HK1003229B - Carrier-bound recombinant protein, process for producing it and its use as an immunogen and vaccine - Google Patents

Description

The invention relates to vector-bound recombinant proteins, a method for their production and use, in particular as immunogens and vaccines.

The main function of the immune system in humans and animals is to prevent and prevent pathological damage caused by degenerate cells of infectious viruses, bacteria, fungi or protozoa. The immune system is characterized by the fact that after repeated infections with pathogens, increasing resistance occurs. The aim of immunization is to build up the immune system's defenses against certain pathogens without causing the corresponding diseases.

The antibodies and cellular T and B lymphocytes, which provide specific defense against pathogens, are essential for the detection of foreign structures such as those found on a bacterial cell.

For the quality of monoclonal and polyclonal antibodies and for the efficacy of vaccines, it is essential that the immune response to the antigen is sufficient. However, often viral antigens or recombinant human proteins, when used without further modification, show a poor or no immune response. For this reason, these antigens are often coupled to carriers (preferably proteins) to enhance the immune response.

To enhance the immune response, it is beneficial to incorporate such antigens into the outer membrane of bacteria and to use these complexes as immunogens (J. Immunol. 139 (1987) 1658 - 1664, Bacterial Vaccines and Local Immunity - Ann. Sclavor 1986, n.1-2, pp. 19-22, Proceedings of Sclavo International Conference, Siena, Italy, 17-19 November 1986).

A disadvantage of using live bacteria or viruses as immunogens for immunization is that an undesirable pathogenic spread of the germs cannot be excluded.

However, killing or fragmenting the bacteria and viruses before use as immunogens or vaccines can alter the antigenic determinant, which can significantly reduce the immune response.

The present invention is therefore intended to provide immunogens and vaccines which do not have these disadvantages.

This task is accomplished by a carrier-bound recombinant protein, which is available by expression of recombinant DNA in these gram-negative bacteria, which encodes a first DNA sequence (DNA target sequence) which is coded for at least one hydrophobic non-lytic membrane-integrating protein domain with an alpha-helical structure consisting of 14 to 20 amino acids, the N and C terminals of which may be flanked by 2 to 30 amino acids each, a second DNA sequence (DNA protein sequence) which is coded for a recombinant protein, and a third DNA sequence (DNA target gene) which is coded for at least one non-lytic membrane-integrating protein domain, which is an alpha-helical structure consisting of 14 to 20 amino acids, the N and C terminals of which may be flanked by any 2 to 30 amino acids, a second DNA sequence (DNA protein sequence) which is coded for a recombinant protein, and a third DNA sequence (DNA-release gene) which is coded for a lytic protein, or a recombinant protein whose active toxin is obtained from the bacterial culture and contains parts of the recombinant protein.

Preferably, the expression of the fusion protein gene and the lysis gene is controlled from two different promoters (Fig. 1).

This type of expression of fusion protein genes and lysis genes results in a large number of fusion proteins being first integrated into the membrane of the gram-negative bacteria used as the host organism and then lyse of these bacteria. The otherwise dense cell wall complex of the bacteria is permeabilised to release the cytoplasmic components of the bacteria (Eur. J. Biochem. 180 (1989), 393-398). The morphology of the cells, such as the rod shape of E. coli cells, is preserved.

The bacterial hosts consist of the cytoplasmic (inner) membrane, the periplasmic space and the outer membrane, while maintaining the integrity of the cell wall complex to a large extent. For bacterial strains that additionally have an S-layer (paracrystalline protein layer outside the outer membrane), this protein layer is also part of the bacterial hosts (Ann. Rev. Microbiol. (((371983), 311-339).

The host organisms are all gram-negative bacteria, preferably gram-negative pathogens such as Escherichia coli, Bordetella pertussis, Campylobacter nijuni, Corynebacterium diphteriae, Legionella pneumophilia, Listeria monocytogenes, Pseudomonas aeruginosa, Shigella dysenteriae, Vibrio cholerae, Yersinia enterolitica, which are suitable for use in the treatment of the disease (Schaechter, M, H. Medoff, D. Schlesinger, Mechanisms of Microbial Disease, Williams and Wilkins, Baltimore (1989)).

The recombinant proteins based on the invention are surprisingly good as immunogens, resulting in strong immune responses and very high antibody titres.

A particular advantage is that the recombinant protein is integrated into the bacterial membrane immediately after expression, thus producing the carrier bond, thus avoiding the need to isolate the recombinant protein as such before producing the immunogen.

