Antigen Binding to Secretory Immunoglobulin A Results in Decreased Sensitivity to Intestinal Proteases and Increased Binding to Cellular Fc Receptors

Abstract

In intestinal secretions, secretory IgA (SIgA) plays an important sentinel and protective role in the recognition and clearance of enteric pathogens. In addition to serving as a first line of defense, SIgA and SIgA·antigen immune complexes are selectively transported across Peyer's patches to underlying dendritic cells in the mucosa-associated lymphoid tissue, contributing to immune surveillance and immunomodulation. To explain the unexpected transport of immune complexes in face of the large excess of free SIgA in secretions, we postulated that SIgA experiences structural modifications upon antigen binding. To address this issue, we associated specific polymeric IgA and SIgA with antigens of various sizes and complexity (protein toxin, virus, bacterium). Compared with free antibody, we found modified sensitivity of the three antigens assayed after exposure to proteases from intestinal washes. Antigen binding further impacted on the immunoreactivity toward polyclonal antisera specific for the heavy and light chains of the antibody, as a function of the antigen size. These conformational changes promoted binding of the SIgA-based immune complex compared with the free antibody to cellular receptors (FcαRI and polymeric immunoglobulin receptor) expressed on the surface of premyelocytic and epithelial cell lines. These data reveal that antigen recognition by SIgA triggers structural changes that confer to the antibody enhanced receptor binding properties. This identifies immune complexes as particular structural entities integrating the presence of bound antigens and adds to the known function of immune exclusion and mucus anchoring by SIgA.

Introduction

Mucosal surfaces comprising the gastrointestinal, respiratory, and urogenital mucosae represent a large port of entry for most of the pathogens and thus have to be efficiently protected. This task is achieved through a combination of constitutive and innate mechanisms relying on mucus, lysozyme, lactoferrin, defensins (1), and induced specific immunity based, at the antibody (Ab)² level, on the production of secretory IgA (SIgA). In secretions, SIgA binds antigen (Ag), thus preventing its adhesion to the luminal epithelial surface and facilitating its elimination by peristalsis or mucociliary movements (2), a phenomenon called immune exclusion (3).

SIgA consists of two or four monomeric IgA linked through a joining (J) chain in association with secretory component (SC) (4). SC is derived from the polymeric immunoglobulin receptor (pIgR) that ensures selective transport of polymeric IgA (pIgA) across the epithelium to the luminal environment (5). SIgA exhibits a remarkable stability in the harsh environment of the gastrointestinal tract (6). Another feature of SIgA is its capacity to adhere selectively to microfold (M) cells in Peyer's patches (7); these latter are subsequently able to transport SIgA Ab across the epithelium and to bring them in contact with dendritic cells in the underlying mucosa-associated lymphoid tissue (8). In the form of SIgA·Ag immune complexes (IC), this translates into the onset of mucosal and systemic responses associated with production of anti-inflammatory cytokines and limited activation of Ag-presenting cells (9).

In face of the large excess of SIgA in the intestinal lumen, intrinsic passage of the Ab or IC remains limited. In this context, we previously postulated that upon binding of the Ag, SIgA experiences conformational changes that result in increased binding to the so far unidentified IgA receptor on the apical surface of M cells (7). Conformational differences in pIgA- or SIgA-based IC compared with the corresponding free Ab may also explain differential mucus-binding properties and immune exclusion (10). To address the question of Ag-mediated structural changes in pIgA and SIgA, we focused our analysis on biochemical approaches comparing pIgA and SIgA either as free Ab molecules or in association with three antigenic structures of increased complexity, namely a protein (Clostridium difficile toxin A), a virus (rotavirus), and a bacterium (Shigella flexneri).

In this study, we found that Ag-driven conformational changes can be evidenced by examining the differential sensitivity to intestinal proteases and immunoreactivity of IC compared with free Ab. Increased selective binding to known cellular receptors (FcαRI, pIgR) further revealed major changes in IC compared with free Ab. The data support the notion that binding to Ag of various sizes affects the structure of pIgA and SIgA molecules in such a manner that it makes Ag-complexed Ab molecules a better substrate for receptors that relay the function of the Ab.

