A pathogen-specific epitope inserted into recombinant secretory immunoglobulin A is immunogenic by the oral route.

Oral administration of rabbit secretory IgA (sIgA) to adult BALB/c mice induced IgA+, IgM+, and IgG+ lymphoblasts in the Peyer's patches, whose fusion with myeloma cells resulted in hybridomas producing IgA, IgM, and IgG1 antibodies to the secretory component (SC). This suggests that SC could serve as a vector to target protective epitopes into mucosal lymphoid tissue and elicit an immune response. We tested this concept by inserting a Shigella flexneri invasin B epitope into SC, which, following reassociation with IgA, was delivered orally to mice. To identify potential insertion sites at the surface of SC, we constructed a molecular model of the first and second Ig-like domains of rabbit SC. A surface epitope recognized by an SC-specific antibody was mapped to the loop connecting the E and F β strands of domain I. This 8-amino acid sequence was replaced by a 9-amino acid linear epitope from S. flexneri invasin B. We found that cellular trafficking of recombinant SC produced in mammalian CV-1 cells was drastically altered and resulted in a 50-fold lower rate of secretion. However, purification of chimeric SC could be achieved by Ni2+-chelate affinity chromatoraphy. Both wild-type and chimeric SC bound to dimeric IgA, but not to monomeric IgA. Reconstituted sIgA carrying the invasin B epitope within the SC moiety triggers the appearance of seric and salivary invasin B-specific antibodies. Thus, neo-antigenized sIgA can serve as a mucosal vaccine delivery system inducing systemic and mucosal immune responses.

The protection of the mucosal surfaces of the digestive, respiratory, and urogenital tracts is in part mediated by secretory IgA (sIgA). 1 This antibody consists of IgA dimers associated with the J chain, which is acquired during the process of polymerization in plasma cells just before secretion, and with the secretory component (SC). SC is derived from the polymeric immunoglobulin receptor (pIgR), which binds and transports polymeric immunoglobulins across mucosal and glandular epithelia. Proteolytic cleavage of the extracellular portion of the receptor releases SC together with the bound IgA into mucosal secretions (1). SC belongs to the immunoglobulin superfamily of proteins and consists of a series of five Ig-like domains corresponding to the ectoplasmic portion of the poly-Ig receptor (2)(3)(4)(5).
In addition to the protection of mucosal surfaces by crosslinking pathogens and promoting their clearance by peristalsis or mucociliary movement, sIgA in the intestinal lumen also selectively adheres to M cells. This was first observed in suckling rabbits as a local accumulation of milk sIgA on M cells of Peyer's patches (6). Subsequently, monoclonal mouse IgA, polyclonal rat IgA, and polyclonal IgG antibodies, radiolabeled or coupled to colloidal gold, were found to bind specifically to rabbit or mouse M cells and to compete with each other for binding sites (7). Selective sIgA binding and transport by M cells may play a role in the regulation of the mucosal immune response, but that role has not yet been defined. Thus, while sIgA would generally prevent the contact of antigens with mucosal surfaces, it could also promote re-uptake of small amounts of antigen by M cells.
The fact that orally administered sIgAs are efficiently transported by M cells in Peyer's patches raises the possibility that sIgA itself could serve as a vaccine delivery vector to target foreign epitopes into mucosa-associated lymphoid tissue. However, this would require that the foreign epitope be inserted without affecting the molecular folding of SC or the assembly and the function of sIgA. Also, the site of insertion must be surface-exposed, and the inserted epitope must be immunogenic.
The purpose of this study was to evaluate the feasibility of such an approach. We selected rabbit SC for epitope insertion because 1) the cDNA for the SC portion of pIgR was available (4), 2) SC was shown not to perturb the binding of sIgA to the antigen (8), 3) SC and IgA d can combine to form sIgA in vitro (9), and 4) a battery of monoclonal and polyclonal antibodies to SC is available to map the possible effects of epitope substitution on protein structure (see Table I). Our results first show a so far not identified function of domain I in the process of SC secretion, which nonetheless did not preclude either overexpression or IgA binding of chimeric SC. Second, oral immunization with "antigenized" sIgA elicits both a systemic and mucosal response against the inserted epitope, thereby suggesting that recombinant SC⅐IgA complexes could serve as a mucosal vaccine delivery system.