Another advantage of the method is that a very large number of antigenic epitopes are present in the cell wall complex of the bacterial hosts. It has been shown that the target sequences for the recombinant proteins favour certain areas within the bacterial cell wall complex for integration. These areas are mainly adhesion sites of the inner and outer membranes and are associated with the cell division of the bacteria. This does not result in a uniform distribution of the recombinant protein but rather in insect-like enrichments within the cell wall complex (see Fig. 2d). The clustered arrangement of the recombinant proteins within a relatively small area (cluster) has the advantage of stimulating lipopolysaccharide production by the cells. This also prevents the production of lipopolysaccharide in other cells.

It has been shown that the recombinant proteins of the invention are integrated in their natural protein structures and thus in active form into the bacterial membrane.

This is particularly surprising as recombinant proteins are usually obtained in inactive form as inclusion bodies (see EP-A0219 874, WO 89/03711) after expression in prokaryotes and can only be converted to the active form by denaturing and renaturing.

The use of human proteins and antigens, especially viral antigens, is preferred. Their size is not limited. The molecular weight of the antigens is preferably 2000 to 200000 Daltons.

The recombinant antigen is particularly preferable if it has antigenic structures of human viruses and retroviruses such as HIV, HBV and EBV.

The hydrophobic, non-lytic and membrane-integrating protein domains are referred to as target sequences, preferably complete or partial sequences of membrane proteins, but may also be modified by amino acid exchange, but such exchange must not alter the structure of the corresponding protein.

Preferably, target sequences that are not cleaved by membrane proteases (e.g. signal peptidase and periplasmic space proteases) as opposed to signal sequences from other membrane proteins are used. For example, target sequences can be derived by protein engineering from naturally occurring sequences of the lysine of the PhiX174 phage group (for N-terminal targeting) and from naturally occurring sequences of the lysine of the MS2 phage group (for C-terminal targeting).

The target sequence is preferably a hydrophobic alpha-helical protein domain of 14 to 20 amino acids, which can be flanked at the N- and C-terminal by 2 to 30 amino acids each. Preferably, at least one other protein domain can be bound to this protein domain. The binding is preferably via flexible link sequences.

The additional protein domains coupled to the first protein domain may be structured analogously to the first protein domain, but it is preferable that at least one of the additional domains has a beta fold sheet structure and is composed of 10 to 16 amino acids, preferably 11 to 13 amino acids. Such beta fold sheet structures preferably resemble in structure and secondary structure amphipathic protein sequences found in pores of the outer membranes. For N-terminal targeting, it is preferable to use target sequences containing amino acids 1 to 54 which are structurally similar to the protein phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen phagen

Membrane proteins of bacteriophages are preferably membrane proteins of bacteriophages of the class Microviridae, preferably of icosahedral, lytic and ssDNA-containing phages, which can infect Enterobacteria. Examples of these are the phages PhiX174, S13, G4, G6, G14, PhiA, PhiB, PhiC, PhiR, which can infect E. coli C strains. Also suitable is alpha3, which can infect E. coli C and E. B strains. Also suitable are the phages K9, PhiK, PhiXtB and U3, which can infect E. coli K12 strains (Sinshe, G. (1968) in the London edition of Cold Spring: N.L. Hayes, Ray Hayes, R.S. Hayes, J.N.S., J.N.S., Richard D. W.S., D.N.S., D.N.S., D.N.S., D.N.S., M.D., D.N.S., D.N.S., D.N.S., D.S., M.D., D.N.S., D., D.N.S., D.S., D.N.S., D., D.N.S., D., D.N.S., D., D.N.S., D., D.S., D.N.S., D., D.N.S., D., D.S., D.S., D., D.N.S., D., D.S., D., D.S., D., D.S., D., D.S., D., D., D.S., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D., D.,

Lytic membrane proteins are preferably lyse proteins from the bacteriophages mentioned and other toxin-releasing genes such as colicin lysegen (Microbiol. Sciences 1 (1984) 168-175 and 203-205).

In another preferred embodiment, a non-covalently binding binding partner for that protein is bound to the recombinant protein, to which other substances may be covalently or non-covalently bound. Examples of binding pairs whose partners are suitable as binding partners include biotin-streptavidin or avidin, hapten-antibodies, antigens-antibodies, concavalin-antibodies, sugar-lectin, hapten-binding protein (e.g. thyroxine binding globulin and thyroxine) or oligopeptide antibodies.

