Recognition of Gram-positive Intestinal Bacteria byHybridoma- and Colostrum-derived SecretoryImmunoglobulin A Is Mediated by Carbohydrates*

Received for publication, December 3, 2010, and in revised form, March 9, 2011 Published, JBC Papers in Press, March 21, 2011, DOI 10.1074/jbc.M110.209015

Amandine Mathias and Blaise Corthesy1

From the R&D Laboratory of the Division of Immunology and Allergy, Centre Hospitalier Universitaire Vaudois, Rue du Bugnon,1011 Lausanne, Switzerland

Humans live in symbiosis with 1014 commensal bacteriaamong which >99% resides in their gastrointestinal tract. Themolecular bases pertaining to the interaction between mucosalsecretory IgA (SIgA) and bacteria residing in the intestine arenot known. Previous studies have demonstrated that commen-sals are naturally coated by SIgA in the gut lumen. Thus, under-standing how natural SIgA interacts with commensal bacteriacan provide new clues on its multiple functions at mucosal sur-faces. Using fluorescently labeled, nonspecific SIgA or secretorycomponent (SC), we visualized by confocal microscopy theinteraction with various commensal bacteria, including Lacto-bacillus, Bifidobacteria, Escherichia coli, and Bacteroidesstrains. These experiments revealed that the interactionbetween SIgA and commensal bacteria involves Fab- and Fc-independent structural motifs, featuring SC as a crucial part-ner. Removal of glycans present on free SC or bound in SIgAresulted in a drastic drop in the interaction with Gram-positivebacteria, indicating the essential role of carbohydrates in theprocess. In contrast, poor binding of Gram-positive bacteria bycontrol IgG was observed. The interaction with Gram-negativebacteria was preserved whatever the molecular form of proteinpartner used, suggesting the involvement of different bindingmotifs. Purified SIgA andSC fromeithermouse hybridoma cellsor human colostrum exhibited identical patterns of recognitionfor Gram-positive bacteria, emphasizing conserved plasticitybetween species. Thus, sugar-mediated binding of commensalsby SIgA highlights the currently underappreciated role of gly-cans in mediating the interaction between a highly diversemicrobiota and the mucosal immune system.

Humanmucosal surfaces comprising themouth, respiratory,digestive, and urogenital tracts represent �400 m2, i.e. 200times more than the global skin area. The human gastrointes-tinal tract is peacefully colonized by a large ecosystem esti-mated to belong to�1800 genera, which represents�1014 bac-teria, exceeding by more than 10 times the body cells (1, 2).Overall, the intestinal immune system has the dual task to pro-tect the sterile core of the organism against invasion and dis-

semination of pathogens and maintain a peaceful relationshipwith commensal microorganisms. To preserve mucosal homeo-stasis, a complex communication needs to be establishedbetween a narrow triptych: the microbiota, the epithelial cells,and the mucosal immune system.Secretory IgA (SIgA)2 produced by plasma cells in the lamina

propria represents the major immunoglobulin found at muco-sal surfaces. The protective role of SIgA has been well estab-lished in the context of infection where the antibody (Ab) actsas a first line of defense through bacterial coating, thus largelypreventing attachment to epithelial surfaces and resulting in aprocess referred to as immune exclusion (3). In contrast, tomaintain an abundant and well balanced gutmicrobiota, such aclearance mechanism must be limited to a level guaranteeinghomeostasis. Evidence is accumulating that emphasizes a com-plex cross-talk between the epithelium and microbiota thattriggers SIgA secretion in the gut lumen of neonates already (4,5). In contrast, SIgA production is reduced at barely detectablelevel in germ-free animals, whereas normal values of IgA can bereached within a few weeks following intestinal recolonizationwith various microbiotas (6–8). Recently, data demonstratedthe induction of strain-specific SIgA secretion following rein-troduction of Enterobacter cloacae in the gut of specific patho-gen-freemice, indicating a direct impact of this microorganismon the subjacent immune cells (9). Furthermore, SIgA has beendescribed to promote biofilm formation at the gut surface,underlying a straight relationship linking mucosal Abs and thegut microorganisms (10, 11). However, the molecular mode ofaction of SIgA in regulating microbiota colonization remainsenigmatic. One can speculate that interaction between SIgAand commensals plays a role in modulating the colonization bythe microbiota in steady-state conditions. Moreover, in vivocoating of commensal bacteria by SIgA has been described inanalysis of human feces (12, 13). Because abundant intestinalsecretion of natural SIgA with unknown specificity has alsobeen described, we speculated that the latter can be involved inbinding to commensals (8, 14–16).SIgA is mostly composed of dimeric IgAmade of twomono-

mers linked together with J chain and secretory component

* This work was supported by Swiss Science Research Foundation Grant3200-122039.

1 To whom correspondence should be addressed: R&D Laboratory, Dept. ofImmunology and Allergy, University State Hospital (CHUV), Rue du Bugnon46, 1011 Lausanne, Switzerland. Tel.: 0041-21-314-07-83; Fax: 0041-21-314-07-71; E-mail: blaise.corthesy@chuv.ch.

