{beta}-1,2- and {alpha}-1,2-Linked Oligomannosides Mediate Adherence of Candida albicans Blastospores to Human Enterocytes In Vitro

Candida albicans is a commensal dimorphic yeast of the digestivetract that causes hematogenously disseminated infections inimmunocompromised individuals. Endogenous invasive candidiasisdevelops from C. albicans adhering to the intestinal epithelium.Adherence is mediated by the cell wall surface, a domain composedessentially of mannopyranosyl residues bound to proteins, theN-linked moiety of which comprises sequences of ${alpha}$ -1,2- and ß-1,2-linkedmannose residues. ß-1,2-linked mannosides are alsoassociated with a glycolipid, phospholipomannan, at the C. albicanssurface. In order to determine the roles of ß-1,2and ${alpha}$ -1,2 oligomannosides in the C. albicans-enterocyte interaction,we developed a model of adhesion of C. albicans VW32 blastospores tothe apical regions of differentiated Caco-2 cells. Preincubationof yeasts with monoclonal antibodies (MAbs) specific for ${alpha}$ -1,2and ß-1,2 mannan epitopes resulted in a dose-dependentdecrease in adhesion (50% of the control with a 60-µg/mlMAb concentration). In competitive assays ß-1,2 and ${alpha}$ -1,2 tetramannosides were the most potent carbohydrate inhibitors,with 50% inhibitory concentrations of 2.58 and 6.99 mM, respectively. Immunolocalizationon infected monolayers with MAbs specific for ${alpha}$ -1,2 and ß-1,2oligomannosides showed that these epitopes were shed from theyeast to the enterocyte surface. Taken together, our data indicatethat ${alpha}$ -1,2 and ß-1,2 oligomannosides are involved inthe C. albicans-enterocyte interaction and participate in theadhesion of the yeasts to the mucosal surface.

INTRODUCTION

Top

Introduction

Candida species are part of the commensal flora of the mucosa andskin in humans and other vertebrates. In immunocompromised or intensive-carepatients, increased mucosal proliferation secondary to use ofbroad-spectrum antibiotics, together with reduced host defenses andphysical alteration of the mucosal barriers, may result in bloodstreaminvasion. Altogether, candidemia accounts for $~$ 10% of nosocomialbloodstream infections, and C. albicans is the causative agentin 50 to 70% of disseminated candidiasis (13, 18, 20, 36, 48).

Molecular typing methods have shown an overall genetic similaritybetween C. albicans strains obtained from blood cultures andcolonizing strains obtained from the gastrointestinal tractsof the same patients, confirming endogenous acquisition as themain source of invasive candidiasis (40, 47). On the basis ofthis model, adhesion of the yeasts to the epithelium of thedigestive tract is a prerequisite for colonization and a criticalstep in the pathogenesis of invasive candidiasis. Characterizationof the adhesins and ligands involved in the C. albicans-enterocyteinteraction thus appears to be a necessary approach to developingstrategies aimed at reducing mucosal colonization and preventingbloodstream invasion.

Interaction of C. albicans with host cell surfaces is mediatedby the yeast cell wall, a complex and dynamic structure containingglucan, chitin and mannoproteins (reviewed in reference 6).The outermost layers of the C. albicans cell wall are made of phosphopeptidomannan(PPM), a polymer of mannose residues and proteins commonly referredto as mannan (3, 6). Mannan has been shown to play a role inadherence (27), immunomodulation (11), and antigenic variability(43). The PPM glycan moiety is composed of O-linked and N-linkedoligomannosides. The N-linked part consists of a backbone of ${alpha}$ -1,6-linked mannopyranose residues with branches composed of ${alpha}$ -1,2- and ${alpha}$ -1,3-linked mannopyranose units and terminal ß-1,2 linkagesin C. albicans serotype A (7). Short branches composed of ß-1,2-linkedmannopyranose residues are linked to PPM through phosphodiesterbridges in C. albicans serotypes A and B. These side chainsare referred to as the acid-labile fraction of PPM, since theyare cleaved by mild acid treatment (45). ß-1,2 oligomannosidicchains have also been identified on a 14- to 18-kDa glycolipid,referred to as phospholipomannan (PLM) (46), that is expressed atand shed from the C. albicans cell wall (25, 39).