Preferably used as a binding pair of streptavidin or avidin and biotin, and particularly preferably used as an immobilized, recombinant protein of streptavidin or avidin and biotinylated antigen bound to it.

Furthermore, it is preferable to use a protein that recognizes a chemical ligand as the recombinant protein, such as β-galactosidase/p-aminophenyl-β-D-thiogalactoside (a structural analogue of lactose), gene 29 (1984) 27-31.

Another subject of the invention is recombinant DNA encoding a first DNA sequence (DNA target sequence) which encodes at least one hydrophobic non-lytic membrane-integrating protein domain having an alpha-helical structure and consisting of 14 to 20 amino acids, the N and C terminals of which may be flanked by any 2 to 30 amino acids, a second DNA sequence (DNA protein sequence) encoding a recombinant protein, and a third DNA sequence (DNA lysogen) under separate control encoding a lytic membrane protein containing bacteriophages or lytic toxin release genes or their lytic active parts, which we describe.

DNA sequences that encode for the L- or E-protein are preferred as target DNA sequences, and DNA sequences that encode for amino acid sequences derived from these proteins with the same secondary structure are also suitable, these sequences are preferably linked by DNA sequences that encode for hydrophilic protein domains with 2 to 30 amino acids and disordered secondary structure.

In a preferred embodiment, the DNA lysis sequence includes the DNA sequence of the E-protein, the DNA sequence of the L-protein or the DNA sequence of the EL hybrid protein (sequences see EP-A 0 291 021).

The DNA protein sequence is preferably the DNA sequence of a viral antigen (e.g. HIV, HBV, EBV) or a recombinant human protein.

Another subject of the invention is a method for the production of a recombinant protein bound to the membrane of gram-negative bacteria, characterized by the fact that in these gram-negative bacteria a recombinant DNA encodes a first DNA sequence (DNA target sequence) which is obtained for at least one hydrophobic non-lytic membrane-integrating protein domain having an alpha-helical structure and consisting of 14 to 20 amino acids, the N and C terminals of which may be flanked by any 2 to 30 amino acids, a second DNA sequence (DNA protein sequence) which is encoded for a recombinant protein, and a third separate DNA sequence (Lytogen DNA), which is obtained for the active culture or for the active protein to be released from the recombinant membrane, and a recombinant protein whose components are encoded for the Lytogen or Lytogen toxin.

Preferably, during fermentation, the activity of the lytic protein is first inhibited or the expression of the lysegen is repressed and only at a desired time, preferably in the late logarithmic phase, is the inhibition or repression reversed.

In another preferred embodiment, the recombinant carrier-bound protein obtained is incubated with a derived binding partner for the protein, if any, and the resulting conjugate is isolated.

In another preferred embodiment, the genes of at least two different recombinant proteins are expressed according to the invention. This allows for the production of immunogens or vaccines that have multiple antigenic structures. In particular, it is preferred to use the antigen determinants of different viruses or retroviruses (e.g. HIV1, HIV2, HBV and EBV) as recombinant proteins. For expression, these can be arranged in an expression vector either as open reading frames in the 3′ direction of the target gene sequence or each recombinant protein to be expressed can have its own vector.

Another subject of the invention is a method for the production of antibodies characterized by immunizing a mammal with a carrier-bound recombinant protein obtained by expression of a fusion protein in gram-negative bacteria and containing at least one hydrophobic non-lytic membrane-integrating protein domain and the recombinant protein, where appropriate delayed expression of a lytic membrane protein from bacteriophages or lytic toxin release gas or lytic toxin partial sequences thereof, and the antibodies are obtained from serum or spleen by known methods.

In a preferred embodiment, B lymphocytes from immunised animals are fused with a suitable cell line in the presence of transforming agents, the cell line that produces the desired antibodies is cloned and cultured, and the monoclonal antibodies are obtained from the cells or culture residue.

It has been shown that the method of the invention is particularly suitable for the production of viral immunogens, such as HIV immunogens, HBV immunogens.

It was also surprisingly shown that recombinant antigens, which are usually expressed in inactive form in prokaryotes as refractory bodies (e.g. human proteins such as tPA or G-CSF), are preserved in their activity and thus in their antigenic structures when expressed in accordance with the method of the invention, which proves the method of the invention to be particularly advantageous in the production of immunogenic recombinant human proteins.

