 are known. Thus, it is expected that any protein or protein fragment, of eukaryotic or prokaryotic origin, can be fused to PEΔ553. Furthermore, although all of the P- glycoprotein fragments ranged between about 40 and 100 amino acid residues, it is conceivable that proteins or protein fragments of any size could be fused to P- glycoprotein. Artificial peptide sequences, not known to exist in nature, could also be fused to PEΔ553. This would require backtranslating the peptide sequence, synthesizing oligonucleotides that encode the peptide sequence, and cloning the annealed oligonucleotides into pVC4D. Finally, for some purposes it might be desirable to make other modifications in the sequence of PEΔ533 itself.

Expression of the fusion proteins can be affected by several factors that can be varied to obtain the optimal conditions. For instance, composition of the growth medium, the absence or presence of specific nutrients, the quality of aeration, the point at which the culture is induced, the time of induction, and the amount of IPTG used to induce, may all affect the total amount of fusion protein produced by this method. Furthermore, many variations in the isolation of inclusion bodies and the subsequent purification of the fusion protein are possible. For the present purposes, it was not necessary to purify the fusion proteins after isolating the inclusion bodies. For other purposes, however, it might be necessary to purify the fusion protein, or even perform some chemical or enzymatic modification before using it.

Furthermore, in the present invention, the fusion proteins were used to immunize rabbits and polyclonal sera were obtained against P-glycoprotein. However, any animal could be used for the immunization and subsequent production of antibodies. In particular, fusion proteins produced by the present method may be used to immunized mice to produce monoclonal antibodies. Furthermore, Freund's adjuvant is not necessarily required to stimulate an immune response against PE fusion proteins. Antibodies have been obtained, without adjuvant, by immunizing rabbits with some PEPG fusion proteins in PBS (unpublished data) . The use of the resulting fusion proteins is not limited to the production of antibodies. For example, the inclusion bodies may be denatured and the fusion proteins renatured. Structural or functional studies may then be carried out on the protein or protein fragment that is fused to PEΔ553. The fusion proteins may also be useful for affinity purification of antibodies, such as antibodies produced either with the above method or by some other method.

The strategy used in the present method has at least two technical advantages over the traditional methods for producing fusion proteins. (1) One is no longer limited to the naturally occurring restriction sites that are available in a particular gene. Oligonucleotide primers may be synthesized that introduce the desired restriction sites and make it possible to clone out exactly the fragment that is desired for expression. (2) The tedious process of subcloning and modifying the fragment with linkers and adapters so that it may be introduced into the expression vector is eliminated. The amplified fragment from PCR may be directly introduced into the expression vector. After the desired fragments have been cloned into the fusion expression system, it is important to consider the amount of fusion protein produced by the bacteria and the ease with which it may be purified to a point suitable for immunizations. With the present PE fusions, one can consistently produce large amounts of fusion protein. More specifically, one can produce approximately 50 mg of protein from a 500 ml culture. These fusion proteins can be easily recovered in a form that is approximately 95% pure, simply by isolating the inclusion bodies. The other fusion protein systems have the ability to produce as much protein (Nilsson et al., EMBO J. 4:1075-80 (1985)). Yet, considerably less protein is often produced (Caroll et al., in Glover, D.M. (ed.), DNA Cloning: A Practical Approach. Vol. Ill, pp. 89-111, IRL Press, Oxford (1987), Hardy et al., J. Virol. 62:998-1007 (1988), Lόwenadler et al., EMBO J. 5:2398-98 (1986), and Kim et al., Gene 68:315- 21 (1988)). Furthermore, while purification of soluble fusion proteins by affinity chromatography can work well (Lόwenadler et al., Gene 58:87-97 (1987), Nilsson et al., EMBO J. 4:1075-80 (1985) and Terry et al., Arch. Virol. 104:63-75 (1989)), it is not always successful, and degradation of the fusion proteins during purification can be a problem (Carroll et al., supra, Morinet et al., J. Gen. Virol. 70:3091-97 (1989), Nilsson, supra, and Stahl et al., J. Immunol. Meth. 124:43-52 (1989)). In the experiments related to the present invention, every fusion that was attempted with PE and P-glycoprotein gave large amounts of protein that were easily purified without substantial degradation. A final point is whether or not the fusion proteins successfully stimulate a strong immune response and raise positive antibodies to the desired protein. The present PEPG fusions were mostly successful, although several did not illicit any measurable response in the immunoprecipitation screening assay.