2 The abbreviations used are: SIgA, secretory IgA; Ab, antibody; BL, Bifidobac-terium lactis; Bt, B. thetaiotaomicron DSM 2079; Cy3, indocarbocyanine-3;D2241, E. coli strain D2241; dg suffix, deglycosylated; hSC, human SC; LPR,Lactobacillus rhamnosus; LSCM, laser scanning confocal microscopy; mSC,mouse SC; Nissle, E. coli strain Nissle 1917; pIgA, polymeric IgA; SC, secre-tory component; SCcol, colostrum-derived SC; SIgAcol, colostrum-derivedSIgA; ST11, Lactobacillus paracasei.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 286, NO. 19, pp. 17239 –17247, May 13, 2011© 2011 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

MAY 13, 2011 • VOLUME 286 • NUMBER 19 JOURNAL OF BIOLOGICAL CHEMISTRY 17239

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

(SC), the extracellular degradation product resulting fromcleavage by the epithelial cells of the precursor polymeric Igreceptor ensuring transcytosis of the Ab (17). In secretions, SCis bound covalently, as well as noncovalently, to IgA, and isfound also as free SC (18). Both polymeric IgA (pIgA) and SCare heavily glycosylated (19, 20); remarkably, N- and O- carbo-hydrate residues have already been implicated in the interac-tion with bacteria, participating in the protection againstenteric pathogens (20–25) either as adhesion competitors or inanchoring SIgA in themucus to ensure optimal biological func-tion. In face of the increasing involvement of glycans in thefunction of SIgA, we sought to determine their potential role incommensal binding (25, 26). As substitute of natural SIgA andSC, we used nonspecificmouse hybridoma-derived SIgA, pIgA,and free SC, as well as their deglycosylated counterpart, toexamine the role of carbohydrates in binding to a selectionof bacteria (Lactobacillus, Bifidobacteria, Bacteroides, or Esch-erichia coli strain). The results obtained with recombinant pro-teins were validated with SIgA and SC purified from humancolostrum.

EXPERIMENTAL PROCEDURES

Microorganisms and Growth Conditions—Lactobacillusparacasei ST11 (NCC2461 provided by Nestle Research Cen-ter, Lausanne, Switzerland) and Lactobacillus rhamnosus LPR(NCC4007 provided by Nestle Research Center, Lausanne,Switzerland) were grown inMan-Rogosa-Sharpe broth at 37 °Cwithout shaking (27). Bifidobacterium lactis BL818 (BL)(NCC2818 provided by Nestle Research Center, Lausanne,Switzerland) was cultured in Man-Rogosa-Sharpe comple-mented with 0.05% L-cysteine under microaerophilic condi-tions (AnaerogenTM; Oxoid) (28). E. coliD2241 (29) and Nissle1917 (Nissle) (27) strains were grown in brain heart infusionbroth at 37 °Cwith shaking.Bacteroides thetaiotaomicronDSM2079 (Bt) was culture in anaerobic conditions as described (30).Microorganisms from frozen stocks were incubated in theirrespective growth media for 16 h, with the exception of Bt keptfor 48 h. Bacteria were washed three times in phosphate-buff-ered saline (PBS) by successive centrifugations at 2000� g for 5min. Assessment of colony-forming unit (cfu)/ml was obtainedby multiplying the optical density (O.D.) measured at 600 nmby their respective conversion factor: one O.D. corresponds to1 � 108 cfu/ml for Lactobacillus strains, 1.5 � 108 cfu/ml forE. coli strains, and 5 � 107 cfu/ml for Bifidobacteria. Bt wasnumbered in Neubauer chambers.Cell Lines and Protein Production—Hybridoma cells produc-

ing IgAC5 (31) or IgGC20 (32) specific for Shigella flexneri sero-type 5a LPS, IgASal4 specific for Salmonella typhimurium LPS(33), and IgAHNK20 specific for the respiratory syncytial virusfusion (F) glycoprotein (34) were cultured in RPMI 1640medium supplemented with 10% fetal calf serum, 2 mM gluta-mine, 2 mM sodium pyruvate, 10 mM HEPES (pH 7.0), 0.1 mM

folic acid, 100 units/ml penicillin, and 100 �g/ml streptomycinas described (31, 32). Cell cultures were conducted in Celline-350 bioreactors (Intregra Biosciences AG). Supernatants wereharvested twice a week, filtered through 0.22-�m cartridges,and separated by size-exclusion chromatography to recoverpIgA from other molecular forms and undesired ingredients

(35). Fractions containing pIgA forms were pooled, concen-trated using a Pellicon XL filter unit (100-kDa cut-off; Milli-pore) coupled to a Labscale system (Millipore), and finallystored at 4 °C. IgGC20 was purified by affinity chromatographyusing protein G-Sepharose 4 fast flow beads (GE Healthcare)according to the manufacturer’s instruction. Prior to storage at4 °C, the sample buffer was exchanged against PBS by filtrationover an Amicon Ultra 100K cartridge (Millipore). Mouse SC(mSC) was produced and purified as described (36). The puri-fied protein was stored in PBS at 4 °C until use. The bicin-choninic acid protein assay kit (Pierce) was used for proteinmeasurement. Fluorescent molecules (pIgA, IgG, mSC) wereobtained using the FluoroLink mAb indocarbocyanine-3 (Cy3)labeling kit (Amersham Biosciences) according to the proce-dure provided by themanufacturer. SIgA-Cy3were obtained bycombining pIgA-Cy3 molecules with mSC (5/1 (w/w)) in PBSaccording to the conditions defined in Crottet and Corthesy(37). The effective reassociation between pIgA and mSC pro-teins was checked by SDS-PAGE under nonreducing condi-tions using sera specific for mouse � chain, J chain, or SC (38).Preparation of SIgA and SC from Colostrum—Aliquots of

individual frozen human colostral samples were centrifuged at13,000� g for 20min at 4 °C to remove lipids and cells (39). Thelower layer containing proteins was recovered and sterile-fil-tered. Delipidized colostrum was applied on a Superdex 200column (100 � 2.6 cm; GE Healthcare) equilibrated in PBS.Colostrum-derived SIgA (SIgAcol) and SC (SCcol) were inden-tified, respectively, in the first and the third eluting peaks andcomigrated with leftovers of unresolved IgG and lactoferrin.Pooled fractions were concentrated using Amicon Ultra 30Kcartridge to reach a final volume of 10 ml. Contaminating IgGwas removed using batch incubation with 2 ml of proteinG-Sepharose 4 fast flow beads (GE Healthcare) equilibrated inPBS; SIgA and SC were collected as unbound material. Thebuffer of the SIgA or SC solutions was exchanged for 20 mM

