INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In recent years there has been considerable progress in the development of expression systems for the display of heterologous peptides and proteins on the surfaces of bacteria and yeasts (6, 16, 31). Cells displaying peptides and proteins such as receptors, antibodies, and enzymes are of considerable value for various biotechnological applications, such as bioseparations, vaccine development, and combinatorial library screening. Numerous anchor proteins that mediate the translocation of passenger proteins through the cytoplasmic and outer membranes of Escherichia coli and their exposure on the cell surface have been used. Short peptides (less than approximately 50 amino acids [aa]) were successfully displayed on the cell surface by insertion into surface-exposed loops of fimbrial proteins (24) or outer membrane proteins like LamB (7) or PhoE (2). Larger passenger domains could be presented on the E. coli cell surface by insertion into a surface-exposed domain of the E. coli flagellin FliC (50), by carboxy-terminal fusion to Lpp-OmpA (a hybrid protein consisting of parts of the E. coli lipoprotein Lpp and OmpA protein [9, 12]), by using the peptidoglycan-associated lipoprotein PAL as an outer membrane anchor (15), by amino-terminal fusion to the -domain of immunoglobulin A (IgA) protease of Neisseria gonorrhoeae and other autotransporters (22, 30, 35), or by C-terminal fusion to InaZ, the ice nucleation protein of Pseudomonas syringae (18).

Pathogenic gram-negative bacteria have developed several distinct secretion mechanisms for the efficient surface display of binding domains, which specifically interact with their complementary receptors on host cell surfaces (23). Among them, intimins and invasins are members of a family of bacterial adhesins which specifically interact with diverse eukaryotic cell surface receptors, thereby mediating bacterial adherence and invasion (48). Enteropathogenic E. coli and enterohemorrhagic E. coli (EHEC) produce attaching and effacing lesions in the intestinal mucosa. The intimate bacterial adhesion associated with attaching and effacing lesion formation is promoted by intimin, an EHEC surface protein. Intimin targets the translocated intimin receptor (Tir), which is exported by the bacteria and integrated into the host cell membrane (20). At least five different subtypes of intimin have been described (1, 37). They are integrated into the E. coli outer membrane with their amino-terminal region, while the carboxy-terminal 280 amino acids are surface exposed (13).

The EaeA intimin from EHEC O157:H7 is composed of 939 amino acids. The cell binding activity of EaeA intimin has been localized to its C-terminal 280 residues (13), and the structure of the carboxy-terminal domains has been determined recently by both X-ray crystallography and nuclear magnetic resonance (3, 19, 32). It is assumed that the amino-terminal 550 residues of intimin form a porin-like structure (43) and are folded into an antiparallel -barrel (32). The entire extracellular segment of intimin is an elongated and relatively rigid rod made up of three immunoglobulin-like domains and a C-terminal lectin-like domain to interact with the receptors (Fig. 1A) (19, 32). This domain resides on a rigid extracellular arm, which is most likely anchored to the amino-terminal transmembrane domain through a flexible hinge made by two glycine residues, allowing mechanical movement between the extracellular rod and the bacterial outer membrane (32).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   (A) Schematic drawing of intimin and intimin' used in this study for the display of heterologous passenger domains. The arrow indicates the site of truncation at residue 659 of intimin. The passenger domain is flanked at its amino-terminal and carboxy-terminal ends by short epitope sequences and fused to carboxy-terminally truncated intimin. OM, outer membrane; PS, periplasmic space; CM, cytoplasmic membrane. (B) Schematic representation of the display vector pASKInt100 harboring the structural gene of intimin'. f1, f1 replication origin; cat, chloramphenicol resistance marker; tetR, tetracycline repressor encoding gene; colE1, ColE1 replication origin; intimin', truncated eaeA gene of EHEC O157:H7 from codon 1 to 659; E, Etag epitope-encoding sequence. Unique SmaI and BamHI restriction sites allow the in-frame fusion of genes encoding various passenger domains (P).

Apparently, intimin provides a structural scaffold ideally suited to the cell surface display of receptor binding domains remote from the bacterial cell surface. This prompted us to investigate whether a truncated intimin that lacks the carboxy-terminal receptor binding domain could be used as a platform for the display of heterologous passenger domains on the surface of E. coli K-12 cells. We found that protein fusion to truncated intimin results in cell surface exposure of foreign passengers at high copy numbers. Furthermore, intimin'-based display of a foreign peptide ligand enables E. coli K-12 cells to adhere to eukaryotic target cells that present the corresponding receptor and also allows the isolation of peptide-presenting cells by fluorescence-activated cell sorting (FACS).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. All E. coli strains used in this study are listed in Table 1. Bacteria were grown at 37°C. The medium used for growth and maintenance of E. coli strains was dYT (1% yeast extract, 1.6% Bacto tryptone, 0.5% NaCl). Anhydrotetracycline (Acros, Morris Plains, N.J.) was added to liquid media at a final concentration of 0.2 µg/ml. Antibiotics were used at the following final concentrations if required: chloramphenicol, 25 µg/ml; kanamycin, 50 µg/ml; spectinomycin, 50 µg/ml; tetracycline, 12.5 µg/ml.

