WO2017186784A1

WO2017186784A1 - Autodisplay of antibodies

Info

Publication number: WO2017186784A1
Application number: PCT/EP2017/059910
Authority: WO
Inventors: Joachim Jose; Wilhelmine WECKENBROCK
Original assignee: Autodisplay Biotech Gmbh
Priority date: 2016-04-26
Filing date: 2017-04-26
Publication date: 2017-11-02

Abstract

The invention relates to a method for the surface display of at least one first polypeptide comprising a light chain variable domain V_L or/and at least one second polypeptide comprising a heavy chain variable domain V_H on the surface of a host cell, said method comprising the steps: (a) providing a host cell comprising at least one first polynucleotide or/and at least one second polynucleotide, wherein said at least one first polynucleotide comprises: (i) a portion encoding a signal peptide, (ii) a portion encoding said first polypeptide to be displayed, (iii) a portion encoding a transmembrane linker, and (iv) a portion encoding the transporter domain of an autotransporter, and wherein said at least one second polynucleotide comprises: (i) a portion encoding a signal peptide, (ii) a portion encoding said second polypeptide to be displayed, (iii) a portion encoding a transmembrane linker, and (iv) a portion encoding the transporter domain of an autotransporter, and (b) culturing the host cell under conditions wherein said first polynucleotide or/and said second polynucleotide are expressed and said at least one first polypeptide or/and said at least one polypeptide are displayed on the surface of the host cell.

Description

Autodisplay of antibodies - Description

Field of the Invention

Background of the invention

Antibody molecules or the functional fragment thereof are widely used, for example as therapeutics, diagnostics or for passive immunization. Methods allowing the production or/and selection of antibody molecules or functional fragments thereof are therefore of high relevance. Methods for antibody molecules including both variable and constant domains, such as full-length antibody molecules, are of particular interest.

Several techniques have been described for producing or screening antibodies. The present invention aims at overcoming disadvantages of the state of the art techniques. Hybridoma technology

The hybridoma technology is one of the most commonly used methods to produce large amounts of monoclonal antibodies with single antigen specificity using an immortalized hybridoma cell line.

To generate a hybridoma cell line, an antigen of interest is injected into a mouse to provoke an immune response. Subsequently, B cells producing antibodies capable of binding to the antigen of interest are isolated from said mouse. The isolated B cells are fused with immortal B cell cancer cells (myeloma cells) to produce a hybrid cell line called hybridoma. The myeloma cell line used in this process is selected for its ability to grow in tissue culture and for absence of antibody synthesis. Hybridoma populations can be screened for their affinity to the antigen of interest, e.g. by enzyme-linked immunosorbent assay (ELISA) or Fluorescence-activated cell sorting (FACS).

The hybridoma technology is time-consuming, expensive and requires the use of laboratory animals. Moreover, the choice of the antigen is limited, since toxic, labile or highly conserved antigens cannot be used in the hybridoma technique. Labile antigens are often degraded after immunization. Highly conserved antigens may not trigger an immune response. Additionally, the murine antibodies obtained may be immunogenic and can elicit an immune response in a human organism which renders these antibodies unsuitable for certain therapeutic approaches. Humanizing the antibodies requires additional time and costs.

Phage display

Phage display was developed to overcome the disadvantages of the hybridoma technology and is widely used in research and industry. Phage display enables presentation of libraries of antibody fragments on the surface of phages.

To this end, a gene encoding a phage coat protein is fused to a gene encoding the antibody fragment, such as for example a single chain variable fragment (scFv). After maturation of the phage in E.coli, the antibody fragment capable of binding the antigen is displayed on the capsid of the phage. Phage libraries can be enriched for phages displaying an antibody fragment with affinity to a particular antigen of interest by a process called "biopanning". Here, the antigen of interest is immobilized and bound by the antibody fragments displayed by the phages of the library. After removing non-bound phages by washing, the bound phages are eluted and used for production of a further phage population. Antigen-binding antibody fragments can be enriched in an iterative selection process.

In phage display antibody fragments with the highest binding affinity can be discriminated during "biopanning" of phage libraries, because the mild elution conditions applied cannot detach highly potent binders from the immobilized target. Further, due to their small size phages are not suitable for flow cytometry, which is a useful high throughput method used to determine antigen-antibody fragment binding. Finally, the size of proteins presented by phage display is limited, and antibody molecules, in particular full-length antibodies, are too large to form functional capsid proteins. Consequently, phage display cannot be used for presentation of larger antibody molecules, such as full-length antibodies. Phage display is only suitable for display of antibody fragments, which mostly lacking the constant domains forming the fragment crystallizable region Fc. For most technical and pharmaceutical applications, antibody molecules including the constant domains, e.g. "full-length antibodies", are required, because they have a prolonged half-life (Nelson 2010) and the ability to stimulate secondary effector functions due to the F_c-region (Zhang 2013). Therefore, new antibody variants which have been identified by phage display in a shortened format need to be expanded by the constant domains prior to antibody production.

Cellular display technologies

Cellular display techniques have been developed in an attempt to overcome the drawbacks of phage display.

In cellular display, antibody molecules are displayed on particular host cells, such as yeast cells, mammalian cells or bacterial cells. Cellular display systems have one main advantage over phage display. In contrast to phages, they are compatible with high-throughput screening via fluorescence activated cell sorting (FACS). This screening method is not biased, while biopanning of phages tends to discriminate the most potent binders.

Screening of libraries of yeast cells displaying antibody fragments on their surface has been described by Boder and Wittrup, 1997. By immobilizing the Z domain of protein A on the surface of a yeast cell, presentation of secreted full-length antibodies on the surface of yeast cell could be achieved (Rhiel et al. , 2014). Glycosylation patterns in yeast and humans are however largely different.

Surface display of a scFv library in mammalian cells has been described (Ho et al., 2006). Display of full-length antibodies on human embryonal kidney cells HEK-293 was achieved using the transmembrane domain of PDGFR (platelet-derived growth factor receptor) (Zhou et al., 2010). Growing human cell lines is however time- consuming and expensive.

To date, several approaches to display antibody fragments on bacteria have been described.

A quantitative system for screening combinatorial scFv antibody libraries was developed utilizing surface display on E.coli and fluorescence-activated cell sorting (FACS) (Daugherty et al., 1998). Here, scFV fragments, i.e. fusions of the variable heavy chain domain V_H and the variable light chain domain V_L, are displayed using a chimera of E.coli lipoprotein and the outer membrane protein OmpA.

Blasshofer et al., 2006, Blasshofer et al., 2007 and Thommes et al. , 2010 describe separate expression of the variable heavy chain domain V_H and the variable light chain domain V_L forming a twin chain variable fragment (tcFv) in E.coli using AIDA-I as autotransporter.

Salema et al., 2013 disclose selection of single domain antibodies from immune libraries displayed on the surface of E.coli with β-domains of EhaA autotransporter and intimin. Antibody molecules with separate heavy and light polypeptide chains (e.g. Fabs, IgGs) could not be displayed in this study.

Mazor et al. 2007 discloses the isolation of full-length antibodies from libraries using a technology called anchored periplasmic expression (APEx). In APEx antibodies are display on the inner membrane of E.coli. Heavy and light chains of the antibody are secreted into the periplasm, where they assemble into IgGs, which are in turn captured by an Fc-binding protein that is tethered to the inner membrane. After disruption of the outer membrane, spheroblasted cells associated with the antibody molecules are obtained. Spheroblasts are physiochemically labile. Working with these sensitive spheroplasts therefore comprises some difficulties in handling. Direct use of the E.coli cells displaying an antibody on the inner membrane is not possible due to the presence of the outer membrane.

Autotransporter

Autodisplay is the recombinant surface display of proteins or polypeptides by means of the autotransporter in any Gram-negative bacterium. The transporter domain is preferably capable of forming a β-barrel structure.

Among other systems for the secretion of proteins in Gram-negative bacteria, the autotransporter pathway represents a solution of impressing simplicity. It is possible to transport a protein, regardless whether it is recombinant or the natural passenger, to the actual outer membrane, as long as its coding region lies between a typical signal peptide and a C-terminal domain called β-barrel. Based on these findings the autodisplay system has been developed by the use of the natural E. coli autotransporter protein AIDA-I (the adhesin involved in diffuse adherence) in a homologous E. coli host background (Jose et al. 2007). With the aid of a typical signal peptide, the precursor is transported across the inner membrane. Arrived in the periplasm, the C terminal part of the precursor forms a porin-like structure, a so- called β-barrel, within the outer membrane and through this pore the N terminally attached passenger (the actual protease) is translocated to the cell surface. To obtain full surface exposure of the passenger, a linker peptide is required in between the β- barrel and the passenger.

EhaA is an autotransporter protein derived from E. coli strain 0157:H7. It has been demonstrated that EhaA is located at the cell surface and resulted in the formation of large cell aggregates, promoted significant biofilm formation and mediated adhesion to primary epithelial cell of the bovine terminal rectum (Wells et al., 2008). EhaA has an identity to AIDA-I on the nucleic acid level of about 43% only. On level of the amino acid sequence, the identity is only about 34%.

Objective problem of the invention

The objective problem to be solved can be considered as providing a flexible, fast and inexpensive method for displaying antibody fragments and antibody molecules on the surface of a host cell. Such a method can in turn for example be used for developing and screening new antibody molecule variants.

The problem was solved by using the autotransporter domain for display.

The present inventors have now discovered that autotransporters can be used to functionally express antibodies containing a light chain and a heavy chain on the surface of a bacterial host cell.

The examples show surface display of separately expressed antibody polypeptide light and heavy chains, which assemble into a functional antibody molecule on the surface of E.coli using the AIDA-I autotransporter system and the MATE system using the EhaA autotransporter. Successful expression and co-expression of separated full-length chains of an antibody on the surface of E. coli using the AIDA-I- autotransporter technology (Jose and Meyer 2007) and the EhaA MATE system (WO2014/139862 A1 ) is shown by SDS-PAGE analysis of outer membrane protein isolation (Fig. 3 and Fig. 8). The surface accessibility and display of the antibody chains on E. coli is demonstrated by a protease accessibility assay and flow cytometry (Fig. 4). For E. coli cells co-displaying both immunoglobulin polypeptide chains, functionality of the antibody is shown via flow cytometry based antigen binding assays (Fig. 5). Results of co-immunoprecipitation (Fig. 6) prove that the individually displayed antibody chains interact with one another on the bacterial surface to form a fully assembled antibody molecule (Fig. 7). The technique developed by the inventors can be used to identify new antibodies against any desired target directly in the full-length format. Full-length antibodies in particular comprise a constant domain.

This invention particularly aimed at displaying antibody molecules including the constant domains, such as functional full-length immunoglobulins, on the surface of a bacterial host cell. Constant domains result in a secondary immune response in vivo and prolong the serum half-life compared to shortened fragments. By providing a system for display of antibody molecules including the constant domain, time consuming steps for subsequent addition of constant domains to the variable fragments screened for by e.g. phage display can be omitted from the work flow. The screening method of the present invention allows screening and production of antibody molecules in a time and cost effective way. In contrast to the hybridoma technique, no laboratory animals have to be used. Further, according to method of the present invention the antibody molecules are displayed on the surface of intact host cells. This allows the usage a variety of analytical methods which cannot be applied to spheroblast displaying antibody molecules, obtained by the Apex technology. Advantageously, the method of the present invention is compatible with flow cytometry, such as FACS, which allows the displaying cells to be used in high through-put methods, such as flow cytometry. Bacterial display systems are especially attractive due to their low costs and short process times, e.g. in comparison to yeast or mammalian cells.

The method of the present invention can be used as an easy and versatile tool for screening of antibody fragments and molecules. The present invention is also suitable for bacterial tumor cell targeting or for the production of aglycosylated antibodies, which are not accessible with the prior art methods, i.e. the hybridoma technique or phage display.

Detailed description of the invention

Display method

A first aspect of the present invention relates to a method for displaying at least one first polypeptide comprising a light chain variable domain V_L or/and at least one second polypeptide comprising a heavy chain variable domain V_H on the surface of a host cell, said method comprising the steps:

(a) providing a host cell comprising at least one first polynucleotide or/and at least one second polynucleotide,

wherein said at least one first polynucleotide comprises:

(i) a portion encoding a signal peptide,

(ii) a portion encoding said at least one first polypeptide to be displayed,

(iii) a portion encoding a transmembrane linker, and

(iv) a portion encoding the transporter domain of an autotransporter,

and wherein said at least one second polynucleotide comprises:

(i) a portion encoding a signal peptide,

(ii) a portion encoding said at least one second polypeptide to be displayed,

(iii) a portion encoding a transmembrane linker, and

(iv) a portion encoding the transporter domain of an autotransporter,

and

(b) culturing the host cell under conditions wherein said at least one first polynucleotide or/and said at least one second polynucleotide are expressed and said at least one first polypeptide or/and said at least one polypeptide are displayed on the surface of the host cell.

Preferably,

(i) said at least one first polypeptide further comprises a light chain constant domain C_L, or/and

(ii) said at least one second polypeptide further comprises at least one heavy chain constant domain C_H.

Step (a)

By the method of the present invention, at least one first polypeptide comprising a light chain variable domain V_L or/and at least one second polypeptide comprising a heavy chain variable domain VH can be displayed on the surface of a host cell. The surface of the host cell is preferably an outer membrane of the host cell.

In the context of the present invention, a polypeptide can be a protein or a fusion protein. The first or/and second polypeptide includes "expression product", "polypeptide fusion", "antibody chain", or "passenger", "passenger polypeptide" or "passenger protein", or "amino acid sequence". The first polypeptide includes "light chain" of an antibody or "light chain fragment" of an antibody and the second polypeptide includes "heavy chain" of an antibody or "heavy chain fragment" of an antibody. Most preferably, the first and second polypeptide form an antibody molecule as defined herein.

The host cell preferably comprises at least one first polynucleotide as described herein or/and at least one second polynucleotide as described herein. Preferably the host cell comprises at least one first polynucleotide and at least one second polynucleotide.

The host cell in the method of the present invention is preferably transformed or/and transfected with at least one first polynucleotide or/and at least one second polynucleotide.

In the context of the present invention, a polynucleotide includes "nucleic acid", "nucleic acid fusion", "nucleic acid portion", "nucleotide sequence" or "nucleic acid sequence". The first or/and second polynucleotide can be a nucleic acid fusion comprising nucleic acid sequences encoding portions (i) to (iv) as described herein. The polynucleotide can be RNA or DNA, preferably mRNA or cDNA.

The components (i) to (iv) are fused in frame. The components (i) to (iv) in polynucleotide of the present invention are preferably oriented from 5' to 3'. In the expression product, i.e. the first or second polypeptide obtained in step (b), the amino acid sequences encoded by nucleic acid sequences (i) to (iv) are preferably arranged N terminal to C terminal.

The first or/and second polynucleotide comprises (i) a portion encoding a signal peptide, preferably a portion coding for a Gram-negative signal peptide allowing the transport into the periplasm through the inner cell membrane. The signal peptide may be a signal peptide homologous to the host cell. The signal peptide may also be a signal peptide heterologous to the host cell. An example of a suitable signal peptide is the CtxB signal peptide. The signal peptide can be cleaved off during maturation of the polypeptide fusion.

As described herein above, the first or second polynucleotide comprises (ii) a portion encoding the recombinant first or second polypeptide to be displayed, respectively. The first polypeptide can be a light chain of an antibody as described herein. The second polypeptide can be a heavy chain of an antibody as described herein. By the method of the present invention, at least one first polypeptide comprising a light chain variable domain V_L or/and at least one second polypeptide comprising a heavy chain variable domain V_H can be displayed on the surface of a host cell.

The first or second polynucleotide comprises (ii) a portion encoding a transmembrane linker. Transmembrane linker includes the expression "linker domain".

Additionally, the first or/and second polynucleotide comprise (iv) a portion encoding the transporter domain of an autotransporter.

The at least one first or/and at least one second polynucleotide is optionally operatively linked with an expression control sequence, preferably a promoter, and optionally further comprises sequences required for gene expression in the respective host cell. The person skilled in the art knows suitable expression control sequences and promoters, in particular for expression in the host cell species as described herein. The promoter or/and expression control sequence may be homologous or heterologous to the host cell.

Preferably, the first polynucleotide or/and the second polynucleotide is present in at least one plasmid, such as a plasmid vector. The first and second polynucleotide can be present in one plasmid, also referred to as bicistronic plasmid. More preferably the first polynucleotide is present in a first plasmid and said second polynucleotide is present in a second plasmid. The use of a plasmid is preferred for inducible expression of the first polynucleotide or/and second polynucleotide. Alternatively, the first polynucleotide or/and second polynucleotide is integrated into the genome of the host cell. Integration into the genome of the host cell is done by methods known to a skilled person. Expression of integrated polynucleotides of the invention can be inducible or stable, e.g. by using suitable promoters known to a skilled person. Integration into the genome is preferred for stable expression of the first polynucleotide or/and second polynucleotide.