sodiumphosphate buffer (pH7.3) using filtration throughAmi-con Ultra 30K cartridge. To get rid of residual lactoferrin, theSIgA and SC materials were then fractionated, respectively,onto 2 ml of Q-Sepharose fast flow beads (GE Healthcare) andSP-Sepharose fast flow (GE Healthcare) equilibrated in 20 mM

sodium phosphate buffer. Stepwise elution with a range of KClconcentration (0.05–0.5M)was performed, allowing for elutionof SIgAcol between 0.05 M and 0.2 M KCl and SCcol between0.05 M and 0.1 M KCl. Pooled SIgAcol- or SCcol-containingfractions were desalted by addition of 10 volumes of PBS andconcentrated using an Amicon Ultra 30K cartridge prior tostorage at 4 °C. To ensure that Q- and SP-chromatographyproperly separated them, the resulting absence of lactoferrinwas assessed by ELISA (Lactoferrin ELISA kit; Calbiochem).The integrity of recovered SIgAcol and SCcol was checked bySDS-PAGE under nonreducing conditions. SIgAcol and SCcolwere labeled with Cy3 as described above for hybridoma-de-rived proteins.Protein Deglycosylation—Five units of N-glycosidase F (EC

3.5.1.52; Roche Applied Science) were added to either 5 �g ofreassociated SIgA-Cy3 or 1 �g of mSC-Cy3 and incubated at37 °C for 4 h under gentle agitation, resulting in deglycosylatedSIgAC5 (SIgAC5dg), deglycosylated SIgASal4 (SIgASal4dg),

SIgA-Commensal Interactions Mediated by Glycan Residues

17240 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 19 • MAY 13, 2011

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

and deglycosylated mSC (mSCdg), respectively. Deglycosyla-tion was examined by SDS-PAGE under reducing conditions,and immunodetection was carried out using antisera againsteither � chain or SC. To obtain fully deglycosylated SIgAcol(SIgAcoldg) or deglycosylated SCcol (SCcoldg), further addi-tion of 5 units of N-glycosidase F was carried out after 3 h ofincubation, and the mixture was incubated for 3 more hours.Protein Analysis—2 �g of protein was used for silver staining

(35), 200 ng for � chain detection, and 600 ng for other detec-tions. Samples weremixed with gel loading buffer (100mMTrisbase, 4% SDS, 0.2% bromphenol blue, and 20% glycerol), and forreducing conditions, dithiothreitol (DTT; Applichem) wasadded to a final concentration of 100 mM. Samples were heatedto 95 °C for 3 min and applied onto polyacrylamide gels ofappropriate polyacrylamide percentages as indicated in eachfigure legend. Western blot assay was performed as described(38), with proteins detected by incubation for 1 h with the fol-lowing primary and secondary Abs in PBS containing 0.05%Tween 20 and 0.5% nonfat dry milk: goat anti-mouse � chain(1/3000; Sigma), goat anti-human� chain (1/3000;Cappel), andrabbit anti-goat IgG conjugated to horseradish peroxidase(HRP) (1/3000; Sigma); rabbit anti-human � chain (1/2000;Dako), rabbit anti-human SC (hSC; 1/3000; Dako), rabbit anti-mSC (1/3000) (36) or rabbit anti-J chain (1/1000) (40) followedbyHRP-conjugated goat anti-rabbit IgG (1/5000; Sigma); rabbitanti-human � chain HRP-conjugated (1/5000; Dako). Afterfinal washing in PBS containing 0.05% Tween 20, proteins weredetected by chemiluminescence using Uptilight detection kit(Interchim) and exposed on autoradiographic films (Konica).Protein Association with Bacteria—2 � 107 bacteria were

mixed with 200 ng of SIgA, SIgAdg, or pIgA, or with 40 ng ofmSC and mSCdg, in a final volume of 400 �l of PBS and incu-bated for 1 h at room temperature under gentle agitation. Bac-teria-protein complexes were washed three times in PBS, laidonto 8-well slides (Marienfeld), fixed in 2% paraformaldehydein PBS for 25 min, and mounted in Vectashield (Vector Labo-ratories). Complexes were observed using a Zeiss LSM 510Meta confocal microscope (Carl Zeiss) with a 63 � objective(Cellular Imaging Facility, Lausanne University, Switzerland)and processed using the Zeiss LSM 510 Meta software.Quantification of Bacterial Coating by Antibodies—Quanti-

fication of bacterial coating by antibodies was performed withImageJ software 1.41 (National Institutes of Health). We mea-sured on each series of pictures the area covered by bacteria(bacterial area) using the differential interference contrastchannel. In parallel, in each area associated with bacteria, wequantify the area containing Ab- or SC-linked pixels with afluorescence intensity superior to 15 units (fluorescent area)defined in our experimental settings as the background signal(signal range extends from 0 to 256 units using transformationinto 8-bit gray scale). The following formula was used to quan-tify the percentage of bacterial surface coveredwith fluorescentmolecules: 100 � (fluorescent area)/(bacterial area).Statistical Analysis—The results are given as means � S.E.

Two-tailed nonparametric Mann-Whitney U test analysis wasperformed using the GraphPad 5 Prism software. Differenceswere considered as significant when p values �0.05 wereobtained.

RESULTS

SIgA Binds Bacteria in a Fab- and Fc-independent Manner—Coating of bacteria has been previously observed in feces sam-ples recovered from human andmouse (12, 41), suggesting thatthis takes place at all times in the gut lumen. We made theassumption that nonspecific SIgA serving as substitute of nat-ural SIgA can bind to commensals in a Fab-independent fash-ion.We tested this hypothesis by incubating in vitro a battery of

FIGURE 1. LSCM imaging of the association between fluorescentlylabeled proteins and LPR. Colocalization of bacteria (visualized by differ-ential interference contrast) with nonspecific proteins including SIgAC5,SIgASal4, SIgAHNK20, mSC, and IgGC20 (seen as red dots) is shown. Con-trol panels are obtained upon differential interference contrast (DIC)-me-diated visualization of LPR alone. One representative field obtained from10 different observations following analysis of 5 different slides isdepicted. Scale bars, 10 �m.