Monoclonal Antibodies (mAb) and Polyclonal Antisera (pAb)
The mAb and pAb used in this study and their properties are listed in Table I.
Western Blot Analysis-IgA d , SC, or pIgR immunoprecipitated with the appropriate antibody (Table I) were separated by SDS-PAGE prior to transfer onto blotting membranes. pIgR was detected using mAb 166 (1:1,000 dilution), whereas SC was detected with the guinea pig antiserum (1:500). Primary antibody binding was detected using horseradish peroxidase-conjugated secondary antibodies (1:3,000) and the chemioluminescence detection reagent (Amersham Life Sciences).
Enzyme-linked Immunosorbent Assay (ELISA)-The concentrations of purified SC and IgA d /IgA m concentrations were measured as described in Rindisbacher et al. (9) and Lü llau et al. (8), respectively. Standard curves for IgA were obtained using purified IgA d from ascitic fluid, and SC standards were generated using rabbit pIgR purified from liver.
Immunofluorescence and Confocal Laser Scanning Microscopy-MDCK cells grown to confluence on Transwell filters were washed twice with PBS and fixed in 3% paraformaldehyde for 30 min. After two PBS washes, free aldehyde groups were quenched with 50 mM NH 4 Cl in PBS for 10 min. The cells were permeabilized with 0.05% saponin in PBS, 0.2% bovine serum albumin for 10 min; this solution was used to dilute the antibodies in all subsequent steps. The preparations were incubated with IgG (20 g/ml) purified from anti-SC pAb 982 for 30 min. After extensive washes with PBS, filters were incubated with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (Amersham; 1:100), and finally mounted on glass slides in PBS, 90% glycerol, 3% n-propyl gallate. Preparations were examined under a Zeiss photomicroscope and/or by confocal laser scanning microscopy.
Intestinal Immunization with Rabbit sIgA-Four 6-week-old BALB/c mice were immunized three times at days 1, 14, and 21 by injection into the lumen of a jejunal loop containing a Peyer's patch. The injectate consisted of rabbit whey (50 l diluted 1:1 with PBS) containing about 50 g of both IgA-bound and free SC. Before the first injection, the mice were treated for 3 days with Escherichia coli lipopolysaccharide (10 ng/ml) in the drinking water to stimulate mucosal inductive sites. At day 23, mice were sacrificed, and lymphocytes from Peyer's patches were recovered and fused to P3UI myeloma cells using the procedure of Weltzin et al. (7). The supernatants of hybridomas were screened by ELISA and Western blotting.
Molecular Modeling of the First and Second Domain of the Rabbit pIgR-A model for the first and second domain of the rabbit pIgR was constructed using the approach described by Coyne et al. (12). The quality of the model was assessed by the three-dimensional/one-dimensional profile matching procedure of Lü thy (13).
Construction of Chimerized SC cDNA-A rabbit pIgR cDNA fragment (allele t61) was recovered from plasmid pSC-11 (4) by HindIII/ SauI digestion. The 1082-bp fragment served as a template for two PCRs aiming at the insertion of two unique restriction sites for subsequent sequence replacement within domain I of the rabbit SC clone. In the first reaction, an AccI site (bold characters) was created at position 418 using oligonucleotides 5Ј-GGGAAGATCTTCTAAGTATACCACA-AACTCCCCTTTTTCAGGGAAGTC-3Ј (hybridizing sequence, 429 -394) and 5Ј-AAGAGGAGAGCGGCCGCTGCGTGA-3Ј (hybridizing sequence, 314 -337). In order to facilitate subsequent ligation of PCR products, the 5Ј-primer comprised the unique NotI site at position 324, and the 3Ј-primer carried a 5Ј-end BglII site. In the second reaction, an MroI site (bold characters) was inserted at position 453 using oligonucleotides 5Ј-GGGAAGATCTTCGACTCCGGAAGCTACAAGTGTGGCGT-GGGAGTC-3Ј (hybridizing sequence, 448 -480) and 5Ј-GCACATTCAA-AGGTCACCGAGCCCC-3Ј (hybridizing sequence, 905-881). To allow subsequent cloning steps, a BglII site was added at the 5Ј-end of the 5Ј-primer, and the 3Ј-primer was selected because it contained the unique BstEII site at position 888. The 128-and 469-bp PCR products digested with BglII were ligated together, then digested with NotI and BstEII, and finally inserted into pSC-11 treated with the same two enzymes, resulting in the obtention of pSC-11(A/M). PCR-derived inserts were sequenced by the method of Sanger et al. (14).