An advantage of fusions with PE is that PE is a secreted toxin from a pathogenic bacterium and might act as a general adjuvant to stimulate an immune response against any protein sequence that is associated with it. In fact, several groups are using PE conjugates to develop human vaccines based partially on this idea (Cryz et al., Antibiot. Che other. 42:177- 83 (1989), Cryz et al. , J. Lab. Clin. Med. 111:701-07 (1988) ) . Protein A, which is a cell surface antigen of a pathogenic strain, may offer the same advantage (Lόwenadler et al., EMBO J. 5:2383-98 (1986), Terry et al., Arch. Virol. 63-75 (1989) and Sjόlander et al., Infect. Immunol. 58:854-59 (1990)).

Antibodies against specific regions of a protein are useful reagents to elucidate the structure and the function of that protein. The present invention allows for the production of a comprehensive panel of antibodies against P-glycoprotein.

Moreover, the strategy described here offers flexibility and speed over the older techniques. Large amounts of different fusion proteins may be produced easily and quickly. Additionally, only one expression vector is necessary to produce many different fusion proteins. This strategy will be useful to raise antibodies against other eukaryotic proteins that are difficult to prepare in large amounts. The present invention can be illustrated by use of the following non-limiting examples. Example I Construction of the Parent Vector The expression vector pVC4D was constructed from pVC4 (Chaudhary et al., Proc. Natl. Acad. Sci. 85: 2439-43 (1988)) and pVC45Df(+)T (Jinno et al., J. Biol. Chem. 264: 15953-59 (1989)). Both pVC4 and PVC45Df(+)T were digested with Xhol and EcoRI. The 417 bp fragment from pVC45Df(+)T was ligated to the 3.9 Kbp fragment from pVC4 to create pVC4D. A map of pVC4D is shown in Figure 1. To clone MDRl sequences onto the 3' end of PE, pVC4D was digested with PpuMI (New England Biolabs) and EcoRI, and the 105 bp fragment was removed by gel purification.

Plasmid pVC4, which carries a gene encoding mature, full length PE, was not suitable for the present purposes because the PE it encodes is functional. Plasmid pVC45Df(+)T, on the other hand, carries a PE gene from which the codon for Glu553 has been deleted. This amino acid residue is essential for the catalytic activity of PE, but apparently affects no other function of the protein (Caroll, supra, Douglas et al., J. Bacteriol. 169:4967-71 (1987) and Lukac et al., Infect. Immun. 56:3095-98 (1988)). The 417 bp Xhol/EcoRl fragment of pVC45Df(+)T that was cloned into pVC4 to make pVC4D contains this deletion. The final expression vector pVC4D contains a T7 promoter (Rosenburg et al., Gene 56:125-35 (1987) and Studier et al., J. Mol. Biol. 189:113-30 (1986)), and an initiating methionine codon starting at base pair 2432 (Figure 1) . Following this is the coding sequence for the mature PE gene from which the codon for Glu553 has been removed. The final protein is 613 amino acid residues long. It may be expressed by transforming the plasmid into E^. coli strain BL21(λDE3) , which contains the gene for T7 RNA polymerase under control of a lac promoter, and inducing with IPTG (Rosenburg et al., Gene 56:125-35 (1987) and Studier, supra) . The insertion of the plasmid into the host cell will be discussed more fully in Example II.

Example II Use of the Vector to Create Fusion Proteins MDRl sequences were prepared by PCR from plasmid pMDR2000XS, which contains the full length MDRl cDNA (Pastan et al. , Proc. Nat l Acad. Sci. 85:4486-90 (1988) and Ueda et al-, Proc. Nat'l Acad. Sci. 84:3004- 08 (1987)).

At the 3' end of the PEΔ553 gene is a PpuMI site, base pair 4262, that is suitable for cloning. There is a second PpuMI site within the PEΔ553 gene, but this site is not recognized by the enzyme if the plasmid is prepared from E_j. coli strain HB101. Presumably, this second PpuMI site is modified but the nature of the modification is not known. The 5^Λ PCR primers used to amplify the MDRl sequences each contain a PpuMI recognition site that reconstructs all but the final codon of PEΔ553,Lys613. Thus, in the final PE-P- glycoprotein fusion (PEPG) , PEΔ553 is complete except for the final amino acid residue.