SIgA-Commensal Interactions Mediated by Glycan Residues

MAY 13, 2011 • VOLUME 286 • NUMBER 19 JOURNAL OF BIOLOGICAL CHEMISTRY 17241

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

bacterial strains including Gram-positive LPR, ST11 and BL, aswell as Gram-negative E. coli strains D2241 andNissle andBac-teroides Bt with fluorescently labeled reconstituted SIgA, pIgA,SC, and IgGC20. Panels of Fig. 1 show the data obtained withLPR; similar, if not identical results were produced with thestrains ST11 and BL. Complexes with bacteria were detectedby laser scanning confocal microscopy (LSCM) after 1 h ofincubation in the presence of either SIgAC5, SIgASal4, orSIgAHNK20 (Fig. 1). Bacteria were almost entirely coated bynonspecific SIgA molecules, indicating that the formation ofcomplexes relied on Fab-independent binding mechanisms.Interestingly, incubation with SC alone led to the same obser-vation, pointing out that Fab- but also Fc-independent associ-ations sustain the coating of bacteria by SIgA.Nobacterial coat-ing could be detected using pIgA alone (data not shown).Together, this demonstrates the primordial role of SC in theprocess and indirectly indicates that constituents involved inbinding are available in both bound and free SC. Moreover,weak colocalization with IgGC20 demonstrated discriminatingproperties of SIgA and SC in the capacity to coat bacteria andfurther suggests that the variety of glycosylation patternsresults in modulable Fab- and Fc-independent binding proper-ties. Contrary towhatwas observedwith commensal bacteria ofthe Lactobacillus and Bifidobacterium genera, we found thatthe commensal Gram-negative nonpathogenic strains E. coliD2241 and Nissle and of the Bacteroides phylum were recog-nized by IgGC20 to the same extend as with either SIgA ormSC

(data not shown), indicating that different types of interactionsare involved as a function of the bacterial surface.Glycan Residues Are Involved in the Interaction between Bac-

teria and SIgA—Glycosylation is known to be involved in theinnate defense against pathogens (25). Because SIgA and SC arehighly glycosylated proteins (19, 20), we performed deglycosy-lation cleaving all types ofN-branched glycan residues bound toasparagine. Contrary tomost of othermethods,N-glycosidase Fremoves carbohydrate residues by preserving the conforma-tional structure of the protein (23). Clipping of all sevenbranched glycans carried on pIgA-bound or free SC was con-firmed by analysis on polyacrylamide denaturing gels (Fig. 2A).The shift of the apparent molecular mass from 80 kDa down to62–65 kDa is indicative of fully deglycosylated SC (24). In con-trast, carbohydratemoieties present on the � chain were insen-sitive to the action of the enzyme, suggesting a nonaccessibleanchoring of the carbohydrates in pIgA (Fig. 2B). Representa-tive pictures of association between deglycosylated proteinsand LPR are depicted in Fig. 2C. Deglycosylation of either freeSC or bound to pIgA led to a drastic drop in colocalization withbacteria: only a few isolated red dots of weak fluorescenceintensity were observed by LSCM. Experiments performedwith ST11 or BL yielded to the same dramatic decline in SIgAand mSC reactivity toward bacteria (data not shown). Takentogether, these data underline the paramount role of glycanresidues and in particular those present on SC in ensuring bac-terial coating by SIgA and SC. In sharp contrast toLactobacillus

FIGURE 2. Impact of SIgA and SC deglycosylation on the association with LPR. A, Western blot analysis of mSC in reassociated SIgAC5, SIgASal4, orSIgAHNK20 and mSC was performed after 0, 2 h, or 4 h of incubation with 5 units of N-glycosidase F (see “Experimental Procedures”). Samples were separatedonto an 8% polyacrylamide gel under reducing conditions. The molecular mass (kDa) of detected protein is marked alongside the lanes. B, immunodetectionwas performed using anti-� chain Ab of the same samples as in A. Proteins detected at 62 kDa represent nondeglycosylated polypeptides. C, LSCM imaging ofthe association of LPR formed with native or deglycosylated Cy3-labeled proteins is carried out as in Fig. 1. One representative field obtained from 10 differentobservations after analysis of 5 different slides is depicted. Scale bars, 10 �m.

SIgA-Commensal Interactions Mediated by Glycan Residues

17242 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 19 • MAY 13, 2011

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

and Bifidobacterium, interaction of the three testedGram-neg-ative bacteria with deglycosylated or native proteins resulted insimilar high level binding, indicative of a selective role of car-bohydrates in the binding of Gram-positive bacteria only (datanot shown).Quantitative Analysis of Bacterial Coating by SIgA—LSCM

observations drove us to analyze at the quantitative level thebacterial surface covered by fluorescently labeled proteins.Data in Fig. 3 represent the quantification of the signal associ-ated with Ab- or SC-bacteria complexes using the same set ofbacteria incubated with either SIgA, SIgAdg, pIgA, mSC,mSCdg, or IgGC20. All Gram-positive bacteria mixed withSIgA or free SC resulted in almost full coating by fluorescentlylabeled proteins of LPR (95%) and high values of surface label-ing for ST11 (88.5%) and BL (73.8%) (Fig. 3,A–C). Interestingly,Lactobacillus strains LPR and ST11 displayed the most con-trasted interaction profiles, with deglycosylated moleculesleading to a drop in coating of at least 4-fold, confirming at the