The insertion plasmids for generating rVVs were constructed as follows. Plasmid pSC-11 was digested with NcoI at position 122 and SalI at position 1834 (putative C terminus (2)). The NcoI and SalI sites were blunted using Mungbean exonuclease, and the fragment was ligated into the blunted EcoRI site of pHGS1 (16) containing the strong p11K late promoter inserted in the body of the viral thymidine kinase (TK) gene. This resulted in the in-frame fusion of the 11K ATG to the second codon of the SC coding region. Addition of a C-terminal 6-histidine tag was carried out by annealing of oligonucleotides 5Ј-CTCAT-CACCATCACCATCACTAAAGATCTCGAGTGTAC-3Ј (coding strand) and 5Ј-ACTCGAGATCTTTAGTGATGGTGATGGTGATGAGGTAC-3Ј, followed by insertion into the KpnI site located 3 bp upstream of the SalI site of pHGS1-SC. The sequence encoded 6 histidine residues, a translational termination codon, as well as XhoI and BglII sites to determine the orientation. The resulting plasmid was designated pHGS1-SC6xHis. Plasmid pHGS1-SC(IpaB)6xHis encoding antigenized SC was obtained by cleaving plasmid pHGS1-SC6xHis with NotI and BstEII and replacing the fragment by the corresponding mutated fragment coming from plasmid pSC(IpaB).
Generation of Vaccinia Virus Recombinants-Recombinant viruses vvHGS1-SC6xHis and vvHGS1-SC(IpaB)6xHis were generated by homologous recombination into the TK gene of constructs pHGS1-SC6xHis and pHGS1-SC(IpaB)6xHis as described (9). Integration of the SC gene construct in the viral genome was checked by PCR after each round of plaque purification using the upstream primer 5Ј-GC-TACGCTAGTCACAATCACCAC-3Ј and the downstream primer 5Ј-GTCCCATCGAGTGCGGCTAC-3Ј. The amplification of a 1907-bp band confirmed the presence of the recombinant gene.
Large Scale Production and Purification of Recombinant Rabbit SC in CV-1 Cells-Stationary phase CV-1 cells at a density of 2 ϫ 10 7 cells/175 cm 2 T-flask were washed with PBS and infected with rVVs expressing SC or SC(IpaB) at a multiplicity of infection of 2. Infected cells were cultured in Dulbecco's modified Eagle's medium in the absence of serum and antibiotics for 24 h. The culture supernatant was recovered by centrifugation at 120 ϫ g for 15 min, and purification of secreted recombinant SC proteins was performed by Ni 2ϩ -chelate affinity chromatography according to the procedure given in Rindisbacher et al. (9). Abbreviations are: pAb, polyclonal antiserum; mAb, monoclonal antibody; IP, immunoprecipitation; ELISA, enzyme-linked immunosorbent assay; K, lysine; D, aspartic acid; R, arginine; T, threonine; L, leucine; I, isoleucine; E, glutamic acid; Q, glutamine.
Binding Assay-The interaction between recombinant SC and IgA was determined by dot blot reassociation assay as described in Rindisbacher et al. (9). Murine IgA d and IgA m were obtained from hybridoma ZAC3 (8).
Measurement of IgA Transport by MDCK Cells in the Co-culture Assay-The procedure is based on the technique developed by Hirt et al. (10). MB-2/B7 hybridoma cells secreting IgA were cultivated as described (7). MDCK cells stably transfected with a pLKneo vector containing the wild-type or antigenized rabbit pIgR cDNAs were induced to express the receptor with 1 ϫ 10 Ϫ6 M dexamethasone or left unstimulated. Accumulation of IgA and SC in both the apical and basolateral compartments was measured 24 h post-induction by ELISA. Transcytosis was expressed in ng, respectively, g of protein/24 h/filter.