The invention also relates to the use of the recombinant proteins bound to the carrier of the invention as vaccines and for the stimulation of T lymphocytes.

The vaccines of the invention can be manufactured and used in the usual way.

The invention also relates to a method for the manufacture of vaccines using the recombinant proteins of the invention which can be manufactured by the known methods, but preferably the recombinant protein of the invention is first lyophilized and then suspended, if necessary with the addition of auxiliary substances.

It is further preferable to formulate the vaccine as a multivalent vaccine, for which the recombinant carrier-bound protein of the invention may contain several recombinant antigens immobilized on the membrane of the bacterial host.

Vaccination with the vaccine of the invention may be carried out by methods familiar to any professional, such as intradermal, intramuscular, intraperitoneal, intravenous, subcutaneous, oral and intranasal.

For intramuscular or subcutaneous administration, the vaccine may be suspended in saline, for example.For intranasal or intraocular administration, the vaccine may be administered, for example, in the form of a spray or aqueous solution.For local, for example, oral administration, it is often necessary to temporarily protect the immunogens against infection,for example, from sucrolytic enzymes in the mouth or from proteolytic enzymes in the stomach.One type of temporary protection may be provided, for example, by encapsulation of the immunogen.This encapsulation may be achieved, for example, by coating the immunogen with a protective agent (microencapsulation) or by inserting a multi-make-up capsule into a protective vessel (activation capsule).

The encapsulation material may be semi-permeable or become semi-permeable when introduced into the human or animal body.

The following examples, illustrations and sequence protocols further explain the invention. Fig.1 shows schematic representations of the plasmids pkSELS, pML1 and pMTV1Fig.2 Schematic representation of a bacterial host as a carrier of recombinant proteins (a) longitudinal cut by a gram-negative bacterium (om: outer membrane; pp: periplasmic space, in: inner (cytoplasmic) membrane, cp: cytoplasm). (b) formation of a transmembrane lysetunnel. (c) cytoplasm flowing out through the lysetunnel. (d) bacterial host with fusion proteins anchored in the cell wall complex via target sequences.

Example 1 N-terminal membrane targeting for HIV 1 gp41.

From the plasmid pHF14, an HIV 1 specific DNA fragment is isolated by a partial Hincll/Pvull Digest as a 1445bp DNA fragment. The fragment contains the entire sequence of gp41, (345 codons of gp41) Left sequence, the last 45 codons of gp120. It corresponds to the nucleotides 4 to 1448 from SEQ ID NO: 1.

After linearization of the pKSEL5 plasmid (SEQ ID NO:6) with Accl and filling of the overlying DNA ends with Klenow polymerase, the HIV1 specific DNA fragment is ligated with this linearized plasmid. The resulting plasmid is designated pHIE1 and contains in frame a fusion of a sub-sequence of the E-gene (E'-target sequence) of PhiX174 with the above-mentioned HIV1 fragment, while retaining the natural stop codon of the HIV1 env gene.

Example 2 N- and C-terminal membrane targeting of HIV1 gp41.

From plasmid pHF14 a 1059 bp HIV1 specific DNA fragment is isolated by Hincll digestion. This fragment contains 5'-sided left-handed sequences followed by 45 codons from gp120 and 301 codons from gp41. It corresponds to nucleotides 4 to 1062 from SEQ ID NO:1. After linearizing plasmid pKSEL5 with Accl and filling the overlying DNA ends with Klenow polymerase, the HIV1 specific DNA fragment was ligated with this vector. The resulting plasmid pHIE3 contains a frame in fusion of a partial sequence of the E-gene (target sequence) with an HIV partial sequence and a partial sequence of the L-Te-G (target sequence).

Example 3 C-terminal membrane targeting of HIV1 gp41.

From plasmid pHF14 a 1061 bp DNA fragment is isolated with Sall and Hincll. This fragment contains 5'-sided left-handed sequences followed by 45 codons gp120 and 301 codons gp41. It corresponds to nucleotides 2 to 1062 from SEQ ID NO: 1. After removal of the E-sequence from the plasmid pKSEL5 by Xhol/Accl digestion, the overlying DNA ends of the vector and the isolated HIV1 fragment are replenished and ligated by Klenow polymerase. The resulting plasmid pHIE5 contains an in frame fusion of the first 5 codons of the lacZ gene, polylinker codons, gp120/gp41 codons and polylinker codons followed by the codon's target.