EXPERIMENTAL PROCEDURES

Source of Protein

Chimeric IgAPCG-4 specific for C. difficile toxin A was produced as described (11). Ascites fluid (30 ml) was obtained from BALB/c mice inoculated intraperitoneally with 2 × 10⁶ IgA7D9 hybridoma cells (anti-VP6 rhesus monkey rotavirus) (12). IgAC5 hybridoma (anti-S. flexneri 5a lipopolysaccharide) was cultured in a T flask (BD Falcon) in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mm glutamine, 2 mm sodium pyruvate, 10 mm HEPES (pH 7.0), 0.1 mm folic acid, 100 units/ml penicillin, and 100 μg/ml streptomycin. Supernatants were collected twice a week, filtered through 0.22-μm cartridges, and concentrated to a final volume of 10 ml. Both preparations were loaded onto two successive 1-m long columns filled with high resolution Sephacryl S-300 beads (Amersham Biosciences) equilibrated and run in sterile phosphate-buffered saline (PBS: 116.3 mm NaCl, 10.4 mm Na₂HPO₄, 3.2 mm KH₂PO₄ (pH 7.4)) and coupled to an Äktaprime automated fraction collector (13). After separation by size exclusion chromatography, the fractions containing pIgA were pooled and concentrated using Centriplus 100 devices (Amicon, Millipore) and stored at 4 °C until further use. Purified pIgAPCG-4, pIgA7D9, and pIgAC5 were labeled with fluorescein using a commercial kit (NHS-fluorescein; Pierce).

Mouse secretory component (mSC) was recovered from a stable cell line (clone 2H2) by affinity chromatography on a 2-ml M2-agarose column (14). The source of human SC (hSC) was published previously (10). C. difficile toxin A was purchased from Calbiochem.

In Vitro Reassociation of pIgA and hSC or mSC

SIgA molecules were obtained by combining in vitro 10 μg of mouse pIgA molecules with 2 μg of hSC or mSC, respectively. Reassociation was performed in PBS for 30 min at room temperature as described previously (6). Integrity and proper assembly of the molecule were examined by SDS-PAGE under nonreducing conditions and immunodetection using antisera specific for the α chain, the J chain, and SC (14, 15).

Culture Conditions for S. flexneri

M90T, an invasive isolate from S. flexneri serotype 5a lipopolysaccharide (16), was grown overnight at 37 °C on BBL^TM-agar plates (3% BBL^TM Trypticase^TM soy broth (Difco), 1.5% Bacto^TM Agar (BD Biosciences), and 0.1‰ Congo red). One colony was picked and cultured in Luria-Bertani medium (1% Bacto^TM tryptone, 0.5% Bacto-yeast extracts (Sigma-Aldrich)) and 1% NaCl for 1 h at 37 °C before being spread onto Luria-Bertani agar plates (1.5% Bacto^TM Agar in Luria-Bertani medium). Bacteria grown overnight as a lawn were recovered in 0.9% NaCl, and their concentration was determined by spectrophotometry (Ultrospec 300; Pharmacia Biotech) using the following correlation: 1 A₆₀₀ corresponds to 5 × 10⁸ bacteria/ml (17).

Preparation of Rotavirus

The bovine rotavirus (RF strain) was propagated in monkey kidney MA104 cells in presence of trypsin as described previously (18). Virus titer present in the lysed cell supernatants was determined by plaque assay and expressed as plaque-forming units/ml. The virus was purified on a cesium chloride gradient as described before (19) and conserved at 4 °C until further use. For biochemical analyses, purified rotavirus solutions were desalted immediately before use by gel filtration on Sephadex G-25 coarse beads (Amersham Biosciences), and the concentration was calculated by spectrophotometry (Ultrospec 300; Pharmacia Biotech) using the following correlation: 5.3 A₂₆₀ corresponds to 1 mg/ml virus.

Formation of IC

IC were formed as follows. 100 ng of C. difficile toxin A was mixed with 50 ng of either pIgAPCG-4 or SIgAPCG-4 permitting neutralization in in vitro assays (11). 60 ng of purified rotavirus was incubated with 100 ng of either pIgA7D9 or SIgA7D9 in PBS for 2 h at room temperature, corresponding to ∼160 Ab molecules/virus (19). 10⁸ S. flexneri were incubated with 120 ng of either pIgAC5 or SIgAC5 in PBS for 25 min on ice, corresponding to ∼2000 Ab molecules/bacterium (20). These conditions guarantee that no free pIgA/SIgA are found in the IC preparations.