quantitative level the results obtained on Figs. 1 and 2.Although resulting in a less marked 1.7-fold decrease, dataobtained for the Bifidobacterium BL (Fig. 3C) allowed us toreach the same conclusion: glycan residues carried by SC arethe keystone of the SIgA Fab- and Fc-independent coating ofbacteria. In addition, pIgA and SIgAdg exhibited a low level ofcoating in the case of LPR (3.7%), yet remaining close to 20% forST11 and varying from3% to up to 41% for BL. This later observa-tionmay reveal amodulable role for extremely heterogeneous gly-cans present on monoclonal pIgA as a function of the interactingbacterium (Fig. 3, A–C). In sharp contrast, E. coli strains D2241and Nissle, as well as Bt, were coated to at least 63% followingincubationswith all forms of proteins tested (Fig. 3,D–F), indicat-ing differentmolecular patterns involved in binding toGram-neg-ative bacteria. The very similar results obtained with deglycosy-lated and native proteins allowed us to conclude on the absence ofdirect involvement of carbohydrates present on SC in the coatingof Gram-negative bacteria by SIgA or SC.

FIGURE 3. Quantification of bacterial coating by fluorescently labeled proteins. Using ImageJ software, changes in the percentage of bacterial areacovered by nonspecific labeled proteins including SIgAC5, SIgAC5dg, pIgAC5, SIgASal4, SIgASal4dg, pIgASal4, SIgAHNK20, SIgAHNK20dg, pIgAHNK20,mSC, mSCdg, or IgGC20 were assessed in the case of LPR (A), ST11 (B), BL (C), D2241 (D), Nissle (E), and Bt (F) bacterial strains (see “ExperimentalProcedures”). Control quantifications were carried out on pictures obtained with bacteria alone. Bars represent the mean values � S.E. Statisticallysignificant differences are indicated above the brackets for intragroup tests: ***, p � 0.0001. Data were obtained from 5 different fields of oneexperiment repeated 5 times.

SIgA-Commensal Interactions Mediated by Glycan Residues

MAY 13, 2011 • VOLUME 286 • NUMBER 19 JOURNAL OF BIOLOGICAL CHEMISTRY 17243

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Purification of SIgA and SC fromHumanColostrum and Bio-chemical Characterization—Colostrum and human milk areprovidential sources of SIgA containing, respectively, up to 12 gand 1 g of the Ab/liter (4, 42). We thus purified SIgA and SCfrom human colostrum to examine the interaction betweennatural SIgA and commensal bacteria. The complete purifica-tion scheme is depicted in Fig. 4A. Silver staining andWesternblot analysis revealed the coelution of SIgAwith lactoferrin andIgG (data not shown), two glycoproteins that can potentiallyinterfere with nonspecific binding properties of SIgA (43). IgGwas removed by affinity chromatography on protein G-Sepha-rose resin, whereas elimination of lactoferrin was achieved byion-exchange chromatographies. Silver staining and Westernblot analysis confirmed the integrity and purity of isolatedSIgAcol and SCcol (Fig. 4, B and C).Bacterial Coating with Natural Purified Colostrum-derived

SIgA—To confirm the validity of results obtained with mousereassociated SIgA, we combined native and deglycosylatedSIgAcol and SCcol with the same set of commensal bacteria asillustrated in Figs. 1 and 2. Treatment of SIgAcol withN-glyco-sidase F resulted in the same deglycosylation pattern of SC asfor mouse hybridoma-derived SIgA, although with some delay

(Fig. 5A), whereas sugars associated with the � chain remainedundigested (Fig. 5B). The lag of sugar clipping observed is prob-ably due to slight structural differences between reassociatedSIgA and “naturally” associated SIgAcol in secretions. WhenSIgAcol was incubated with Gram-positive bacteria, we ob-served by LSCM bright dots colocalizing with LPR, whereasdeglycosylated molecules bound only marginally to bacteria(Fig. 5C and data not shown). Quantification of the bacterialpercentage covered by fluorescently labeled proteins confirmedthat the use of SIgAcol, SCcol, and control SIgAC5 resulted insimilar bacterial coating, whereas deglycosylation of colostrum-isolated molecules triggered the loss of efficient binding to Lacto-bacillus LPR (Fig. 5D). The demonstration that either humanSIgAcol or SCcol and reconstituted mouse SIgA molecules yieldsimilar binding data suggests that the heterogeneity of carbohy-drate side chains affords combinational possibilities that are pre-served between species. Selective deglycosylation of SC in bothmurine and human molecules further allows us to conclude as tothe critical roleof sugarmoieties on thispolypeptide in the contextof commensal binding by the whole SIgA protein.

DISCUSSION

The intestinal microbiota have been demonstrated to beinvolved in the proper development of the intestinal immunesystem and in particular in the production of a diverse reper-toire of SIgA (4) composed of at least 90% of natural SIgA withlargely unknown specificity (8, 15, 16). The biological functionof such SIgA remains speculative; however, various studies haveunderscored the natural coating of commensal bacteria bySIgA, a process that may be involved in the homeostatic gutsensing of the microbiota (12, 41, 44).Nevertheless, the precise biochemical mechanisms underly-

ing the relation between intestinal microbiota and the immunesystem, in particular, SIgA, remain enigmatic. To tackle thisissue, various strains isolated from the human intestinal tractwere incubated with a battery of nonspecific SIgA, and thenature of the interaction was investigated qualitatively andquantitatively. The use of purifiedmurine SIgA reconstituted inthe test tube or SIgA isolated from human colostrum identifieda Fab- and Fc-independent pattern of bacterial coating. Degly-cosylation of both SC and SIgA allowed us to conclude as to thepivotal role of carbohydrates, and in particularN-branched gly-cans, in the interaction with a selection of Gram-positive bac-teria. Remarkably, SC alone was able to coat such commensalsto the same extent as thewhole SIgA.Moreover, deglycosylatedSC in the context of SIgA prevented bacterial coating by theAb,featuring surface-exposed carbohydrate moieties as crucialpartners in the process of binding.It is conceivable that the association of Gram-positive bac-

teria and SIgA is due to recognition between Ab-boundN-branched glycans and abundant cell wall components suchas peptidoglycans, lipoteichoic acid, or teichoic acid available tothe environment.Lactobacillus lactisharbors a particular, thickouter layer composed of polysaccharides acting as a protectivecapsule (45). This polysaccharide envelope conferring to thebacterium a protective barrier against phagocytosis by murinemacrophages appears to be restricted toGram-positive bacteriaand might as well serve as a particular constituent involved in

