Mapping the Interaction Between Murine IgA and Murine Secretory Component Carrying Epitope Substitutions Reveals a Role of Domains II and III in Covalent Binding to IgA*

We have identified sites for epitope insertion in the murine secretory component (SC) by replacing individual surface-exposed loops in domains I, II, and III with the FLAG sequence (Crottet, P., Peitsch, M. C., Servis, C., and Corthésy, B. (1999) J. Biol. Chem. 274, 31445–31455). We had previously shown that epitope-carrying SC reassociated with dimeric IgA (IgAd) can serve as a mucosal delivery vehicle. When analyzing the capacity of SC mutants to associate with IgAd, we found that all domain II and III mutants bound specifically with immobilized IgAd, and their affinity for IgAd was comparable to that of the wild type protein (IC50 ∼ 1 nm). We conclude that domains II and III in SC are permissive to local mutation and represent convenient sites to antigenize the SC molecule. No mutant bound to monomeric IgA. SC mutants exposing the FLAG at their surface maintained this property once bound to IgAd, thereby defining regions not required for high affinity binding to IgAd. Association of IgAd with SC mutants carrying a buried FLAG did not exposede novo the epitope, consistent with limited, local changes in the SC structure upon binding. Only wild type and two mutant SCs bound covalently to IgAd, thus implicating domains II and III in the correct positioning of the reactive cysteine in SC. This establishes that the integrity of murine SC domains II and III is not essential to preserve specific IgAd binding but is necessary for covalency to take place. Finally, SC mutants existing in the monomeric and dimeric forms exhibited the same IgAdbinding capacity as monomeric wild type SC known to bind with a 1:1 stoichiometry.

The first line of defense against pathogens consists of mucosal secretions, with secretory IgA being one of the main effectors. The latter is transported by the polymeric immunoglobulin receptor (pIgR) 1 from submucosal sites into the lumen. The pIgR is expressed by epithelial cells in a variety of mucosal tissues and in rodents is further implicated in transport into bile through the liver (1). The receptor can bind dimeric IgA (IgA d ), polymeric IgA (IgA p ), and IgM (which are produced by subepithelial plasma cells) but not monomeric IgA (IgA m ) or IgG. The complex is internalized at the basolateral surface of the epithelium and transcytosed to the apical plasma membrane, where the extracellular portion of the receptor called secretory component (SC) is released by proteolytic cleavage and remains bound to IgA within a complex termed secretory IgA (sIgA) (2,3).
All three loops, referred to as B-C, D-E, and F-G in domain I are particularly implicated in binding to IgA (4,5); in addition, some residues of loop E-F in the opposite face appear to be important for IgA binding. In contrast, deletion of domains II and III or individual deletion of domain IV and V of rabbit pIgR did not prevent binding to IgA in a qualitative cell-ligand binding assay (5). A natural variant of rabbit SC lacking domains II and III can bind to IgA, although only in a noncovalent fashion (6,7). In human sIgA, a disulfide bridge within domain V is also involved in covalent binding to the C␣2 domain of IgA through a disulfide exchange mechanism (8 -10). Species variations in the level of covalency between pIgR and IgA have been reported (11,12). Thus, it is tempting to speculate that although not critical for the initial recognition of IgA, domains II to IV appear to position domain V such that disulfide exchange with IgA can take place.
The mechanism for the selective recognition of IgA d over IgA m remains an additional open question. We have generated murine SC mutants with predicted exposed loops of domains I, II, and III replaced with the FLAG epitope (13) to evaluate their IgA binding properties. We show that although the affinity of these mutants for IgA and the selectivity for IgA d over IgA m are both preserved, covalent binding is lost in all mutants but one. Given the pinpointing strategy designed here, this suggests that domains II and III are essential to the proper positioning of domain V. However, because binding to the anti-FLAG mAb is preserved in both free and IgA d -bound reactive SC mutants, we postulate that minor structural changes occur upon IgA d -mSC interactions. Dimeric SC-FLAG mutants were shown to be as good IgA d binders as their monomeric counterparts. In the same assay, we found by direct comparison of mSC and pIgR mutants that mSC binds to IgA d with much reduced stringency. We conclude that mSC-FLAG mutants are useful 1) to study the topology of SC⅐IgA d complexes, 2) as short epitope carrier once combined with IgA d , 3) to tackle the binding specificity and stoichiometry of SC/pIgR IgA d complexes.