ß-1,2 mannosidic linkages are uncommon structureswhose presence has been reported in only few bacterial and yeastspecies (30, 35). In C. albicans, their presence was first identifiedon PPM by Shibata et al. (41). These oligomannoside sequencesare involved in the adhesion of C. albicans to the macrophagemembrane, at least in part through binding to galectin 3, amember of a family of carbohydrate binding proteins implicatedin a variety of biological functions (17, 25). ß-1,2 oligomannosidesalso generate protective antibodies (22) and induce cytokine production(26). These unique carbohydrate sequences thus appear to playa key role in the C. albicans-host balance.

Despite the importance of the digestive tract as the main sourceof invasive candidiasis, few groups have reported analyses ofthe C. albicans-enterocyte interaction at the cellular and molecularlevels (44, 49, 50). In the present paper, we describe a modelof adhesion of C. albicans blastospores to the human enterocytecell line Caco-2 and its use to analyze the role of oligomannosideswith distinct anomere-type linkages, either ${alpha}$ or ß,in the attachment of C. albicans blastospores to Caco-2 cells.Indeed, galectin-3, which binds ß-1,2 oligomannosideson the macrophage (17), is also expressed in intestinal epithelialcells (1, 8). Moreover, recent studies with a mouse model ofcandidiasis showed that oral administration of synthetic ß-1,2oligomannosides could reduce colonization of the gut, presumablyby competing with the natural flora for binding to enterocytes (12).We were thus interested in understanding the basis of this phenomenonat the cellular and molecular levels.

MATERIALS AND METHODS

Top

Materials AND METHODS

Growth and preparation of C. albicans strain for adherence assay. C. albicans strain VW32 (serotype A) (5) was used throughoutthis study. This strain, which was originally isolated from apatient with human renal candidiasis, has been employed in prior studiesof C. albicans-macrophage interactions (16, 17, 25) and in thechemical characterization of C. albicans ß-mannosidicepitopes (15, 46). Stock cultures were maintained at -20°Con Sabouraud dextrose agar. For adhesion experiments, stockcultures were plated on Sabouraud dextrose agar and incubatedat 37°C. After 18 to 20 h of culture, the yeasts were recovered,washed with phosphate-buffered saline (PBS), and resuspendedin PBS. The yeast concentration was determined by hemocytometerand growth on Sabouraud dextrose agar. For experiments usingheat-killed yeasts, blastospores were treated at 100°C for15 min, washed in PBS by centrifugation, and resuspended inPBS.

Growth and differentiation of Caco-2 cells. Caco-2 cells were obtained from the American Type Culture Collection(HTB 27) and were cultivated in the absence of antibiotics andantifungal agents. Cells (passages 4 to 15) were grown to confluencein 25-cm² flasks at 37°C and 5% CO₂ in Dulbecco's modifiedEagle's medium (3 volumes) with Ham F12 (1 volume) containing10% fetal calf serum, glutamine, and 1 g of glucose/liter. Foradhesion experiments, $~$ 4 x 10⁵ cells were seeded on each 11-mm-diameterglass coverslip in 24-well plastic dishes. The cultures weremaintained at 37°C and 5% CO₂, and the medium of each culturewell was replaced every other day until it was used.

Fluorescent probes and antibodies. Monoclonal antibody (MAb) EBCA1 is a rat immunoglobulin (IgM)that reacts with ${alpha}$ -linked mannose (24). MAb 5B2 is a mouse-ratchimeric IgM that reacts with ß-linked mannose (23).The control MAb A255 is a mouse IgM that recognizes a proteinof the respiratory syncytial virus. It was provided by P. Pothier(Dijon, France). Uvitex 2B was purchased from LD Bio Diagnostics(Lyon, France). For immunofluorescence experiments, the secondaryantibodies Alexa Fluor 488 goat anti-mouse IgM (µ chain)and Alexa Fluor 488 goat anti-rat IgM (µ chain) (1/200dilutions in PBS; Molecular Probes, Leiden, The Netherlands)were employed for MAbs 5B2 and EBCA1, respectively. Detectionof actin in confocal-microscopy experiments was performed withAlexa Fluor 568-phalloidin (1 U of phalloidin per well; MolecularProbes). For double-labeling experiments, the secondary antibodyAlexa Fluor 594 goat anti-rat IgM (1/200 dilution in PBS; MolecularProbes) was employed for MAb EBCA1.