The light chain variable domain V_L preferably comprises three complementary determining regions (CDR) CDRL1 , CDRL2, and CDRL3. Furthermore, the heavy chain variable domain V_H preferably comprises three complementary determining regions CDRH1 , CDRH2, and CDRH3. Preferably, each variable domain comprises three complementary regions. Complementary determining regions are the most variable part of an antibody and are primarily responsible for antigen-binding and antigen-specificity. Preferably, the complementary determining regions form an antigen-binding site, more preferably all six complementary determining regions CDRL1 , CDRL2, CDRL3 and CDRH1 , CDRH2, and CDRH3 form an antigen-binding site. In particular, a set of CDRs constitutes a paratope, i.e. the antigen-binding site of an antibody, which recognizes an antigen. The light or/and heavy chain variable domains can further comprise framework regions (FR) before and after each CDR. Framework regions are also part of the variable domains, but are less variable than CDRs.

In a further aspect, the antibody molecule or a functional fragment thereof can be a naturally occurring antibody molecule or non-naturally occurring antibody molecule. A non-naturally occurring antibody or functional fragment thereof can for example be a chimeric antibody. A chimeric antibody comprises domains from different organisms, e.g. from mouse and human. In chimeric antibodies, the constant domains are e.g. from human, whereas the variable domains are e.g. from mouse. The antibody molecule or functional fragment thereof can also be a humanized antibody. In a humanized antibody all constant domains, including the framework regions are human. The skilled person knows chimeric and humanized antibodies and methods for the production thereof. In a preferred embodiment, the first polypeptide further comprises a light chain constant domain C|_, i.e. the first polypeptide preferably comprises a light chain variable domain V_L and a light chain constant domain C_L, also referred to as full- length light chain. The V_L domain and the C_L domain are preferably arranged from N- terminus to C-terminus. The first polypeptide is preferably a light chain of an antibody. The isoform of the light chain can be κ or λ.

In another preferred embodiment, the second polypeptide further comprises at least one heavy chain constant domain CH- The second polypeptide preferably comprises at least one heavy chain constant domain, selected from C_H , C_H2, C_H3, C_H4 and combinations thereof. In particular the second polypeptide comprises one, two, three or four heavy chain constant domains selected from CH1 , CH2, CH3, C_h4 and combinations thereof. CH is also referred to as first heavy chain constant domain, CH2 also referred to as second heavy chain constant domain, CR3 also referred to as third heavy chain constant domain, and CR4 also referred to as forth heavy chain constant domain. More preferably, the second polypeptide comprises three heavy chain constant domains C_H1 , C_H2 and C_H3, or four heavy chain constant domains C_H1 , C_H2, C_H3, and C_H4.

The antibody molecule or functional fragment thereof is preferably an antibody of an isotype selected from the group consisting of IgG, IgA, IgD, IgE, and IgM. The isoform of the heavy chain can be γ, α, δ, ε, or μ for IgG, IgA, IgD, IgE, or IgM respectively. The antibody is preferably of isotype IgG or an antibody of isotype IgG.

IgG-type antibody molecules preferably comprise up to three heavy chain constant domains. IgM- and IgE-type antibody molecules preferably comprise up to four heavy chain constant domains.

A fragment crystallisable Fc comprises the second and the third heavy chain constant domains C_H2 and C_H3.

The second polypeptide preferably further comprises a hinge region. The hinge region is preferably located between C_H1 und C_H2. A combination of second and third heavy chain constant domains (C_H2 and C_H3) and at least a part of the hinge region is also termed fragment crystallisable Fc'. Fc' fragments are known in the art. The second polypeptide preferably comprises a heavy chain variable domain V_Hl a heavy chain constant domain C_H1 , a hinge region and heavy chain constant domains C_H2 and C_H3, also referred to as full-length heavy chain. The V_H domain, the C_H1 domain, the hinge region, the C_H2 domain and C_H3 domain are preferably arranged from N-terminus to C-terminus. The second polypeptide is preferably a heavy chain of an antibody. The isoform of the heavy chain can be γ, α, δ, ε, or μ.

In a preferred embodiment, the at least one first polypeptide is a light chain of an antibody or/and the at least second polypeptide is a heavy chain of an antibody.

In a particularly preferred embodiment at least two first polypeptides and at least two second polypeptides are displayed on the surface of the host cell. Preferably at least two light chains and at least two heavy chains are displayed on the surface of the host cell.

Particularly preferred is presentation of a full-length antibody, which in particular comprises two light chains comprising a light chain variable domain V_L and a light chain constant domain C_L and two heavy chains comprising a heavy chain variable domain V_H, a heavy chain constant domain C_H1 , a hinge region and heavy chain constant domains C_H2 and C_H3.

The first or/and second polynucleotide can further comprise at least one nucleic acid sequence encoding an affinity tag. The nucleic acid sequence encoding the affinity tag can flank the portion (ii) encoding the recombinant polypeptide to be displayed. The polynucleotide encoding the affinity tag can be separated from portion (ii) by a sequence encoding at least one protease recognition sequence. The at least one protease recognition sequence can be a first protease recognition sequence. The at least one protease recognition sequence can be any protease recognition sequence as described herein. Preferably, the at least one protease recognition sequence is independently selected from factor Xa cleavage site, OmpT cleavage site, and TEV protease cleavage site.

The affinity tag can independently be selected from His₆ and epitopes. In particular, the epitope is recognised by a specific antibody, for example a monoclonal antibody. An example of a suitable epitope is the amino acid sequence PEYFK which is recognized by antibody D i 142 (Spohn et al, 1992).

Furthermore, the polynucleotide can comprise a nucleotide sequence encoding at least one protease recognition sequence. Said nucleotide sequence can be located between portions (ii) and portion (iii). Two recognition sequences can present. The at least one protease recognition sequence can be any protease recognition sequence as described herein. Preferably, the at least one protease recognition sequence is independently selected from factor Xa cleavage site, OmpT cleavage site, and TEV protease cleavage site.

The protease recognition sequence, as used herein, may be a recognition site for an intrinsic protease, i.e. a protease naturally occurring in the host cell, or an externally added protease. For example, the externally added protease may be an IgA protease (cf. EP-A-0 254 090), thrombin or factor X (factor Xa). The intrinsic protease may be e.g. selected from OmpT, OmpK or protease X. The protease may also be TEV.

Step (b)

Step (b) of the method of the present invention refers to culturing the host cell under conditions wherein the first or/and second polynucleotide are expressed and said at least one first polypeptide or/and at least one second polypeptide are displayed on the surface of the host cell. The person skilled in the art knows suitable culture conditions, in particular for the host cell species as described herein. Step (b) of the method according to the present invention is preferably be carried out in a culture medium.

In another aspect of the invention, the at least one first polynucleotide or/and said at least one second polynucleotide are expressed in step (b) in the presence of an agent capable of reducing a disulfide bond between two cy stein residues or/and protecting an SH group of a cystein residue. The agent capable of reducing disulfide groups or/and protecting an SH group of an amino acid residue is preferably β- mercaptoethanol or DDT. The reducing agent, in particular β-mercaptoethanol or DDT, is preferably present in a concentration range of about 2-50 mM, preferably about 5-25 mM, more preferably about 10-15 mM. The reducing agent can be present in the culture medium. The presence of a reducing agent prohibits premature formation of disulfide bonds between and within the polypeptides expressed. In particular the presence of a reducing agents prevents formation of disulfide bridges in the plasma or/and periplasmatic space. Prevention of premature formation of disulfide bonds is particularly important to ensure transport of the polypeptides from the periplasmic space into the outer membrane of the host cell.

The method according to the invention allows for an efficient expression of passenger proteins on the surface of host cells, particularly E. coli or other Gram-negative bacterial cells up to 100 000 or more molecules per cell by using a liquid medium of the following composition: 5 g/l to 20 g/l, preferably about 10 g/l trypton or/and peptone, 2 g/l to 10 g/l, preferably about 5 g/l yeast extract, 5 g/l to 20 g/l, in particular about 10 g/l NaCI and the remaining part water. The medium should possibly contain as little as possible divalent cations, thus preferably Aqua bidest or highly purified water, e.g. Millipore water is used. The pH is preferably adjusted to about 6.0 - 8.0, in particular about 7.0. The liquid medium may contain in addition preferably EDTA in a concentration of 2 μΜ to 20 μΜ, in particular 10 μΜ. Moreover, it contains preferably reducing reagents, such as 2-mercaptoethanol or dithiothreitol (DTT) in a preferred concentration of about 2-50 mM, more preferably about 5- 25 mM, most preferably about 10-15 mM. The reducing reagents favour a non-folded structure of the polypeptide during transport and they prevent premature formation of disulfide bonds. The liquid medium can further contain additional C-sources, preferably glucose, e.g. in an amount of up to 10 g/l, in order to favour secretion i.e. transfer of the passenger to the surrounding medium. For surface display preferably no additional C-source is added. Preferred culture conditions for Gram-negative cells, such as E. coli in the Examples.

If the host cell is a Gram-negative bacterium, the polypeptide synthesized in the cytoplasma can be translocated from the cytoplasm into the periplasmic space by crossing the inner membrane. This can be affected by the signal peptide. During maturation, the signal peptide can be cleaved off. The first or second polypepide fusion to be displayed on the surface of a host cell can therefore comprise an amino acid sequence encoded by the first or/and second nucleic acid components (ii) to (iv), as described herein. Optional step (c)

The method of the present invention can optionally further comprises the step:

(c) culturing or/and incubating said host cell under conditions, wherein the at least one first polypeptide and the at least one second polypeptide form an antibody molecule or a functional fragment thereof, which is displayed on the surface of the host cell. Step (c) is preferably carried out in a medium, particularly in a culture medium or an incubation solution. A functional fragment of an antibody is in particular capable of binding an antigen.

Step (c) is preferably carried out to enable the formation of disulfide bonds. The antibody molecule, or the functional fragment thereof, preferably comprises at least one disulfide bond. The antibody molecule or the functional fragment thereof is preferably formed by at least one disulfide bond.

Disulfide bonds are covalent bonds, which are formed between two thiol groups (SH), particularly of cysteins, by oxidation. Reducing agents are capable of dissolving disulfide bonds. Disulfide bonds can be formed within one polypeptide (intrachain disulfide bonds) or/and between two polypeptides (interchain disulfide bonds). At least one disulfide bond, particularly an intrachain disulfide bond, can be formed within the first polypeptide or/and within the second polypeptide. Intrachain disulfide bonds between two polypeptides can be formed between one first polypeptide and one second polypeptide or between two second polypeptides.

Where the first polypeptide comprises a light chain constant domain C_L and the second polypeptide comprises a heavy chain constant domain C_H1 , at least one disulfide bond can be formed between the light chain constant domain C_L of the first polypeptide and the heavy chain constant domain C_H1 of the second polypeptide. Further, where two second polypeptides comprising a hinge region are displayed, at least one disulfide bond can be formed between the hinge regions of said two second polypeptides.

Step (c) is preferably carried out in a medium which is substantially free of agents capable of reducing a disulfide bond between two cystein residues or/and protecting an SH group of a cy stein residue. Agents capable of reducing a disulfide bond are in particular β-mercaptoethanol or DTT. The medium used is step (c) is preferably substantially free of reducing agents, such as β-mercaptoethanol and DTT. Substantially free is a concentration, which is low enough to allow formation of disulfide bonds. In particular, the concentration of the reducing agent is below 0.1 mM, or below 10 μΜ, or below 1 μΜ. Most preferably, the culture medium or incubation solution is free of such reducing agents.

The reducing agents present in step (b) of the method according to the present invention can be removed by washing the host cells with a medium or incubation solution, which is free of reducing agents. To this end, the host cells obtained in step

(b) are washed before step (c) to remove reducing agents. Host cells are preferably washed at least once, more preferably at least twice after step (b) and before step

(c) . Most preferably, host cell are washed twice, preferably with PBS buffer.

Culturing or/and incubating in step (c) is preferably performed for about 8-24 hours, preferably about 14-18 hours, most preferably about 16 hours. Culturing or/and incubating in step (c) at a temperature in the range of about 4-25X, more preferably of about 4-10°C, most preferably about 4-6°C. Step(c) is preferably performed at 4°C for 16 hours. The incubation is preferably carried out under conditions which are substantially free of reducing agents. Step (c) allows the formation of disulfide bonds, which in turn are involved in the formation of an antibody molecule or the functional fragment thereof, which is displayed on the surface of the host cell.

Antibody molecule

An antibody molecule according to the present invention can be any molecule comprising at least one first polypeptide as described herein and at least one second polypeptide as described herein or the functional fragment thereof. The antibody molecule is preferably an antibody or a functional fragment thereof. In a particularly preferred embodiment, the antibody molecule displayed by the method of the present invention is capable of binding an antigen.

The antibody molecule may be a full-length antibody as described herein. A full- length antibody particularly comprises at least one constant domain in the light or/and heavy chain. A full-length antibody particularly comprises two light chains comprising a light chain variable domain V_L and a light chain constant domain C_L and two heavy chains comprising a heavy chain variable domain V_H, a heavy chain constant domain C_H1 , a hinge region and heavy chain constant domains C_H2 and C_H3. A full-length antibody in particular comprises a fragment crystallisable domain (Fc). Fc fragments are known to a person skilled in the art. The F_c-domain particularly consists of the constant regions of the heavy immunoglobulin chains, in particular the heavy chain constant domains C_H2 and C_H3.

In a preferred embodiment, the antibody molecule or the functional fragment thereof comprises two first polypeptides as described herein and two second polypeptides as described herein, i.e. a heterotetramer. More preferably, the first polypeptide comprises a light chain variable domain V_L and a light chain constant domain C|_, and the second polypeptide comprises a heavy chain variable domain V_H) a heavy chain constant domain C_H , a hinge region and heavy chain constant domains C_H2 and C_H3. Most preferably, the antibody molecule or the functional fragment thereof comprises two light chains of an antibody and two heavy chains of an antibody. A full- length antibody molecule preferably consists of two first polypeptides as defined herein and two second polypeptides as defined herein.

In a preferred embodiment, the antibody molecule or the functional fragment thereof is not a single variable fragment scFv. The skilled person knows scFv fragments. scFv fragments are known in the art and commonly relate to a single polypeptide comprising the light chain variable domain V_L and the heavy chain variable domain V_H. A scFv fragment is an artificial antibody fragment obtained by fusion of both the light chain variable domain V_L and the heavy chain variable domain V_H into one single polypeptide. The light chain variable domain V_L and the heavy chain variable domain V_H are often connected via a flexible linker peptide. Mostly, scFv fragments do not include constant domains.

Preferably, the antibody molecule is not a twin chain variable fragment (tcFv). tcFv comprises a light chain variable domain V_L and a heavy chain variable domain V_H on separate chains. In a preferred embodiment the antibody molecule is a fragment antigen binding Fab. A skilled person knows Fab fragments. A Fab fragment commonly refers to an antibody fragment resulting from papain cleavage of an antibody comprising two polypeptides. Papain cleaves an antibody into a Fab fragment and a Fc fragment. A Fab fragment comprises a first polypeptide consisting of a light chain variable domain VL and a light chain constant domain C_L and a second polypeptide consisting of a heavy chain variable domain V_H and a first heavy chain constant domain C_H1 . The first and second polypeptide of a Fab fragment are preferably linked by at least one disulfide bond between the light chain constant domain C_L and the first heavy chain constant domain C_H1 .

In another preferred embodiment the antibody molecule or the functional fragment thereof is a fragment antigen binding F(ab')₂. A skilled person knows F(ab')₂ fragments. A F(ab')₂ fragment commonly refers an antibody fragment resulting from pepsin cleavage of an antibody comprising four polypeptides. Pepsin cleaves an antibody into a F(ab')₂ fragment and a Fc' fragment. A F(ab')₂ fragment comprises two first polypeptide a light chain variable domain V_L and a light chain constant domain C_L and two second polypeptide consisting of a heavy chain variable domain V_H, a first heavy chain constant domain C_H1 and a hinge region. The first and second polypeptide are preferably linked by at least one disulfide bond between the light chain constant domain C_L and the first heavy chain constant domain C_H1. Additionally, the two second polypeptides are preferably linked by at least one additional disulfide bond between the hinge regions of the second polypeptides.

In another preferred embodiment the antibody molecule or the functional fragment thereof of the present invention is not a single domain antibody. Single chain antibodies are known in the art. A single domain antibody comprises only a polypeptide comprising a heavy chain. Single domain antibodies are found in camelids and sharks and are also referred to as nanobodies. A single domain antibody can comprise a heavy chain variable fragment V_H and optionally at least one heavy chain constant domain C_H. Single domain antibody fragments comprise only the heavy chain variable fragment VH. Single domain antibodies and fragments thereof comprise at least one second polypeptide without comprising a first polypeptide. In a most preferred embodiment, the antibody molecule or the functional fragment thereof of the present invention comprises at least one first polypeptide comprising a light chain variable domain V_L and light chain constant domain C_L and at least one second polypeptide comprising a heavy chain variable fragment V_H, three or four heavy chain constant domains CH and a hinge region. More preferably, the antibody molecule or the functional fragment thereof of the present invention comprises at least two first polypeptides comprising a light chain variable domain V_L and light chain constant domain C_L and at least two second polypeptides comprising a heavy chain variable fragment V_H, three or four heavy chain constant domains C_H and a hinge region.

Autotransporter

Furthermore, the polynucleotide comprises (iii) a portion encoding a transmembrane linker and (iv) a portion encoding the transporter domain of an autotransporter, which are required for the presentation of the passenger polypeptide (ii) on the outer surface of the outer membrane of the host cell.

The transmembrane linker (iii) and the transporter domain (iv) can be homologous or heterologous. In this context "homologous" means that the transmembrane linker (iii) and the transporter domain (iv) are obtained from the same autotransporter protein. In this context "heterologous" means that the transmembrane linker (iii) and the transporter domain (iv) are obtained from the different autotransporter proteins.