Reagents. Restriction enzymes and DNA-modifying enzymes were purchased from MBI Fermentas and New England BioLabs. Tfl polymerase was obtained from Promega. Biotinylated goat anti-mouse antibody was obtained from Sigma. The monoclonal antibody (VII-E-7) to a 13-residue C-terminal epitope of Sendai virus L protein (DGSLGDIEPYDSS) (11) was a gift from H. Einberger and H. P. Hofschneider (Max-Planck-Institut für Biochemie, Martinsried bei München, Germany). Anti-E monoclonal antibody was obtained from Pharmacia Biotech. A streptavidin-R-phycoerythrin conjugate was purchased from Molecular Probes. All other chemicals were of analytical grade and obtained from Sigma.

Oligonucleotides. Synthetic oligonucleotides were purchased from Metabion (Martinsried, Germany), SigmaARK (Darmstadt, Germany), and MWG Biotech (Ebersberg, Germany). The oligonucleotides were as follows: IL-4-up, 5'-GCGCCCCGGGCACAAGTGCGATATCACC-3'; IL-4send-lo, 5'-GCGCGGATCCTTATGATGAATCGTATGGTTCGATATCACCTAATGATCCATCGCTCGAACACTTTGAATATT-3'; intimin-amber-lo, 5'-GCGCGAATTCTAATTAACATAAAAAAACAATCC-3'; intilo1, 5'-GCGCCAATTGCGCTGGCCTTGGTTTGATC-3'; intiminup, 5'-GCGCTCTAGATAACGAGGGCAAAAAATGATTACTCATGGTTGTTATAC-3'; pASK-cat5'-seqlo, 5'-TATCAACAGGGACACCAGG-3'; REI-lo-BglII, 5'-GCGCAGATCTCCTAGTGATTTGAAGCTTAG-3'; RGD-up, 5'-GCGCCCCGGGTGCATCCCTCGAGGGGACTACCGTTGCWAACAGGACTCCGACTG-3'; RSPX, 5'-GTGAATTTCGACCTCTAG-3'; SupE2-Eco-up, 5'-GCGCGAATTCACCAGAAAGCGTTGTACGG-3'; SupE2-Mlu-lo, 5'-GCGCACGCGTAAGACGCGGCAGCGTCGC-3'; EheI-up, 5'-CAGCTGTTGCCCGTCTCG-3'; and AWcatlo, 5'-CGCGTCGACAAGCTTGAAAACGTTTCAGTTTGC-3'.

DNA procedures. Standard DNA procedures, such as plasmid isolation, ligation, restriction analysis of plasmids, and isolation of DNA fragments, were as described elsewhere (42). PCR using Tfl polymerase was as follows: 30 s of denaturation at 94°C, 30 s of annealing at 53°C, and 30 s of elongation at 72°C for 30 cycles. Plasmids constructed and used in this study are listed in Table 1.

Construction of plasmid pASKInt100. For construction of plasmid pASKInt100, the following cloning strategy was applied. Codons 1 to 659 of the eaeA gene were PCR amplified directly from heat-inactivated cells of EHEC O157:H7 strain 933 (a gift from J. Hacker, University of Würzburg) using the oligonucleotides intiminup and intilo1. The resulting PCR product was digested with XbaI and MunI and ligated to XbaI/EcoRI-digested vector pASKC21-EETI to give pASKInt1. pASKC21-EETI is a derivative of plasmid pASK21-EETI (9), where the coding sequence for an E epitope was placed 5' to the Ecballium elaterium trypsin inhibitor II (EETI-II) coding sequence. It contains a chloramphenicol resistance marker instead of the ampicillin resistance gene which has codons 8 to 281 deleted. To place an amber codon at codon position 35 of the eaeA gene, part of the eaeA coding sequence was PCR amplified with the primers intiminup and intimin-amber-lo. The resulting PCR product was digested with XbaI and EcoRI and ligated to similarly digested vector pASKInt1, yielding pASKInt100. The complete nucleotide sequence of this plasmid is available on our website (http://www.gwdg.de/~hkolmar).