Carbohydrates. D-Glucose, D-galactose, galactosamine, N-acetylglucosamine, D-mannose,D-fucose, L-fucose, N-acetylgalactosamine, N-acetylneuraminicacid, D-xylose, D-lactose, N-acetyllactosamine, D-mannosamine,mannan from Saccharomyces cerevisiae, hyaluronic acid, fetuin, asialofetuin,laminarin, heparin, fucoidan, and chondroitin sulfate A andC were purchased from Sigma-Aldrich Chimie (Saint Quentin Fallavier,France) and stored according to the recommendations of the manufacturer.D-Rhamnose and ${alpha}$ -methylmannoside were purchased from Acros Organics(Noisy le Grand, France) and stored at 4°C. ${alpha}$ -1,2 tetramannosidesand ß-1,2 tetramannosides were synthesized accordingto a previously described protocol (12) and stored at 4°Cuntil they were used. All carbohydrates were diluted extemporaneouslyin PBS for attachment inhibition assays.

Adherence of C. albicans blastospores to Caco-2 cells. Monolayers for adhesion experiments were used 15 to 21 daysafter being seeded. The medium in each well was aspirated andreplaced by 300 µl of PBS preincubated at 37°C andcontaining 10³ blastospores. After 30 min at 37°C, the monolayerswere washed three times to remove nonadhering blastospores,fixed in 2% glutaraldehyde for 10 min, and washed twice. Fordetection of adherent blastospores, Uvitex 2B (1/100 dilutionin PBS) was added to each well for 5 min in the dark at roomtemperature. The coverslips were rinsed, counterstained with 0.2%Evans Blue in PBS for 30 s, rinsed three times, and mountedinverted on microscope slides. In all the experiments described,the Caco-2 monolayers were washed with PBS at 37°C. Attachmentwas quantified by immunofluorescence microscopy using a ZeissAxioscop 2 microscope with an excitation filter at 360 nm anda suppression filter at 460 nm. The percentage of adhesion ineach culture was determined as the ratio of the number of adherentyeasts on the entire surface of the coverslip to the inoculumevaluated by quantitative culture. Each condition was testedin triplicate, and three separate experiments were performed.

Attachment inhibition assays. The effects of MAbs 5B2 and EBCA1 on the attachment of C. albicansblastospores to Caco2 cells were studied by using the adherenceassay described above. Yeasts were suspended in PBS and preincubatedwith purified MAbs at 60, 6, or 3 µg/ml for 1 h at 37°Cunder mild agitation. Agglutinated blastospores were pelletedby a 3-min centrifugation at 200 x g. Supernatants containing nonagglutinatedblastospores were diluted with PBS, and 10³ yeast cells weretransferred to Caco-2 monolayers. The adhesion experiment wasthen performed as described above. In parallel, 20-µlfractions of each supernatant were recovered, placed on immunofluorescenceslides, air dried, reacted with secondary antibodies, and examinedwith a Zeiss axioscope 2 microscope to assess the labeling ofthe blastospores with primary MAbs. All experiments includedtwo positive control wells (cocultures in PBS alone without inhibitingantibody) and duplicate testing of attachment with each antibodyconcentration. Three separate experiments were performed.

To determine the effect of synthetic ß-1,2 and ${alpha}$ -1,2tetramannosides and control carbohydrates on adhesion, Caco-2monolayers were incubated at 37°C in PBS containing the followingconcentrations of the carbohydrates tested. D-Glucose, D-galactose,galactosamine, N-acetylglucosamine, D-mannose, D-fucose, L-fucose, N-acetylgalactosamine,N-acetylneuraminic acid, D-xylose, D-lactose, N-acetyllactosamine,D-mannosamine, D-rhamnose, and ${alpha}$ -methylmannoside were testedat 10, 50, and 250 mM. Mannan from S. cerevisiae, hyaluronicacid, fetuin, asialofetuin, laminarin, heparin, fucoidan, andchondroitin sulfate A and C were tested at 0.2, 1, and 5 mg/ml.For ß-1,2 mannotetraoses and ${alpha}$ -1,2 mannotetrasoses,the concentrations tested were 0.3, 1.5, and 7 mM. After 30min, 10³ yeast cells were added to each well and coincubatedwith Caco-2 cells for 30 min at 37°C. The adhesion experimentwas then performed as described above. All experiments includeda series of positive control wells (cocultures in PBS alonewithout inhibiting carbohydrates) and triplicate testing ofattachment with each carbohydrate concentration. Three separateexperiments were performed for each carbohydrate tested.