Example 4 C-terminal membrane targeting of streptavidin.

The 498 bp Klenow polymerase-filled Xbal fragment (FXaStrpA, nucleotide 2 to 499 of SEQ ID NO:2) from pFN6 is ligated into the plasmid pKSEL5 from which the E-gene fragment was deleted by cutting with Hincll/Xhol at the filled interfaces. The resulting plasmid is called pAV5. This results in an in frame fusion of the first 5 codons of the LacZ gene, 26 amino acid codons from the remaining polylinker sequence, and the amino acid sequences of the FXaStrpA component in the plasmid pAV5 followed by the L'Target sequence.

Example 5 N-terminal membrane targeting of streptavidin.

From plasmid pFN6, the 5'-side streptavidin gene, extended by a factor Xa protease interface, is isolated as a 511 bp fragment from BamHI digestion. It contains nucleotides 14 to 524 from SEQ ID NO.2. This DNA fragment is integrated into the filled Xbal interface of the pKSEL5 vector between the E' and L' target sequences after filling the ends with Klenow polymerase. In plasmid pAV1, a gene fusion from the E'-target sequence and the FXaStrpA sequence is thus carried out in frame. The 3'-side of the streptavidin occurs by the occuring stopcodons.

Example 6 N- and C-terminal membrane targeting of streptavidin.

The 5'-TAATAA-3' stop codons in the pAV1 plasmid behind the streptavidin gene are removed by deletion of a 33bp DNA fragment produced by partial Hincll and subsequent Xbal digestion. The streptavidin-specific DNA sequence contains nucleotides 14 to 499 from SEQ ID NO.2. After replenishment of the plasmid endpoints with Klenow polymerase and ligation, the L'-target sequence on the vector fuses in frame to the E'-target sequence and the FXaPA-sequence (pAV3 plasmid). The corresponding gene product thus has an N- and C-terminal sequence.

Example 7 N-terminal membrane targeting of beta-galactosidase.

From the plasmid pMC1403 (J. Bacteriol. 143 (1980) 971-980), a 3124 bp DNA fragment (SEQ ID NO:3) is isolated by Pstl and Dral and ligated to the Pstl and Nrul restriction sites of the plasmid pKSEL5. The resulting plasmid pLZ1 contains the first 54 codons of the E-target sequence, 13 linker codons and 1015 codons of the LacZ gene. The Pstl/Dral fragment used for plasmid pLZ1 extends from Nucleotide NO 26 to 3149 including 3124p in the sequence protocol SEQ ID: 3 of Nucleotide NO 26 and includes 3124b.

Example 8 N- and C-terminal membrane targeting of beta-galactosidase.

From plasmid pMC1403 the 3010 bp LacZ DNA fragment (Pstl - EcoRI, nucleotides 26 to 3035 from SEQ ID NO: 3) is isolated and integrated into the Pstl/Hindlll restriction site of pKSEL5 after replenishment of the EcoRI and Hindlll ends respectively, resulting in a fusion in frame of the E' target sequence with the LacZ gene and the L' target sequence in the resulting plasmid pLZ3.

Example 9 C-terminal membrane targeting of beta-galactosidase.

Plasmid pLZ3 is digested with EcoRI and partially with Accl. This removes the E'-target sequence. The fragment contains the nucleotide 29-3035 from SEQ ID NO:3 and is 3007 bp long (after EcoRI interface filling). After filling the overlapping DNA ends of the vector and religation, the vector pLZ5 is obtained, which contains a lacZ-L'-fusion gene and whose gene product has a C-terminal membrane target sequence.

Example 10 The test chemical is a chemical that is used to produce the protein.

The plasmids pMTV1 and pML1 contain a lysescassette consisting of the lambda c1857 repressor gene, the right-hand lambda promoter/operator system pR, and PhiX174 lysegen E. Integration of the foreign gene can be performed in the multiple cloning site mcs 2 for pMTV1 or pkSEL5 (Fig. 1) in an analogous manner as described in examples 1-9.

Example 11 Fermentation and lyse

The plasmid is incorporated into E. coli K12 (DSM 2093) and the culture is drawn to OD 0.8 - 1.2 at 600 nm in the shaker flask, whereby the expression of lysegen E is repressed by c1857 repressor molecules.

Example 12 Modified protein E-lysis.