Digestion of Ab and IC with Mouse Intestinal Washes

A laparotomy was performed on BALB/c mice (4–6 weeks old) purchased from Harlan (Den Horst, The Netherlands). The gut was cut, and the intestinal lumen was washed with 300 μl of PBS. Final aspiration resulted in the recovery of ∼200 μl of intestinal wash, which was immediately aliquoted and frozen in liquid nitrogen (6). For in vitro digestion, 120 ng of purified pIgA or SIgA (pIgAPCG-4, SIgAPCG-4, pIgAC5, SIgAC5, pIgA7D9, SIgA7D9) or IC was mixed or not with 1 or 2 μl of intestinal washes in a final volume of 20 μl of PBS and incubated at 37 °C for either 3 or 21 h. Reactions were immediately stopped by the addition of 2 μl of Complete^TM protease inhibitor mixture (Roche Applied Science).

Western Blot Analysis

In Fig. 1, up to 400 ng of purified pIgA or SIgA was mixed in a final volume of 20 μl with SDS-sample buffer (100 mm Tris base, 4% SDS, 0.2% bromphenol blue, and 20% glycerol) containing 100 mm dithiothreitol (Applichem, Darmstadt, Germany). Samples were heated to 95 °C for 3 min and subjected to electrophoresis in 6% polyacrylamide gels. For digests originating from the free or Ag-bound pIgA or SIgA, 40 ng (α chain detection) or 80 ng (κ chain detection) per lane was separated onto 10% or 18% polyacrylamide gels, respectively. After transfer onto polyvinylidene difluoride membranes (Millipore) and blocking for 1 h in PBS and 0.05% Tween 20 (PBS-T) containing 5% nonfat dry milk, proteins were detected by incubation for 1 h with the following Ab or combination of primary and secondary Ab in PBS-T 0.5% nonfat dry milk: goat anti-human α chain IgG conjugated with horseradish peroxidase (1/3000; Sigma); goat anti-mouse IgA diluted 1/3000 (α chain-specific; Sigma) and rabbit anti-goat IgG conjugated to horseradish peroxidase (1/3000; Sigma); rabbit anti-human κ chain antiserum (1/2000; Dako, Glostrup, Denmark) and horseradish peroxidase-conjugated goat anti-rabbit IgG (1/3000; Sigma); goat anti-mouse κ chain IgG conjugated to horseradish peroxidase (Exalpha Biologicals, Watertown, MA); rabbit anti-hSC diluted 1/3000 (15) or rabbit anti-mSC diluted 1/3000 (14) followed by horseradish peroxidase-conjugated goat anti-rabbit IgG; rabbit anti-mouse or human J chain diluted 1/1000 (21) followed by horseradish peroxidase-conjugated goat anti-rabbit IgG. After washing with PBS-T, membranes were developed with appropriate secondary Ab coupled to horseradish peroxidase. Proteins on membranes were detected by chemiluminescence using Uptilight detection kit (Interchim, Montluçon, France) and exposed on autoradiographic films (Konica, Hevac Product, Villeneuve, Switzerland).

FIGURE 1.

IgA molecules used in the study. Western blot analysis under nonreducing conditions of the various pIgA and SIgA Ab is shown. Proper assembly of chimeric pIgAPCG-4 was assessed with antisera specific for human α chain (lane 1), the J chain (lane 2), and reassociated SIgAPCG-4 assayed with anti-hSC antiserum (lane 3). Mouse pIgA7D9 (lanes 4 and 5), pIgAC5 (lanes 7 and 8), and SIgA forms (lanes 8 and 9) were revealed using antisera specific for mouse α chain, J chain, and SC, respectively. The molecular mass (kDa) of immunoreactive species is shown alongside the lanes. In lanes 3, 6, and 9, the band at 80 kDa represents free SC not covalently bound to pIgA.

Enzyme-linked Immunosorbent Assay (ELISA)

96-well plates (Nunc-immuno Plate Maxisorp surface; Nalge Nunc International, Roskilde, Denmark) were coated with a range of 20–40 ng of either Ab, IC, or Ag in a final volume of 100 μl of coating buffer (PBS or NaHCO₃ (pH 9.6)) for 2 h at room temperature. After washing three times with PBS-T, wells were blocked with PBS-T containing 1% of bovine serum albumin (Fluka, Buchs, Switzerland) for 1 h at room temperature. Detection of human or mouse α- and κ chains was carried out for 1 h at room temperature with 100 μl/well of the same primary/secondary Ab/antisera in PBS-T and 1% bovine serum albumin used in Western blot assays. After final washes, revelation was performed with a 0.1 m citrate sodium solution (pH 5.0) containing 1 mg/ml o-phenylenediamine (Sigma) and 0.01% H₂O₂. The reaction was stopped by the addition of 2 m sulfuric acid, and the absorbance was measured at 492 nm with 620 nm as reference.