EXPERIMENTAL PROCEDURES
The structure and production of the various recombinant proteins and most analytical procedures as well as the source of most antibodies and reagents are described in the accompanying article (13).
Recovery of Recombinant mSC and mpIgR-COS cell culture supernatants containing mSC proteins were harvested as described (13). In some experiments, cells were incubated for 36 h with fresh medium containing 2.5 mM 2-mercaptoethanol (Life Technologies, Inc.), and the culture supernatants were supplemented with 30 mM iodoacetamide. Murine SC proteins were recovered from the supernatants by concanavalin A-agarose chromatography (Vector Laboratories; Ref. 13) and stored at 4°C in PBS with 0.02% (w/v) sodium azide. Such preparations were used for all the experiments. Cell lysates containing mpIgR proteins were obtained as described (13).
Antibodies-Purified IgA d and IgA m from the murine hybridoma ZAC3 (14) were kind gifts of Dr. E. Lü llau (Ecole Polytechnique Fédérale, Lausanne, Switzerland). Murine IgG1 MOPC-31c, biotinylated goat IgG to murine ␣ chain, mAb to human SC, and horseradish peroxidase (HRP)-conjugated reagents were obtained from Sigma. Rabbit antiserum to mSC and murine mAb to the FLAG were as described (13).
Biotinylation of Proteins-To specifically label sialic acid residues, 250 g of affinity-purified mSC-FLAG:Cterm (36) in 0.5 ml of 0.1 M sodium acetate (pH 5.5) were incubated with 2 mM sodium periodate on ice in the dark for 30 min (15). Oxidation was quenched with 15 mM glycerol, which was then removed by several washes with 0.1 M sodium acetate (pH 5.5) using a Centricon-50 filtration unit (Amicon). Biotin-LC-hydrazide (Pierce) in dimethyl sulfoxide was added to a 5 mM final concentration and incubated at room temperature for 2 h. The buffer was changed to PBS containing 0.02% (w/v) sodium azide using a Centricon-50 filtration unit. Proteins were stored at 4°C and used within 2 weeks following biotinylation. Purified IgA d from hybridoma ZAC3 (250 g) was biotinylated using the same procedure with the exception that 10 mM sodium periodate was used for oxidation of the sugar moieties.
Two mg each of purified rabbit IgG against unfolded mSC, respectively against the native form of mSC, were biotinylated using sulfo-NHS-LC-biotin (Pierce) as described by the manufacturer and recovered and stored as indicated above for the mSC and IgA d proteins.
Binding of mSC-FLAG Mutants to Immobilized IgA-The wells of Nunc MaxiSorp ELISA plates were coated with 50 l of either purified IgA d (5 g/ml) or IgA m (2.5 g/ml) dissolved in PBS. Control wells were coated with murine IgG (Sigma, 2 g/ml) or left uncoated. Wells were blocked with 0.2 ml of Tris-buffered saline (25 mM Tris-HCl, 137 mM NaCl, 2.7 mM KCl (pH 7.5)) containing 5% (w/v) nonfat dry milk and 0.05% (w/v) Tween 20 (Bio-Rad). Murine SC prepared by concanavalin A lectin chromatography (13) at saturating concentration (500 ng/ml) in 0.1 ml of PBS were incubated for 1 h at room temperature. After three washes with Tris-buffered saline containing 0.05% Tween 20, either rabbit IgG was applied to native mSC (10 g/ml) or mAb M2 was applied to the FLAG sequence (10 g/ml) for 1 h at room temperature. Bound antibody was detected using HRP-conjugated secondary antibodies (1:1,000 dilution) with 1,2-phenylenediamine as the chromogen. Reactions were stopped with one volume of 2 M H 2 SO 4 , and plates were read at 492 nm using 620 nm as the reference wavelength.