In both series of inhibition experiments, the results of adhesionwithout inhibiting antibodies or competing carbohydrates were setto 100%, and adhesion in the presence of antibodies or carbohydrateswas expressed as the ratio of the percentage of adhesive yeastsin the presence of a given concentration of antibody or carbohydrateto the percentage of adhesive yeasts in positive control wells.

Immunofluorescence confocal microscopy. Cultures for immunofluorescence confocal microscopy were performedin 24-well culture dishes containing removable porous inserts(0.1-µm pore diameter; BD Falcon). Approximately 0.5 x10⁵ cells were seeded in the upper compartment. The cultureswere maintained at 37°C and 5% CO₂, and the medium in eachwell was replaced every other day. The monolayers were used15 to 21 days postseeding. The adhesion assays were performedas described above, with an inoculum of 10⁵ yeast cells perwell for cocultures of 30 to 60 min. When longer interactionswere analyzed, an inoculum of 10³ yeast cells per well was used.Following incubation, the monolayers were washed three timesto remove nonadhering blastospores. They were then fixed for10 min at room temperature in a 3.7% formaldehyde solution inPBS. After two more washes, the monolayers were extracted with0.1% Triton X-100 in PBS for 3 to 5 min. Each well was thenwashed twice, and the monolayers were reacted with Alexa Fluor568-phalloidin for 20 min at room temperature for F-actin labelingin Caco-2 cells. After two washes, the monolayers were incubatedwith MAb 5B2 or EBCA1 (30 µg/ml; 1 h at 37°C). Themonolayers were washed twice and reacted with Alexa Fluor-labeledsecondary antibodies. The preparations were examined by confocallaser scanning microscopy on a Leica TCS 4000 microscope. The cellmonolayers were optically sectioned in horizontal (x-y) or vertical(x-z) planes every 0.8 µm. Five separate experiments wereperformed to verify the absence of internalization of reactiveyeasts at 30 or 60 min postinfection, and 20 microscopic fieldswere examined for each culture.

Double-labeling experiments on infected monolayers. Caco-2 monolayers grown on glass coverslips were infected with10³ blastospores per well 15 to 21 days postseeding. At 12 h postinfection,the monolayers were washed to remove nonadherent blastospores.They were then fixed for 10 min at room temperature in a 3.7%formaldehyde solution in PBS. After two washes, the monolayerswere reacted sequentially with MAb EBCA1, Alexa Fluor 594 goatanti-rat IgM, MAb 5B2, and Alexa Fluor 488 goat anti-mouse IgMat the concentrations given above. All incubations were performedfor 1 h at 37°C and were followed by two washes. The monolayerswere examined on a Nikon E600 Eclipse microscope.

Statistical analysis. The within-day and between-day repeatabilities were studiedby one-way analysis of variance. In attachment inhibition assays,the differences among 5B2, A255, and EBCA1 antibodies, as wellas those between ${alpha}$ -1,2 and ß-1,2 tetramannosides, werestudied, along with the concentrations, by two-way analysisof variance. The concentration associated with a 50% inhibition(IC₅₀) was estimated with its standard error by modeling theinhibition curve according to the following equation: y = a/(a +x), where y is the percentage of inhibition, x is the concentration,and a is the value of x associated with a y of 50%; the minimizationalgorithm used for this nonlinear regression was the Gauss-Newton type.

RESULTS

Top

Results

Characterization of a model of adhesion of C. albicans blastospores to Caco-2 cells. In order to identify the molecular partners involved at theinitial stage of the C. albicans-enterocyte interaction, wedeveloped a model of adhesion of C. albicans blastospores toCaco-2 cells. The binding of blastospores to monolayers increasedfrom 10 to 60 min of incubation. After 60 min, clusters of blastosporesindicating multiplication of the inoculum were observed, whichprecluded accurate evaluation of the initial attachment process.Moreover, germ tubes, hyphae, and pseudohyphae were detected,and these fungal elements express surface molecules, i.e., putativeadhesins, distinct from blastospores (31). Our model was thus optimizedto analyze C. albicans attachment after a 30-min interaction.Since the degree of maturation of Caco-2 cells increases untilthe cells are differentiated 14 to 21 days after being seeded (37),we evaluated the attachment of C. albicans to Caco-2 monolayersfrom 5 to 40 days postseeding. The binding of blastospores increasedfrom 5 to 15 days and reached a plateau after 15 days. Whenthe monolayers were infected with 10³ blastospores 15 to 21days postseeding, an average of 120 to 250 adherent yeast cellswere detected after a 30-min interaction. Altogether, the meanwithin-day coefficient of variation of the model was 9%, whereasthe between-day variance gave a coefficient of variation of29%.