FIGURE 4. Purification and characterization of colostrum-derived SIgAand SC. A, fractionation scheme leading to the purification of SIgAcol andSCcol (for details, see “Experimental Procedures”). B, characterization of puri-fied SIgAcol performed by silver staining (lane 1) and by Western blot analysisusing antisera specific for human � chain (lane 2), hSC (lane 3), � chain (lane 4),and J chain (lane 5). C, characterization of purified SCcol performed by silverstaining (lane 1) and by Western blot analysis using antisera specific for hSC(lane 2). Denaturing 8% polyacrylamide gels were run under nonreducingconditions.

SIgA-Commensal Interactions Mediated by Glycan Residues

17244 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 19 • MAY 13, 2011

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

recognition by SIgA or SC. In support of the importance ofbacterial surface in selective binding, the presence in theouter membrane of lipopolysaccharides in Gram-negativebacteria masking the underlying peptidoglycan layer couldexplain why both “native” and deglycosylated Ab and SCexhibit the same pattern of association. This indirectly sug-gests that other molecular patterns besides glycans on SCcontribute to genera-dependent association, opening a newavenue of research at the interface between immunology andmicrobiology.The different binding pattern between Gram-positive bacte-

ria LPR, ST11, and BL with respect to Gram-negative D2241,Nissle, and Bt revealed a fine tuning of the bacterial coating by

the Ab and SC. Perrier et al. (24) have demonstrated thatrecombinant SC prevents target HEp2 cells from being dam-aged by enteropathogenic E. coli, a property partly lost afterdeglycosylation, indicating an important contribution of carbo-hydrates in this process. These data are in sharp contrast withour results obtained with the nonpathogenic E. coli commensalstrains D2241 and Nissle for which maintenance of bindingcould be observed following SC deglycosylation. Because onlypathogenic E. coli strains including enteropathogenic E. coliexpress outer membrane intimin (46) involved in the interac-tion with glycans present on SC (24), this may explain differ-ences in SIgA binding. The sum of these data implies that theFab- and Fc-independent molecular features responsible for

FIGURE 5. Deglycosylation of colostrum-derived SIgA or SCcol and defective association with LPR observed by LSCM. A, immunodetection of the hSCwas performed after 0, 2 h, 4 h, or 6 h of incubation with 5 units of N-glycosidase F from 0 to 3 h and a further 5 units of enzyme for the last 3 h. Samples wereseparated onto a 10% polyacrylamide gel under reducing conditions. The molecular mass (kDa) of detected polypeptides is marked alongside the lanes.B, samples were separated as in A with immunodetection of the � chain. The unique signal detected at 62 kDa represents nondeglycosylated � chains. C, LSCMimaging of complexes formed between LPR and native or deglycosylated SIgAcol or SCcol was performed as in Fig. 1. SIgAC5 were used as a positive controlfor the interaction of bacteria with proteins. One representative field obtained from 10 different observations after analysis of 5 different slides is depicted. Scalebars, 10 �m. D, quantification of LPR coating by fluorescently labeled proteins was carried out on pictures of LPR alone (control) or pictures of complexes withSIgAC5, SIgAcol, SIgAcoldg, Sccol, and SCcoldg. Bars represent the mean values � S.E. Statistically significant differences are indicated above the brackets forintragroup tests: ***, p � 0.0001. Data were obtained from 5 to 10 different fields of one experiment repeated 5 times.

SIgA-Commensal Interactions Mediated by Glycan Residues

MAY 13, 2011 • VOLUME 286 • NUMBER 19 JOURNAL OF BIOLOGICAL CHEMISTRY 17245

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

bacterial coating are highly plastic and flexible and result indiverse biological consequences.As SIgA carbohydrates harbor highly heterogeneous struc-

tures, these latter can contribute to combinatory-like, multi-affinity interactions with gut microorganisms (25). This wasillustrated in neonatal mice in which antibody immunity fixedto the hapten nitrophenol protected against bacterial challenge(47). Recently, the assessment of the crystal structure of amucus-binding protein present at the surface of the commensalstrain Lactobacillus reuteri identified motifs with potentialimmunoglobulin binding activity (48). The triad mucus-bind-ing protein–bacteria–SIgA may serve as an organized latticeprecluding massive commensal penetration from the intestinallumen to the underlying tissue, thus ensuring proper host-mi-crobial mutualism. Noteworthy, drug-immunosuppressed andimmunodeficient individuals have increased susceptibility tosepsis or chronic intestinal inflammation mediated by themicrobiota (49, 50), and this holds particularly true for individ-uals lacking SIgA.We found that carbohydrates present on the � chain could

not be removed following exposure to N-glycosidase F. Thisobservation is consistent with themasking ofN-glycan residuesby SC in the context of the whole SIgA molecule (20, 51). It isplausible that these noncleavable carbohydrates would beresponsible for the various residual interaction observed uponincubation of Gram-positive bacteria with SIgAdg. However, asany removal of these glycan residues occurs unless harsh con-ditions (use of reducing agents, exposure to heat) are used (52),the function of residual sugarmoieties present on pIgA in bind-ing to bacteria cannot be addressed by this approach.Our study provides strong evidence thatN-branched glycans