Construction of plasmids pASKInt100-EETI-CK^Send, pASKInt100-IL-4, and pASKInt100-REI. The EETI-CK^Send gene was obtained from pASK21-EETI-CK^Send (9) by digestion with SmaI and BamHI and ligated to similarly digested pASKInt100 to give vector pASKInt100-EETI-CK^Send. The interleukin 4 (IL-4)-encoding gene was amplified by PCR using the vector pRPR9IL-4FD (29) as a template and the primer pair IL-4-up and IL-4send-lo. IL-4send-lo hybridizes at the 3' end of the IL-4 gene and introduces the nucleotide sequence encoding the 13-residue epitope from the Sendai virus L protein (9). The resulting PCR product was digested with SmaI and BamHI and ligated to similarly digested pASKInt100. For the construction of an eaeA'-rei gene fusion, the vector pASKInt100-TmDegP, a derivative of pASKInt100 which contains a BglII restriction site preceding the coding sequence for the Sendai virus L-protein epitope, was digested with SmaI and BglII and ligated to the EcoRV- and BglII-digested PCR product containing the rei gene, which was obtained by PCR amplification with vector pHKREI (26) as the template and the oligonucleotide primer pair EheI-up and REI-lo-BglII to give pASKInt100-REI. To obtain pASKInt100-REISend, the Sendai virus epitope-encoding sequence, which is flanked in pASKInt100-REI by a BglII and a BamHI restriction site, was removed by digestion of pASKInt100-REI with BglII and BamHI followed by ligation of the vector fragment.

Construction of plasmids pASKInt100-RGD and pASKInt100-EETI-CK^RGD. pASKInt100-EETI-CK^Send was used as template DNA for PCR with the oligonucleotide pair RGD-up and AWcatlo. RGD-up introduces the coding sequence for CIPRGDYRC, which replaces the EETI-II inhibitor loop. Due to its degeneration, it contains either a TAA stop codon or an AAA (Lys) codon after the nonameric peptide coding sequence. The resulting PCR product was cloned as described above for the construction of pASKInt100-EETI-CK^Send to give pASKInt100-RGD and pASKInt100-EETI-CK^RGD. These coding sequences were introduced in a similar manner into pMX-EETI (9) to give plasmids pMX-RGD and pMX-EETI-CK^RGD. The nucleotide sequences of all cloned genes were confirmed by nucleotide sequence analysis.

Construction of plasmid pREP4supE. To place the supE tRNA coding sequence under P_Llac promoter/operator control, a DNA segment containing the coding sequence for tRNA^Gln(CAA), tRNA^Gln(CCT), tRNA^Met, supE tRNA, and tRNA^Gln(AAT) was amplified by PCR using chromosomal DNA of 71-18 as a template and the oligonucleotide pair SupE2-Eco-up and SupE2-Mlu-lo. The resulting PCR product was digested with EcoRI and MluI and ligated to the similarly digested vector pZA22-MCS1 (33). The resulting vector was digested with XhoI and XbaI to give a DNA fragment with the tRNA gene cluster under P_Llac promoter/operator control followed by a fill-in reaction using T4 DNA polymerase. The fragment was ligated to SmaI-digested vector pREP4 (Qiagen) to give pREP4supE.

Flow-cytometric analysis and FACS. For flow-cytometric analysis, cultures of E. coli strains containing the appropriate expression plasmid were grown overnight at 37°C and subcultured 1:50 at 37°C until they reached an optical density at 600 nm (OD₆₀₀) of 0.2. After induction with anhydrotetracycline (0.2 µg/ml) for 60 min, cells (200 to 500 µl) were pelleted by centrifugation in a tabletop centrifuge for 1 min and resuspended in 10 µl of phosphate-buffered saline (PBS). After the addition of 1 µl of the respective antibody (1 mg/ml), cells were incubated for 5 min at room temperature. After the addition of 500 µl of PBS, cells were centrifuged for 1 min and resuspended in 10 µl of PBS containing biotinylated goat anti-mouse immunoglobulin (1:10 dilution). After 5 min of incubation at room temperature and addition of 500 µl of PBS, cells were pelleted again and resuspended in 10 µl of PBS containing streptavidin-R-phycoerythrin conjugate (1:10 dilution) followed by 5 min of incubation at room temperature. Finally, after the addition of 500 µl of PBS, cells were pelleted and resuspended in 100 µl of PBS for flow cytometry analysis. A total of 300,000 events were collected on a Cytomation MoFlo cell sorter. Parameters were set as follows: forward scatter, side scatter, 730 (LIN mode, amplification factor 6); FL1 (fluorescein isothiocyanate), 600 (LOG mode); FL2 (phycoerythrin), 600 (LOG mode); trigger parameter, side scatter. The sample flow rate was adjusted to an event rate of approximately 30,000 s¹.

Titration of E. coli cell surface-exposed antibody binding sites. A 100-µl aliquot of induced cells from DH5Z1 containing pASKInt100-EETI-CK^Send was diluted with PBS to 3 ml. The total number of cells was determined by flow cytometry and found to be 4.5 × 10⁷. In parallel, 100 µl of the induced culture were centrifuged, resuspended in 10 µl of PBS, and incubated with serial dilutions of anti-E antibody ranging from 1 ng (6.7 fmol) to 2 µg (13.3 pmol). Cells were incubated for 5 min at room temperature. After addition of 500 µl of PBS, cells were centrifuged for 1 min and successively incubated with an excess over the anti-E antibody of biotinylated goat anti-mouse immunoglobulin and with streptavidin-R-phycoerythrin conjugate, as described above. Cells were analyzed by flow cytometry analysis. A total of 300,000 events were collected on a Cytomation MoFlo cell sorter.