Mucosal Immunization with Reconstituted dIgA-SC(IpaB)-350 g of SC(IpaB) and SC(wt) were associated in vitro with 1.75 mg of IgA d purified from ZAC3 hybridoma (8). Reconstituted IgA d ⅐SC complexes were purified by FPLC onto a 140 ϫ 1.6 cm Superdex 200 column (8). Fractions containing the reconstituted sIgA were pooled and concentrated by filtration using Centriprep-50 units (Amicon, Inc.) equilibrated in PBS. IgA d ⅐SC antibodies were sterilized by passage through a 0.2-m pore size Acrodisc filter (Gelman Sciences) and stored at 4°C until use. 100 g of wild-type and chimerized IgA d -SC antibodies were mixed with 10 g of cholera toxin (Calbiochem) and administered in 200 l of PBS into germ-free BALB/c mice by intragastric intubation via PE10 polyethylene tubing connected to a 1-ml syringe. Two groups of three mice were challenged 4 times at day 1, 10, 44, and 51, and blood and saliva were collected 10 days after the last immunization. Biological fluids were kept frozen until use.

RESULTS
Rabbit sIgA Orally Administered to Mice Is Immunogenic-Secretory IgA is known to bind to M cells and to be transported into the underlying organized mucosal lymphoid tissue (7). To determine whether orally administered sIgA from rabbit whey was taken up and processed in mouse Peyer's patches, we recovered Peyer's patch lymphoblasts from immunized animals, fused them with myeloma partners (7), and screened the hybridoma cells for production of immunoglobulins specific for rabbit SC and sIgA. Out of 60 hybridoma clones, 10 reacted with SC by ELISA, out of which 4 recognized free and bound SC in rabbit whey on Western blots. One clone produced polymeric IgM, two clones secreted IgG1, and one clone secreted IgA. IgA from clone 36.4 recognized both free and IgA-bound SC blotted on nitrocellulose membrane (Fig. 1, lane 3), but only recognized immunoprecipitated free SC when immobilized on Sepharose beads (Fig. 1, lane 4). This indicates that mucosally administered sIgA is able to elicit both a mucosal and systemic immune response against SC.
Computer Modeling of SC-In order to identify sites in the SC polypeptide that would be suitable for epitope insertion, we constructed a molecular model for the first and second immunoglobulin-like domain of rabbit SC (Fig. 2). The three-dimensional structure of two IgG light chains and the first domain of CD4 T cell accessory molecule were used to build the molecular model, which was subsequently used to select a site at the surface of rabbit SC to insert a foreign epitope.
Selection of an Epitope Replacement Site on SC-In rabbit, three distinct SC alleles, t61, t62, and t63, have been identified (17,18). The molecular model indicates that an 8-amino acid long sequence extending from leucine 79 to aspartic acid 86 ( Fig. 3A) with the strongest allelic variation (Ͼ60% divergence when compared with other regions (2, 4)) is located within the loop connecting the E and F ␤ strands. The loop is opposite to the stretches corresponding to the immunoglobulin-like complementary determining regions (CDR) involved in IgA binding (12). We reasoned that since the loop tolerates important sequence variations, it might represent an ideal site for foreign epitope insertion.
In order to confirm that the loop indeed corresponded to the allele-specific epitope, we tested the ability of a synthetic peptide corresponding to amino acids 76 -87 to compete with binding of a t61 allele-specific antiserum raised in t62 rabbits (17) against purified rabbit SC. The antiserum was incubated with 0, 140, or 280 ng of peptide, and the degree of competition was assessed by Western blotting and quantitated by scanning (Fig.  3B). Under non-reducing conditions, binding of the t61 allelespecific antibody to rabbit t61 SC was only partially competed by the peptide. Competition was complete under reducing conditions, yet the affinity of the antibody for reduced SC is known to be lower (17). This suggests that the anti-t61 antiserum recognizes a discontinuous surface epitope that includes all or part of the loop.