It is cultured as in example 11, but 30 minutes before raising the temperature from 28°C to 42°C, the culture medium is raised to 0.2 mol/I magnesium sulphate by adding magnesium sulphate solution, which prevents the bacteria from lysing despite expression of gene E.

The cells are harvested by centrifugation after 30 min. By resuspension of the cell pellet in a low-molar buffer (PBS, 1 mmol/1 phosphate buffer, 1 to 10 mmol/I Tris - HCI pH 6-8) or water, an immediate lysis of the cells takes place. The resulting cell shells are called bacterial hosts.

For cleaning, wash the bacterial host twice with PBS or 0,9% NaCI (resuspend and centrifuge) and freeze-dry.

Example 13 Immunisation

109 Germs (equivalent to 1 mg of bacterial host dry weight) per mouse are administered at 0.9% NaCI 4 times per month for immunisation.

Example 14 Binding of biotinylised HBc antigen

Following example 4, manufactured bacterial hosts incorporating streptavidin are lyophilised via target sequences. 1 mg of this lyophilisate is incubated in 10 ml of a 20 Ig/ ml solution of a conjugate of hepatitis B core antigen and biotin (produced by reaction of HBcAg with N-hydroxysuccinimide activated biotin) in 40 mmol/1 phosphate buffer, pH 7.4, for 30 min, and then washed repeatedly with 40 mmol/1 phosphate buffer, pH 7.4, to obtain a carrier-bound HBcAg immunogen that can be used for immunity and antibody production.

Claims (16)

1. Process for the production of a recombinant protein bound to the membrane of gram-negative bacteria, wherein a recombinant DNA which contains a first DNA sequence (DNA target sequence) that codes for at least one hydrophobic, non-lytically active, membrane-integrating protein domain which has an alpha-helical structure and is composed of 14 to 20 amino acids which can be flanked N- and C-terminally by 2 to 30 arbitrary amino acids in each case, a second DNA sequence (DNA protein sequence) that codes for a recombinant protein as well as a third DNA sequence (DNA lysis gene) under separate control from this that codes for a lytically-active membrane protein from bacteriophages or a lytically-active toxin release gene or for their lytically active parts is expressed in these gram-negative bacteria and the carrier-bound recombinant protein is isolated from the culture broth.

2. Process as claimed in claim 1, wherein at least one further gene of a recombinant protein is expressed.

3. Process as claimed in claims 1 or 2, wherein during the culture the activity of the lytic membrane protein is inhibited or the expression of the lytically active membrane protein or toxin gene is repressed and the inhibition or repression is abolished at a desired time.

4. Process as claimed in claims 1 to 3, wherein the protein is incubated with a binding partner for the protein which is derivatized if desired and the conjugate bound to the membrane of gram-negative bacteria which is formed is isolated.

5. Process as claimed in claim 1, wherein the fusion protein contains at least one further protein domain which consists of 10 to 16 amino acids and has a ß-pleated sheet secondary structure.

6. Process as claimed in claim 1, wherein for the protein domain one uses the amino acids 1 to 54 from the protein E of the phage PhiX174 or the amino acids 21 to 75 from the protein L of the phage MS2 and/or an amino acid sequence which is obtainable from these sequences by amino acid substitution and has an analogous protein secondary structure.

7. Process as claimed in claim 1, wherein the protein domains and the recombinant protein are linked by a hydrophilic amino acid sequence with 2 to 100 amino acids and a disordered secondary structure.

8. Process as claimed in claim 1, wherein the recombinant protein has an antigenic structure.

9. Process as claimed in claim 1, wherein a non-covalently binding binding partner for this protein to which, if desired, further substances are bound covalently or non-covalently, is bound to the recombinant protein

10. Process as claimed in claim 9, wherein streptavidin or avidin is used as the recombinant protein.

11. Protein as claimed in claims 9 and 10, wherein the non-covalent binding partner is a biotinylated antigen.

12. Process for the production of antibodies, wherein a mammal is immunized with a protein which was obtained according to claims 1 to 11 and the antibodies are isolated from the serum or the spleen.

13. Process for the production of antibodies as claimed in claim 12, wherein the B lymphocytes of the immunized animals are fused with a suitable cell line in the presence of transforming agents, the cell line producing the desired antibodies is cloned and cultured, and the antibodies are isolated from the cells or the culture supernatant.

14. Use of the carrier-bound proteins which are obtained according to claims 1 - 11 for the production of vaccines.