Competitive Binding of mSC Proteins to Immobilized IgA-Microwells were coated with IgA d and blocked as indicated above. The following steps were performed at 4°C. 0.1 ml of biotinylated mSC-FLAG:Cterm (100 ng/ml) was added to each well, and serial 2-fold dilutions of mSC-FLAG mutants were rapidly pipetted from a 20 g/ml stock before incubation for 1 h. Control wells were incubated without competitor or without biotinylated SC. After washing, the bound biotinylated mSC was detected with HRP-conjugated ExtrAvidin diluted 1:1,000 (Sigma). After the last wash, the plate was brought to room temperature and developed as above. Results in Fig. 2 are expressed as percent of maximal biotinylated SC binding, which was obtained in the absence of competitor. The IC 50 of the various proteins were defined as the concentration of competitor that inhibited by 50% the binding of biotinylated mSC-FLAG:Cterm.
Binding to Immobilized IgA of mSC-FLAG Mutants after Incubation with mAb M2-500 ng of mSC-FLAG mutants were incubated for 1 h on ice in 50 l of PBS in the presence of either mAb M2 or the murine IgG MOPC-31c as a control, both at 40 g/ml. The SC-FLAG⅐Ab complexes were then applied to microwells coated with IgA d , blocked as described above, and incubated for 1 h at 4°C. After washing, 50 l of biotinylated IgG was applied to native mSC (40 g/ml, protein A, purified) for 1 h at 4°C and detected with HRP-coupled ExtrAvidin at 1:1,000 dilution.
Murine SC⅐IgA Co-immunoprecipitation-Identical amounts (corre-sponding to 200 ng of protein) of mSC mutants purified by concanavalin A-agarose chromatography were incubated with 1 g of biotinylated IgA d in 50 l of PBS for 1 h at room temperature. Such conditions have been shown to permit full covalent association of SC⅐IgA d in vitro (16). The volume was adjusted to 500 l, 80 l of streptavidin-agarose slurry (Amersham Pharmacia Biotech) in PBS was added, and the tubes were rotated for 4 h at 4°C. Beads were pelleted by gentle centrifugation and washed four times with TENT buffer (50 mM Tris-HCl (pH 7.5), 5 mM EDTA (pH 8.0), 150 mM NaCl, 1% Triton X-100) before boiling in gel-loading buffer (100 mM Tris-HCl (pH 6.8) 4% (w/v) sodium dodecyl sulfate, 0.2% (w/v) bromphenol blue, 20% (v/v) glycerol). Samples were separated in nonreducing, denaturing 6% polyacrylamide gels before analysis by immunoblotting using rabbit antiserum to mSC and HRPconjugated mouse anti-rabbit IgG and the chemiluminescence assay (Amersham Pharmacia Biotech).
Cell lysates containing mpIgR or mpIgR-FLAG:D2:BC mutant were subjected to the same protocol except that biotinylated IgA d was incubated in the overlay in place of IgA d .

mSC-FLAG Mutants Keep Binding Selectivity for IgA d over
IgA m -We have generated a series of mSC molecules wherein loops predicted to be surface-displayed had been replaced with the FLAG epitope (13). The IgA binding capacity of the various mSC-FLAG mutants was determined by ELISA. IgA d or IgA m purified from the mouse hybridoma ZAC3 (14) was coated onto microwells and incubated with a 20-fold molar excess of the various SC-FLAG. Equilibrium was achieved within 1 h of incubation. 2 Fig. 1A shows that all mutants and wild type mSC were able to recognize IgA d , with the exception of mSC-FLAG: D1:FG, used as a negative control. Given that mutation of rabbit pIgR domain I loop F-G abolishes binding to IgA (5), the selective binding of other mutants validates the option to select for substitution sites in domains II and III and reflects the sensitivity of the assay. Thus, the loops of mSC that carry the FLAG do not contain essential determinants necessary for specific recognition of IgA d despite their superficial localization in the molecular models. Furthermore, the lack of binding of the mutants to immobilized IgA m indicates that specific recognition of IgA d is preserved (Fig. 1A).