The length of the transmembrane linker (iii) is preferably 30-160 amino acids. The transmembrane linker is preferably a transmembrane linker obtained from an autotransporter protein, particularly an AIDA-I or EhaA protein as described herein.

The transmembrane linker domain may be encoded by a nucleic acid portion directly 5' to the autotransporter domain in the polynucleotide construct of the present invention.

The transporter domain as described herein include variants which can e.g. be obtained by altering the amino acid sequence in the loop structures of the β-barrel not participating in the transmembrane portions. Optionally, the polynucleotide coding for the surface loops can be deleted completely. Also within the amphipathic β-sheet conserved amino exchanges, i.e. the exchange of an hydrophilic by another hydrophilic amino acid or/and the exchange of a hydrophobic by another hydrophobic amino acid may take place. Preferably, a variant has a sequence identity of at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% on the amino acid level to the respective native sequence of the autotransporter domain, in particular in the range of the β-sheets.

The sequence of the nucleic acid fusion can have a codon usage adapted to the host cell. In particular, the codon usage of the transmembrane linker sequence, the autotransporter domain or/and the passenger can be adapted to the host cell. More particular, the codon usage of the transmembrane linker sequence or/and the autotransporter domain can be adapted to the host cell. This is can improve expression if the autotransporter domain, particularly the autotransporter domain, is heterologous to the host cell. Optimisation of codon usage does usually not affect the amino acid sequence. The polynucleotide being codon-optimized can have an identity of about 80%, 85%, 90%, 95%, 98% or 99% identity with a natural sequence.

Examples of codon-optimized nucleotide sequences are described in SEQ ID NO: 10, and 13.

AIDA-I

Further, the polynucleotide comprises (iv) a transporter domain of an autotransporter. In the context of the present invention, autodisplay may be the recombinant surface display of proteins or polypeptides by means of an autotransporter in any Gram- negative bacterium.

The transporter domain of the autotransporter according to the invention can be any transporter domain of an autotransporter and is preferably capable of forming a β- barrel structure. A detailed description of the β-barrel structure and preferred examples of β-barrel autotransporters are disclosed in WO97/35022 incorporated herein by reference. Henderson et al. (2004) describes autotransporter proteins which comprise suitable autotransporter domains (for summary, see Table 1 of Henderson et al., 2004). The disclosure of Henderson et al. (2004) is included herein by reference. For example, the transporter domain of the autotransporter may be selected from Ssp (P09489, S. marcescens), Ssp-hi (BAA33455, S. marcescens), Ssp-h2 (BAA1 1383, S. marcescens), PspA (BAA36466, P. fluorescens), PspB (BAA36467, P. fluorescens), Ssa1 (AAA80490, P. haemolytica), SphB1 (CAC44081 , B. pertussis), AspA/NalP (AAN71715, N. meningitidis), VacA (Q48247, H. pylori), AIDA-I (Q03155, E. coli), IcsA (AAA26547, S. flexneri), isL (AAD16954, S. enterica), TibA (AAD41751 , E. co//), Ag43 (P39180, E. coli), ShdA (AAD251 10, S. enterica), AutA (CAB891 17, Λ/. meningitidis), Tsh (I54632, E. co//), SepA (CAC05786, S. flexneri), EspC (AAC44731 , E. co//), EspP (CAA66144, E. co//), Pet (AAC26634, E. co//), Pic (AAD23953, E. co//), SigA (AAF67320, S. flexneri), Sat (AAG30168, E. co//), Vat (AAO21903, E. co//), EpeA (AAL18821 , E. co//), EatA (AA017297, E. co//), Espl (CAC39286, E. co//), EaaA (AAF63237, E. co//), EaaC (AAF63038, E. co//), Pertactin (P14283, B. pertussis), BrkA (AAA51646, B. pertussis), Tef (AAQ82668, B. pertussis), Vag8 (AAC31247, B. pertussis), PmpD (084818, C. trachomatis), Pmp20 (Q9Z812, C. pneumoniae), Pmp21 (Q9Z6U5, C. pneumoniae), lgA1 protease (NP_283693, Λ/. meningitidis), App (CAC14670, Λ/. meningitidis), lgA1 protease (P45386, H. influenzae), Hap (P45387, H. influenzae), rOmpA (P15921 , R hckettsii), rOmpB (Q53047, R. hckettsii), ApeE (AAC38796, S. enterica), EstA (AAB61674, P. aeruginosa), Lip-1 (P40601 , X. luminescens), McaP (AAP97134, M. catarrhalis), BabA (AAC38081 , H. py/or/), SabA (AAD06240, H. py/on), AlpA (CAB05386, H. py/or/), Aae (AAP21063, A actinomycetemcomitans) , NanB (AAG35309, P. haemolytica), and variants of these autotransporters. Given in brackets for each of the exemplary autotransporter proteins are examples of suitable genbank accession numbers and species from which the autotransporter may be obtained. Preferably the transporter domain of the autotransporter is the E. coli AIDA-I protein or a variant thereof, such as e.g. described by Niewert U. , Frey A. , Voss T., Le Bouguen C , Baljer G. , Franke S. , Schmidt MA. The AIDA Autotransporter System is Associated with F18 and Stx2e in Escherichia coli Isolates from Pigs Diagnosed with Edema Disease and Postweaning Diarrhea. Clin. Diagn. Lab. Immunol. 2001 Jan, 8(1 ): 143- 149;9.

Variants of the above indicated autotransporter sequences can e.g. be obtained by altering the amino acid sequence in the loop structures of the β-barrel not participating in the transmembrane portions. Optionally, the nucleic acid portions coding for the surface loops can be deleted completely. Also within the amphipathic β-sheet conserved amino exchanges, i.e. the exchange of a hydrophilic by another hydrophilic amino acid or/and the exchange of a hydrophobic by another hydrophobic amino acid may take place. Preferably, a variant has a sequence identity of at least 70%, at least 90%, at least 95% or at least 98% on the amino acid level to the respective native sequence of the autotransporter domain, in particular in the range of the β-sheets.

The skilled person knows suitable methods to determine the degree of identity of nucleic acid sequences and amino acid sequences. Known algorithms, such as BLAST (for nucleic acids) or P BLAST (for amino acid sequences) may be used. A nucleic acid or polypeptide comprising sequences having at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identity to a given sequence includes fragments of the given nucleic acid or polypeptide.

In a preferred embodiment, the autotransporter domain (iv) is derived from an autotransporter selected from the group consisting of Ssp, Ssp-h1 , Ssp-h2, PspA, PspB, Ssa1 , SphB1 , AspA/NalP, VacA, AIDA-I, IcsA, MisL, TibA, Ag43, ShdA, AutA, Tsh, SepA, EspC, EspP, Pet, Pic, SigA, Sat, Vat, EpeA, EatA, Espl, EaaA, EaaC, Pertactin, BrkA, Tef, Vag8, PmpD, Pmp20, Pmp21 , lgA1 protease, App, Hap, rOmpA, rOmpB, ApeE, EstA, Lip-1 , McaP, BabA, SabA, AlpA, Aae, NanB, and variants thereof.

In another preferred embodiment, the transporter domain (iv) is the transporter domain of an AIDA-I protein, or a variant thereof.

The amino acid of AIDA-I is described in SEQ ID NO:2. The nucleic acid sequence of AIDA-I is described in SEQ ID NO:1 .

The transporter domain (iv) of the AIDA-I protein is particularly encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO:5,

(b) a nucleotide sequence encoding SEQ ID NO:6,

(c) nucleotide sequences comprising a sequence being at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to SEQ ID NO:5 or/and a sequence encoding SEQ ID NO: 6, and (d) nucleotide sequences which encodes the polypeptides encoded by (a), (b) or/and (c) within the scope of the degeneracy of the genetic code.

The transporter domain (iv) of the AIDA-I protein in particular comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO:6, and

(b) sequences which are at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to the sequences of (a).

The transmembrane linker (iii) is preferably is a transmembrane linker from an autotransporter selected from the group consisting of Ssp, Ssp-h1 , Ssp-h2, PspA, PspB, Ssa1 , SphB1 , AspA/NalP, VacA, AIDA-I, IcsA, MisL, TibA, Ag43, ShdA, AutA, Tsh, SepA, EspC, EspP, Pet, Pic, SigA, Sat, Vat, EpeA, EatA, Espl, EaaA, EaaC, Pertactin, BrkA, Tef, Vag8, PmpD, Pmp20, Pmp21 , lgA1 protease, App, Hap, rOmpA, rOmpB, ApeE, EstA, Lip-1 , McaP, BabA, SabA, AlpA, Aae, NanB, and variants thereof.

In the method of the present invention the transmembrane linker (iii) is preferably a transmembrane linker from an AIDA-I protein, or a variant thereof.

In particular, the transmembrane linker (iii) is encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO:3,

(b) a nucleotide sequence encoding SEQ ID NO:4,

(c) nucleotide sequences comprising a sequence being at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to SEQ ID NO: 3 or/and a sequence encoding SEQ ID NO: 4, and

(d) nucleotide sequences which encodes the polypeptides encoded by (a), (b) or/and (c) within the scope of the degeneracy of the genetic code.

In another embodiment, the transmembrane linker (iii) comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO:4, and (b) sequences which are at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to the sequences of (a).

EhaA

In yet another preferred embodiment, the transporter domain (iv) is the transporter domain of an EhaA protein, or a variant thereof.

The EhaA protein of the present invention and a nucleic acid encoding therefore can be obtained from E. coli. The amino acid sequence of the E. coli EhaA protein is in particular described by YP_003498036 (for example, Version YP_003498036.1 , genbank identifier Gl:291281218). This amino acid is described in SEQ ID NO:8. The nucleic acid sequence is described in SEQ ID NO:7. The identity of EhaA and AIDA-I is about 43 % on the nucleic acid level and about 34% on the amino acid.

The expression system as described herein employing the EhaA autotransporter domain is also termed MATE (maximized autotransporter expression) system of pMATE system. Plasmids (in particular expression plasmids comprising the polynucleotide as described herein) to be used in the MATE system are also termed by the prefix "pMATE".

The transporter domain (iv) of the EhaA protein is particularly encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO: 12 or 13,

(b) a nucleotide sequence encoding SEQ ID NO:14,

(c) nucleotide sequences comprising a sequence being at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to SEQ ID NO: 12 or 13 or/and a sequence encoding SEQ ID NO: 14, and

The transporter domain (iv) of the EhaA protein comprises in particular a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO: 14, and (b) sequences which are at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to the sequences of (a).

In one embodiment of the invention, the transmembrane linker (iii) is a transmembrane linker from an EhaA protein, or a variant thereof.

The transmembrane linker (iii) is preferably a transmembrane linker from an EhaA protein or a variant thereof.

The transmembrane linker (iii) is in particular encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO: 9 or 10,

(b) a nucleotide sequence encoding SEQ ID NO: 1 1 ,

(c) nucleotide sequences comprising a sequence being at least 70%, at least 80%, at least 90%, at least 95%, at least 98%, or at least 99% identical to SEQ ID NO: 9 or 10 or/and a sequence encoding SEQ ID NO: 1 1 , and

Particularly, the transmembrane linker (iii) comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO: 1 1 , and

Host cell

As described herein, step (a) of the methods of the present invention refers to the provision of a host cell. The host cell used in the method of the present invention is preferably a bacterium, more preferably a Gram-negative bacterium.

The Gram-negative bacterium can be selected from from the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., Cupriavidus spp., Rhodobacter spp., Acinetobacter spp., Gluconobacter spp.,

Gluconacetobacter spp., Acidomonas spp., Acetobacter spp., Paracoccous spp., Rhizobium spp., Xanthomonas spp., Halomonas spp., Vanovorax spp., Alcanivorax spp., Sphingomonas spp., and Mannomonas spp.

Another preferred selection of the Gram-negative bacterium is the selection from the group consisting of E. coli, Salmonella spp. , Zymomonas spp., Zymobacter spp., Pseudomonas spp., and Halomonas spp.

Another preferred selection of the Gram-negative bacterium is the selection from Salmonella spp., Zymobacter spp., and Pseudomonas spp.

In another preferred embodiment, the host cell is E.coli.

In a further preferred embodiment, the host cell is not E.coli.

If the autotransporter is selected from the group consisting of Ssp, Ssp-h1 , Ssp-h2, PspA, PspB, Ssa1 , SphB1 , AspA/NalP, VacA, AIDA-I, IcsA, MisL, TibA, Ag43, ShdA, AutA, Tsh, SepA, EspC, EspP, Pet, Pic, SigA, Sat, Vat, EpeA, EatA, Espl, EaaA, EaaC, Pertactin, BrkA, Tef, Vag8, PmpD, Pmp20, Pmp21 , IgA1 protease, App, Hap, rOmpA, rOmpB, ApeE, EstA, Lip-1 , McaP, BabA, SabA, AlpA, Aae, NanB, and variants thereof, preferably AIDA-I or a variant thereof, the Gram-negative bacterium is preferably E. coli.

If the autotransporter is the transporter domain of an EhaA protein, or a variant thereof, the host cell is preferably selected from the group consisting of E. coli, Salmonella spp. , Zymomonas spp., Zymobacter spp., Pseudomonas spp., Cupnavidus spp., Rhodobacter spp., Acinetobacter spp., Gluconobacter spp., Gluconacetobacter spp., Acidomonas spp., Acetobacter spp. , Paracoccous spp., Rhizobium spp., Xanthomonas spp., Halomonas spp., Variovorax spp., Alcanivorax spp., Sphingomonas spp., and Mannomonas spp., more preferably the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., and Halomonas spp., even more preferably Salmonella spp., Zymobacter spp., and Pseudomonas spp.. Here, the host cell is preferably not E. coli. In this case, the host cell is heterologous to the transporter domain of the EhaA protein. In one aspect, the transporter domain (iv) of the autotransporter can be heterologous with respect to the host cell.

A preferred Salmonella species is Salmonella entenca. A preferred Zymomonas species is Zymomonas mobilis. More preferred is Zymomonas mobilis strain DSM 3580. A preferred Zymobacter species is Zymobacter palmae. A preferred Pseudomonas species is Pseudomonas putida or Pseudomonas fluorescens. A preferred Cupriavidus species is Cupriavidus necator or Cupriavidus metallidurans. A preferred Rhodobacter species is Rhodobacter capsulatus. A preferred Acinetobacter species is Acinetobacter baylyi ADP1 . A preferred Gluconobacter species is Gluconobacter oxydans. A preferred Acetobacter species is Acetobacter xylinum. A preferred Paracoccous species is Paracoccous denitrificans. A preferred Rhizobium species is Rhizobium me I i lot i. A preferred Xanthomonas species is Xanthomonas campesths. A preferred Halomonas species is H. mehdiana, and H. elongate. A preferred Variovorax is Variovoras paradoxus.

Expression in bacteria is the method of choice for the commercial production of pharmaceutical and industrial proteins. E. coli is a versatile lab organism for the expression of recombinant proteins, their investigation and in some issues for even their production in preparative scales. For such purposes, a wide variety of vectors e.g. plasmids, suitable mutants and protocols are available. For being used in industrial applications or crude biotechnological production processes E. coli has several disadvantages. It usually does not grow to high cell densities in fermentation processes and needs high concentrations of glucose to gain growth energy. Being a natural gut inhabitant, it is not used to crude environments as water or soil, rather sensible to harsh treatments and is not resistant to organic solvents. The present invention can also be used for surface display of recombinant proteins in host organisms different from E. coli, e.g. the natural soil bacterium Pseudomonas (putida), resistant to a variety of organic solvents, organisms that exploit other sources than glucose to gain growth energy e.g. Rhodobacter (light) or Cuphavidus necator (oxyhydrogen). For this purpose an optimized second generation surface display system is used, the MATE (maximized autotransporter expression) system, shown to be superior in E. coli and proven to be a versatile tool for the surface display of recombinant proteins in a wide variety of gram negative bacteria different from E. coll.

The transporter domain of EhaA can be successfully used for surface-display of recombinant passenger proteins in a host cell heterologous to E.coli. Examples of such heterologous cells are described herein. These cells expressing a construct of the present invention on the cell surface do not form aggregates or a biofilm. Aggregation or biofilm formation would make such cells unsuitable for biotechnological applications (for example, suspension culture).

By the heterologous expression, autodisplay of first and second polypeptides on a bacterial cell surface can be employed in species being more suitable in biotechnological applications than E. coli. Such cells include the cells described herein but are not limited thereto.

In a further preferred embodiment the host cell is OmpT deficient. OmpT is located on the outer membrane of gram-negative bacteria and are capable of cleaving the displayed polypeptide through an Omptin recognition sequence in the transmembrane linker (iii). The omptin protease of E.coli is OmpT. Where E.coli is used as a host cell, the host cell is preferably OmpT deficient (AompT). An example is E.coli K12 UT5600.

Antibody molecules or the functional fragment thereof displayed by these host cells can be released into the medium by cleavage of the OmpT recognition site in the transmembrane linker (iii) by OmpT.

A further aspect of the present invention is a recombinant vector comprising at least one first polynucleotide or/and at least one second polynucleotide as defined herein, operatively linked to an expression control sequence.