Cells

The human promyelocytic cell line U937 was obtained from the American Type Culture Collection. The cells were plated in 75-cm² plastic flasks containing RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mm l-glutamine, 100 μg/ml streptomycin, and 100 units/ml penicillin, at 37 °C in a humidified 5% CO₂/air incubator. The cells were subcultured every third day, at a density of 3 × 10⁵ cells/ml. 18 h prior to binding experiments, cells were stimulated with 10⁻⁸ m phorbol myristate acetate because this was shown to increase surface expression of FcαRI (22). Transfected Madin-Darby canine kidney (MDCK) cells overexpressing the human pIgR were cultured to 90% confluence on 6-well plastic culture dishes as described previously (23).

Flow Cytometry

U937 or MDCK cells were washed twice with PBS. 2 × 10⁵ cells in 100 μl of PBS containing 1% fetal calf serum (PBS-S) were preincubated with 50 μg/ml human or mouse IgG for 30 min at 4 °C to mask FcγRs. After washing with cold PBS, cells were incubated for 1 h at 4 °C with 100 μl of PBS-S containing 2 μg of fluorescein-labeled pIgA or SIgA or the same amount of IgA in the form of IC thereof. Unbound Ab or complexes were washed three times with ice-cold PBS. Cells were gently detached with a rubber policeman, allowing for viability above 90%, as determined by trypan blue staining. For binding inhibition studies, U937 cells were preincubated with 1 μg of anti-FcαRI mAb MIP8a (AbD Serotec, Düsseldorf, Germany), and MDCK cells with 10 μl of rabbit antiserum to hSC prior to the addition of fluorescent IgA or IC. Cell-associated fluorescence intensity was evaluated by flow cytometry (FACScan flow cytometer; Becton-Dickinson).

Measurement of Binding Affinity

U937 and MDCK cell-associated radioactivity was determined. Moles of bound pIgAPCG-4, SIgAPCG-4, or IC formed with the toxin A were calculated, and equilibrium dissociation constants (K_d) and number of binding sites per cell were determined by Scatchard analysis (24).

Statistical Analysis

All statistical analyses were performed using GraphPad Prism version 5. For all data, an unpaired Student' t test was applied, and the limit of significance was set at p = 0.05.

RESULTS

Proper Assembly of pIgA and SIgA for Biochemical and Binding Studies

To examine the consequences of the binding of Ag of various nature (protein, virus, bacterium) on IgA structure, we generated cognate monoclonal Ab in the two molecular forms relevant for mucosal immunity, i.e. pIgA and SIgA. Purified pIgA (IgAPCG-4, IgA7D9, and IgAC5) was analyzed by immunodetection following separation on polyacrylamide gel under nonreducing conditions (Fig. 1). Proper assembly and integrity of the molecules were checked using antisera specific for the α chain (lanes 1, 4, and 7) and the J chain (lanes 2, 5, and 8), which is essential for the association of the pIgA with SC. A combination of equimolar amounts of purified pIgA and hSC (IgAPCG-4) or mSC (IgA7D9 and IgAC5) for 1 h at room temperature resulted in the formation of covalent heterodimer complexes, as indicated by the detection of shifted SC to the position of pIgA (lanes 3, 6, and 9) when the protein was analyzed under nonreducing conditions. As reported previously, the percentage of covalence varied from one pIgA to another (11, 25) as reflected by the appearance of free SC migrating as a band of 80 kDa. When loaded onto a molecular sieving column run in PBS, co-elution of SC and pIgA confirmed that the two proteins were indeed associated in the form of covalent and noncovalent SIgA complexes (15) (data not shown).

Impact of a Protein Ligand on IgA Structure

We have reported that binding of SC to pIgA protects the Ab from in vitro degradation by intestinal washes mimicking the protease-rich environment found in vivo (6). We used the same technical approach to evaluate the impact of C. difficile toxin A binding on the sensitivity of pIgAPCG-4 and SIgAPCG-4, as reflected by the appearance of α and κ chain degradation products assayed by immunodetection (Fig. 2, A and B). In time course experiments, the susceptibility of SIgAPCG-4 to intestinal washes was reduced compared with pIgAPCG-4, with integral α chain (62 kDa band) recovered after 21 h of incubation in addition to degradation products of 40 and 36 kDa (Fd fragment). Upon association with the toxin A Ag, the pattern of digestion of the two molecular forms of the Ab was the same, with undegraded α chain detected at the 21-h end point. The C-terminal cleavage of the α chain, yielding the 36- and 40-kDa fragments, was shown to occur upon incubation with intestinal washes (6), some distance away from the Fab domains. This suggests that the susceptibility of the protein is reduced through the preferential action of conformational changes rather than masking of sensitive sites. In contrast, the integrity of the κ chain was not affected under any experimental conditions, arguing intramolecular and toxin A-mediated steric hindrance ensuring protection against the action of proteases (Fig. 2B).