The 5' primers contained the sequence AGGACCTG, which introduces, a PpuMI site, flanked on each side by five complete codons of the appropriate MDRl sequence. The 3' primers contained the sequence TGATGAATTC, which introduces two stop codons and an EcoRI site. This sequence was also flanked on each side by five complete codons of the appropriate MDRl sequence. PCR was performed with reagents and equipment from Perkin Elmer-Cetus. After PCR, the amplified MDRl fragments were digested with PpuMI (or Avail (Bethesda Research Laboratories) , which is more reliable than PpuMI) and EcoRI. To clone the MDRl sequences onto the 3' end of the PEΔ553 gene, the 105 bp PpuMI/EcoRI fragment of pVC4D was removed (as described above) and replaced with the PpuMI/EcoRI MDRl fragment that had been amplified by PCR. The P-glycoprotein sequences in each PEPG fusion are shown in Figure 2 and summarized in Table 1. The fragments were then transformed into E. coli strain HB101, and selected on ampicillin. The constructions were screened by sequencing the MDRl region with Sequenase (United States Biochemicals) to ensure that the correct reading frame was maintained and that no mutations had been introduced by the primers, PCR itself, or the cloning steps.

Example III Expression and Isolation of the Proteins The fusion proteins were expressed in

E. coli strain BL21(/1DE3) (Rosenburg et al., Gene 56:125-35 (1987) and Studier, supra) ) . A 500 ml culture in LB was induced at OD₆₅₀ = 0.8 with 1 mM IPTG and harvested one hour later. Inclusion bodies were prepared by the following protocol (Marston, F. , "The Purification of Eukaryotic Polypeptides Expressed" In Escherichia Coli, In DNA Cloning: A Practical Approach, Glover, D. (ed.), Vol. Ill, IRL Press, Oxford, pp.59-88 (1987)). The cell pellet was resuspended with 20 ml 50 mM TrisHCl pH 8.0, 1 mM EDTA,

100 mM NaCl (TEN) . 20μg of lysozyme was added, and the suspension was incubated at room temperature for 30 minutes with gentle agitation. The cells were centrifuged and resuspended with 20 ml TEN plus 0.1% deoxycholate. 160 μl 1 M MgCl and 200 μl 1 mg/ml DNase were added, and the suspension was incubated at room temperature for 30 minutes with gentle agitation. The suspension was centrifuged 10 minutes at 12000 x g, and the pellet was resuspended by sonication in 20 ml cold TEN plus 0.5% Triton X-100. The suspension was centrifuged again, 10 minutes at 12000 x g, resuspended by sonication in 20 ml cold PBS, and centrifuged again as before. This pellet was resuspended by sonication in 2 ml cold PBS. The final suspension of inclusion bodies typically contains about 50 mg of protein. (See Figure 3A for examples of the whole bacterial extracts and Figure 3B for examples of isolated inclusion bodies.)

Approximately 50 mg of protein were routinely obtained from a 500 ml culture, and it is estimated that the fusion proteins make up approximately 95% of the protein in the inclusion bodies. There are some contaminating bands in the inclusion body preparations that contain fusion proteins that are not present in the control with PEΔ553. It is believed that these contaminates are degradation products from the fusion protein. In any case, the contaminants represent only a small fraction of the total protein.

Example IV Immunization and Production of Antiserum To immunize rabbits, the fusion proteins were diluted to 2 mg/ml in PBS and then mixed with one volume of either complete Freund's adjuvant (CFA) or incomplete Freund's adjuvant (IFA) . The fusion proteins were used either directly as inclusion bodies, or they were first denatured with 2% SDS and 5% b- ercaptoethanol. The final concentration of SDS for those samples that were denatured was 0.05%. For each immunization or boost, 1 mg of protein was injected at multiple subcutaneous sites according to the following schedule: Day 1, immunize with CFA; Days 22 and 43, boost with IFA; Day 50, bleed and test serum. Female white New Zealand rabbits were used to raise antiserum. Sera were screened by attempting to immunoprecipitate P-glycoprotein that had been photoaffinity labeled with ³H-azidopine (Brugge ann et al., J. Biol. Chem. 254:15483-88 (1989)).