Another aspect of the present invention is a recombinant host cell comprising the first or/and second polynucleotide as described herein. The host cell may be any host cell as described herein. The recombinant host cell particularly comprises a first polynucleotide encoding a first polypeptide and a second polynucleotide encoding a second polypeptide comprising a heavy chain variable domain VH, as defined herein. The host cell is in particular capable of displaying the first and second polypeptide on the surface. The recombinant host cell preferably displays an antibody molecule, or a functional fragment thereof, on the surface of said host cell. The antibody molecule, or the functional fragment thereof, displayed, is particularly capable of binding an antigen.

Yet another aspect of the present invention relates to a recombinant host cell comprising at least one recombinant vector as described herein. The host cell may be any host cell as described herein. The recombinant host cell particularly comprises a first polynucleotide encoding a first polypeptide and a second polynucleotide encoding a second polypeptide comprising a heavy chain variable domain V_H, as defined herein. The host cell is in particular capable of displaying the first and second polypeptide on the surface. The recombinant host cell preferably displays an antibody molecule, or a functional fragment thereof, on the surface of said host cell. The antibody molecule, or the functional fragment thereof, displayed, is particularly capable of binding an antigen.

Yet another aspect of the present invention relates to a recombinant host cell comprising the first or/and second polypeptide as described herein. The recombinant host cell particularly comprises at least one first polypeptide and at least one second polypeptide comprising a heavy chain variable domain V_H, as defined herein. The host cell in particular displays the first or/and second polypeptide on the surface. The recombinant host cell preferably displays an antibody molecule, or a functional fragment thereof, on the surface of said host cell. The antibody molecule, or the functional fragment thereof, displayed, is particularly capable of binding an antigen.

The polypeptide fusion displayed on the cell surface in particular comprises:

(I) a portion comprising the first polypeptide or second polypeptide to be displayed,

(II) a portion comprising a transmembrane linker, and

(III) a portion comprising the transporter domain of an autotransporter.

The portion (I) is also termed "passenger" or "light chain" or "heavy chain", as described herein. The portions (!) to (II!) of the first or/and second polypeptide (polypeptide fusion) displayed by the host cell of the present invention are encoded in particular by the components (ii), (iii), and (iv) of the polynucleotide, as described herein. The first and second polypeptide preferably form an antibody molecule or a functional fragment thereof.

The antibody molecule or functional fragment thereof displayed on the cell surface in particular comprises:

(a) at least one first polypeptide comprising a

(II) a portion comprising a transmembrane linker, and

(III) a portion comprising the transporter domain of an autotransporter, and

(b) at least one second polypeptide comprising

(II) a portion comprising a transmembrane linker, and

(III) a portion comprising the transporter domain of an autotransporter.

The at least first polypeptide fusion particularly comprises a light chain of an antibody or/and the at least second polypeptide fusion particularly comprises a heavy chain of an antibody. The first and second polypeptides preferably form a functional antibody molecule or functional fragment thereof by disulfide bonds as described herein.

Yet another aspect of the present invention is a membrane preparation comprising at least one first or/and at least one second polypeptide as described herein. The membrane preparation of the present invention may comprise membrane particles, as described herein. The membrane preparation may be obtained from a host cell as described herein. The first or/and second polypeptide of the may be any recombinant polypeptide or fusion polypeptide as described herein.

A further aspect of the present invention relates to a membrane preparation which is derived from a host cell as described herein. The membrane preparation particularly comprises the outer membrane of the host cell. Preferably, the membrane preparation as described herein comprises the at least one first polypeptide or/and the at least one second polypeptide as defined herein. Particularly, the membrane preparation according to the present invention comprises an antibody molecule or a functional fragment thereof as described herein. The antibody molecule or the functional fragment thereof is preferably presented on the surface of the membrane preparation.

The method of the present invention may comprise preparing a membrane preparation from the cell obtained in step (b) or/and step (c). The membrane preparation may comprise membrane particles. The membrane particles may be membrane vesicles. Preferred membrane particles are outer membrane particles. In particular the method of the present invention may comprise preparing outer membrane particles of cells displaying a recombinant polypeptide on the surface, e.g. of Gram-negative bacterial cells. The person skilled in the art knows suitable conditions (Park et al., 2015). Outer membrane particles from a host cell as described herein may be performed by a method comprising the steps:

1 . treating the host cell with a hydrolase (such as lysozyme) and optionally with a DNAse. This enzymatic treatment may be performed at room temperature. The hydrolase hydrolyses the cell wall within the periplasmic space. The cell wall comprises peptidoglycans to be hydrolyzed.

2. optionally solubilizing the preparation of (a) with a tenside, such as Triton X- 100, or/and with sarcosine, followed by optional centrifugation of cell debris. The thus obtained preparation of outer membrane particles may be centrifuged, washed and resuspended.

The diameter of the membrane particles may be in the range of 1 nm to 1000 nm, in the range of 50 nm to 500 nm, in the range of 75 to 200 nm, or in the range of 90 to 120 nm. At least 80%, at least 90%, at least 95 %, or at least 98% of the membrane particles may have a diameter in a range selected from the ranges described herein.

In a host cell being a Gram-negative bacterium, after translocation, the recombinant passenger remains attached to the surface of the outer membrane by the β-barrel, which is serving as an anchor within the outer membrane. Due to the controlled integration of the β-barre! within the outer membrane, the C terminal part of the β- barrel is directed to the inner side of the outer membrane, whereas the N-terminal part of the linker, to which the recombinant passenger protein is covalently bound, is directed to the outer surface of the outer membrane, i.e. the environment. The recombinant passenger protein has an oriented location after transport, namely it is directed to the cellular surface. The recombinant passenger protein has the identical orientation as the lipopolysaccharide (LPS) layer which may be present in the outer membrane.

Membrane particles of the present invention prepared from the host cell of the present invention comprise the at least one first or/and at least one second polypeptide at the surface directed to the environment. In contrast to the inner membrane which is a unit membrane, the outer membrane of Gram-negative bacteria, in particular E. coli, is asymmetric. The outer membrane may comprise an inner layer comprising phospholipids and an outer layer comprising LPS. LPS is hydrophilic and may contain several negative charges. By using outer membrane particles with anchored passenger proteins by a β-barrel for the coating of carriers, the outer side of the outer membrane, in particular the LPS side will be directed to the surface distal to the carrier. As a consequence the recombinant protein or a domain thereof, which are integrated in the outer membrane, will be directed to the surface distal to the carrier as well. The core part of the membrane particles may stabilize the interaction of the outer membrane layer obtained by applying outer membrane particles to the carrier by hydrophobic interactions and may contain lipoproteins or peptidoglycans.

Yet another aspect of the present invention is the use of a membrane preparation comprising at least one first or/and at least one second polypeptide in the manufacture of a carrier comprising a recombinant polypeptide.

The membrane preparation of the present invention may be employed for coating a carrier. The carrier may comprise a membrane preparation of the present invention, as described herein.

The carrier may comprise a hydrophobic surface. The hydrophobic surface may have a contact angle of more than 90°. A increasing surface angle of more than 30° indicates a gradually increasing hydrophobicity of a surface. In the present context, a hydrophobic surface may have a contact angle of at least 40°. The surface preferably has a hydrophobicity described by a contact angle of at least 40°, at least 50°, at least 60°, at least 65°, at least 70°. Contact angles are preferably determined by the sessile drop method. The sessile drop method is a standard method for determining contact angles. Measurements may be performed with a contact angle goniometer. Preferred contact angles of the hydrophobic surface are in a range of 40° to 100°, 50° to 90°, or 60° to 80°.

The surface of the carrier may be a metal surface. A suitable metal surface has a contact angle e.g. in the range of 50° to 80°. A suitable metal may be selected from gold, silver, titanium, aluminium and alloys such as brass. A preferred surface is a gold surface. The gold surface may be employed as it is. An untreated gold surface has a hydrophobicity suitable for the carrier as described herein. A treatment of the gold surface with thiolated hydrocarbons or hydrocarbons with functional groups such as carboxylic acids or hydroxyl groups is not required.

Another preferred surface of the carrier comprises a polymer, for instance a surface usually employed in disposable materials for use in biochemical or/and medical science. The polymer may be an artificial polymer. Examples of artificial polymers include a polymer selected from polystyrenes, polypropylenes, and polycarbonates. The polystyrene may be produced from [2,2]paracyclophane monomers. Polystyrene surfaces may be treated with oxygene plasma introducing OH or/and methylene groups in order to modify the hydrophobicity. Examples of such modified surfaces include Maxi-sorp, Medi-sorp, Multi-sorp, and Poly-sorp surfaces. Another suitable polystyrene surface is Parylene N produced from [2,2]paracyclophane monomers. Yet another suitable surface is Parylene A [Poly(monoamino-p-xylene)]. Especially suitable are surfaces comprising a polymer having a hydrophobicity described by a contact angle of at least 50°. Suitable surfaces are selected from polystyrene, Parylene A, Parylene N, Maxi-sorp, Medi-sorp, Multi-sorp, and Poly-sorp. Preferred surfaces are selected from polystyrene, Parylene A, Parylene N, Maxi-sorp, Medi- sorp, and Poly-sorp. The surface may comprise a natural polymer. Suitable natural polymers include polybutyrate and cellulose and derivatives thereof. A further surface is provided by latex particles, in particular latex beads.

Yet another surface is provided by C18-modified particles, in particular C18-modified monolithic silica particles. C18 refers to an alkyl group comprising 18 carbon atoms. C18-modified particles are known in the art.

Yet another suitable surface is a glass surface.

The surface may be modified is order to adjust the hydrophobicity. Modification may be performed by chemical treatment (i.e. by oxygen plasma), physical treatment (e.g. by laser irradiation or/and radioactive irradiation), or by mechanical treatment.

Still a further aspect of the present invention relates to an antibody molecule, which is displayed on a host cell as described herein or present in a membrane preparation as described herein. The antibody molecule, or the functional fragment thereof, particularly comprises at least one first polypeptide comprising at least a light chain variable domain V_L and at least one second polypeptide comprising at least a heavy chain variable domain V_H. The first and second polypeptide can be any fusion polypeptide as described herein. In a particular embodiment, the signal peptide (i) may be cleaved off, such that the fusion polypeptide particularly comprises portions (ii)-(iv) as described herein.

The method according to the present invention, the host cells and membrane preparations according to the present invention can be used for a variety of different applications.

One aspect of the present invention relates to the use of a host cell according the present invention, a membrane preparation according to the present invention, or an antibody according to the present invention for screening antibody molecules or functional fragments thereof. The antibody molecules are particularly full-length antibodies. Screening can be carried out be high-throughput methods, such as FACS (Fluorescence-activated cell sorting). Screening of antibodies or antibody fragments can be achieved in a particular embodiment by varying the amino acid sequence of the first or second polypeptide, as described herein, via site-specific or random mutagenesis and by testing variant carrying cells or membrane preparations or libraries containing variant carrying cells or membrane preparations thereof for their affinity to an antigen of interest with the help of suitable screening methods, in particular high throughput screening methods, such as flow cytometry.

A further aspect of the present invention is a method for screening an antibody library expressed on a plurality of host cells, said method comprising the steps:

(a) producing a plurality of host cells according to a method as described herein, each host cell expressing one of a plurality of different antibodies, or a functional fragment thereof, and

(b) selecting a host cell expressing an antibody, or a functional fragment thereof, on the surface, by the specificity of the antibody, or the functional fragment thereof.

Yet another embodiment is a method for producing an antibody library expressed on a plurality of host cells, wherein said method comprises the production of a plurality of host cells according to the method of the present invention, each host cell expressing one of a plurality of different antibodies, or functional fragments thereof. The expressed antibodies or functional fragments thereof are in particular displayed on the surface of the plurality of host cells.

In yet another preferred embodiment libraries of variants of an antibody molecules or functional fragments thereof, as described herein, are examined in view of the role of defined amino acids for antigen binding or/and affinity.

In general, these particular embodiments concern the production of variants of antibody molecules or functional fragments thereof and the production of antibody molecules or functional fragments thereof which are screened in view of a certain characteristic, i.e. one or optionally several variants fulfilling this desired (predetermined) characteristic are selected. By selecting the variant the cell is selected, too, and carries the polynucleotide coding the variant. Thus, at the same time both the amino acid sequence and the structural information of the variant can be determined via the nucleic acid sequence. The characteristics in question are for example antigen affinity, antigen binding and antigen stability.

Furthermore, the cell or the membrane preparation of the invention may be used for a directed evolution procedure, e.g. for the development of new antibody molecules or fragments thereof.

Still a further embodiment is a method for producing an antibody, or a functional fragment thereof, displayed on the surface of a host cell, comprising the method of the present invention. The antibody, or the functional fragment thereof, is a particularly an aglycosylated antibody, or a functional fragment thereof.

Another embodiment relates to the use of a host cell as described herein according for producing an aglycosylated antibody, or a functional fragment thereof. In this context the host cell is preferably a bacterial host cell as described herein. Bacterial cells lack the enzymes for posttranslational glycosylation of proteins. Aglycosylated antibodies have been described to have useful characteristics (Jung et al., 201 1 , Leabman et al., 2013). Incorporation of artificial modifications into the "naked" antibody for example may influence the immune response of the antibody. Furthermore, proteins displayed in the surface are stable and easy to isolate, e.g. by centrifugation. The bacterial host cell could thus be used as matrix for chemical modifications of the "naked" antibody displayed on its surface.

Another aspect of the present invention relates to the use of a host cell as described herein, a membrane preparation as described herein, or an antibody as described herein for bacterial tumor targeting. In this context, the host cell preferably displays an antibody molecule or a fragment thereof on the surface of the host cell. Bacterial tumor targeting has been described for targeted infection of tumors, without infection of healthy cells (Malmgren und Flanigan, 1955; Pawelek et al., 1997; Lemmon et al., 1997; Patyar et al., 2010; Zhao et al 2006; Yazawa et al., 2000).

According to the present invention, E.coli cells, which display an antibody directed against a tumor-specific antigen may lead to tumor-specific targeting of the bacterial host to the tumor by the specific affinity of the antibody displayed to the tumor antigen. The bacterial cell could for example then induce death of the tumor cell. This could for example be achieved by co-expression of a peptidic cytostatic, which could be specifically transported to the desired site of action by the bacterial cell. Further, co-expression of an enzyme, which actives a prodrug, i.e. a prodrugs converting enzyme could provide to a local and selective cytotoxic effect.

Sequence Listing

The sequence listing includes the following sequences.

SEQ ID NO: 1 represents the wild type DNA sequence of the E.coli AIDA-I autotransporter.

SEQ ID NO:2 represents the protein sequence of the E.coli AIDA-I autotransporter

SEQ ID NO:3 represents the wild type DNA sequence of the E.coli AIDA-I linker domain.

SEQ ID NO:4 represents the protein sequence of the E.coli AIDA-I linker domain.

SEQ ID NO:5 represents the wild type DNA sequence of the E.coli AIDA-I transporter domain.

SEQ ID NO:6 represents the protein sequence of the E.coli AIDA-I transporter domain.

SEQ ID NO:7 represents the wild type DNA sequence of the E.coli EhaA autotransporter.

SEQ ID NO:8 comprises the protein sequence of the E.coli, EhaA autotransporter.

SEQ ID NO:9 represents the wild type DNA sequence of the E.coli EhaA linker domain.

SEQ ID NO: 10 represents the cod on optimized DNA sequence of the EhaA linker domain.

SEQ ID NO:1 1 represents the protein sequence of the E.coli EhaA linker domain. SEQ ID NO: 12 represents the wild type DNA sequence of the E.coli EhaA transporter domain.

SEQ ID NO:13 represents the codon optimized DNA sequence of the E.coli EhaA transporter domain.

SEQ ID NO: 14 represents the protein sequence of the E.coli EhaA transporter domain.

SEQ ID NO 15 represents the DNA sequence encoding the fusion polypeptide comprising the signal peptide, the light chain and the AIDA-I autotransporter (linker domain and autotransporter domain).

SEQ ID NO 16 represents the protein sequence of the fusion polypeptide comprising the signal peptide, the light chain and the AIDA-I autotransporter (linker domain and autotransporter domain).

SEQ ID NO 17 represents the DNA sequence encoding the fusion polypeptide comprising the signal peptide, the heavy chain and the AIDA-I autotransporter (linker domain and autotransporter domain).

SEQ ID NO 18 represents the protein sequence of the fusion polypeptide comprising the signal peptide, the heavy chain and the AIDA-I autotransporter (linker domain and autotransporter domain).

SEQ ID NO 19 represents the DNA sequence encoding the fusion polypeptide comprising the signal peptide, the light chain and the EhaA autotransporter (linker domain and autotransporter domain).

SEQ ID NO 20 represents the protein sequence of the fusion polypeptide comprising the signal peptide, the light chain and the EhaA autotransporter (linker domain and autotransporter domain).

SEQ ID NO 21 represents the DNA sequence encoding the fusion polypeptide comprising the signal peptide, the heavy chain and the EhaA autotransporter (linker domain and autotransporter domain).

SEQ ID NO 22 represents the protein sequence of the fusion polypeptide comprising the signal peptide, the heavy chain and the EhaA autotransporter (linker domain and autotransporter domain). Brief description of the figures

The present invention shall be further illustrated by the following figures and examples.

Figure 1 - Fusion proteins according to the present invention.

Fusion proteins according to the present invention comprise the heavy or light chain sequence of the chimeric antibody T84.66 embraced by an N-terminal signal peptide and a C-terminal linker region which connected the antibody chain to the β-barrel.