Selection of a Pathogen-specific Epitope-Several of the genes involved in invasion of cultured cells by Shigella flexneri have been identified and designated as ipa for "invasion plasmid antigen" (19,20). The IpaB protein is exposed on the bacterial surface and results in antibody production in infected animals (21). For insertion into SC, we selected the immunodominant linear B cell epitope consisting of residues Lys-Asp-Arg-Thr-Leu-Ile-Glu-Gln-Lys of IpaB (22). This sequence changed the length of the original loop by only a single amino acid, is specifically recognized by mAb H10 (22), and remains accessible in the context of a ␤-galactosidase-IpaB fusion protein containing the entire IpaB polypeptide (Fig. 3C, lane 3).
Cloning of SC cDNA Sequences in a Vaccinia Virus Insertion Vector-Rabbit SC cDNA was engineered to contain the coding sequences for amino acids 1-571 fused to a C-terminal histidine hexamer tag (Fig. 4A). Following cloning into VV insertion plasmid pHGS1 (16), the fragment was introduced by homologous recombination into the TK locus of the viral genome. The presence of the rabbit SC cDNA insert in the viral genome was ascertained by PCR utilizing transfer vector-specific primers that generated unique amplification products of the expected sizes (Fig. 4B). Insertion of the IpaB epitope into the E-F loop of SC domain I was performed using PCR as described under "Experimental Procedures." Insertion of a Pathogen-specific Epitope into the First Domain of SC Reduces Its Secretion Rate by Infected CV-1 Cells-Heterologous production of SC was performed by infecting CV-1 cells with rVVs vvHGS1-SC6xHis and vvHGS1-SC(IpaB)6xHis at a multiplicity of infection of 2 for 24 h. These conditions have been shown to be optimal for the expression of human SC using an identical approach (9). Secretion of wildtype or chimeric SC was assessed by immunoblotting. After 24 h of culture, wild-type SC could be detected in as little as 5 l of cell culture supernatant (Fig. 5A, lane 3), whereas no chimeric SC was detected under the same conditions (Fig. 5A,  lane 1). 100-fold concentration of the medium was required to generate a chimeric SC signal of similar intensity to that obtained with cells infected with SC(wt) (Fig. 5A, lane 2), suggesting that processing and/or secretion was blocked within the cells. To test this hypothesis, cell extracts were prepared in a final volume corresponding to that of the culture medium and analyzed for the presence of SC(IpaB)6xHis. 5 l of extract generated a signal intensity on Western blot comparable with that of supernatant of vvHGS1-SC6xHis-infected cells. The chimeric intracellular SC migrated significantly faster than secreted wild-type SC on SDS-PAGE (Fig. 5A, lane 4), reflecting incomplete glycosylation and subsequent block in transport as a consequence of the alteration of the E-F loop in domain I. However, affinity purification of recombinant SC(IpaB) could be achieved on a Ni 2ϩ -NTA-agarose column (Fig. 5B) using the conditions given in Rindisbacher et al. (9).
Trafficking of the pIgR in MDCK cells Is Altered by the Insertion of the Foreign Epitope-We next examined whether the dramatic effect on secretion of replacing the E-F loop by an IpaB epitope observed in recombinant SC would hold true for the full pIgR. Receptor biosynthesis and its routing along the secretory pathway were analyzed in MDCK cells stably transfected with wild-type and chimeric receptors under the control of a glucocorticoid-inducible promoter (15). Transfected MDCK cells were grown to confluency on Transwell filters, and pIgR expression was induced by incubation with dexamethasone for 24 h. Intracellular accumulation of the receptor over the induction period was similar for the wild-type and chimeric proteins (Fig. 6A, lanes 1-4). In contrast, secretion of the chimeric SC into the apical medium was drastically reduced when compared with wild-type SC (Fig. 6A, lanes 5 and 6), in agreement with the results obtained for secretion of chimeric SC by CV-1 cells. When the stably transfected MDCK cells were examined by confocal laser scanning microscopy, wild-type pIgR exhibited strong perinuclear and cytoplasmic, as well as cell surface labeling (Fig. 6B). In contrast, chimeric pIgR was exclusively intracellular with a complete absence of surface staining (Fig.  6B). These data are consistent with an early block along the secretory pathway resulting in accumulation of endoglycosidase H-sensitive chimeric pIgR or SC in the rough endoplasmic reticulum or the cis-Golgi compartment (data not shown). Pulse-chase experiments were performed to compare the kinetics of wild-type and chimeric pIgR secretion into the apical medium. Intracellular production of pIgR(wt) and pIgR(IpaB) was followed over time (Fig. 6C, upper part). The receptor was recovered from cytoplasmic extracts chased for 0, 40, and 160 min by immunoprecipitation with antibody mAb 166. The pool of newly synthesized receptor following the pulse was the same for the wild-type and chimeric protein. With increasing chase times, however, no shift in the apparent molecular weight of the chimeric receptor was observed. No secreted chimeric SC protein could be immunoprecipitated with pAb 982 in the supernatant of MDCK cells expressing the protein, indicating that secretion was affected in an epithelial cell line as it was in fibroblast-like CV-1 cells (Fig. 6C, lower panel). The same antibody precipitated secreted wild-type SC in supernatants taken after a 160-min chase.