Accessibility of the FLAG Epitope in IgA d -bound SC Resembles That of Unbound SC-We next tested using ELISA whether the FLAG epitope was still displayed on the surface when the mutants are bound to IgA d . Fig. 1B shows that the epitope was detectable in mSC-FLAG:Cterm, mSC-FLAG:D2-D3, or mSC-FLAG:D2:BC associated with IgAd to an extent similar to the free form (13). Other mutants not reacting with the anti-FLAG mAb M2 as unbound material remained undetectable. No signal was obtained with mSC-FLAG:D1:FG, since the latter does not bind to coated IgA d . This observation together with the absence of anti-FLAG reactivity following coating with IgA m 2 demonstrates the specificity of the assay. Results in Fig. 1A and these data show that we have designed mutant SC proteins that can both carry epitopes exposed or buried in recombinant sIgA. In terms of assembly, this indicates that the environmental changes induced by the epitope do not alter the binding of the mutants to IgA d , most likely reflecting a remote location with respect to the IgA polypeptides.
Epitope Replacements In Domains II and III of mSC Preserve High Affinity Binding to IgA d -To quantitatively determine the relative affinity of the various SC mutants for IgA d , we next performed competition ELISAs (16) using mixtures of a fixed amount of biotinylated mSC-FLAG:Cterm and increasing amounts of each individual mutant. Fig. 2 shows that mutants displaying the FLAG sequence within domains II and III or at the carboxyl terminus were as good inhibitors as wild type recombinant mSC. Murine SC-FLAG:D1:FG was not inhibitory at the highest concentration tested, in agreement with data in Fig. 1. The concentration needed to reach 50% inhibition of binding of biotinylated mSC-FLAG:Cterm (IC 50 ) was about 1 nM for wild type mSC and 1-2 nM for all the mutants except mSC-FLAG:D2-D3, which had a slightly lower affinity (6 nM; Table I). These results are consistent with domains II and III not being required for high affinity recognition of IgA d (Fig. 1,  A- Fig. 3 shows that binding was not inhibited in any case and was even significantly enhanced for mutant mSC-FLAG: D3:DE. The lack of inhibition seen for mSC-FLAG:Cterm, mSC-FLAG:D2-D3, and mSC-FLAG:D2:BC is consistent with the FLAG being at the surface before and after binding to IgA d and domains II and III exhibiting a degree of flexibility not affecting IgA d recognition. As expected, mutants not recognized by the mAb M2 as free proteins keep binding IgA d efficiently. The increased binding of mSC-FLAG:D3:DE in the presence of mAb M2 can be explained by the formation of a ternary complex (IgA d ⅐SC⅐M2), as detected by sieving chromatography under native (PBS) conditions, 3 with M2 having induced conformational changes favoring association.
The Presence of the FLAG Impairs the Covalent Binding to IgA d in Most Mutants-In sIgA isolated from mouse milk, 70 -80% of mSC is covalently linked to IgA d (19). We thus examined to which extent SC-FLAG mutants can form covalent complexes with IgA d . Equimolar amounts of biotinylated IgA d and wild type mSC or mSC-FLAG mutants were incubated for 1 h at ambient temperature before precipitation with streptavidin beads. Bound material was resolved by SDS-PAGE under nonreducing conditions, and the presence of free and IgA dassociated mSC-FLAG was determined by immunoblotting (Fig. 4). Monomers and dimers of all mutants were co-immunoprecipitated with IgA d , confirming that data in Fig. 1   binding, long range effects cannot be excluded that are not compatible with covalent binding through a reactive disulfide bond in domain V of human SC. It is likely then that despite a preserved capacity to recognize each other, mSC mutants and IgA d polypeptides do not optimally position domain V for the disulfide interchange reaction to take place (9). Dimerization could be an additional barrier to the flexibility of the mSC molecule, and significantly, no mSC-FLAG mutant that makes dimers ( Fig. 4 and Table I) associates with IgA d in a covalent manner, yet they exhibit the same binding capacity as their monomeric counterparts. (13) suggests that partial renaturation of SC occurs on the blotting membrane. This raises the possibility of examining, using an overlay assay, whether IgA d in solution reassociates preferentially with either mutant SC monomers, dimers, or both. Fig. 5A shows that several mSC mutants could be recognized by IgA d . No signal was detected when IgA m was present in the overlay. 2 IgA d recognized both dimeric and monomeric mSC-FLAG:D2:BC, monomeric mSC-FLAG:D2:DE, dimeric and to a lesser extent monomeric mSC-FLAG:D2-D3, mSC-FLAG:Cterm, and wild type mSC. The pattern of proteins present was verified after stripping of the membrane and immunoblotting with IgG to mSC. 2 The partial discrepancy with binding data obtained using mSC-FLAG proteins in solution (Figs. 1 and 2) together with the efficiency of the assay when using wild type or mSC-FLAG:Cterm suggests that the mere presence of the FLAG precludes renaturation of a selection of mSC mutants. The assay is, however, reliable and convenient to rapidly test supernatants containing recombinant SC or mutant pIgR in cell extracts.