The structure of the autotransporter fusion proteins with the heavy (A) or the light chain (B) of anti-CEA antibody T84.66 as passenger domains is illustrated. The cloning sites enveloping the antibody chains are expanded and the DNA and amino acid sequences of these regions are given. Restriction sites of relevant restriction enzymes are underlined. The white boxes represent the antibody chains which are embraced by the N-terminal signal peptide (SP) shown in dark grey and the C- terminal linker region which is depicted in light grey.

Figure 2 - Full amino acid sequences of the chimeric antibody

The amino acid sequences of anti-CEA antibody T84.66 variable domains were taken as translations from the NCBI Nucleotide Database (CAA36980.1 and CAA36979.1 ). Amino acids contributing to the immunoglobulin signal peptide were identified by a Kyte-Doolittle plot (Kyte and Doolittle 1982) and eliminated. The amino acid sequence of the human lgG1 heavy chain constant region (UniProt no.: P01857) was attached to the sequence of the heavy chain variable domain. The variable domain of the light chain of T84.66 was expanded by the constant region of human κ light chain (UniProt no.: P01834).

Figure 3 - Expression and Surface Display of IgG heavy and light chains of the antibody T₈4.66 in E.coli

Outer membrane fractions of E. coli featuring the plasmids coding for the AIDA-I autotransporter fusion proteins with the heavy or light chain of anti-CEA T84.66 as passengers were prepared (A), E. coli UT5600 (DE3) pWWOOS with the genetic information for the heavy chain autotransporter fusion protein is termed ^' H T84_.66- · E. coli UT5600 (DE3) pWW004 carrying the plasmid coding for the light chain autotransporter fusion protein is described as Ί_ Τ84 66' · Outer membrane preparation of E. coli carrying the genetic information for both immunoglobulin chains was also performed (Ή _Ts4.66 + L _T84.6e (B).

Protein fractions were separated by SDS-PAGE (7.5% acrylamide gel) and stained with Coomassie Brilliant Blue G250. Addition of IPTG for induction of protein expression is marked by JPTG +". Non-induced samples are indicated by JPTG -". Digestion with Proteinase K is marked by„Prot K +", while undigested samples are labelled "Prot K -". As a control, the outer membrane preparation of the host strain without a plasmid was also analyzed ("UT5600 (DE3)"). The apparent molecular weights of the fusion proteins are indicated by asteriskes. The bands of OmpF and OmpA/C, which served as control for cell integrity in this experiment, are marked with arrows.

Figure 4 - Surface Display of IgG chains of the antibody T₈ .66 on E. coli

E. coli cells displaying both the heavy and the light chain (H _T84.66 + L T84.66) and cells presenting either the separated heavy chain (H T84.66) or the light chain (L _{T84 66}) of anti-CEA antibody T84.66 were labelled by treatment with a FITC-conjugated polyclonal anti-human antibody. Results of flow cytometry are shown as histograms plotting the fluorescence intensity (FITC-A) on a biexponential x-axis against the cell count. E. coli displaying one (H _T84 ee / L _{Ύ8Λ 66}) or both (H _T84 66 + L _Ts4 ee) antibody chains are depicted in dark grey. Cells of the same strains which have been submitted to proteinase K treatment prior to application of the anti-human antibody are shown in medium grey. As a control, E. coli cells displaying an unrelated protein were labelled and analyzed. The control cells are depicted in light grey histograms.

Figure 5 - Binding of IgG chains presented on the surface to antigen CEA The ability of surface displayed antibody T84.66 to bind the antigen CEA was investigated via flow cytometry. E. coli cells displaying both the heavy and the light chain (H _T84.66 + L Τ84.6β) and cells presenting either the separated heavy chain (H Τ84.6β) or the light chain (L T84.66) °f anti-CEA antibody T84.66 were incubated with CEA obtained from SW-403 cells. The binding of the antigen was detected by subsequent treatment with a commercial mouse anti-CEA antibody (clone COL-1 ) and a secondary Dyl_ight633-conjugated anti-mouse antibody. Samples from E. coli displaying the immunoglobulin chains are shown in dark grey. As a control, cells from the same strains were digested with proteinase K prior to the binding assay. Those digested cells are shown in medium grey. Cells of E. coli UT5600 (DE3) pKP006 displaying an unrelated protein (shown in light grey) served as an additional control. The histograms depict the fluorescence intensity (FITC-A) plotted on a biexponential x-axis against the cell count.

Figure 6 - Experimental set-up of co-immunoprecipitation

A co-immunoprecipitation experiment was designed as depicted in this illustration to demonstrate the assembly of the immunoglobulin chains on the surface of E. coli. Magnetic beads coated with protein A were used. Protein A binds the F_c-region of human antibodies with high affinity. The F_c-region consists of the constant domains of the heavy immunoglobulin chain. The light antibody chain does not contribute to the Fc-region. The protein A presenting beads were incubated with outer membrane preparations from E. coli displaying the heavy (H T84.66), the light (L _Τ84 6β) or both antibody chains (H _T84.66 ⁺ L T84.66) - Afterwards, the supernatants were collected as "unbound" fractions and the beads were washed and e luted. The e luted proteins represent the "bound" fraction. For samples containing only the heavy chain, the heavy chain was expected to be found in the bound fraction (H Τ84.66) · In case of the separated light chain, no antibody chains are supposed to be present in the bound fraction, because the light chain cannot interact with protein A (L T84.66) - For the samples containing both the heavy and the light chain, both immunoglobulin chains are expected to be found in the bound fraction. If the heavy chain interacts with the light chain, the light chain will be precipitated along with the heavy chain on the beads (H _T84.66 + L T84.66) - Figure 7 - Assembly of IgG chains on the surface of E. cols

Outer membrane preparations of E. coli UT5600 (DE3) pWW003 pWW004 (lane 1 , H T84.66 + L T84.66) , E. coli UT5600 (DE3) pWW003 (lane 2, H _T84 ₆₆) , E. coli UT5600 (DE3) pWW004 (lane 3, L _T84.6₆) and E. coli UT5600 (DE3) (lane 4) as control were submitted to co-immunoprecipitation. They were incubated with protein A coated magnetic beads. The supernatant was collected as "unbound" fraction and the proteins eluted from the beads were termed "bound" fraction. After co- immunoprecipitation, protein fractions were separated by SDS-PAGE (7.5 % acrylamide gel). Proteins were detected by staining with Coomassie Brilliant Blue G250 (a) or Western Blotting (b) with an anti-human antibody. The apparent molecular weight of the antibody chain fusion proteins is indicated by arrows.

Figure 8 - Expression and Surface Display of IgG heavy and light chains of the antibody T₈4.66 in E.coli using the MATE autotransporter system.

SDS-PAGE of outer membrane isolations to examine the expression and co- expression of the MATE autotransporter fusion proteins with the heavy and the light chain of T84.66 as a passenger.

Outer membrane fractions of E. coli featuring the plasmids coding for the EhaA autotransporter fusion proteins with the heavy or light chain of anti-CEA T84.66 as passengers were prepared. E. coli UT5600 (DE3) modified pPQ33 (pPQ33 is described in Quel et al., 2016) with the genetic information for the heavy chain EhaA autotransporter fusion protein is termed TH Τ84 66^' · E. coli UT5600 (DE3) modified pPQ29 (pPQ 29 is described in Quel et al., 2016) carrying the plasmid coding for the light chain autotransporter fusion protein is described as Ί_ Τ84.66- · Outer membrane preparation of E. coli carrying the genetic information for both immunoglobulin chains was also performed (Ή _T84.66 + L Τ84.66^' )·

Protein fractions were separated by SDS-PAGE (10% acrylamide gel) and stained with Coomassie Brilliant Blue G250. Addition of L-Arabinose for induction of protein expression is marked by „Arabinose +". Non-induced samples are indicated by „ Arabinose -". As a control, the outer membrane preparation of the host strain without a plasmid was also analyzed ("UT5600 (DE3)"). The apparent molecular weights of the fusion proteins are indicated by asteriskes. ^* = H Τ84 66 with MATE (120 kDa), ^** = L _T84.66 with MATE (77 kDa).

Examples

The present invention is supported by the following examples.

The anti-CEA antibody T84.66 was used exemplarily as the antibody molecule to be displayed by the method of the present invention. The anti-CEA antibody T84.66 used is a humanized, monoclonal antibody (mAb) of subtype lgG1 , which is directed against the carcinoembryonal antigen (CEA), a glycoprotein, and which serves as tumor marker. The AIDA-I autotransporter system with AIDA-I as autotransporter and the MATE system with EhaA as autotransporter were exemplarily tested. The heavy chain (HT84.66) and light chain (LT84.66) passengers are separately cloned into two vectors as autotransporter fusion proteins. The heavy and light chains are displayed on the surface on E.coli and assemble into a functional antibody molecule after co- transformation of both vectors into the same bacterial cell.

Example 1 : Surface display of antibodies with the AlDA-l autotransporter

Materials and Methods Media and buffers

Luria-Bertani (LB) medium consisted of 10 g/L tryptone/peptone, 5 g/L yeast extract and 10 g/L NaCI and was adjusted to pH 7.0. Solid media were prepared by addition of 1.5 % (w/v) agar. For standard growth conditions, LB medium was supplemented with 10 mM β-mercaptoethanol, 10 μΜ ethylenediaminetetraacetic acid (EDTA) and 50 mg/L carbenicillin and/or 15 mg/L kanamycin, depending on the antibiotic resistance of the respective plasmid. Phosphate buffered saline (PBS) for bacterial applications was composed of 137 mM NaCI, 2.7 mM KCI, 10 mM Na₂HP0₄, 2 mM KH₂P0₄. Sodium dodecylsulfate (SDS) sample buffer contained 100 nM Tris-HCI (pH 6.8), 4 % SDS, 0.2 % bromophenol blue, 20 % glycerol and 30 mg/mL dithiothreitol (DTT). Western Blot transfer buffer featured 25 mM Tris base, 192 mM glycine and 20 % (v/v) methanol and pH 8.3-8.4. Western Blot blocking solution contained 5 % (w/v) milk powder and 0.5 % (w/v) Tween20 in PBS. Dulbecco^'s Modified Eagle Medium with high glucose concentration (4.5 g/l) (DMEM, obtained from Gibco by Life Technologies) was supplemented with 2 mM L-glutamate and 10 % (v/v) fetal calf serum (FCS, obtained from Gibco by Life Technologies). Dulbecco^'s phosphate buffered saline (Dulbecco^'s PBS) for mammalian cells featured 137 mM NaCI, 2.7 mM KCI, 8.1 mM Na₂HP0₄ and 1.5 mM KH₂P0₄ and pH 7.4. Radioimmunoprecipitation assay buffer (RIPA buffer) was used to lyse mammalian cells. RIPA buffer contained 25 mM Tris-HCI pH 8.0, 137 mM NaCI, 10 % (v/v) glycerol, 0.1 % (v/v) SDS, 0.5 % (v/v) sodium deoxycholate (Na-DOC), 1 % (v/v) octylphenoxypolyethoxyethanol (IEGPAL) and 20 mM sodium pyrophosphate. Co-immunoprecipitation binding buffer consisted of 25 mM Tris-HCI (pH 7.4), 1 mM EDTA, 150 mM NaCI, 1 % (v/v) NP40 and 5 % (v/v) glycerol.

DMA Techniques

Genes were designed according to the required amino acid sequences. First, the amino acid sequences of anti-CEA antibody T84.66 variable domains were taken as translations from the NCBI Nucleotide Database (CAA36980.1 and CAA36979.1 ). Amino acids contributing to the immunoglobulin signal peptide were identified by a Kyte-Doolittle plot (Kyte and Doolittle 1982) and eliminated. The amino acid sequence of the human lgG1 heavy chain constant region (UniProt no.: P01857) was attached to the sequence of the heavy chain variable domain. The variable domain of the light chain of T84.66 was expanded by the constant region of human κ light chain (UniProt no.: P01834). Full amino acid sequences of the chimeric antibody are given in Figure 2. The amino acid sequences of the two antibody chains were translated to nucleic acid sequences with the GeneArt® online tool (by Life Technologies, Darmstadt, Germany). The optimal codon usage of E. coli was taken into account. These DNA sequences were custom synthesized by GeneArt® (Life Technologies). Nucleic acid sequences and amino acid sequences of the chimeric antibody chains are given in SEQ ID NO: 15 and 19 (nucleic acid sequence of the light chain from nucleotide 88 to 750, or 82 to 732, respectively), SEQ ID NO: 16 and 20 (amino acid sequence of the light chain from amino acid 30 to 250, or 28 to 244, respectively), SEQ ID NO: 17 and 21 (nucleic acid sequence of the heavy chain from nucleotide 67 to 1440, or 82 to 1443, respectively), SEQ ID NO: 18 and 22 (amino acid sequence of the heavy chain from amino acid 23 to 480, or 28 to 481 , respectively), respectively.

DNA coding for the heavy or the light chain of monoclonal antibody (mAb) T84.66 was amplified and extended by the Xhol and Kpnl restriction sites via PGR. Primers were custom synthesized by Eurofins Genomics (Ebersberg, Germany). Restriction enzymes Xhol and Kpnl were obtained from Thermo Scientific (Dreieich, Germany). After restriction digest, the genes coding for the antibody chains were inserted into vector backbones pST003 (for heavy chain sequence) or pST006 (for light chain sequence), which contained the genetic information for the AIDA-l-autotransporter. Ligation was done using T4 ligase (from Thermo Scientific) for 20 min at 37 °C. This resulted in plasmids pVWVOOS and pWW004. Plasmid pWW003 featured kanamycin resistance and the sequence coding for a fusion protein consisting of the CtxB signal peptide, the heavy chain of T84.66 mAb, a linker region and the autotransporter β- barrel under the control of a T7/lac promotor. Plasmid pWW004 featured carbenicillin resistance and a T7/lac promotor. It encoded the same fusion protein with the light chain of T84.66 mAb as a passenger. The nucleic acid sequences and amino acid sequences for the autotransporter fusion constructs including (i) the signal peptide, (ii) the antibody chain, (iii) the transmembrane linker of AIDA-I and (iv) the transporter domain of AIDA-I are given in SEQ ID NO: 15 (nucleic acid sequence of the light chain as passenger and AIDA-I as the autotransporter SEQ ID NO: 16 (amino acid sequence of the light chain as passenger and AIDA-I as the autotransporter), SEQ ID NO: 17 (nucleic acid sequence of the heavy chain as passenger and AIDA-I as the autotransporter), and SEQ ID NO: 18 (amino acid sequence of the heavy chain as passenger and AIDA-I as the autotransporter).

Bacterial strains, cell line and growth conditions

E. coli UT5600 (DE3) (F-, ara-14. leuB6. secA6. lacY1. proC14, tsx-67, A(ompT- fepC)266. entA403, trpE38, rfbD1 , rpsL109(Str_r), xyl-5. mtl-1. thi-1 , X(DE3)) was used for expression of the autotransporter fusionproteins. E. coli DH5a (F-, (j>80dlacZAM15, A(lacZYA-argF)U169, deoR, recA1 , endA1 , hsdR17(rK-mK₊), phoA, supE44, λ-, thi-1 , gyrA96, relA1 ) (Invitrogen, Darmstadt, Germany) was used for cloning and cryostorage.

Plasmids pWW003 and pVWV004 were separately transformed into E. coli UT5600 (DE3) by electroporation resulting in one E. coli strain for each of the single antibody chains: UT5600 (DE3) pVWV003 contains the genetic information for the autotransporter fusion protein with the heavy antibody chain (Η_Τ84.6θ) as a passenger and UT5600 (DE3) pWW004 for the light antibody chain (L_T84.66) - Furthermore, E. coli cells UT5600 (DE3) pWW003 were transformed with pWW004, resulting in the strain UT5600 (DE3) pVWVOOS pWW004, which was used for co-expression of both autotransporter fusion proteins (H_T84.66 ⁺ L_T84 66) on the same bacterial cells. UT5600 (DE3) pKP006 displaying an unrelated peptide and the host strain UT5600 (DE3) without a plasmid served as controls in this study.

Precultures of E. coli were grown overnight from a single colony picked from an agar plate in 20 mL LB medium containing only the required antibiotics at 37 °C under shaking (200 rpm). For the main culture 100 mL LB medium containing 10 mM β- mercaptoethanol, 10 pm EDTA and antibiotics were inoculated with 1000 μΙ of the preculture which had previously been washed twice in 1 mL LB. The cultures were grown at 37 °C under shaking (200 rpm) until an optical density (OD₅7₈nm) of 0.6 was reached. For induction of protein expression, 1 mM isopropyl-p-d- thiogalactopyranoside (IPTG) was added. Induction was carried out at 30 °C under shaking (200 rpm) for 1 hour. Afterwards, cells were harvested by centrifugation (3910 xg, 5 min, 4 °C) for subsequent treatment.

The human colon adenocarcinoma cell line SW-403 (DSMZ no. ACC-294) was obtained from Leibniz-lnstitut DSMZ (German Collection of Microorganisms and Cell Culture, Braunschweig, Germany). Eucaryotic cells were grown in DMEM High Glucose with 2 mM L-glutamate and 10 % (v/v) FCS in a 37 °C atmosphere containing 5 % C0₂.