A key problem in setting up the model was to verify that yeastscounted after washings were adherent yeasts, i.e., yeasts thatwere attached to the surfaces of the cells as opposed to blastosporesthat could have been internalized by enterocytes. The Uvitex2B dye met this requirement. This marker is excluded from livephagocytes (29) and has been employed to identify nonphagocytizedyeasts in a model of Candida-endothelial-cell interaction (19).Moreover, fluorescence confocal microscopy analysis of permeabilizedmonolayers at 30 and 60 min of coculture confirmed that C. albicans blastosporeswere present at the apical surfaces of the cells only (Fig. 1).No internalized yeasts were detected in these experiments.

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. Confocal laser scanning microscopy of Caco-2 monolayers at 30 min postinfection with C. albicans blastospores. Monolayers sectioned in vertical planes (x-z) were processed for double labeling using Alexa fluor 568-phalloidin (red) and MAb 5B2 (green), and a merged image is presented. The apical region of the cell is identified by strong reactivity of phalloidin with the actin network of the brush border. Yeast cells are present at the surfaces of the cells. Bar, 5 µm.

Inhibition of C. albicans adhesion to Caco-2 cells by MAbs specific for ${alpha}$ -linked or ß-1,2-linked oligomannosides. To establish the role of ${alpha}$ - and ß-mannosidic sequencesin the C. albicans-Caco-2 interactions, we first performed neutralizingexperiments with 5B2 and EBCA1, two MAbs that bind ß-1,2and ${alpha}$ -1,2 mannoside epitopes, respectively (23, 24). An isotype-matched irrelevantMAb was used as a control. Since 5B2 and EBCA1 agglutinated C.albicans blastospores, yeasts treated with the MAbs were centrifugedand adhesion experiments were performed with the nonagglutinatedblastospores present in the supernatants. Fractions of thesenonagglutinated yeasts were assayed in parallel by immunofluorescenceto verify that they bound the MAbs (Table 1). In experimentsperformed with 60 µg of MAbs/ml, the percentages of adhesionof C. albicans to Caco-2 cells were 91.4 ± 7.17, 53 ±6.21, and 49 ± 5.56% for A255 (control), 5B2 (anti-ß-1,2mannoside), and EBCA1 (anti- ${alpha}$ -1,2 mannoside) antibodies, respectively(Fig. 2), and statistical analysis showed that both anti-mannosideantibodies differed significantly from the A255 control antibody(P < 0.05). The increased adherence of blastospores treatedwith MAb 5B2 at 6 µg/ml (114 ± 18.2%) was significant (P< 0.05) compared to that of A255 (95.0 ± 6.08%) (Fig.2). These results reflect a significant interaction betweenantibody and concentration, as the difference between antibodiesis much higher at 60 than at 6 µg/ml. Other attachmentpercentages did not differ significantly.

View this table:
[in this window]
[in a new window]
TABLE 1. Immunofluorescence reactivities of fractions of yeast suspensions exposed to MAbs 5B2, EBCA1, and A255 in attachment inhibition assays^a

View larger version (21K):
[in this window]
[in a new window]
FIG. 2. Effects of MAbs 5B2 (anti-ß-1,2 mannose), EBCA1 (anti- ${alpha}$ -1,2 mannose), and A255 (control antibody) on attachment of C. albicans blastospores to Caco-2 monolayers. Yeast cells were incubated with 3, 6, and 60 µg of antibodies/ml or with PBS alone prior to and during adhesion. Attachment is expressed as a percentage of that of the controls to which no antibody was added. The values are the means plus standard deviations of three independent experiments, including testing in duplicate wells.