Trypsin treatment of intact cells and membrane preparation. Cultures of E. coli strains containing the respective expression plasmid were grown overnight at 37°C and subcultured 1:50 at 37°C until they reached an OD₆₀₀ of 0.8. After induction with anhydrotetracycline (0.2 µg/ml) for 180 min, cells (50 ml) were pelleted by centrifugation and resuspended in PBS. The bacterial suspension was adjusted to an OD₆₀₀ of 10 and incubated for 10 min at 37°C with trypsin at a final concentration of 50 µg/ml. To remove trypsin after the reaction, cells were centrifuged and washed with PBS. The membrane fraction was prepared as described previously (17), with minor modifications. The cells were lysed by passage through a French pressure cell at 1,000 lb/in². Remaining large bacterial fragments were sedimented by centrifugation at 5,000 × g for 10 min. After incubation of the lysate on ice in 100 mM Tris-HCl (pH 8.0)-10 mM EDTA-1% (wt/vol) Triton X-100 for 30 min, the membrane fraction was obtained by centrifugation of the cleared solution for 120 min at 100,000 × g and 15°C. The membranes were solubilized in sample buffer (50 mM Tris-HCl [pH 6.8], 100 mM dithiothreitol, 2% [wt/vol] sodium dodecyl sulfate [SDS], 0.1% (wt/vol) bromophenol blue, 10% [vol/vol] glycerol) and subjected to SDS-polyacrylamide gel electrophoresis (PAGE) (10% acrylamide-bisacrylamide, 30:0.8) followed by immunoblotting.

Protein purification. The MalE-RGD fusion protein was affinity purified (34) using an amylose column (Pharmacia). MalE-EETI-CK^RGD, which contains six carboxy-terminal histidines, was purified by metal chelate affinity chromatography as described elsewhere (9). The protein concentration was determined from the A₂₈₀ (38).

Determination of IC₅₀s for inhibition of platelet aggregation. Human platelets were obtained from the Göttingen University hospital. The platelet suspension was diluted to a concentration of approximately 250,000 U/µl and preincubated with various concentrations of MalE-RGD and MalE-EETI-CK^RGD proteins. Aggregation was stimulated by adding 25 µM ADP, and the change in OD₆₅₀ was monitored by photometry. Values were expressed as percent aggregation, which represents the percentage of light transmission standardized to fully and not aggregated samples (8). The IC₅₀ is the concentration of the protein at which the platelet aggregation is inhibited by 50%.

Enrichment of RGD peptide-displaying E. coli cells via binding to human platelets. SilanePrep slides (Sigma) were coated with glutaraldehyde (6.25% [vol/vol]) for 30 min. After extensive washing with water, they were immediately used for platelet capture. Platelets were resuspended in Tris-buffered saline (TBS; 100 mM Tris-HCl [pH 7.5], 150 mM NaCl) containing 5 mM MgCl₂ and 5 mM CaCl₂ and sequestered to the glutaraldehyde-treated slides for 30 min. Remaining aldehyde groups were blocked with TBS containing 2% (wt/vol) bovine serum albumin. Slides were preincubated with untransformed 71-18 to saturate unspecific binding sites. Induced cells from a 50-ml culture of 71-18 containing pASKInt100-RGD or pASKInt100-EETI-CK^RGD were harvested by centrifugation and resuspended in 50 ml of TBS containing 5 mM MgCl₂ and 5 mM CaCl₂.The bacterial cell suspension was transferred into a 50-ml plastic tube, and the slide was immersed in the cell suspension and agitated for 30 min at room temperature. The slides were carefully but thoroughly washed with TBS containing 5 mM MgCl₂ and 5 mM CaCl₂. The bacteria were propagated by placing the slides into flasks containing rich medium with the appropriate antibiotic and overnight incubation at 37°C.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Targeting of intimin' to the surfaces of E. coli K-12 cells. To determine whether the amino-terminal fragment of intimin was sufficient for outer membrane translocation of a heterologous passenger domain using E. coli K-12 as the expression host, a derivative of EETI-II was used as a reporter. This protein is a member of the cystine knot family of protease inhibitors and provides a stable framework for the display of conformationally constrained peptides of various length and sequence (9, 49). The variant used here (EETI-CK^Send) contains a 13-residue epitope sequence from Sendai virus L protein in place of the original inhibitor loop and can easily be detected using a monoclonal anti-Sendai virus antibody (11).

View larger version (99K): [in this window] [in a new window]   FIG. 2.   Phase-contrast and fluorescence micrographs of the same field of DH5Z1 and DH5Z1(pASKInt100-EETI-CKSend) after consecutive incubation of induced cells with anti-E antibody, biotinylated anti-mouse antibody, and streptavidin-R-phycoerythrin conjugate.