FIGURE 2.

In vitro digestion pattern and changes in ELISA reactivity of IgAPCG-4 molecules bound or not to their cognate antigen. A, comparative digestion pattern after a 3- or 21-h incubation with intestinal washes of pIgAPCG-4 and SIgAPCG-4, complexed or not with C. difficile toxin A, assayed by immunodetection of the α chain. The initial intact α chain is detected as a single band of 62 kDa as shown on the right panel. B, same samples as in A, with immunodetection carried out with of anti-κ chain antiserum. The single band at 24 kDa corresponds to intact κ chain. In A and B, the numbers just above the panels indicate the duration of digestion performed at 37 °C; one representative experiment of three is depicted. No cross-reactivity of the detection Ab and toxin A is observed (right lanes). C, change in α chain and κ chain reactivity of pIgAPCG-4 or SIgAPCG-4 molecules in the absence or presence of toxin A, respectively. Data are expressed compared with the absorbance mean values obtained for pIgAPCG-4 fixed arbitrarily at 100. Bars represent mean values ± S.D. (error bars) (n = 4). Significant statistical differences are indicated above the bars.

Changes in conformation and topography in pIgA and SIgA molecules have been assessed by analyzing the immunoreactivity of the constituting polypeptides in the Ab (26). Hence, we compared the reactivity of free and toxin A-bound pIgAPCG-4 and SIgAPCG-4 with antisera specific for either the α or the κ chains by ELISA (Fig. 2C). We chose to use polyclonal antisera to permit to map the overall effect of toxin A binding on the availability of linear and conformational epitopes. For both chains, a weak (16–18%) yet highly reproducible reduction in reactivity was detected upon toxin A binding. The effect was slightly reinforced in the presence of SC in the Ab molecule (23–25% reduction), possibly accounted for by the capacity of SC to bind with toxin A (19) or masking of epitopes on the α chain. Together, this reflects changes directly associated with the impact of the bound Ag. Furthermore, the assay is complementary to that based on exposure to protease because it also implicates the κ chain in the process.

Impact of a Whole Virus Ligand on IgA Structure

We next investigated whether the variability in protease sensitivity and immunoreactivity of pIgA/SIgA we observed after association with a protein Ag would similarly occur in the context of Ag of much bigger sizes and thus epitopic complexity. In comparison with the “naked” Ab that remained degradable into low molecular mass α chain products, incubation of rotavirus with VP6 (surface-exposed virus protein 6)-specific pIgA7D9 or SIgA7D9 led to the demonstration that the Ab in either molecular form was almost fully protected against the action of proteases in intestinal washes (Fig. 3A). However, for this particular IgA, the addition of SC did not offer a better protection against proteases; this might be due to the low percentage of covalent interaction between SC and the α chain, or alternatively, to the intrinsic stability of IgA7D9 in the assay (Fig. 1). Together, these features may partially mask the impact of Ag binding in this assay. In support of a less reduced protective effect of SC associated with IgA7D9, reduction of the κ chain signal was observed after 21 h of incubation both in the absence and in the presence of SC. Tight interaction with globular rotavirus in a Fab-dependent manner maintained a stable level of κ chain comparable between lanes (Fig. 3B).

FIGURE 3.

In vitro digestion pattern and changes in ELISA reactivity of IgA7D9 molecules bound or not to their cognate antigen. A and B, pattern of digestion of pIgA7D9 and SIgA7D9 complexed or not with rotavirus after a 3-h or 21-h incubation with intestinal washes. Same line content as in Fig. 2, A and B, is shown, with the exception that pIgA7D9 and SIgA7D9, complexed or not with rotavirus, is assayed by immunodetection. The background signal due to cross-reactivity with viral antigens is shown on the two last lanes. One representative experiment of three is depicted. C, change in α chain and κ chain reactivity of pIgA7D9 or SIgA7D9 molecules in the absence or presence of rotavirus, respectively. Data are expressed compared with the absorbance mean values obtained for pIgA7D9 fixed arbitrarily at 100. Bars represent mean values ± S.D. (error bars) (n = 4). Significant statistical differences are indicated above the bars.