The Chimeric pIgR Is Unable to Transport Dimeric IgA across MDCK Cells-Transport of IgA d across stably transfected MDCK cells and the recovery of secreted sIgA were analyzed using a system in which the transfected MDCK cells were co-cultivated with IgA producing hybridoma cells (10). The amounts of SC (SC ϩ sIgA) and of IgA (IgA ϩ sIgA) were measured in the apical medium by ELISA (Fig. 7A). The rate of IgA d transported by cells expressing the wild-type pIgR increased linearly to a plateau at 24 h, and ϳ150 ng IgA/filter were recovered from the apical medium. In contrast, no IgA or sIgA was detected in apical media from cells transfected with the chimeric receptor cDNA. The susceptibility of the chimeric receptor to cleavage was not increased, as reflected by the absence of chimeric SC in the basal medium. IgA d production by the hybridoma cells was not rate-limiting (Fig. 7B).

Insertion of a Foreign Epitope into the First Domain Did Not Compromise Specific
Binding to IgA-The binding of wild-type SC or chimeric SC to mouse IgA d or IgA m was measured by dot blot reassociation assay (9). Both SC and SC(IpaB) maintain their ability to interact with IgA d immobilized on filters (Fig.  7C) or free in solution (data not shown). This indicates that while epitope insertion within the E-F loop affects pIgR trafficking in the cell, it does not preclude interaction with IgA d . We thus concluded that although domain I carries information for IgA binding (12,23), it contains at least one region whose substitution dramatically impairs intracellular maturation of the pIgR without affecting IgA binding.
The Pathogen-specific Epitope Inserted in SC Is Recognized by Its Cognate Monoclonal Antibody-To determine whether the amino acid substitution in the E-F loop altered the local structure in domain I, we tested whether the chimeric SC could still be recognized by the IpaB-specific mAb H10. Wild-type or antigenized SC in supernatants from rVV infected CV-1 cells were concentrated by immunoprecipitation using antibodies pAb 982, mAb 303, or mAb H10 and by trichloroacetic acid precipitation. Western blotting analysis showed that mAb H10 recognized antigenized SC in solution with an efficiency similar to pAb 982 and mAb 303. H10 failed to precipitate the wild-type protein, which was however immunoprecipitated by incubation with pAb 982 and mAb 303. Trichloroacetic acid-precipitated samples showed the same patterns when Western blots were probed with anti-SC antibodies, confirming that immunoprecipitation of recombinant SC was quantitative (Fig. 8A).