Influence of intimin' expression on cell growth and survival. Induction of gene expression by addition of the inducer anhydrotetracycline to the growth medium of DH5Z1(pASKInt1-EETI-CK^Send) resulted in an arrest of cell growth and a drastic reduction of cell viability (data not shown). Overproduction of outer membrane-linked proteins often results in changes in the structure of the outer membrane and, as a consequence, in periplasmic enzyme leakage and cell death (9, 16). To restore cell viability, net accumulation of fusion protein in the outer membrane was reduced to a tolerable level by introducing an amber codon at position 35 of the eaeA' sequence and utilizing the amber suppressor strain DH5Z1 as the expression host. DH5Z1 contains a glutaminyl amber suppressor tRNA, which by recognition of the TAG codon gives rise to translational readthrough. Due to the reduced efficiency of in vivo nonsense suppression compared to the translation of a CAG codon at the same position (36), yields of the encoded protein are diminished.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   (A) Survival of DH5Z1 harboring pASKInt100-EETI-CKSend after induction of intimin'-EETI-CKSend production by addition of anhydrotetracyline (time zero). At each data point, single cells were spotted with a MoFlo cell sorter onto agar plates. The number of colonies after overnight incubation at 37°C was compared with the number of untransformed DH5Z1 colonies, which was set at 100%. Values are means and standard deviations from three independent experiments. (B) Growth competition of induced 71-18mutS harboring pASKInt100-EETI-CKSend compared to 71-18mutS harboring pASK75cat. A 1:1 mixture of both plasmid-containing cells was propagated in the presence of tetracycline for the number of generations indicated. The proportion of intimin'-displaying E. coli cells was determined by fluorescent labeling of culture aliquots with anti-E antibody. Values are means and standard deviations from three independent experiments.

Regulation of cellular net accumulation of the intimin'-EETI-CK^Send fusion protein. In order to obtain an estimate of the net accumulation of the intimin'-EETI-CK^Send fusion protein in the outer membrane of a single bacterial cell, the number of anti-E antibody molecules required to obtain saturation of the immunofluorescent staining of 4.5 × 10⁷ induced bacterial cells was determined. Saturation of antibody binding was reached at approximately 0.4 µg (2.67 pmol) of anti-E antibody, which corresponds to approximately 36,000 molecules per bacterial cell (Fig. 4).

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Titration of anti-E antibody binding to induced cells. Induced cells of DH5Z1 containing pASKInt100-EETI-CKSend were incubated with increasing amounts of anti-E antibody and immunofluorescently labeled with excess biotinylated anti-mouse antibody and streptavidin-R-phycoerythrin conjugate. Saturation of mean cellular fluorescence is reached at approximately 0.4 µg of anti-E antibody.

View larger version (20K): [in this window] [in a new window]   FIG. 5.   Regulation of intimin' outer membrane accumulation. (A) Schematic outline of IPTG-induced synthesis of supE tRNA that acts as an amber suppressor tRNA at codon 35 (UAG) of the eaeA' transcript. The eaeA' mRNA is synthesized upon induction of transcription of the eaeA' gene, which resides in vector pASKInt100 under tetA promoter/operator control. (B) Dose-response curve depicting the mean relative cellular fluorescence of intimin'-Etag-EETI-CKSend-displaying cells labeled with anti-E antibody as a function of IPTG concentration.

Probing the surface display of various passenger domains. Besides the 35-aa cystine knot protein EETI-CK^Send, two other protein domains were chosen as passengers to probe surface display via intimin' fusion, namely, human IL-4, a 128-aa four-helix-bundle protein, and REI_v, a 108-aa immunoglobulin variable light chain domain. The corresponding genes were introduced into pASKInt100-EETI-CK^Send by replacement of the EETI-CK^Send coding sequence. The 13-residue Sendai virus epitope tag was placed at the carboxy terminus of both domains. Intact cells containing pASKInt100-EETI-CK^Send, pASKInt100-IL-4, or pASKInt100-REI were labeled by indirect immunofluorescence with anti-E antibody or anti-Sendai virus antibody. The fluorescently labeled cells were analyzed with a MoFlo flow cytometer (Fig. 6A to C). All three constructs with different passenger domains could be fluorescently labeled with both antibodies, corroborating the cell surface exposure of the epitope sequences flanking the various passenger domains.