Protease treatment of E. coli For whole-cell protease treatment, cells were grown, protein expression was induced and cultures were harvested as described above. Cells from 50 mL culture volume were suspended in 1 mL PBS, 10 μΙ proteinase K solution (5 mg/ml, corresponds to an activity of approximately 2 mAU, enzyme purchased from AppliChem, Darmstadt, Germany) were added and digestion was performed at 37 °C under shaking (200 rpm) for 40 min. Digestion was stopped by addition of 5 mL Tris-HCI containing 5 % PCS. Afterwards the protease treated cells were washed two more times, once in 5 mL Tris-HCI containing 5 % FCS, 0.7 mM aprotinin and 1 mM PMSF, and once in 5 mL PBS.

SDS-PAGE and Western Blotting

For the protease accessibility assay, E. coli cultures were grown. The outer membrane protein fractions of proteinase K digested cells and undigested cells were prepared as previously described (Park et al. 2015). As a control, outer membrane fractions of the same E. coli strains which had not been induced were also isolated. Outer membrane preparations were mixed with the one and a half volume of SDS sample buffer and separated by SDS-PAGE (7.5 % polyacrylamide gel). Proteins were stained with Coomassie Brilliant Blue R250 (Fig. 3). The protein ladder PageRuler® (Thermo Scientific) was used as a standard for the molecular weight (MW). Western Blotting was conducted at 10 V for 1 hour using a semi-dry blotting device. After blocking for 2 hours at room temperature, the antibody chains were detected using a polyclonal FITC-coupled goat anti-human primary antibody (Thermo Scientific, Cat. No 31529) at a concentration of 0.3 g/mL in Western Blot blocking solution for 16 hours at 4 °C. A horseradish peroxidase (HRP) coupled donkey anti- goat antibody (Santa Cruz Biotechnology, Dallas, Texas USA, Cat. No sc2020) at a concentration of 0.08 pg/rnL in Western Blot blocking buffer was applied as secondary antibody for 1 hour at room temperature. Luminol reagent for detection was also purchased from Santa Cruz Biotechnology (ImmunoCruz®).

Flow cytometric analysis of surface accessibility

E. coli cultures of UT5600 (DE3) pWWOOS pWW004 (H_T84.66 + L _Τ84.6δ), UT5600 (DE3) pWWOOS (H_T84.66) , UT5600 (DE3) pWW004 (L_{T84 6}6) and of the control strains UT5600 (DE3) pKP006 and UT5600 (DE3) were grown as described above. The pelleted cells were stored at 4 °C for 16 hours. Then E. coli cells were washed twice in 5 mL PBS and blocked with 5 mL PBS supplemented with 5 % (w/v) bovine serum albumin (BSA) for 25 min on ice. The OD₅7₈nm of this solution was adjusted to 2 with PBS containing 5 % (w/v) BSA. 200 pL E. coli suspension for each sample were incubated with 3 pg of the FITC-conjugated goat anti-human antibody for 50 min at 37 °C under shaking (300 rpm) (Fig. 4). Afterwards, cells were washed twice in 100 pL particle-free PBS and resuspended in 1 mL particle-free PBS. Flow cytometry was performed with a FACSAria III™ (BD Biosciences, Heidelberg, Germany) using a 488 nm laser for excitation and a 530/30 nm filter for detection.

Flow cytometric analysis of binding ability

SW-403 cells were cultivated in 75 cm² cell culture flasks in DMEM High Glucose supplemented with 10 % (v/v) FCS and 2 mM L-glutamine. Two flasks containing a confluently grown monolayer of SW-403 cells were washed with 2 mL Dulbecco^'s PBS each. Afterwards, 1 mL RIPA buffer containing a set of protease inhibitors (1 mM Na₃V0₄, 0.2 mM phenylmethanesulfonylfluoride (PMSF), 1 .4 μΜ aprotinin, 2 μΜ benzamidine) was added to each flask. Cells were incubated with RIPA buffer on ice for 10 min. The cell lysates were centrifuged at 4 °C for 10 min at 10,000 xg. The supernatants were pooled and transferred to an Amicon® centrifugal filter tube. Ultrafiltration was performed at 4 °C for 20 min at 4,000 xg. The resulting concentrated SW-403 extract was dialyzed for 48 hours in PBS, in order to remove remains of the detergents from the RIPA buffer. PBS for dialysis was changed twice during this procedure.

Cultures of E. coli strains UT5600 (DE3) pWW003 pWW004 { _ΎΜ66 + L_T84.₆6), UT5600 (DE3) pWW003 (H_T84.₆₆), UT5600 (DE3) pWW004 (L_{T84 6}6) and of the control strains UT5600 (DE3) pKP006 and UT5600 (DE3) were grown as described above. After washing the cells twice in 5 mL PBS, the pelleted cells were stored at 4 °C for 16 hours. Then cells were washed twice in 5 mL PBS and blocked in 5 mL PBS containing 5 % (w/v) BSA for 1 hour on ice. The OD₅7₈nm of this cell suspension was adjusted to OD 2 with PBS + 5 % (w/v) BSA. 200 pL of these E. coli suspensions were mixed with 0.2 pL freshly prepared and dialyzed SW-403 extract and incubated at 37 °C under shaking (300 rpm) for 45 min. Then 10 pL of 1 mg/mL DNAse solution (approx. 30 U) were added to the cell suspensions. Cells were harvested by centrifugation. This and all following centrifugal steps were performed at 8,000 xg for 2 min at room temperature. Then cells were suspended in 100 μΐ_ particle-free PBS and 0.4 pg mouse anti-CEA antibody (clone COL-1 , Cat. No. MA5-13714, Thermo Scientific) were added. From this step on only particle-free PBS was used. The samples were mixed and incubated at 37 °C under shaking (300 rpm) for 30 min. Again, cells were harvested by centrifugation and washed once in 1 ml PBS. Cells were then suspended in 100 pl_ PBS. 2 pg of a Dylight633 coupled goat anti-mouse antibody (Thermo Scientific, Cat. No. 35512) were added and samples were incubated at 37 °C under shaking (300 rpm) for 25 min. Subsequently, the cells were washed twice in 100 pL PBS and suspended in 1 ml_ PBS. Flow cytometry was performed with a FACSAria III™ (BD Biosciences, Heidelberg, Germany) using a 633 nm laser and a 660/20 nm filter set (Fig. 5).

Co-immunoprecipitation

E. coli cultures of the strains UT5600 (DE3) pWW003 pWW004 (Η_Τ84.66 + L_T84.66) , UT5600 (DE3) pWW003 (H_T84 6₆) , UT5600 (DE3) pWW004 (L_T84.66) and of the control strain UT5600 (DE3) were grown. After induction, harvest and two washing steps in 5 mL PBS, cells were stored at 4 °C for 16 hours and washed again twice in 5 mL PBS. Outer membrane protein isolation was performed as previously mentioned. Magnetic beads coated with F_c-region binding protein A were obtained from Pierce (Thermo Scientific, Cat. No. 88845). Beads can be collected on the reaction tube wall in a magnetic stand. 25 pL of beads suspension were required for each outer membrane sample. Beads were washed according to the manual, first in 175 pl_ and then in 1 mL binding buffer. The outer membrane fractions were suspended in 500 pL binding buffer and mixed with the beads. The magnetic particles were incubated with the outer membrane preparations for 1 hour at 24 °C under shaking (300rpm). Beads were collected on the tube wall and the supernatant was kept as "unbound fraction" (Fig. 6). Beads were washed three times, twice in 500 μΐ_ binding buffer and once in 500 μΐ_ purified water. Finally, the beads were suspended in 60 pL SDS sample buffer and boiled at 95 °C for 30 min. This step eluted the bound proteins from the surface of the beads and the obtained suspension was kept as "bound fraction" (Fig. 6). The collected samples were separated by SDS-PAGE (7.5 % polyacrylamide gel). Subsequent Coomassie staining or Western Blot analysis was performed as described above (Fig. 7).

Amino acid sequences

Amino acid sequence for the heavy chain variable domain of anti-CEA mAb: GenBank no. CAA36980.1 . Amino acid sequence for the light chain variable domain of anti-CEA mAb: GenBank no. CAA36797.1 . Amino acid sequence for the constant region of human lgG1 (heavy chain): UniProt no.: P01857. Amino acid sequence for the constant region of human immunoglobulin κ light chain UniProt no.: P01834. Amino acid sequences of the chimeric antibody chains are given in SEQ ID NO: 16 and 20 (amino acid sequence of the light chain from amino acid 30 to 250, or 28 to 244, respectively), and SEQ ID NO: 18 and 22 (amino acid sequence of the heavy chain from amino acid 23 to 480, or 28 to 481 , respectively), respectively.

Results

Expression and surface display of full-length antibody chains

In order to display heavy and light chains of a full-length antibody on the surface of E. coli, autotransporter fusion proteins were designed. These fusion proteins consisted of the heavy or light chain sequence of the chimeric antibody T84.66 embraced by an N-terminal signal peptide and a C-terminal linker region which connected the antibody chain to the β-barrel (Fig. 1 ). The plasmids encoding these constructs were transformed into E. coli UT5600 (DE3), the cultures were grown and induced with IPTG and the outer membrane protein fractions were prepared. SDS- PAGE analysis of the outer membrane preparations showed distinct bands of the expected molecular weight (MW) of the antibody chains (Fig. 3). The induced sample of UT5600 (DE3) pWW003 (coding for the autotransporter fusion protein with the heavy chain of T84.66 as passenger, H_T84.66) showed a substantial band at approximately 100 kDa (calculated MW H_T84.66 : 98 kDa) (Fig. 3A). The outer membrane preparation of induced cells of UT5600 (DE3) pWW004 (coding for the autotransporter fusion protein with the light chain of T84.66 as a passenger, L_T8 .66) featured an intense band at around 70 kDa (calculated MW L_T84.66: 72 kDa) (Fig. 3A). For the co-expression strain carrying both relevant plasmids both bands were found (Fig. 3B). Cells treated with proteinase K before outer membrane preparation did not exhibit the relevant bands. This indicated that the passenger had been digested by the protease. Since proteinase K is not able to penetrate the outer membrane of E. coli, we concluded that the antibody chains were actually present on the surface of the bacterium where they were accessible for the protease. As a control, the band corresponding to outer membrane protein OmpA was examined. OmpA has a C- terminal domain which is susceptible to periplasmic protease cleavage. Since the OmpA band of the digested samples was not perceptibly digested, it could be deduced that proteinase K had not entered the periplasm. The results of this protease accessibility assay indicated that the antibody chains were surface displayed on E. coli.

In order to confirm the proof of surface exposure of the antibody chains by a second method, flow cytometry was performed. For this purpose, E. coli cells expressing full- length antibody chains on their surface were incubated with a polyclonal FITC- coupled anti-human antibody, because the constant domains of the surface displayed antibody were human (Fig. 4). Flow cytometric analysis of these samples showed a considerable increase of mean fluorescence intensity for the cells displaying one or both antibody chains compared to E. coli cells displaying an unrelated peptide (ratio H_T84.66⁺l-T84.66/control: 23; His control: 38; b^ control: 8.5). Digestion of cells with proteinase K prior to labelling with FITC-conjugated anti- human antibody led to a loss of the effect on mean fluorescence intensity (ratio H_T84.66⁺LT84.66/∞ntrol: 3.7; Hr^ control: 7.8; Lra control: 1 .5). This proved that the antibody chains were presented on the bacterial surface and accessible for immunological reactions.

Antigen binding ability

To show that the displayed antibody was functional, binding assays based on flow cytometry were carried out. Antibody T84.66 was directed against the tumor associated antigen CEA. An extract containing CEA was prepared from the human colon adenocarcinoma cell line SW-403. E. coli cells displaying the separated antibody chains (H_T84.66 or L_T84.66) , E. coli cells presenting both chains (H_T84.66 + LT84_.66). control cells displaying an unrelated peptide (UT5600 (DE3) pKP006) and cells of E. coli UT5600(DE3) without a plasmid were incubated with this extract. Afterwards, the cells were incubated with a commercial anti-CEA antibody (clone COL-1 ) which targets an epitope that is different from the epitope of the autodisplayed antibody (Kuroki et al. 1989). Finally, a secondary Dylight633 coupled antibody was applied and the cells were analyzed by flow cytometry (Fig. 5). Dylight633 fluorescence indicated CEA binding in this ^"sandwich-like assay^". The cells displaying the light chain of T84.66 (Fig. 5, "LTSW) featured a small increase in fluorescence intensity by a factor of 26 compared to E. coli autodisplaying a random peptide. For E. coli displaying the heavy chain of T84.66 (Fig. 5, "H_T84_.66") a slight increase in fluorescence intensity in comparison to the control strain could be observed (ratio h^ controI: 10). But a clear rise in fluorescence intensity compared to the control cells could be detected for the strain co-expressing both antibody chains on the bacterial surface (Fig. 5, "H_T84_.66 + L_T84 66")- The mean fluorescence intensity of those cells was increased by a factor of 40 compared to control cells. As an additional control, the same E. coli strains displaying the antibody chains were submitted to proteinase K treatment prior to this antigen binding assay. Protease treatment antagonized the increase in fluorescence intensity compared to the control strain (ratio 18 / 1 1 / 15 for H_T84.66 "LT84.66 / H_T84.66 / L_T84 66)- This indicated once again, that the displayed antibody chains on the surface of E. coli were accessible to both immunoreactions and enzymatic digest. The fact that the co- expression of both antibody chains on the same cells resulted in considerable CEA binding, while the display of the separated chains entailed only marginal antigen binding, is explained by the assembly of the separated chains to form a full antibody the surface of E. coli. In order to confirm the assembly of the chains, a co- immunoprecipitation assay was subsequently carried out.

Assembly of antibody chains on bacterial surface

To verify that the separated antibody chains assembled on the bacterial surface, co- immunoprecipitation experiments were performed with protein A coated magnetic beads. The magnetic particles could easily be separated from reaction solutions by collecting them on the reaction tube wall in a magnetic stand. Protein A is able to bind the F_c-domain of human immunoglobulins very well. The F_c-domain consists of the constant regions of the heavy immunoglobulin chains. The light chain does not contribute to the F_c-domain and has therefore no affinity to protein A. Outer membrane preparations of E. coli displaying one or both of the separated antibody chains (HTS4.66 or LTS4.66 or H_T84.66 + Lj84.66) and E. coli without a plasmid were incubated with protein A carrying beads (Fig. 6). After washing of the beads, the protein fraction which had bound to the protein A coated beads was eluted. This bound fraction and the unbound protein fraction were separated by SDS-PAGE. The unbound fraction served as a control. As expected (Fig. 6), the samples of the unbound fraction contained their respective antibody chains: H_Ts4 66 (lane 2), L_T84.66 (lane 3) or both (lane 1 ) (Fig. 7, "unbound"). For the bound fraction of E. coli displaying the heavy chain, a protein band with an apparent MW close to the calculated MW of the heavy chain fusion protein was detected by Coomassie staining (Fig. 7A, "bound", lane 2) and Western Blotting (Fig. 7A, "bound", lane 2). This was expected, because the heavy chain constant domains can bind to protein A (Fig. 6). For the bound fraction of E. coli displaying the light chain, no detection of the light immunoglobulin chain was expected for the bound fraction, because the light chain does not have affinity towards protein A (see Fig. 6). In fact, the light chain was not present in the bound fraction (Fig. 7, "bound" lane 3). However, for the co-expression of both chains in one organism, both immunoglobulin chains were expected to be found. If the light chain interacted with the heavy chain, the light chain should be co- precipitated with the magnetic beads. Thus, both the heavy and the light chain fusion proteins should be found in the eluted samples of the bound protein fractions (see Fig. 7). Actually, the bands corresponding to the MW of the two immunoglobulin chains were detected in the sample of E. coli co-displaying both chains (Fig. 7, "bound", lane 1 ). This co-precipitation of the light chain is explained by the interaction of the heavy and the light chain on the surface of E. coli. In combination with the flow cytometer data, which revealed a particular well antigen binding ability for the bacterial cells co-displaying both antibody chains compared to cells displaying only one chain, these results prove that the separated immunoglobulin chains assemble on the surface of E. coli to form a full antibody.

Discussion For the first time, the surface display of a functional full-length antibody on E. coli is reported. Results obtained by two independent methods, protease accessibility assay and flow cytometry, proved that the separately expressed antibody chains were present on the surface of intact bacterial cells. Functionality of the displayed antibody was demonstrated in binding assays conducted by flow cytometry. The assembly of the chains on the surface of E. coli was shown in a co-immunoprecipitation experiment. The free motility of the β-barrel in the outer membrane of £. coli (Jose et al. 2002) enables the antibody chains to be drawn towards each other due to their natural affinity (Hakim and Benhar 2009). The interaction of different autodisplayed passengers on the surface of the same cells has been described before (Gratz et al. 2015).

One possible application of this technology is the generation and screening of antibody libraries on the surface of E. coli. Cellular display systems have one main advantage over phage display. In contrast to phages, they are compatible with high- throughput screening via fluorescence activated cell sorting (FACS). This screening method is not biased, while biopanning of phages tends to discriminate the most potent binders (Levin and Weiss 2006). Screening of antibody fragment libraries displayed on yeast (Hoogenboom 2005) and E. coli (Salema et al. 2013) has previously been described. But, as described above, this approach requires the extension of the identified new antibody variants by the constant immunoglobulin domains prior to antibody production. The technique developed in this study can be used to identify new antibodies against any desired target directly in the full-length format.