Differential IgA d Binding to mSC-FLAG Molecular Forms Previously Separated by SDS-PAGE-The relative inaccessibility of the FLAG to immunodetection after SDS-PAGE under nonreducing conditions and transfer
IgA d Recognizes SC and pIgR in a Different Manner-We then tested the binding of IgA d to wild type mpIgR and the mpIgR-FLAG:D2:BC mutant. Although the SC counterpart of both proteins was recognized by IgA d , only the wild type pIgR can serve as a binding site on the membrane (Fig. 5B). After stripping, immunodetection of the same blot with IgG to mSC detected both proteins, including monomeric and dimeric mpIgR-FLAG:D2:BC. 2 This result is at variance with the recognition by IgA d of both monomeric and dimeric mSC-FLAG: D2:BC (Fig. 5A) and argues in favor of a reduced binding stringency of SC as compared with pIgR. Thus, it appears that the mode of recognition of SC and of the complete receptor may differ.

Biosynthesis of mSC-FLAG Proteins in the Presence of 2-Mercaptoethanol Modifies Their
IgA-binding Properties-The presence of the reducing agent 2-mercaptoethanol during culture did not alter the ability of mSC-FLAG mutants to form dimers FIG. 3. IgA binding properties of mSC-FLAG mutants preincubated with mAb M2. mSC-FLAG proteins were mixed with anti-FLAG mAb M2 or with a control IgG antibody as described under "Experimental Procedures" (see inset for bar identification). After application to microwells coated with IgA d , binding was detected using biotinylated IgG to native mSC and HRP-coupled streptavidin. Data are the mean Ϯ S.D. of triplicate wells and are representative of two independent experiments.

FIG. 4. Composition and covalency of complexes formed in solution between mSC-FLAG mutants and IgA.
Murine SC-FLAG mutants were incubated with an equimolar amount of biotinylated IgA d , followed by precipitation with streptavidin-agarose beads. Proteins fractionated on a 6% nonreduced SDS-PAGE were blotted and probed with IgG to mSC and detected using a HRP-labeled anti-rabbit IgG antibody. The nature of the polypeptides is marked along the figure. but changed the immunoreactivity of the epitope in some instances (13). We determined the effect of this treatment on IgA d binding properties of the mutants. As seen in Fig. 6A, mSC-FLAG:D2-D3 was the only mutant whose ability to bind immobilized IgA d was affected significantly by this treatment. The control mutant mSC-FLAG:D1:FG did not bind to IgA d , and no mutants bound above background levels to IgA m . Fig. 6B shows a similar experiment using mAb M2 for detection of the FLAG in mSC mutants bound to IgA d . Strikingly, the FLAG was well detected in IgA d -bound mSC-FLAG:D2-D3 despite the poor binding detected using the antibody to mSC, suggesting an exacerbated surface exposure of the FLAG. A strong signal was also obtained with mSC-FLAG:D3:CCЈ, in sharp contrast to what was seen when the protein was produced under normal conditions (Fig. 1B). In this case, however, the treatment also increased the immunoreactivity of the FLAG in the free protein (13). It is possible, therefore, that the C-CЈ loop is no longer constrained by a disulfide bond under these conditions. Thus, reduction has selectively altered the structure of some mutants in a manner that is unraveled upon IgA d binding. Fig. 6C shows a comparative overlay analysis of proteins that had been synthesized in the presence or absence of reducing agent. Monomeric mSC-FLAG:D2:BC and both forms of mSC-FLAG:D2-D3 have lost their capacity to be recognized in this assay. Therefore, the effect of exposure to 2-mercaptoethanol was more pronounced on the capacity of the mutants to renature (13) than on their ability to bind IgA under native conditions (Fig. 5A). The mere alkylation of the samples during the isolation procedure (see "Experimental Procedures") cannot account for this effect because the same proteins in solution were able to bind IgA d . DISCUSSION We have studied the IgA binding properties of a panel of mSC mutants with the FLAG octapeptide replacing SC sequences (13). To our knowledge, this is the first report studying the interaction between murine IgA and murine SC. In terms of the structure-function relationship, we demonstrate that the integrity of domains II and III of mSC is not essential for IgA d recognition. In contrast, it is necessary to permit formation of covalent binding between mSC and IgA d through optimal positioning of domain V in the molecule. Furthermore, although the mSC carrying sequence changes in domain II and III remains an excellent IgA d binder, mutated pIgR cannot accom-modate such changes; this suggests that the multi-stage process leading to specific, stable recognition of IgA d is different for the two forms of the molecule. Together, these data shed new insights into the complex nature of the interaction between IgA d and SC/pIgR.