The results that were obtained with two fusion proteins are shown in Figure 4, and the results from all of the antigens are summarized in Table 1. Inclusion bodies are generally believed to be good antigens, but better results were obtained with the denatured, solubilized fusion proteins. Although inclusion bodies frequently gave positive antisera, soluble fusion proteins often gave stronger antisera. In no case have better antisera been obtained with rabbits immunized with particulate inclusion bodies.

Table 1 summarizes the antigens utilized and antisera obtained by the present inventors.

Table 1. Summary of Antigens and Antisera

The amino acid residues of P-glycoprotein that are incorporated into each fusion protein, PEPG, are shown. Yield is the amount of inclusion bodies in mg isolated from a 500 ml culture of bacteria. The results of the rabbit immunizations and the quality of the best serum are listed for each PEPG antigen . Excellent antisera, which immunoprecipitate approximately 80% to 100% of the azidonpine- labeled P-glycoprotein, are indicated with +++. Weak antisera, which precipitate less than 20% of P-glycoprotein in the same assay, are indicated with +.

SUBSTITUTE SHEET

Claims

WHAT IS CLAIMED IS:

1. A method of producing a fusion protein comprising the steps of:

(i) cloning at least one DNA sequence which encodes a protein or a portion thereof, into a vector, said vector comprising a gene which encodes an inactivated form of a bacterial toxin;

(ii) transforming a host cell with the resulting vector of step (i) , thereby allowing for expression of said fusion protein.

2. The method of claim 1 wherein said DNA sequence which encodes said protein or portion thereof is a human drug resistance gene.

3. The method of claim 2 wherein said DNA sequence which encodes said protein or portion thereof is MDR 1.

4. The method of claim 3 wherein said protein is human P-glycoprotein.

5. The method of claim 1 wherein said vector of step (i) is pVC4D.

6. The method of claim 1 wherein said gene comprises a DNA sequence which encodes an inactivated form of Pseudomonas exotoxin.

7. The method of claim 6 wherein said gene comprises a DNA sequence having a codon deletion for one amino acid of Pseudomonas exotoxin (PE) .

8. The method of claim 7 wherein said gene is PEΔ553.

9. The method of claim 7 wherein said amino acid is Glu553.

10. The method of claim 1 wherein said fusion protein is PE-p-glycoprotein (PEPG) .

11. A recombinant DNA molecule comprising: (i) a DNA segment which encodes a protein or a portion thereof;

(ii) a vector for introducing said DNA segment into host cells, said vector comprising a gene which encodes an inactivated form of a bacterial toxin.

12. The recombinant DNA molecule of claim 11 wherein said vector is pVC4D.

13. The recombinant DNA molecule of claim 11 wherein said protein is human P-glycoprotein.

14. The recombinant DNA molecule of claim 11 wherein said gene comprises a DNA sequence which has a codon deletion for one amino acid of Pseudomonas exotoxin.

15. The recombinant DNA molecule of claim 14 wherein said gene is PEΔ553.

16. The recombinant DNA molecule of claim 14 wherein said amino acid is Glu553.

17. A host cell stably transformed with the recombinant DNA molecule of claim 11 in a manner allowing expression of a fusion protein encoded by said recombinant DNA molecule.

18. The host cell of claim 17 wherein said cell is a procaryotic cell.

19. The host cell of claim 18 wherein said procaryotic cell is an Escherichia coli cell.

20. A recombinantly produced fusion protein consisting of a toxin protein encoded by a parent vector and a eucaryotic protein or fragment thereof.

21. The recombinantly produced fusion protein of claim 20 wherein said toxin protein is Pseudomonas exotoxin (PE) protein, and said PE protein lacks the amino acid Gly553.

22. The recombinantly produced fusion protein of claim 20 wherein said eucaryotic protein or portion thereof is human P-glycoprotein.

23. An antibody raised against the fusion protein of claim 20.

24. The antibody of claim 23 wherein said antibody is raised against a recombinantly produced fusion protein, said protein consisting of a eucaryotic protein or fragment thereof and a Pseudomonas exotoxin (PE) protein, said PE protein lacking the amino acid Gly553.

25. The antibody of claim 24 wherein said antibody is polyclonal.

26. The antibody of claim 24 wherein said antibody is monoclonal.