Moreover, the described technology can be used to investigate aglycosylated antibodies and their derivatives, because bacteria do not have any enzymatic machinery for protein glycosylation. Aglycosylated antibodies are currently under close investigation (Jung et al. 201 1). In some cases a complete loss of glycosylation is needed in order to make the antibody therapeutically applicable, for example, if the mechanism of action of the antibody is purely agonistic or antagonistic and does not require the activation of the immune system (Eder et al. 2009). Furthermore, the introduction of specific, artificial secondary modifications to ^"naked^" antibodies obtained from bacteria can be useful: antibodies featuring altered patterns of glycosylation have been engineered in order to create predictable, selective immune responses (Natsume et al. 2009). Autodisplayed antibodies are attached to the bacterial surface. This is an advantage for the introduction of chemical alterations to the amino acid sequence: the bacterial cell provides a matrix for the antibody, which allows the simple separation of the antibody from the reagents, e.g. by centrifugation. This benefit applies for any artificial secondary modification. Interestingly, the deficiency of glycosylation does not necessarily impair the interaction of the displayed immunoglobulin with F_c-region binding proteins or secondary antibodies. The constant domains of the model antibody investigated in our study were recognized by both protein A and an anti-human polyclonal serum.

Finally, E. coli displaying antibodies could be applied as tools for bacterial tumor targeting. Bacterial tumor cell targeting is defined as the application of bacterial cells with selective affinity towards tumorous tissue for cancer diagnosis or treatment (Forbes 2010; Lee 2012). Lipopolysaccharide (LPS) induced toxicity only applies for systemic use, not for intestinal application. Therefore, in gastrointestinal cancer, orally or rectally applied bacteria with specific affinity for malignant cells could serve as drug carriers delivering a drug precisely to the spot of interest. The design of tumor targeting bacteria which express a prodrug converting enzyme has also been accomplished (Lemmon et al. 1994; Pawelek et al. 1997). This allows the conversion of a prodrug into the active agent selectively within tumorous tissue.

The presented technology is not only interesting because of its various potential applications. These findings also overcome what we used to regard as the limitations of the autotransporter technology. Cysteine rich protein sequences with the ability to form multiple disulfide bonds were regarded as problematic passengers for autotransporter mediated surface display. The formation of disulfide bonds within the periplasm was generally considered an obstacle for transport across the inner membrane (Jong et al. 2007; Leyton et al. 201 1 ). In 2012, an investigation on the limits of autotransporter mediated surface display was published and chymotrypsin containing 4 disulfide bonds was presented as an autotransporter passenger with high cysteine content (Ramesh et al. 2012). Now we succeeded in displaying an lgG1 antibody heavy chain which contains 1 1 cysteine residues which form 4 intrachenary and 3 interchenary disulfide bonds. This pushes the limits of the system even further. It should be mentioned, that the concentration of β-mercaptoethanol and the choice of bacterial strain was observed to be crucial for successful display of this cysteine rich structure. Furthermore, a prolonged storage time (16 hours at 4 °C) after induction turned out to enhance the functionality of the antibody in binding assays.

In conclusion, we present a promising new approach in antibody engineering research. Bacterial surface display of full-length antibodies could become a valuable tool in design, screening and manufacturing of novel antibody variants. The technique combines the enormous potential of aglycosylated therapeutic antibodies with the benefits of a cost-effective and time-saving tool for unbiased antibody screening.

Example 2: Surface display of antibodies with the MATE autotransporter

For the presentation of full length antibodies on the surface of E.coli the heavy and the light chain of the chimeric antibody T84.66 directed against the human carcinoembyronic antigen (CEA) were used (c.f. Example 1 ).

The genes for the autotransporter fusion protein with the heavy (H_T84.66) or light chain (L-T84_.66) as a passenger were created by amplifying the codon optimized antibody sequences from the plasmids pWW003 and pWW004 (cf. Example 1). Using InFusion® HD Cloning technique (Clontech Laboratories) the genes were introduced into two different plasmid backbones which were suitable for co-expression. For the heavy chain the modified backbone of the plasmid pPQ33 was utilized (pPQ33 is described in Quehl et al. , 2016). In comparison to the original plasmid the rhamnose inducible promoter (RhaP) was replaced by the L-Arabinose inducible araBAD promoter. Furthermore, the MATE autotransporter was used instead of AIDA-I. The light chain gene was inserted into a modified pPQ29 backbone (pPQ29 is described in Quehl et al., 2016) which also showed the replacement of the RhaP promoter by the araBAD promoter (Quehl et al., 2016).

The nucleic acid sequences and amino acid sequences for the autotransporter fusion constructs including (i) the signal peptide, (ii) the antibody chain, (iii) the transmembrane linker of EhaA and (iv) the transporter domain of EhaA are given in SEQ ID NO: 19 (nucleic acid sequence of the light chain as passenger and EhaA as the autotransporter), SEQ ID NO: 20 (amino acid sequence of the light chain as passenger and EhaA as the autotransporter), SEQ ID NO: 21 (nucleic acid sequence of the heavy chain as passenger and EhaA as the autotransporter), and SEQ ID NO: 22 (amino acid sequence of the heavy chain as passenger and EhaA as the autotransporter).

Insertion of these plasmids into the E.coli strain UT5600(DE3) was conducted according to standard electroporation protocols. UT5600(DE3) carrying already the plasmid with the information for the heavy chain was transformed with the plasmid for the light chain in order to co-express the antibody chains.

All strains were grown under similar conditions. As a preculture 20 mL of Luria Bertani (LB) medium containing the required antibiotics (50 pg/mL carbenicillin and/or 30 pg/mL kanamycin) were inoculated with a single colony from an agar plate and cultivated overnight. For the main cultures 100 mL LB medium containing 10 μΜ EDTA, 15 mM β-Mercaptoethanol and the appropriate antibiotics were inoculated with 1 mL preculture which was washed twice with fresh LB medium previously. The cultures were cultivated at 37 °C under shaking (200 rpm) until they reached an optical density (OD₅7₈nm) of 0.5. Afterwards the protein biosynthesis of the heavy and light chain fusion proteins was induced by adding L-arabinose to a final concentration of 0.2 % (w/v). Induction was performed for 1 hour at 30 °C, 200 rpm. Afterwards cells from 40 mL culture were harvested by centrifugation (39 0 xg, 10 min, 4 °C) for subsequent treatment.

The outer membrane protein isolation was conducted according to the modified protocol from Park et al. 2015. As a control, outer membrane fractions of the same E.coli strains which had not been induced with L-arabinose were isolated as well. The outer membrane preparations were heated for 30 min at 95 °C in twofold SDS sample buffer (100 mM Tris-HCI, pH 6.8; containing 4% sodium dodecylsulfate, 0.2 % bromphenole blue, 20 % glycerol and 30 mg/mL dithiothreitol) to be separated electrophoretically in a 10% polyacrylamide gel afterwards. Finally, proteins were stained with Coomassie Brilliant Blue R250. SDS-PAGE analysis of the outer membrane preparations showed distinct bands at the expected molecular weights of 102 kDa for the heavy chain construct and 77 kDa in case of the light chain fusion protein. Both bands could be found at the mentioned molecular weights for the co-expression strains. Neither in the not induced samples nor in the UT5600(DE3) strain without any plasmid comparable bands were found (Fig. 8).

This experiment confirms the successful expression of both antibody chains separately but also co-expressed simultaneously in one cell.

Example 3: Development of a full-length antibody library

Full-length antibodies, which are displayed on the surface of bacterial cell may be used for development and screening of antibody libraries. The host cell may be a cell capable of expressing an autotransporter, for example Eschenchia coli for AIDA-I or E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., or Halomonas spp. for EhaA.

For the generation of a full length antibody library the following method can be used.

Antibody's hypervariable areas can be divided into three complementarity determining regions (CDR) for each chain. Out of these six CDRs the CDR-H3 for the heavy chain and CDR-L3 of the light chain show the highest variability and the strongest antigen contact (Wu und Kabat, 1970; Morea et al. , 1998). Therefore, these two are the most interesting regions for generating a library. To randomize these CDRs an oligonucleotide is constructed which contains an area with arbitrarily randomized bases in the middle. Its 3'- and 5^'-end consist of two well defined regions, which show no mutations and are complementary to the framework region surrounding the CDR3. Whereas the randomized part is as long as the CDR3, the two surrounding constant regions consist of 15 to 30 bases. Prerequisite for the application of this primer in a linear amplification reaction is the concomitance of another complementary oligonucleotide with exactly the same randomization. Therefore, a small second primer is used, which binds at the non-randomized 3'-end of the library primer. After the annealing step the long library oligonucleotide is filled up to a double strand by the large (Klenow) fragment of the DNA-polymerase I. The resulting primer pair is used for a linear amplification reaction of plasmids carrying the genetic information for the heavy or the light antibody chain. Each primer pair introduces a different CDR3 into the antibody sequence. The heavy and the light chain are randomized separately because of their occurrence on two different plasmids. For both constructs the methylated parenteral plasmid is removed by digestion with the restriction enzyme Dpn\. The nicked amplification product remains and is introduced into an E.coli expression strain by electroporation. For co-expression of both libraries, one for the heavy and one for the light chain, E.coli is transformed with both nicked plasmids simultaneously or successively in two electroporation steps. The resulting E.coli strain is going to co-express a library for the CDR-H3 and a library for CDR-L3 randomly combined.

To generate a library of antibodies on bacterial host cell, the polypeptides which are encoded by the randomized nucleic acids obtained as described above are expressed and displayed on the surface of a plurality of bacterial host cells such that each host cell displays one of a plurality of the different antibodies, by the method of the present invention, in particular by an AID A- 1 or EhaA autotransporter.

After inducing gene expression the antibody library can be incubated with any fluorescently labeled target of interest and screened for binding antibodies by flow cytometry of the host cells.

Literature

Blasshofer F, Klotz LO, Sauerbier E, Jose J (2007) Tumour cell targeting by cells of E. coli displaying functional single chain and "twin chain" antibody fragments against EGFR. Frontiers in Medicinal Chemistry. Joint GDCH and DPhG Spring Meeting, Berlin, March 18-21 . ON009.

Blasshofer F and Jose J (2006) Autodisplay of variable fragments of different antibodies in E. coli. DPhG Jahrestagung, Marburg, 4.-7. Oktober 2006, P99 Boder ET und Wittrup KD (1997) Yeast surface display for screening combinatorial polypeptide libraries. Nat Biotech 15:553-557.

Daugherty PS, Chen G, Olsen MJ, Iverson BL, Georgiou G (1998) Antibody affinity maturation using bacterial surface display Protein Engineering 1 1 :825-832

Eder JP, Woude GFV, Boerner SA, LoRusso PM (2009) Novel therapeutic inhibitors of the c-Met signaling pathway in cancer Clinical Cancer Research 15:2207- 2214

Forbes NS (2010) Engineering the perfect (bacterial) cancer therapy Nat Rev Cancer 10:785-794

Gratz A, Bollacke A, Stephan S, Nienberg C, Le Borgne M, Gotz C, Jose J (2015) Functional display of heterotetrameric human protein kinase CK2 on Escherichia coli: a novel tool for drug discovery Microbial cell factories 14:74

Hakim R, Benhar I (2009) "Inclonals" mAbs 1 :281-287

Henderson I et al., 2004. Type V protein secretion pathway: the autotransporter story.

Microbiology and Molecular Biology Reviews, 68(4), 692-744. Ho M, Nagata S und Pastan I (2006) Isolation of anti-CD22 Fv with high affinity by Fv display on human cells. Proc Natl Acad Sci 103:9637-9642.

Hoogenboom HR (2005) Selecting and screening recombinant antibody libraries Nat Biotech 23:1 105-1 1 16

Jong WS et al. (2007) Limited tolerance towards folded elements during secretion of the autotransporter Hbp Molecular microbiology 63:1524-1536

Jose J, Bernhardt R, Hannemann F (2002) Cellular surface display of dimeric Adx and whole cell P450-mediated steroid synthesis on E. coli Journal of Biotechnology 95:257-268

Jose J, Meyer TF (2007) The Autodisplay Story, from Discovery to Biotechnical and Biomedical Applications Microbiology and Molecular Biology Reviews : MMBR 71 :600-619

Jung ST, Kang TH, Kelton W, Georgiou G (201 1 ) Bypassing glycosylation: engineering aglycosylated full-length IgG antibodies for human therapy Current Opinion in Biotechnology 22:858-867

Kuroki M, Greiner JW, Simpson JF, Primus FJ, Guadagni F, Schlom J (1989) Serologic mapping and biochemical characterization of the carcinoembryonic antigen epitopes using fourteen distinct monoclonal antibodies International journal of cancer 44:208-218

Kyte J, Doolittle RF (1982) A simple method for displaying the hydropathic character of a protein Journal of Molecular Biology 157:105-132 # Leabman MK, Meng YG, Kelley RF, DeForge LE, Cowan KJ, Iyer S (2013) Effects of altered FcyR binding on antibody pharmacokinetics in cynomolgus monkeys. mAbs, 5: 896-903.

Lee C-H (2012) Engineering bacteria toward tumor targeting for cancer treatment: current state and perspectives Appl Microbiol Biotechnol 93:517-523

Lemmon M, Elwell J, Brehm J, Mauchline M, Minton N, Giaccia A, Brown J Anaerobic bacteria as a gene delivery system to tumors. In: Proc. Am. Assoc. Cancer Res, 1994. pp 879-891

Levin AM, Weiss GA (2006) Optimizing the affinity and specificity of proteins with molecular display Molecular BioSystems 2:49-57

Leyton DL et al. (2011 ) Size and conformation limits to secretion of disulfide-bonded loops in autotransporter proteins Journal of Biological Chemistry 286:42283- 42291

Malgren RA, Flanigan CC (1955) Localization of the vegetative form of Clostridium tetani in mouse tumors following intravenous spore administration. Cancer Res, 15: 473-478.

Mazor Y, Van Blarcom T, Mabry R, Iverson BL, Georgiou G (2007) Isolation of engineered, full-length antibodies from libraries expressed in Escherichia coli Nat Biotech 25:563-565

Morea V, Tramontano A, Rustici M, Chothia C und Lesk AM (1998) Conformations of the third hypervariable region in the VH domain of immunoglobulins . Journal of Molecular Biology 275:269-294.

Natsume A, Shimizu-Yokoyama Y, Satoh M, Shitara K, Niwa R (2009) Engineered anti-CD20 antibodies with enhanced complement-activating capacity mediate potent anti-lymphoma activity Cancer Science 100:241 1 -2418

Nelson AL (2010) Antibody fragments: Hope and hype mAbs 2:77-83

Park M, Yoo G, Bong J-H, Jose J, Kang M-J und Pyun J-C (2015) Isolation and characterization of the outer membrane of Escherichia coli with autodisplayed Z-domains. Biochimica et Biophysica Acta (BBA) - Biomembranes 1848:842- 847.

Patyar S, Joshi R, Prasad Byrav DS, Prakash A, Medhi B, Das BK (2010) Bacteria in cancer therapy: a novel experimental strategy. J Biomed Sci, 17:21.

Pawelek JM, Low KB, Bermudes D (1997) Tumor-targeted Salmonella as a novel anticancer vector Cancer research 57:4537-4544

Ramesh B, Sendra VG, Cirino PC, Varadarajan N (2012) Single-cell Characterization of Autotransporter-mediated Escherichia coli Surface Display of Disulfide Bond-containing Proteins The Journal of Biological Chemistry 287:38580- 38589

Rhiel L, Krah S, Gunther R, Becker S, Kolmar H, Hock B (2014) REAL-Select: Full- Length Antibody Display and Library Screening by Surface Capture on Yeast Cells PLoS ONE 9:e1 14887

Quehl P, Hollender J, Schuurmann J, Brossette T, Maas R und Jose J (2016) Co- expression of active human cytochrome P450 1A2 and cytochrome P450 reductase on the cell surface of Escherichia coli. Microbial Cell Factories 15:1 - 15.

Salema V et al. (2013) Selection of Single Domain Antibodies from Immune Libraries Displayed on the Surface of E. coli Cells with Two β-Domains of Opposite Topologies PLoS ONE 8

Spohn et al. (1992) B-cell epitopes of the Nef protein. Research in Virologyl 43:70- 81 .

Thommes S, Blasshofer F, Jose J (2010) Autodisplay of combinatorial antibody libraries in E. coli. DPhG Jahrestagung, 4.-7. Oktober 2010, Braunschweig. C108, published in : ISBN 978-3-9271 15-65-1 ).

Wells TJ et al. (2008) EhaA is a novel autotransporter protein of enterohemorrhagic Escherichia coli 0157:H7 that contributes to adhesion and biofilm formation. Environmental Microbiology 10:589-604

Wu TT und Kabat EA (1970) An analysis of the sequences of the variable regions of Bence Jones proteins and myeloma light chains and their implications for antibody complementarity. J Exp Med 132:21 1 -250.

Zhang H (2013) Empowering scFv with effector cell functions for improved anticancer therapeutics Oncoimmunology 2:e24439

Zhao M, Yang M, Ma H, Li X, Tan X, Li S, Yang Z, Hoffman RM (2006) Targeted therapy with a Salmonella thyphimurium leucine-arginine auxotroph cures orthotopic human breast tumors in nude mice. Cancer Res, 66: 7647-7652.