A clearer picture could be drawn from the immunoreactivity assay. Compared with IgAPCG-4, more marked changes in the capacity to be recognized by a chain-specific antiserum were measured for pIgA7D9 (30% reduction) and SIgA7D9 (34%) (Fig. 3C). The decrease in α chain reactivity can be explained by reduced access to the IgA-associated epitopes as well as by conformational effects limiting productive interactions with the antiserum, both resulting from coating of the virus particle by the Ab. In pIgA7D9, κ chain in close proximity with the virus Ag displayed a reduction in reactivity (18%), favoring the masking hypothesis over distant structural changes in this case. In SIgA7D9, the drop in immunoreactivity reached 22%, in support of the spatial distribution of SC wrapping the constant domains of the protein (27).

Impact of a Whole Bacterium Ligand on IgA Structure

Comparison of free pIgA/SIgA and the corresponding IC made of a protein or a viral Ag indicates that close interaction between partners on one hand and induced subtle modifications in the Ab structure on the other hand are differently affected as a function of the Ab·Ag couple. We further challenged this notion by extending our analysis to the effect of associating S. flexneri to IgAC5 specific for the bacterial surface lipopolysaccharide. Upon Ag binding, both molecular forms of the IgA Ab behaved the same, with a mixture of undigested α chain and 36- and 40-kDa degradation products detected at the two time points (Fig. 4A). This contrasted with the pattern obtained in the absence of S. flexneri, where unexpectedly, it appeared that pronounced degradation of the Ab took place at 21 h irrespective of bound SC (Fig. 4A). Although some cross-reactivity between the anti-α chain Ab and S. flexneri proteins could be observed, its contribution to the specific signal could be neglected in terms of interpretation of the results. The stability of the κ chain was demonstrated under all experimental conditions (Fig. 4B).

FIGURE 4.

In vitro digestion pattern and changes in ELISA reactivity of IgAC5 molecules bound or not to their cognate antigen. A and B, pattern of digestion of pIgAC5 and SIgAC5 complexed or not with S. flexneri after a 3-h or 21-h incubation with intestinal washes. Same line content as in Fig. 2, A and B, is shown, with the exception that pIgAC5 and SIgAC5, complexed or not with S. flexneri, is assayed by immunodetection. The background signal due to cross-reactivity with bacterial antigens is shown on the two last lanes. One representative experiment of three is depicted. C, change in α chain and κ chain reactivity of pIgAC5 or SIgAC5 molecules in the absence or presence of S. flexneri, respectively. Data are expressed compared with the absorbance mean values obtained for pIgAC5 fixed arbitrarily at 100. Bars represent mean values ± S.D. (error bars) (n = 4). Significant statistical differences are indicated above the bars.

The immunoreactivity of the α chain was about the same between pIgA and SIgA (Fig. 4C), suggesting a loose, although covalent association of SC, consistent with the limited protection observed in Fig. 4A. In contrast, a 26% reduction of κ chain reactivity occurred upon association with SC; this might result from a particular folding of the Fab domains in SIgAC5 because the first molecular model for SIgA shows surface exposure of two of four κ chains (27). The combination of the bacterium with the Ab led to a marked drop in α chain immunoreactivity for both the polymeric (91%) and secretory (92%) forms of IgAC5. A slightly less pronounced reduction for κ chain was measured for pIgAC5 (81%) and SIgAC5 (86%). In complexes with S. flexneri, these changes could be caused by enhanced epitope density recognized by the IgAC5 Ab or, alternatively, by masking of proteolytic sites in the α and κ chain.