present on SCare themain component responsible for Fab- andFc-independent specific binding to Gram-positive commensalbacteria. Due to the presence of LPS on Gram-negative bacte-ria, motifs other than carbohydrates on SIgA account for bind-ing. It is worth mentioning that introduction of commensals isassociated with the induction in the intestine of both strain-specific SIgA (9) and natural SIgAwith unknown specificity (8).Based on these observations, one can argue that SIgA present inhuman colostrum is a mixture of specific SIgA from maternalorigin and natural SIgA. Despite the potential presence of spe-cific SIgA for commensals, identical conclusions were reachedwith SIgA purified from human colostrum, compared withcoating data obtained with monoclonal Ab. Altogether, thiscontributes to explain at the biochemical level the role of colos-trum or milk SIgA with a broad maternal repertoire in shapingthe newborn microbiota before maturation of the mucosalimmune system (4). Commensal coating by colostrum-derivedSIgA would translate into controlled primary gut colonizationand the early education of the newborn immune system towardnovel antigens prone to become symbiotic partners. We pro-pose that coating of commensals by SIgA adds to the function ofmucosal SIgA inmaintaining gut homeostasis, and this processintegrates potentiation of binding to epithelial cells (28), pres-entation under noninflammatory conditions (26), and luminalsequestration or limited penetration of host tissues (53, 54).

REFERENCES1. Frank, D. N., St. Amand, A. L., Feldman, R. A., Boedeker, E. C., Harpaz, N.,

and Pace, N. R. (2007) Proc. Natl. Acad. Sci. U.S.A. 104, 13780–137852. Hooper, L. V., and Gordon, J. I. (2001) Science 292, 1115–11183. Mestecky, J., Russell, M. W., and Elson, C. O. (1999) Gut 44, 2–54. Newburg, D. S., and Walker, W. A. (2007) Pediatr. Res. 61, 2–85. Sjogren, Y. M., Tomicic, S., Lundberg, A., Bottcher, M. F., Bjorksten, B.,

Sverremark-Ekstrom, E., and Jenmalm,M. C. (2009)Clin. Exp. Allergy 39,1842–1851

6. Moreau, M. C., Ducluzeau, R., Guy-Grand, D., and Muller, M. C. (1978)Infect. Immun. 21, 532–539

7. Shroff, K. E., Meslin, K., and Cebra, J. J. (1995) Infect. Immun. 63,3904–3913

8. Talham,G. L., Jiang,H.Q., Bos, N. A., andCebra, J. J. (1999) Infect. Immun.67, 1992–2000

9. Hapfelmeier, S., Lawson, M. A., Slack, E., Kirundi, J. K., Stoel, M., Heiken-walder,M., Cahenzli, J., Velykoredko, Y., Balmer,M. L., Endt, K., Geuking,M. B., Curtiss, R., 3rd, McCoy, K. D., andMacpherson, A. J. (2010) Science328, 1705–1709

10. Bollinger, R. R., Everett, M. L., Palestrant, D., Love, S. D., Lin, S. S., andParker, W. (2003) Immunology 109, 580–587

11. Orndorff, P. E., Devapali, A., Palestrant, S., Wyse, A., Everett, M. L., Bol-linger, R. R., and Parker, W. (2004) Infect. Immun. 72, 1929–1938

12. van der Waaij, L. A., Limburg, P. C., Mesander, G., and van der Waaij, D.(1996) Gut 38, 348–354

13. van derWaaij, L. A., Kroese, F.G., Visser, A.,Nelis, G. F.,Westerveld, B.D.,Jansen, P. L., and Hunter, J. O. (2004) Eur. J. Gastroenterol. Hepatol. 16,669–674

14. Quan, C. P., Berneman, A., Pires, R., Avrameas, S., and Bouvet, J. P. (1997)Infect. Immun. 65, 3997–4004

15. Bos, N. A., Jiang, H. Q., and Cebra, J. J. (2001) Gut 48, 762–76416. Stoel, M., Jiang, H. Q., van Diemen, C. C., Bun, J. C., Dammers, P. M.,

Thurnheer, M. C., Kroese, F. G., Cebra, J. J., and Bos, N. A. (2005) J. Im-munol. 174, 1046–1054

17. Rojas, R., and Apodaca, G. (2002) Nat. Rev. Mol. Cell Biol. 3, 944–95518. Poger, M. E., and Lamm, M. E. (1974) J. Exp. Med. 139, 629–64219. Hughes, G. J., Reason, A. J., Savoy, L., Jaton, J., and Frutiger-Hughes, S.

(1999) Biochim. Biophys. Acta 1434, 86–9320. Royle, L., Roos, A., Harvey, D. J., Wormald, M. R., van Gijlswijk-Janssen,

D., Redwan el-R. M., Wilson, I. A., Daha, M. R., Dwek, R. A., and Rudd,P. M. (2003) J. Biol. Chem. 278, 20140–20153

21. Wold, A. E., Mestecky, J., Tomana, M., Kobata, A., Ohbayashi, H., Endo,T., and Eden, C. S. (1990) Infect. Immun. 58, 3073–3077

22. Schroten, H., Stapper, C., Plogmann, R., Kohler, H., Hacker, J., andHanisch, F. G. (1998) Infect. Immun. 66, 3971–3973

23. Phalipon, A., Cardona, A., Kraehenbuhl, J. P., Edelman, L., Sansonetti, P. J.,and Corthesy, B. (2002) Immunity 17, 107–115

24. Perrier, C., Sprenger, N., and Corthesy, B. (2006) J. Biol. Chem. 281,14280–14287

25. Mestecky, J., and Russell, M. W. (2009) Immunol. Lett. 124, 57–6226. Corthesy, B. (2010) Future Microbiol. 5, 817–82927. Thierry, A. C., Bernasconi, E., Mercenier, A., and Corthesy, B. (2009)Clin.