Orally Administered IgA d -SC(IpaB) Is Immunogenic-In order to demonstrate that chimerized reconstituted sIgA can serve as a mucosal delivery system, we immunized mice 4 times over a period of 2 months with 100 g of IgA d -SC(IpaB) together with cholera toxin as an adjuvant. Controls consisted of immunization with 100 g IgA d -SC(wt)/cholera toxin and PBS buffer. Control and immune sera were collected and tested by ELISA against lysates of Shigella strains containing or lacking IpaB and against recombinant SC(IpaB). High titers of antibodies to SC were obtained in sera from mice challenged with either IgA d -SC(IpaB) or IgA d -SC(wt) (Fig. 8B, left panel), thereby confirming the immunogenicity of SC when delivered mucosally. A strong specific response against the IpaB-containing lysate could be detected by incubation of the sera of the mice immunized with IgA d -SC(IpaB) (Fig. 8B, middle and right  panels). This indicates that serum antibodies in mice immunized with IgA d -SC(IpaB) can recognize the IpaB epitope in its native environment. Background levels were obtained with the sera of animals immunized with IgA d -SC(wt) or PBS alone (Fig. 8B). Sera from mice immunized with IgA d -SC(IpaB) also FIG. 3. Panel A, sequence comparison of the rat and rabbit (alleles t61 and t62) pIgR between between amino acids 76 and 89. Boxed residues form the E-F loop in domain I and exhibit the highest variability in the molecule. Mutations in the loop created by substitution with the IpaB peptide are shown for comparison. Panel B, titration of the anti-allelic t61 antiserum binding to native SC with increasing amounts of a synthetic peptide composed of amino acids 76 -87 from the E-F loop. Western blots of unreduced and reduced SC were incubated with anti-t61 serum mixed with increasing amounts of peptide, and antibody binding was detected with secondary antibody coupled to horseradish peroxidase. The relative intensities of the signals arbitrarily set at 100 in the absence of competitor peptide were quantified by densitometry. Panel C shows that the mAb H10 recognizes a fusion protein consisting of E. coli galactosidase and S. flexneri invasin B. bind to the free IpaB peptide, but not to the unrelated FLAG (Eastman Kodak) peptide (Fig. 8C). Saliva samples of two mice challenged with IgA d -SC(IpaB) exhibited a reproducible, albeit weak, binding to Shigella lysates containing IpaB (Fig. 8D). Antibodies against SC were detected in all saliva samples of animals immunized with IgA d -SC(IpaB) or IgA d -SC(wt) (Fig.  8D). Data in Fig. 8B-D thus demonstrate that gastric delivery of antigenized sIgA can serve as a vaccine system capable of eliciting both a systemic and a mucosal antibody response. DISCUSSION The rationale for using SC as an epitope delivery system was based on the observation that oral administration of rabbit milk sIgA in mice triggered an immune response with antibodies directed against SC. As a basis to identify a site for epitope insertion at the surface of SC, we have predicted the threedimensional structure of domain I. We selected a site mapping to a loop situated between the E and F ␤ strands (residues 79 -86) opposite to the three CDR-like stretches in SC recently shown to be involved in IgA binding (12). This sequence also varies considerably among the different alleles of the rabbit pIgR. To test the hypothesis that epitopes inserted at this site would be immunogenic, this sequence was replaced by a S. flexneri IpaB epitope for which a monoclonal antibody was available (22).
Chimeric SC containing a Shigella invasin-specific epitope in its first domain retained the capacity to bind IgA d but not IgA m both in solution and on membrane support. Although insertion of the epitope did not alter the rate of synthesis of the chimeric SC or pIgR, the intracellular routing was compromised by the insertional mutation. This was reflected by a drastic reduction in the secretion rate of chimeric SC by rVV-infected CV-1 cells and by the absence of polymeric IgA transepithelial transport by MDCK cells expressing the receptor.
It has been established that the determinants required for non-covalent binding of polymeric IgA are restricted to the first most distal of the five immunoglobulin-like domains of pIgR and SC (24,25). Amino acid substitutions in the three CDR-like loops in the first domain, and in the loop bridging strands E and F, drastically reduced IgA binding capacity of the pIgR (12). In contrast, the substitution in the E-F loop of amino acids 79 -86 by the IpaB sequence did not affect binding to IgA d (this study). In addition, the mutation did not seem to induce significant long range structural changes in domain I since mAb 303, specific for a conformational epitope in the first domain (25), was able to bind as efficiently to chimeric and wild-type SC. This is in agreement with the results reported by Coyne et al. (12) in which mutations in the E-F loop of the pIgR did not affect binding of the same monoclonal antibody. Provided that the spatial organization of the E-F loop and CЉ-D loop is stabilized by a salt bridge between the beginning of the D strand (Arg-63) and the end of the E-F loop (Asp-86), this suggests that Asp-86 plays a role in positioning the loop on the surface of the molecule rather than serving as a point of electrostatic interaction between SC and IgA.