View larger version (54K): [in this window] [in a new window]   FIG. 6.   Intimin'-mediated cell surface display of various passenger domains. (A to C) FACS histograms of recombinant DH5Z1 harboring pASKInt100-EETI-CKSend (A), pASKInt100-IL-4 (B), or pASKInt100-REI (C). Cells were successively incubated with anti-Sendai virus antibody (Send) or anti-E antibody (E), biotinylated anti-mouse antibody, and streptavidin-R-phycoerythrin conjugate. Untransformed DH5Z1 labeled with anti-E antibody served as a control. (D and E) SDS-PAGE (D) and Western blot (E) analysis of surface targeting and protease accessibility of intimin'-EETI-CKSend, intimin'-IL-4, and intimin'-REIv fusion proteins. Membrane preparations of E. coli DH5Z1 expressing the respective passenger domains fused to intimin' and of control cells without a plasmid were analyzed by SDS-PAGE (10% gel, Coomassie brilliant blue stain) and Western blots probed with anti-Sendai virus antibody. To verify the surface location of the various fusion proteins, bacteria were either subjected to trypsin treatment or left untreated. The intimin' fusion proteins are indicated by arrows. M, marker proteins (sizes are in thousands).

Mimicking adhesin-host interaction by E. coli surface display of intimin'-RGD peptide fusions. The extracellular portion of intimin comprises a rod of immunoglobulin domains extending from the bacterial surface where the carboxy-terminal domain mediates the intimate attachment of enteropathogenic and enterohemorrhagic E. coli to the translocated intimin receptor on the mammalian host cells. For a number of applications, particularly for the study of receptor-ligand interactions, it is interesting to see whether a heterologous ligand fused to the amino-terminal intimin fragment is able to promote bacterial adhesion via binding to a corresponding receptor on the surface of mammalian cells.

View larger version (35K): [in this window] [in a new window]   FIG. 7.   (A) Amino acid sequences of EETI-CKSend- and RGD-containing peptides. The sequence in brackets corresponds to a short linker region (PPTPL) and the Etag epitope (italic) fused to intimin'. Cysteine residues that are involved in disulfide bond formation are in bold. The amino acid sequence of the RGD-containing peptide is underlined. (B) FACS histograms from enrichment of 71-18(pASKInt100-EETI-CKRGD) from a 1:1,000 mixture with 71-18(pASKInt100-EETI-CKSend). The mixture was labeled with anti-Sendai virus antibody and analyzed by FACS before and after three and six rounds of enrichment of 71-18(pASKInt100-EETI-CKRGD) by panning on immobilized platelets. The bar in each graph represents the sorting gate defined as a negative event, i.e., bacteria which do not display the EETI-CKSend epitope on their surfaces. The percentage of cells counted as negative events is indicated. (C) FACS histograms from enrichment of 71-18(pASKInt100-RGD) from a 1:1,000 mixture with 71-18(pASKInt100-EETI-CKSend).

Enrichment of cells displaying a Sendai virus epitope. To explore whether cells presenting a particular passenger protein can be enriched by FACS from a background of negative cells, the Sendai virus epitope fused to the carboxy terminus of the intimin'-REI_v fusion protein was used as an immunotag for fusion protein detection and FACS. 71-18 cells containing pASKInt100-REI were mixed at a ratio of 1:200,000 with cells containing pASKInt100-REISend, which lack the Sendai virus epitope. Cells were subjected to labeling with anti-Sendai virus antibody, followed by biotinylated anti-mouse antibody and streptavidin-R-phycoerythrin conjugate. The cell population was run through the MoFlo cell sorter at an event rate of 30,000 s¹ and sorted on the basis of fluorescence intensity (Fig. 8). A sorting gate was chosen such that less than 0.5% of the control cells fell within the positive window. After immediate re-sorting, the collected cells were plated onto agar plates, grown overnight, labeled again, and re-sorted. After two consecutive sorting rounds, 10% of the total cell population were found in the positive fraction (Fig. 8B), and after an additional re-sorting, 74% of the cell population were stained with anti-Sendai virus antibody.

View larger version (14K): [in this window] [in a new window]   FIG. 8.   Histogram data from enrichment of 71-18 containing pASKInt100-REI. The bar in each graph represents the sorting gate defined as a positive event. The percentage of cells counted as positive events is indicated. (A) A 1:200,000 mixture of 71-18(pASKInt100-REI) and 71-18(pASKInt100'-REISend) prior to the first sorting round; (B) mixture after two consecutive runs of FACS and overnight cell growth; (C) mixture from round B after re-sorting and overnight cell growth.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report we describe the export of passenger peptides and proteins to the surfaces of E. coli cells by fusion to intimin', a truncated E. coli adhesin. To address the question of whether surface display via fusion to intimin' provides a means of exposing various passenger domains in high copy numbers on the outer surfaces of E. coli cells, three different passenger proteins were probed for cell surface display. A derivative of the Ecballium elaterium trypsin inhibitor, the Bence-Jones protein REI_v, and human IL-4 were efficiently targeted to the surface of E. coli 71-18(pASKInt100) cells. This was demonstrated by successful immunofluorescence labeling using a monoclonal antibody that recognizes a carboxy-terminal epitope and by the accessibility of the fusion proteins residing on intact cells to exogenously added trypsin.