Exp. Allergy 39, 527–53628. Mathias, A., Duc, M., Favre, L., Benyacoub, J., Blum, S., and Corthesy, B.

(2010) J. Biol. Chem. 285, 33906–3391329. Riedel, C. U., Foata, F., Goldstein, D. R., Blum, S., and Eikmanns, B. J.

(2006) Int. J. Food Microbiol. 110, 62–6830. Le Blay, G., Lacroix, C., Zihler, A., and Fliss, I. (2007) Lett. Appl. Microbiol.

45, 252–25731. Phalipon, A., Kaufmann, M., Michetti, P., Cavaillon, J. M., Huerre, M.,

Sansonetti, P., and Kraehenbuhl, J. P. (1995) J. Exp. Med. 182, 769–77832. Phalipon, A., Costachel, C., Grandjean, C., Thuizat, A., Guerreiro, C., Tan-

guy, M., Nato, F., Vulliez-Le Normand, B., Belot, F., Wright, K., Marcel-Peyre, V., Sansonetti, P. J., and Mulard, L. A. (2006) J. Immunol. 176,1686–1694

33. Michetti, P.,Mahan,M. J., Slauch, J.M.,Mekalanos, J. J., andNeutra,M. R.(1992) Infect. Immun. 60, 1786–1792

SIgA-Commensal Interactions Mediated by Glycan Residues

17246 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 286 • NUMBER 19 • MAY 13, 2011

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

34. Weltzin, R., Hsu, S. A., Mittler, E. S., Georgakopoulos, K., and Monath,T. P. (1994) Antimicrob. Agents Chemother. 38, 2785–2791

35. Favre, L. I., Spertini, F., and Corthesy, B. (2003) J. Chromatogr. B Analyt.Technol. Biomed. Life Sci. 786, 143–151

36. Crottet, P., Peitsch,M. C., Servis, C., and Corthesy, B. (1999) J. Biol. Chem.274, 31445–31455

37. Crottet, P., and Corthesy, B. (1998) J. Immunol. 161, 5445–545338. Lullau, E., Heyse, S., Vogel, H., Marison, I., von Stockar, U., Kraehenbuhl,

J. P., and Corthesy, B. (1996) J. Biol. Chem. 271, 16300–1630939. Parr, E. L., Bozzola, J. J., and Parr, M. B. (1995) J. Immunol. Methods 180,

147–15740. Hendrickson, B. A., Rindisbacher, L., Corthesy, B., Kendall, D., Waltz,

D. A., Neutra, M. R., and Seidman, J. G. (1996) J. Immunol. 157,750–754

41. Tsuruta, T., Inoue, R., Nojima, I., Tsukahara, T., Hara, H., and Yajima, T.(2009) FEMS Immunol. Med. Microbiol. 56, 185–189

42. Carbonare, C. B., Carbonare, S. B., and Carneiro-Sampaio, M. M. (2005)Pediatr. Allergy Immunol. 16, 574–581

43. Gonzalez-Chavez, S. A., Arevalo-Gallegos, S., and Rascon-Cruz, Q. (2009)Int. J. Antimicrob. Agents 33, 301.e1–8

44. Fagarasan, S., Muramatsu, M., Suzuki, K., Nagaoka, H., Hiai, H., and

Honjo, T. (2002) Science 298, 1424–142745. Chapot-Chartier,M. P., Vinogradov, E., Sadovskaya, I., Andre, G.,Mistou,

M. Y., Trieu-Cuot, P., Furlan, S., Bidnenko, E., Courtin, P., Pechoux, C.,Hols, P., Dufrene, Y. F., and Kulakauskas, S. (2010) J. Biol. Chem. 285,10464–10471

46. Frankel, G., Phillips, A. D., Trabulsi, L. R., Knutton, S., Dougan, G., andMatthews, S. (2001) Trends Microbiol. 9, 214–218

47. Harris, N. L., Spoerri, I., Schopfer, J. F., Nembrini, C., Merky, P., Massa-cand, J., Urban, J. F., Jr., Lamarre, A., Burki, K., Odermatt, B., Zinkernagel,R. M., and Macpherson, A. J. (2006) J. Immunol. 177, 6256–6262

48. MacKenzie, D. A., Tailford, L. E., Hemmings, A. M., and Juge, N. (2009)J. Biol. Chem. 284, 32444–32453

49. van der Meer, J. W., Guiot, H. F., van den Broek, P. J., and van Furth, R.(1984) Semin. Hematol. 21, 123–140

50. Schenk, M., and Mueller, C. (2007) Semin. Immunol. 19, 84–9351. Bonner, A., Almogren, A., Furtado, P. B., Kerr, M. A., and Perkins, S. J.

(2009)Mucosal Immunol. 2, 74–8452. Busch, N., Lammert, F., and Matern, S. (1998) Gastroenterology 115,

129–13853. Macpherson, A. J., and Uhr, T. (2004) Science 303, 1662–166554. Kadaoui, K. A., and Corthesy, B. (2007) J. Immunol. 179, 7751–7757

SIgA-Commensal Interactions Mediated by Glycan Residues

MAY 13, 2011 • VOLUME 286 • NUMBER 19 JOURNAL OF BIOLOGICAL CHEMISTRY 17247

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Amandine Mathias and Blaise CorthésyColostrum-derived Secretory Immunoglobulin A Is Mediated by Carbohydrates

Recognition of Gram-positive Intestinal Bacteria by Hybridoma- and

doi: 10.1074/jbc.M110.209015 originally published online March 21, 20112011, 286:17239-17247.J. Biol. Chem. 

  10.1074/jbc.M110.209015Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/286/19/17239.full.html#ref-list-1

This article cites 54 references, 32 of which can be accessed free at

by guest on May 23, 2020

http://ww

w.jbc.org/

Dow

nloaded from