Binding of Free IgA and IC to Cellular Receptors

Results obtained by exposure to intestinal washes and immunoreactivity argue in favor of multiple consequences on the structure of the pIgA/SIgA Ab molecules following interaction with Ag of various sizes and nature. At the biochemical level, these combined approaches led us to conclude on the dichotomy between structural changes and steric hindrance. To gain further insight into the impact of Ag binding on pIgA/SIgA molecules, the interaction with two known receptors of the Ab was evaluated using cell lines expressing FcαRI or pIgR. We focused our analyses on toxin A-pIgA/SIgA IC because surface distribution of the fluorescent Ab on the rotavirus and S. flexneri Ag would bias the assessment of the mere contact between the receptor and the Ab ligand. For the duration of the experiment, we initially checked that toxin A assayed alone was not toxic for the cells (data not shown). Fluorescein-labeled chimeric pIgAPCG-4/SIgAPCG-4 carrying human constant domains was examined by flow cytometry for its capacity to recognize FcαRI present on the surface of the monocytic cell line U937 (Fig. 5A). Consistent with previous data, both molecular forms of the Ab bound the cell line (>95% fluorescent cells), reaching uniform fluorescence levels well above background measured with medium alone. Binding, as measured by an increase in mean fluorescence intensity, of either pIgAPCG-4 or SIgAPCG-4 Ab was almost doubled in the presence of bound Ag (Fig. 5A). Association of free or antigen-bound fluorescent Ab molecules to U937 cells was strongly reduced when FcαRI was blocked by preincubation with the specific mAb MIP8a (Fig. 5A). Together, this indicates that binding of the protein Ag somehow alters the structure of either the polymeric or secretory form of the Ab in such a way that it enhances the specific association with U937 cells expressing FcαRI.

FIGURE 5.

Analysis of the binding capacity of pIgA/SIgA antibodies and immune complexes to cellular receptors. A, flow cytometry analysis of the binding capacity of fluorescently labeled pIgAPCG-4/SIgAPCG-4 complexed or not with the toxin A cognate Ag to U937 cells expressing FcαRI. Background level of fluorescence is shown in the first four lanes, and binding specificity is confirmed by inhibition with the anti-FcαRI mAb MIP8a. B, same type of analysis performed with MDCK cells overexpressing human pIgR and fluorescein-labeled pIgAPCG-4/SIgAPCG-4 associated or not with toxin A. Specificity of binding is controlled by inhibition of binding in the presence of anti-human pIgR antiserum and by the absence of binding of SIgA molecular forms. In plots A and B, numbers indicate the mean of the cell-associated fluorescence intensity ± S.D. (error bars) of four to six individual experiments evaluated by flow cytometry. Significant statistical differences are indicated above the bars.

The same set of experiments was conducted using epithelial MDCK cells overexpressing human pIgR (Fig. 5B). As expected, in the absence of toxin A protein, pIgAPCG-4 bound efficiently to MDCK cells, whereas SIgA was unable to do so as a consequence of the SC moiety present on the Ab (Fig. 5B). Similar to the picture with FcαRI, antigen association resulted in a 2-fold increment in pIgAPCG-4 Ab binding to cells expressing human pIgR. SIgA-based IC did not bind, indicating that the toxin A Ag did not perturb the arrangement of SC. Specificity of association between pIgR and pIgAPCG-4 was further confirmed in blocking experiments carried out in the presence of anti-human SC antiserum (Fig. 5B). Both cellular receptors exhibit improved binding properties for IC compared with the Ab alone. This indirectly suggests that accessibility is not limiting, and differences may involve relative affinity. Scatchard analyses revealed binding sites of higher affinity for IC compared with Ab alone (Table 1). Because the same cell lines with the same absolute number of receptors were used, this implies that more effective binding of IC occurred; structural changes triggered by the binding of the antigen are likely to be the cause of the differences observed.

TABLE 1.

Scatchard analysis of binding of pIgAPCG-4, SIgAPCG-4, and IC formed with toxin A to U937 and MDCK cells

Ligand	U937 cells Kd	MDCK cells Kd
	nm
pIgAPCG-4	15.3	9.0
SIgAPCG-4	17.7
pIgAPCG-4 + toxin A	2.4	1.1
SIgAPCG-4 + toxin A	3.3

DISCUSSION

The data reported herein demonstrate that, upon Ag binding, pIgA and SIgA experience structural changes reflected in the differential sensitivity to intestinal proteases and immunoreactivity toward α and κ chain-specific antisera. Furthermore, Ag binding increased the capacity of the Ab molecule to interact with two well documented IgA cellular receptors, namely FcαRI and pIgR. Thus, the conformational changes characterized by epitope recognition, protease sensitivity, and binding to cellular receptor have physiologic consequences particularly relevant to mucosal immunology.

The nature and size of the Ag did not seem to be crucial parameters for these structural changes to occur because we found measurable effects when using a protein, a virus, or a bacterium in association with their cognate pIgA/SIgA Ab. The formation of IC of higher size range due to an increasing number of anchoring sites present on the Ag did indeed reduce enzymatic degradation of the α chain, a phenomenon accompanied by a decrease in epitope recognition. In addition, structural “freezing” of the Ab molecule upon interaction with the Ag might as well explain the differences found with respect to sensitivity to proteases or diminished recognition by antisera. Because interaction with cellular receptors was improved, we believe that conformational changes mediated by Ag binding are principally responsible for the observed effects, with steric hindrance playing a limited role. Moreover, this is consistent with the more artificial situation where aggregated pIgA exhibited a better binding to target cells (28), possibly because of reduced dissociation of polymers as seen in our data and Ref. 29. In the context of IC, Ag recognition brings together Ab molecules and thus increases productive interactions with cellular receptors.