The Ecballium elaterium trypsin inhibitor was chosen as a model passenger domain, since it was shown to be successfully translocated through the outer membrane when fused to Lpp-OmpA or the C-terminal domain of the N. gonorrhoeae IgA protease precursor protein IgA (9, 49). EETI-II is a 28-residue peptide which is stabilized by three intramolecular disulfide bonds. The immunoglobulin variable light chain domain REI_v consists of 102 residues and is also mainly stabilized by a single intramolecular disulfide bond (14). It was shown previously that cell surface exposure of CtxB, a subunit of cholera toxin which contains a single intramolecular disulfide bond, was influenced by the conformational state of the polypeptide chain. Transport of CtxB via fusion to the IgA autotransporter across the outer membrane was blocked by intramolecular disulfide bonds and appeared to proceed only under reducing conditions (22). Protein translocation through IgA is thought to occur via passage of the unfolded linear polypeptide chain of the passenger through a hydrophilic pore in the center of the -barrel formed by the outer membrane-anchored IgA domain (21). The mechanism by which the carboxy-terminal extracellular domains of the bacterial adhesin intimin reach the external surface is currently unknown. The finding that EETI-CK^Send (35 aa), containing three disulfide bonds, IL-4 (128 aa), containing two disulfide bonds, and REI_v (108 aa), containing one disulfide bond, are all efficiently translocated to the E. coli cell surface under oxidizing conditions does not necessarily imply that they are able to pass the outer membrane in a partially or completely folded state. Initial experiments to address this question indicate that passenger proteins are able to fold into their native structure on the surfaces of E. coli cells and that at least for some passengers periplasmic protein folding prevents efficient bacterial surface display (Wentzel, unpublished). Hence, it may be interesting to study whether the outer membrane translocation of passenger proteins depends on the velocity of disulfide bond formation and overall protein folding.

Surface display of recombinant proteins on bacteria and yeast is a promising tool for the analysis of macromolecular interactions. One major advantage over phage display lies in the ability to use FACS for high-throughput screening of polypeptide libraries. FACS screening of a library of single-chain F_v antibody fragments displayed on Saccharomyces cerevisiae allowed the isolation of variants with femtomolar antigen binding affinity, the highest monovalent ligand binding affinity yet reported for a monovalent protein (4). Besides yeast, E. coli is the preferred organism for the display of large populations of variant polypeptides, and numerous procedures have been developed to direct peptides or proteins to and anchor them on the E. coli cell surface (for a review, see reference 16). Several requirements have to be fulfilled by a favorable display system. First, a maximum number of polypeptide molecules per cell should ideally be presented. It can be estimated that at least 10,000 fluorescent ligand molecules bound to the surface of the cell are required to achieve a sufficient fluorescence signal for FACS (5). By indirect staining using a biotinylated second antibody and streptavidin-R-phycoerythrin conjugate, which contains approximately 30 chromophores per molecule, substantial signal enhancement can be achieved, allowing detection of fewer than 1,000 antigenic sites per cell (Fig. 4). However, since the fluorescence signal-to-noise ratio of cells expressing high-affinity polypeptides to nonexpressing cells is a critical parameter for a successful FACS selection, it is desirable to reach a considerably higher level of expression. Second, expression of surface-exposed polypeptides should occur without detrimental effects on the outer membrane and without compromising cell viability. Third, the surface-displayed polypeptides should be remote from the outer membrane and the lipopolysaccharide layer in order to be freely accessible to the respective interaction partner.

These requirements are for the most part met by the intimin'-based surface display of passenger domains. We found by immunofluorescence titration with various amounts of the anti-E antibody, which recognizes an extracellular epitope at the junction of the intimin' fragment and the carboxy-terminal domain, that approximately 35,000 intimin' molecules are displayed per bacterial cell. This number is a minimal approximation, since it was calculated under the assumption of a 1:1 binding of antibody and intimin' molecule. Since one antibody molecule contains two antigen binding sites and may therefore be capable of binding two intimin' molecules, the actual number may even be higher. However, as can be seen from Western blot analysis of membrane preparations (Fig. 6), only small amounts of full-length intimin'-passenger protein fusions are found together with numerous shortened products, which are distributed over a broad size range. Since the epitope tag which is recognized by the anti-Sendai virus antibody is located at the carboxy-terminal end of the intimin'-passenger protein fusion, the observed shortened versions are likely the result of a proteolytic attack of the amino-terminal periplasmatic part and/or of the transmembrane regions connecting loops of intimin'. Despite their amino-terminal truncation, these shortened versions of the fusion protein remain anchored in the bacterial outer membrane and are capable of displaying the foreign passenger domain. Significant differences in the level of net accumulation are seen with the three passenger domains tested, where IL-4 surface accumulation is greatly reduced compared to that of the EETI-CK^Send and the REI_v passenger domain. Whether these differences are the consequences of reduced gene expression, of enhanced proteolysis, or of hampered passenger protein translocation through the cytoplasmic or outer membrane is currently under investigation.