Clearly, this salt bridge plays a crucial role in the folding of the first domain since its loss perturbs the intracellular trafficking of the chimeric SC or pIgR although IgA binding remains intact. For instance, interactions between domains I and II might be affected and hence interfere with SC and pIgR secretion. Alternatively, in the t61 allele-encoded protein, Nglycosylation takes place at a site 14 residues upstream of the E-F loop (26); conformational changes might thus perturb this process and lead to retention by calnexin or BiP (27) of immature forms of the protein. Therefore, while the correct folding of the pIgR first domain is not required for binding to IgA d , it is required for intracellular trafficking, most likely for exit from the endoplasmic reticulum compartment. Based on these considerations one should insert pathogen-specific epitopes elsewhere in SC to try to obtain a secretion-competent chimeric molecule. We are currently addressing this issue by mutating other loops and ␤ strands in domains II and III. Indeed, the alternatively spliced low M r pIgR (28) with deletion of domains 2 and 3 is functional (29) and binds non-covalently IgA d . This suggests that these domains should be ideal alternative sites for peptide insertion or for replacement by larger foreign sequences containing both T and B epitopes.
Oral immunization is the most effective way to stimulate mucosal immunity in the intestine and at more distant sites including the genital tract and the lungs (30,31). Usually, however, most soluble protein antigens are poorly immunogenic when given orally due to the systemic hyporesponsiveness, or oral tolerance, that is naturally generated (32). Mucosal adjuvants, such as the cholera toxin (CT) from Vibrio cholerae or the heat labile toxin from E. coli, are known to break tolerance (33,34). Based on the observation that sIgA can bind to and travel across M cells in Peyer's patches (7) and that CT favors switch differentiation to IgG1-and IgA-secreting cells (35,36), we combined antigenized sIgA with CT to prevent tolerance. Under these conditions, we have been able to demonstrate that an epitope derived from a pathogenic bacte- FIG. 8. Oral immunogenicity in BALB/c mice of antigenized sIgA. Panel A, recognition of pIgR(IpaB) and pIgR(wt) by antibodies directed against SC or the IpaB epitope. Supernatants of CV-1 cells infected with the rVVs vvHGS1-SC(IpaB)6xHis and vvHGS1-SC(wt)6xHis were subjected to immunoprecipitation with pAb 982, mAb 303, and mAb H10 or precipitated with trichloroacetic acid (TCA). Detection of SC was performed by Western blotting with guinea pig pAb against SC domains II and III. Panel B, specific and control sera binding to the indicated antigen measured by ELISA. Groups of mice are: 1, mice immunized with IgA d -SC(wt); 2, mice challenged with IgA d -SC(IpaB); and 3, control mice given PBS. Results are mean Ϯ S.D. of two separate triplicate experiments involving groups of three mice and are expressed as absorbance values at 492 nm. Panel C, binding specificity of the antisera to the IpaB peptide, measured as in panel B. Panel D, specific binding of saliva samples recovered from mice in groups 1, 2, and 3. As stated in the text, only two mice in group 2 responded to the IpaB epitope compared with three against SC, which is reflected by the larger error bar for these samples. rium inserted into recombinant sIgA survives in the proteolytic environment of the gut and elicits both a systemic and mucosal antibody response. Thus, SC associated with IgA d can serve as a delivery vehicle for oral vaccination by preserving the immunogenicity of the inserted epitope. Interestingly, in the absence of adjuvant, keyhole limpet hemocyanin feeding, followed by subsequent subcutaneous immunization, induced systemic T cell tolerance but induced B cell priming at both systemic and mucosal sites (37). Whether a similar mechanism might take place with antigenized sIgA remains to be determined. Since topically administered sIgA can be used for the prevention of viral and bacterial infection (38 -41), it now becomes possible to combine passive immunization with active mucosal vaccination.