Although mSC was produced as a recombinant protein, we have recently shown that, except for its glycosylation, recombinant hSC cannot be distinguished from milk hSC (16,20). All the mSC-FLAG proteins mutated outside domain I bound with high affinity to IgA d , which corroborates with the previous finding that the first domain of rabbit and human pIgR and SC carries the most important information for ligand recognition (5,17,18). Nevertheless, although domain I of rabbit SC is both necessary and sufficient for binding IgA (17), domain III of bovine SC contributes to binding (21), and domains II-V of hSC add to its affinity for IgA (22). Using a competition assay, we found that the concentration for half-maximal inhibition (IC 50 ) was about 1 nM for wild type mSC and 1-6 nM for all mutants except mSC-FLAG:D1:FG (Table I). Hence, the presence of the FLAG in domains II-III or at the carboxyl terminus of domain V of mSC did not impair IgA d binding and neither did the dimerization of some mutants. Thus, domain I of mSC is responsible for high affinity binding to IgA d , and the mutations performed do not interfere with binding determinants. We have measured an IC 50 of 10 nM for the interaction taking place between either recombinant or milk-derived hSC and the same murine IgA d (16). Thus, it appears that mSC has a higher affinity than hSC for murine IgA d . In the same setting an IC 50 of 3-30 nM (4,18,22) has been reported for the interaction of hSC isolated from milk with human IgA p . Furthermore, relative affinities for human IgA p of bovine and rabbit SC were in the range of 1 nM, whereas that of rat SC was about 40 nM (4). Scatchard analysis yielded a K D of 10 nM for the binding of hSC to human IgA d (23), rabbit SC to rabbit IgA d (24), or rabbit IgA d to rabbit pIgR (25) and a K D of 2.5 nM for the binding of human IgA d to rabbit pIgR (26).
Wild type mSC and mSC-FLAG mutants were selective for IgA d over IgA m , as was our mutant of loop E-F in rabbit SC (27). Recombinant hSC domain I was shown to bind IgA m but not IgG (18), suggesting that this domain discriminates among Ig classes, whereas the other domains are responsible for distinguishing the oligomeric state of the Ig. Thus, the loops mutated in domains II and III of mSC do not contain any motif important for the recognition of Ig polymers. In contrast, Bakos et al. (22) showed that tryptic peptides encompassing the first two or first three domains of hSC could bind both IgA p and IgM but did not recognize IgA m or IgG. A peptide comprising domain II alone could not bind to either molecule. Although such discrepancies could be intrinsic to species differences, we favor the hypothesis that our approach, based on specific sequence modifications, does not drastically perturb the overall conformation and allows more accurate pinpointing of the structurefunction relationship in the complex.
As in most species (1), mSC is partially bound in a covalent manner in sIgA isolated from milk (19). Mapping of sIgA isolated from human colostrum indicated that SC was disulfidebridged via domain V to the C␣2 domain of IgA p (10). Most mSC-FLAG mutants bound IgA d in a noncovalent manner only, indicating that the positioning of the reactive bond was not optimal. The absence of covalence was also seen when rabbit SC lacking domains II-III is bound to IgA d (6,7) or when the g subclass of rabbit IgA is bound by full-length SC (11). Notably, none of the mSC-FLAG mutant forming dimers could bind covalently to IgA d , suggesting that their overall flexibility was hindered; this represents further evidence that one SC molecule has to be present per sIgA complex. Interestingly, in the two mSC-FLAG mutants that bound IgA d in a covalent manner, the FLAG was located at the end of domain V or was closest to the carboxyl terminus (loop D-E of domain III).