For maintaining library diversity of passenger domain variants exposed on the cell surface by intimin' fusion, it is necessary to prevent growth competition between different library members. Moreover, in cases where synthesis of the surface-exposed protein results in reduced growth rates and cell viability, randomly occurring clones, where the synthesis of the fusion protein is impeded, can overgrow the culture in only a few generations (10). To avoid these problems, the outer membrane net accumulation of intimin'-passenger fusion proteins was adjusted to a tolerable level by reducing the efficiency of intimin' translation by introducing an amber codon in the eaeA' gene and using an amber suppressor strain as the expression host. Under conditions of fully induced intimin' passenger gene transcription for more than 100 generations, the proportion of fusion protein-producing cells in a 1:1 mixture of producing and nonproducing cells decreased approximately 10-fold, which indicates that cell viability and growth rate were only slightly compromised by permanent overexpression of the fusion protein. This finding contrasts with surface display based on Lpp-OmpA, where display of approximately 30,000 molecules per cell resulted in growth inhibition and strongly reduced culture viability (10; Wentzel, unpublished).

Daugherty and coworkers have used the tetA and araBAD promoter to control the expression of an Lpp-OmpA passenger protein (10). In an attempt to regulate gene expression by addition of submaximal amounts of inducer to the growth medium, they found heterogeneous distribution of cellular expression levels. Variation of inducer concentration yielded mixed populations of uninduced and fully induced cells. Homogenous expression levels could be obtained only under conditions where the length of induction period was varied using saturating concentrations of inducer. With the araBAD promoter, induction times of several hours are required to reach intermediate levels of expression. Expression of Lpp-OmpA fusion proteins and intimin' fusion proteins under tetA promoter control reached a maximum level after only approximately 45 min of induction (A. Christmann, unpublished results). In our hands, adjustment of gene expression levels by varying the induction time was hard to control, since minor differences in growth conditions resulted in larger experimental variations of fusion protein net accumulation. The problem could be overcome by controlling the expression of passenger proteins fused to truncated intimin at the level of translation instead of transcription. To achieve this, a glutaminyl suppressor tRNA gene was placed under P_Llac promoter/operator control on a helper plasmid, which also contained lacI, the gene encoding the Lac repressor. By varying the concentration of the inducer IPTG at saturating concentrations of tetracycline and keeping the induction duration constant, homogenous expression levels in uninduced and fully induced cells were reached by adjustment of suppressor tRNA net accumulation and efficiency of translational readthrough at the amber codon residing in the eaeA' gene. At least for regulation of gene expression by induction of transcription from the ara or tetA promoter, growth at various concentrations of inducer is not well suited to modulation of gene expression as long as expression levels reflect the proportion of cells that are fully induced rather than intermediate expression in any individual cell (10, 45, 46). Translational regulation by induction of suppressor tRNA synthesis allows us to overcome this problem and ensures a very tight repression of gene expression by dual control of target protein synthesis at the levels of both transcription and translation.

In this study we replaced the two carboxy-terminal domains of intimin which mediate the adhesion of enteropathogenic and enterohemorrhagic E. coli to target epithelia by various passenger proteins. By constructing a fusion protein of truncated intimin with a conformationally constrained integrin binding RGD peptide, the bacterial adhesin could be newly functionalized towards an adhesin that mediates binding of E. coli cells to human platelets. This was confirmed by biopanning a 1:1,000 mixture of RGD peptide-presenting cells and control cells on immobilized platelets, which resulted in enrichment of RGD peptide-displaying bacteria. However, substantial unspecific binding of control cells to the coated platelets hampered the efficient enrichment of RGD peptide-presenting cells, and six rounds of panning were required for 1,000-fold enrichment. Nevertheless, this finding confirms that the heterologous passenger domain which is fused to the shortened intimin fragment is sufficiently remote from the bacterial cell surface and lipopolysaccharide layer to allow interaction with a receptor protein residing on eukaryotic cells, which are not natural targets for E. coli cell adhesion. Only three rounds of sorting were required to achieve an enrichment of cells displaying the Sendai epitope fused to the intimin'-REI_v fusion protein from a 1:200,000 to an approximately 1:1 ratio. With modern cell sorters, event rates of up to 100,000 s¹ can be applied, which opens up the possibility of screening large molecular repertoires with over 10⁹ initial bacterial cells displaying different peptide or protein variants fused to intimin' for binders to a particular target molecule in a short time.

In conclusion, carboxy-terminal truncated intimin expressed from the newly constructed pASKInt100 display vector directs fused polypeptides to the extracellular surfaces of E. coli cells in high copy numbers without detrimental effects on the integrity of the cell envelope. It may be useful for various applications, including combinatorial library screening, study of ligand-receptor interaction, and the production of live vaccines and cellular adsorbents.