Zhou C, Jacobsen FW, Cai L, Chen Q, Shen WD (2010) Development of a novel mammalian cell surface antibody display platform mAbs 2:508-518

Yazawa K, Fujimori M, Amano J, Kano Y, Taniguchi S (2000) Bifidobacterium longum as a delivery system for cancer gene therapy: selective localization and growth in hypoxic tumors. Cancer Gene Ther, 7: 269-274. Niewerth U, Frey A, Voss T, 1 Le Bouguenec C,2 Baljer G, Franke S, Schmidt IVIA (2001 ) The AIDA Autotransporter System Is Associated with F18 and Stx2e in Escherichia coli Isolates from Pigs Diagnosed with Edema Disease and Postweaning Diarrhea. Diarrhea. Clin. Diagn. Lab. Immunol. 8(1 ): 143-149;9.

Claims

1 . A method for displaying at least one first polypeptide comprising a light chain variable domain V_L or/and at least one second polypeptide comprising a heavy chain variable domain V_H on the surface of a host cell, said method comprising the steps:

wherein said at least one first polynucleotide comprises:

(i) a portion encoding a signal peptide,

(ii) a portion encoding said at least one first polypeptide to be displayed,

(iii) a portion encoding a transmembrane linker, and

(iv) a portion encoding the transporter domain of an autotransporter,

and wherein said at least one second polynucleotide comprises:

(i) a portion encoding a signal peptide,

(ii) a portion encoding said at least one second polypeptide to be displayed,

(iii) a portion encoding a transmembrane linker, and

(iv) a portion encoding the transporter domain of an autotransporter,

and

wherein

2. The method according to claim 1 , wherein the host cell is transformed or/and transfected with said at least one first polynucleotide or/and at least one second polynucleotide.

3. The method according to claim 1 or 2, wherein said at least one first polynucleotide or/and said at least one second polynucleotide is operatively linked with an expression control sequence.

4. The method according to any one of the preceding claims, wherein the first polynucleotide or/and the second polynucleotide is present in at least one plasmid.

5. The method according to any one of the preceding claims, wherein the first polynucleotide is present in a first plasmid and the second polynucleotide is present in a second plasmid.

6. The method according to any one of the claims 1-3, wherein said the first polynucleotide or/and the second polynucleotide is integrated into the genome of the host cell.

7. The method according to any one of the preceding claims, wherein said light chain variable domain V_L comprises three complementary determining regions CDRL1 , CDRL2, and CDRL3.

8. The method according to any one of the preceding claims, wherein said heavy chain variable domain V_H comprise three complementary determining regions CDRH1 , CDRH2, and CDRH3.

9. The method according to claim 7 or 8, wherein said complementary determining regions form an antigen-binding site.

10. The method according to any one of claims 1-9, wherein said at least one second polypeptide comprises three heavy chain constant domains C_H1 , C_H2 and C_H3.

The method according to any one of the preceding claims, wherein said at least one second polypeptide further comprises a hinge region.

The method according to any one of the preceding claims, wherein the at least one first polypeptide is a light chain of an antibody or/and the at least one second polypeptide is a heavy chain of an antibody.

The method according to any one of the preceding claims, wherein two first polypeptides and two second polypeptides are displayed.

The method according to any one of claims 1-13, wherein said at least one first polynucleotide or/and said at least one second polynucleotide are expressed in step (b) in the presence of an agent capable of reducing a disulfide bond between two cy stein residues or/and protecting an SH group of a cystein residue.

The method according to claim 14, wherein the agent capable of reducing disulfide groups or/and protecting an SH group of an amino acid residue is β- mercaptoethanol or DDT,

The method according to claim 15, wherein β-mercaptoethanol or DDT is present in a concentration range of about 2-50 mM, preferably about 5-25 mM.

The method according to any one of claims 1-16, wherein said at least one first polypeptide comprises a light chain constant domain C_L and said at least one second polypeptide comprises a heavy chain constant domain C_H1 , and wherein at least one disulfide bond is formed between the light chain constant domain C_L of the at least one first polypeptide and the heavy chain constant domain C_H1 of the at least one second polypeptide.

The method according to to any one of claims 1 -17, wherein two of said second polypeptides comprising a hinge region are displayed and wherein at least one disulfide bond is formed between the hinge regions of said two second polypeptides.

19. The method according to any one of the preceding claims, further comprising the step:

(c) culturing or/and incubating said host cell under conditions, wherein said at least one first polypeptide and said at least one second polypeptide form an antibody molecule, or a functional fragment thereof, which is displayed on the surface of the host cell.

20. The method according to claim 19 wherein culturing or/and incubating said host cell in step (c) is performed in a medium which is substantially free of agents capable of reducing a disulfide bond between two cy stein residues or/and protecting an SH group of a cy stein residue.

21 . The method according to claim 19 or 20, wherein cutturing or/and incubating said host cell in step (c) is performed in a medium which is substantially free of β- mercaptoethanol and DTT.

22. The method according to any one of claims 19-21 , wherein the antibody molecule, or the functional fragment thereof, comprises at least one disulfide bond.

23. The method according to any one of claims 19-22, wherein culturing or/and incubating said host cell in step (c) is performed for about 8-24 hours.

24. The method according to any one of claims 19-23, wherein culturing or/and incubating said host cell in step (c) is performed for about 14-18 hours.

25. The method according to any one of claims 19-24, wherein culturing or/and incubating said host cell in step (c) is performed at a temperature in the range of about 4-25°C.

26. The method according to any one of claims 19-25, wherein culturing or/and incubating said host cell in step (c) is performed at a temperature in the range of about 4-10°C.

27. The method of any one of the preceding claims, wherein the at least one first or/and the at least one second polypeptide displayed on the surface of a host cell form an antibody molecule, or a functional fragment thereof.

28. The method according to any one of claims 27, wherein said antibody molecule, or the functional fragment thereof, is capable of binding an antigen.

29. The method according to any one of claims 27-28, wherein said antibody molecule, or the functional fragment thereof, is not an scFv fragment.

30. The method according to any one of claims 27-29, wherein said antibody molecule, or the functional fragment thereof, is not a tcFv fragment.

31. The method according to any one of claims 27-30, wherein said fragment is a Fab functional fragment.

32. The method according to any one of claims 27-31 , wherein said fragment is a F(ab')₂ functional fragment.

33. The method according to any one of claims 27-32, wherein said antibody molecule, or the functional fragment thereof, is not a single domain antibody.

34. The method according to any one of claims 27-33, wherein the antibody molecule, or the functional fragment thereof, comprises at least one first polypeptide comprising V_L and C_L domains and at least one second polypeptide comprising V_H, three C_H domains and a hinge region.

35. The method according to any one of the claims 27-34, wherein the antibody, or the functional fragment thereof, is a chimeric antibody, or a functional fragment thereof.

36. The method according to any one of claims 27-35, wherein the antibody, or the functional fragment thereof, is a humanized antibody, or a functional fragment thereof.

37. The method according to any one of claims 27-36, wherein said antibody molecule is an antibody selected from the group consisting of IgG, IgA, IgD, IgE, and IgM.

38. The method according to any one of the claims 1-37, wherein the transporter domain (iv) is derived from an autotransporter selected from the group consisting of Ssp, Ssp-h1 , Ssp-h2, PspA, PspB, Ssa1 , SphB1 , AspA/NalP, VacA, AIDA-I, IcsA, MisL, TibA, Ag43, ShdA, AutA, Tsh, SepA, EspC, EspP, Pet, Pic, SigA, Sat, Vat, EpeA, EatA, Espl, EaaA, EaaC, Pertactin, BrkA, Tef, Vag8, PmpD, Pmp20, Pmp21 , lgA1 protease, App, Hap, rOmpA, rOmpB, ApeE, EstA, Lip-1 , caP, BabA, SabA, AlpA, Aae, NanB, and variants thereof.

39. The method according to any one of the claims 1 -38, wherein the transporter domain (iv) is the transporter domain of an AIDA-I protein, or a variant thereof.

40. The method according to claim 39, wherein the transporter domain (iv) of the AIDA-I protein is encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO:5,

(b) a nucleotide sequence encoding SEQ ID NO:6,

(c) nucleotide sequences comprising a sequence being at least 70% identical to SEQ ID NO:5 or/and a sequence encoding SEQ ID NO: 6, and

41. The method according to claim 39 or 40, wherein the transporter domain (iv) of the AIDA-I protein comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO:6, and

(b) sequences which are at least 70% identical to the sequences of (a).

42. The method according to any one of claims 1-37, wherein the transporter domain (iv) is the transporter domain of an EhaA protein, or a variant thereof.

43. The method according to claim 42, wherein the transporter domain (iv) of the EhaA protein is encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO: 12 or 13,

(b) a nucleotide sequence encoding SEQ ID NO:14,

(c) nucleotide sequences comprising a sequence being at least 70% identical to SEQ ID NO: 12 or 13 or/and a sequence encoding SEQ ID NO: 14, and

44. The method according to claim 42 or 43, wherein transporter domain (iv) of the EhaA protein comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO: 14, and

(b) sequences which are at least 70% identical to the sequences of (a).

45. The method according to any one of the preceding claims, wherein the host cell is a bacterium.

46. The method according to any one of the preceding claims, wherein the bacterium is a Gram negative bacterium.

47. The method according to any one of the preceding claims, wherein the host cell is selected from the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., Cupriavidus spp., Rhodobacter spp., Acinetobacter spp., Gluconobacter spp., Gluconacetobacter spp., Acidomonas spp., Acetobacter spp., Paracoccous spp., Rhizobium spp., Xanthomonas spp., Halomonas spp., Variovorax spp., Alcanivorax spp., Sphingomonas spp., and Mannomonas spp.

48. The method according to any one of the preceding claims, wherein the host cell is selected from the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., and Halomonas spp.

49. The method according to any one of the preceding claims, wherein the host cell is selected from the group consisting of Salmonella spp., Zymobacter spp., and Pseudomonas spp.

50. The method according to any one of claims 1-48, wherein the host cell is E.coli.

51. The method according to any one of claims 1-48, wherein the host cell is not E. coli.

52. The method according to any one of claims 38-41 , wherein the host cell is E.coli.

53. The method according to claim 42-44, wherein the host cell is selected from the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., Cupriavidus spp., Rhodobacter spp., Acinetobacter spp., Gluconobacter spp., Gluconacetobacter spp., Acidomonas spp., Acetobacter spp., Paracoccous spp., Rhizobium spp., Xanthomonas spp., Halomonas spp., Variovorax spp., Alcanivorax spp., Sphingomonas spp., and Marinomonas spp.

54. The method according to claim 42-44, wherein the host cell is selected from the group consisting of E. coli, Salmonella spp., Zymomonas spp., Zymobacter spp., Pseudomonas spp., and Halomonas spp.

55. The method according to claim 42-44, wherein the host cell is selected from the group consisting of Salmonella spp., Zymobacter spp., and Pseudomonas spp.

56. The method according to claim 42-44, wherein the host cell is not E. coll.

57. The method according to any one of the claims 1-56, wherein the transmembrane linker (iii) is a transmembrane linker from an autotransporter selected from the group consisting of Ssp, Ssp-h1 , Ssp-h2, PspA, PspB, Ssa1 , SphB1 , AspA/NalP, VacA, AIDA-I, IcsA, MisL, TibA, Ag43, ShdA, AutA, Tsh, SepA, EspC, EspP, Pet, Pic, SigA, Sat, Vat, EpeA, EatA, Espl, EaaA, EaaC, Pertactin, BrkA, Tef, Vag8, PmpD, Pmp20, Pmp21 , lgA1 protease, App, Hap, rOmpA, rOmpB, ApeE, EstA, Lip-1 , caP, BabA, SabA, AlpA, Aae, NanB, and variants thereof.

58. The method according to any one of the claims 1-57, wherein the transmembrane linker (iii) is a transmembrane linker from an AIDA-I protein, or a variant thereof.

59. The method according to any one of the claims 1 -58, wherein the transmembrane linker (iii) is encoded by a sequence comprising a sequence selected from the group consisting of:

(a) a nucleotide sequence comprising SEQ ID NO:3,

(b) a nucleotide sequence encoding SEQ ID NO:4,

(c) nucleotide sequences comprising a sequence being at least 70% identical to SEQ ID NO: 3 or/and a sequence encoding SEQ ID NO: 4, and

60. The method according to any one of the claims 1 -59, wherein the transmembrane linker (iii) comprises a sequence selected from the group consisting of: (a) an amino acid sequence comprising SEQ ID NO:4, and

(b) sequences which are at least 70% identical to the sequences of (a).

61 . The method according to any one of claims 1 -56, wherein the transmembrane linker (iii) is a transmembrane linker from an EhaA protein, or a variant thereof.

62. The method according to claim 61 , wherein the transmembrane linker (iii) is encoded by a sequence comprising a sequence selected from the group consisting of;

(a) a nucleotide sequence comprising SEQ ID NO: 9 or 10,

(b) a nucleotide sequence encoding SEQ ID NO:1 1 ,

(c) nucleotide sequences comprising a sequence being at least 70% identical to SEQ ID NO: 9 or 10 or/and a sequence encoding SEQ ID NO: 1 1 , and

63. The method according to any one of the claims 61-62, wherein the transmembrane linker (iii) comprises a sequence selected from the group consisting of:

(a) an amino acid sequence comprising SEQ ID NO: 1 1 , and

(b) sequences which are at least 70% identical to the sequences of (a).

64. The method according to any one of the claims 1 -63, wherein the transporter domain (iv) and the transmembrane linker (iii) are homologous.

65. The method according to any one of the claims 1-63, wherein the transporter domain (iv) and the transmembrane linker (iii) are heterologous.

66. The method according to any one of the preceding claims, wherein the signal peptide (i) is a CtxB signal peptide.

67. The method according to any one of the preceding claims, wherein the at least one first or/and the at least one second polynucleotide further comprises at least one portion encoding an affinity tag.

68. The method according to any one of the claims 67, wherein the affinity tag is independently selected from His₆ and epitopes.

69. The method according to any one of the preceding claims, wherein the at least one first or/and the at least one second polynucleotide further comprises at least one portion encoding at least one protease recognition sequence.

70. The method according to claim 69, wherein in the at least one first or/and the at least one second polynucleotide, the at least one protease recognition sequence is located between portion (ii) and (iii).

71 . The method according to claim 69 or 70, wherein the at least one protease recognition sequence is independently selected from factor Xa cleavage site, OmpT cleavage site, and TEV protease cleavage site.

72. The method according to any one of the preceding claims, wherein the amino acid sequences encoded by the portions (i) to (iv) of the at least one first or the at least one second polynucleotide are arranged from N terminal to C terminal.

73. The method according to any one of the preceding claims, wherein the portions (i) to (iv) of the at least one first or the at least one second polynucleotide are arranged from 5' to 3'.

74. The method according to any one of the preceding claims, wherein the transporter domain (iv) of the autotransporter is heterologous with respect to the host cell.

75. The method according to any one of the preceding claims, wherein the codon- usage of the polynucleotide sequence is adapted to the host cell.

76. A recombinant vector comprising at least one first polynucleotide or/and at least one second polynucleotide as defined in any one of the claims 1 -75, operatively linked to an expression control sequence.

77. A recombinant host cell comprising at least one recombinant vector according to claim 76.

78. A recombinant host cell comprising a polynucleotide encoding at least one first polypeptide and a second nucleic acid molecule encoding at least one second polypeptide comprising a heavy chain variable domain V_H, as defined in any one of the claims 1 to 75.

79. Recombinant host cell according to claim 77 or 78, displaying an antibody molecule, or a functional fragment thereof, on the surface of the host cell.

80. Recombinant host cell according to claim 79, wherein the antibody molecule, or the functional fragment thereof, is capable of binding an antigen.

81 . Membrane preparation which is derived from a host cell according to any one of claims 77-80.

82. The membrane preparation according to claim 81 , wherein the membrane preparation comprises the outer membrane of the host cell.

83. The membrane preparation according to claim 81 or 82, comprising the at least one first polypeptide or/and the at least one second polypeptide.

84. The membrane preparation according to any one of claims 81-83 comprising the antibody molecule.

85. An antibody molecule, displayed on a host cell according to any one of claims 77-80, wherein the antibody molecule, or the functional fragment thereof, comprises at least one first polypeptide comprising at least a light chain variable domain V_L and at least one second polypeptide comprising at least a heavy chain variable domain V_H.

86. Use of a host cell according to any one of claims 77-80, a membrane preparation according to any one of claims 81 -84, or an antibody according to claim 85 for screening for an antibody.

87. A method for screening an antibody library expressed on a plurality of host cells, said method comprising the steps:

(a) producing a plurality of host cells according to the method of any one of the claims 1 to 75, each host cell expressing one of a plurality of different antibodies, or a functional fragment thereof, and

88. A method for producing an antibody library expressed on a plurality of host cells, wherein said method comprises the production of a plurality of host cells according to the method of any one of the claims 1 to 75, each host cell expressing one of a plurality of different antibodies, or functional fragments thereof.

89. Use of a host cell according to any one of claims 77-80, for producing an aglycosylated antibody, or a functional fragment thereof.

90. A method for producing an antibody, or a functional fragment thereof, displayed on the surface of a host cell, comprising the method of any one of the claims 1 -75.

91. The method according to claim 90, wherein the antibody, or the functional fragment thereof, is an aglycosylated antibody, or a functional fragment thereof. Use of a host cell according to any one of claims 77-80, a membrane preparation according to any one of claims 81 -84, or an antibody according to claim 85 for bacterial tumor targeting.