One aim of this study was to engineer mSC-FLAG mutants capable of binding to IgA d and still exposing the epitope on their surface. Our approach relied on the specific replacement of SC sequences with the FLAG epitope that can be traced using a specific mAb. We found that the FLAG was best available in IgA d -bound mutants that already exposed the epitope in their free form (Ref. 13; Fig. 1 and Table I); insertion at the carboxyl terminus of domain V and in between domains II-III represented optimal locations. Excess mAb to the FLAG did not inhibit the binding of mSC proteins to immobilized IgA d and, in the case of mSC-FLAG:D3:DE, even enhanced binding. Studies using mAbs have shown that IgA-induced conformational changes occur in domains II-III of hSC (28). Together, the data reflect the flexibility of the SC in that portion of the molecule and suggest that domain II and domain III do not establish close contacts with the IgA backbone. In contrast, domains II-III might be essential to trigger the "zipper effect" initiated by the interaction of domain I with IgA and progressing to domain V. Given the very different binding properties of mpIgR-FLAG:D2:BC and mSC-FLAG:D2:BC, it is also possible that integral domains II-III are dispensable in SC but manda-tory in pIgR. Together, the data demonstrate that mSC "antigenized" in domains II and III, once combined with IgA d , represents an adequate option to stabilize short antigen sequences for oral administration.
Binding assays based on blot overlay rely on the capacity of proteins exposed to SDS to refold into a functional state and have been applied frequently in studies involving cytosolic proteins. Examples of membrane-bound ecto-proteins that have been probed by overlay include the low density lipoprotein receptor (29), the asialoglycoprotein receptor (30), high density lipoprotein-binding proteins (31), and hyaluronan binding proteins (32,33). Furthermore, the cytosolic domain of pIgR is also bound by calmodulin in overlay experiments (34,35). We have found that wild type mSC and mpIgR as well as several mSC-FLAG mutants could be recognized by IgA d in solution (see Table I). Since all the mutants but mSC-FLAG:D1:FG were able to recognize IgA d under native conditions, we conclude that only the renaturation of mutants mSC-FLAG.D3:CCЈ and mSC-FLAG.D3:DE were impaired following boiling in SDS, fractionation, and blotting. Notably, both hSC denatured in 6 M guanidinium-HCl or rabbit pIgR boiled in 2% SDS can subsequently regain IgA binding activity (22,25). Overlay assays of extracellular domains are sensitive to reduction but favored by blocking the membrane with nonionic detergents (29,30,32). In this respect, wild type mSC and mutants thereof, which contain many essential disulfide bridges, behaved accordingly. In contrast to mSC-FLAG:D2:BC, the mpIgR-FLAG:D2:B2 mutant was not recognized by overlaid IgA d despite sharing the same extracellular domain. Given that wild type pIgR and mSC-FLAG:D2:BC are recognized by IgA d , it is unlikely that refolding of the ectodomain of mpIgR-FLAG:D2:BC is affected by the presence of transmembrane and cytosolic domains. We rather favor the hypothesis that the mode of IgA binding differs between SC and pIgR, yet preserves the preference for IgA d over IgA m .
Mutants that were secreted with culture medium containing 2-mercaptoethanol were still dimerizing, and the exposure of the FLAG was not dramatically affected (13). These proteins were generally able to bind IgA d under native conditions (the binding of mSC-FLAG:D2-D3 was reduced), but their ability to be recognized in the overlay assay was altered. Thus, the assay revealed subtle folding defects in these mutants that were induced by the reducing agent.
In conclusion, we present evidence that surface loops in domains II and III of mSC do not affect binding to IgA d , in contrast to what was observed for the same loops in domain I (5,27). Antibody reactivity of the FLAG motif in domains II and III loops in free and IgA d -bound mSC mutants maps multiple insertion sites for epitope substitution. A simple, rapid binding assay we developed based on FLAG-substituted molecules allowed us to dissect the different binding properties of SC and pIgR. The topological study presented here establishes that the integrity of mSC domains II and III is not essential to preserve specific IgA d binding but is necessary for covalency to take place.