We reported in a previous study that IgA-based IC are taken up by intestinal M cells in Peyer's patches (20). Although the receptor(s) involved is (are) in need of identification, the phenomenon is specific for the IgA isotype (mouse IgA and human IgA2 only) and requires both the Fcα1 and Fcα2 domains of the Ab (7). In face of the large excess of free SIgA in mucosal secretions, one would have expected that sampling by M cells is an unlikely event. We speculated that preferential uptake of SIgA·Ag complexes compared with free SIgA might be due to conformational change(s) that favor(s) binding to the M cell receptor. In support of our working hypothesis, we found that binding to Ag resulted in decreased sensitivity of the Ab to the action of proteases in the intestinal washes, indicating that domains in the Fc region were most likely in a more closed conformation or at least presented less accessible cleavage sites for intestinal enzymes. The effect was limited to the C-terminal portion of the Ab molecule because the κ light chain buried in the Fab was not, or only slightly, affected by the incubation with intestinal washes. An immunoreactivity study led to a decline in α chain and κ chain recognition upon Ag binding, and the effect was even more pronounced for SIgA, as expected from the wrapping of SC around the Fc domains of pIgA (27).

Increased interaction with pIgR and FcαRI of IC further indicates that important structural modifications occurred, which might also have relevance for the sampling of Ag by M cells. If we assume a multivalent nature for the interaction between pIgA/SIgA-based IC and the M cell receptor(s), this would lead to a slow off-rate, allowing stable binding and subsequent internalization, even in the presence of the high free SIgA concentrations found in exocrine secretions. The same mechanism has been postulated to explain FcαRI-mediated phagocytosis by monocytes in serum containing large excess of circulating IgA (30).

We observed identical association of free pIgA and SIgA with FcαRI on the surface of U937 cells. This confirms previous data showing binding of either form of the Ab, yet in a less quantitative manner (31). Although residues involved in the binding to FcαRI seem to also interact with SC (32–34), the expression of the co-receptor Mac-1 (CD11b/CD18, complement receptor 3) (35) on the surface of U937 stimulated with phorbol myristate acetate is sufficient to guarantee recognition of SIgA as well. In the form of IC, both pIgA and SIgA bound more efficiently to U937 cells, indicating that conformational changes most likely relayed by the Fab domains promoted interaction with FcαRI. This is consistent with structural variability mapped in FcαRI upon interaction with the recombinant Fcα fragment consisting of domains Cα2 and Cα3 (30). The importance of the Cα2/Cα3 junction as sites of functional diversity is indirectly reflected by the fact that it appears to be a hot spot for targeting of molecules produced by pathogens; such a mechanism of subverting immunity has been documented for Staphylococcus aureus SSL7 (36), group A streptococcus Sir22 (37), and Arp4 (38), as well as group B streptococcus b Ag (39). We also found enhanced binding to pIgR of pIgA-based immune complexes compared with free pIgA. This may be rationalized by the need to transport pIgA·Ag complexes efficiently from the lamina propria to the lumen to preserve mucosal integrity (5). In contrast to our results, Kaetzel et al. (24) found similar K_d values for binding of pIgA·IC and free pIgA to rabbit pIgR. This could be due to species differences between human and rabbit pIgA-pIgR interactions: rabbit pIgA has a very high affinity for its cognate pIgR, and it may therefore not be necessary to enhance binding of IC to expedite export of foreign Ag. Possible technical explanations for these differences range from the Ag:Ab ratio used, the nature and size of the Ag, the Ab coating by the Ag, and from the number and nature of pIgR.

Together, these data indicate that subtle changes in the overall conformation of pIgA and SIgA take place upon Ag binding, resulting in modified protease sensitivity, epitope exposure, and interaction with cellular receptors. This strongly suggests that similar events can explain the differential binding of Ag-loaded SIgA compared with “free” SIgA to the still to be identified receptor at the surface of M cells in intestinal Peyer's patches. More generally, this raises the concept that when the Ab is complexed to the Ag, improved interactions with cellular receptors are established, which might eventually impact on the biological function of the target cells.