Competitive inhibition of C. albicans adhesion to Caco-2 cells by glycans. We next performed competitive assays using synthetic ß-1,2and ${alpha}$ -1,2 tetramannosides with antigenic profiles that mimicthe antigenicities of their natural counterparts in the C. albicanscell wall (12). A panel of mono- and disaccharides and complexglycans was also evaluated for the ability to inhibit C. albicansbinding to Caco-2 cells. In these experiments, the concentrationassociated with a 50% inhibition was correctly estimated, sinceits coefficient of variation (the standard error divided bythe estimate) ranged from 9 to 19% between products. The concentrationsat which synthetic ß-1,2 and ${alpha}$ -1,2 tetramannosidesinhibited 50% of C. albicans attachment to Caco-2 monolayers(IC₅₀s) were 2.58 (standard error, 0.481) and 6.99 (standarderror, 1.32) mM, respectively (Fig. 3), and these two glycanswere the most potent inhibitors of C. albicans adhesion to Caco-2cells (Fig. 4). Moreover, when the analysis took into accountthe IC₅₀ range (IC₅₀ ± 2 standard errors), the resultsfor ß-1,2 oligomannosides ranged from 1.62 to 3.54mM and differed significantly from those for all other glycanstested, including ${alpha}$ -1,2 oligomannosides (IC₅₀ range, 4.36 to9.64 mM). None of the complex glycans reduced adherence at 1mg/ml or below.

View larger version (23K):
[in this window]
[in a new window]
FIG. 3. Effects of synthetic mannotetraoses (TetraMan) on attachment of C. albicans blastospores to Caco-2 monolayers. The monolayers were incubated with ß-1,2 mannotetraoses and ${alpha}$ -1,2 mannotetraoses at 0.3, 1.5, and 7 mM prior to and during adhesion. Attachment is expressed as a percentage of that of the control cultures to which no competing carbohydrate was added. The values are the means ± standard deviations of three independent experiments, including testing in triplicate wells.

View larger version (27K):
[in this window]
[in a new window]
FIG. 4. Inhibition of adherence of C. albicans blastospores to Caco-2 monolayers by carbohydrates. GlcNac, N-acetylglucosamine; NaNa, N-acetylneuraminic acid; GalNac, N-acetylgalactosamine; Alpha-Me, ${alpha}$ -methylmannoside; LacNac, N-acetyllactosamine; AlphaMan, ${alpha}$ -1,2 tetramannoside; BetaMan, ß-1,2 tetramannoside. The concentration of carbohydrate at which adhesion is 50% of adhesion in the absence of carbohydrate (IC₅₀) (*) was obtained by modeling the inhibition curves (as shown in Fig. 3) following the method described in Materials and Methods. **, standard error of the IC₅₀.

Shedding of antigens containing ${alpha}$ -1,2- and ß-1,2-linked mannose residues during coculture of C. albicans with Caco-2 cells. As antigens containing ß-1,2 mannose residues areshed during the interaction of C. albicans with macrophages (25),we were next interested in determining whether shedding of ${alpha}$ -1,2or ß-1,2 mannosides occurred during the C. albicans-Caco-2interaction. To this end, cocultures of C. albicans with Caco-2cells were subjected to indirect immunofluorescence detectionof ß-1,2 and ${alpha}$ -1,2 mannosides with MAbs 5B2 and EBCA1,respectively. At 30 min postinfection, examination of the monolayersshowed that MAb 5B2 detected islands of reactive material closeto the yeasts (Fig. 5A). A more diffuse localization of thelabeling was observed after 12 h of coculture (data not shown).Upon examination in the x-z orientation, the 5B2-reactive materialwas localized at the surfaces of the monolayers (Fig. 5B). Whenheat-killed blastospores were used, the percentage of adhesionwas low ( $~$ 2.5% of the inoculum), and no reactive material was detectedat the surfaces of the monolayers with MAb 5B2 (data not shown).Similar experiments were performed with MAb EBCA1, and clusters ofreactive material were detected at the surfaces of the monolayersin the vicinity of the yeasts (data not shown). Similarly, noreactivity of MAb EBCA1 was detected upon incubation of themonolayers with heat-killed blastospores (data not shown). Doublelabeling of infected monolayers with 5B2 and EBCA1 MAbs (Fig. 6)showed that both antibodies essentially bound to the same domainsof the monolayers (Fig. 6D). However, within a reactive domain,not all spots of 5B2-reactive material colocalized with spotsof EBCA1-reactive material(Fig. 6D).

View larger version (23K):
[in this window]
[in a new window]
FIG. 5. Immunofluorescence confocal microscopy analysis of ß-1,2 oligomannosidic epitopes after a 30-min (A) or 12-h (B) culture of Caco-2 cells with C. albicans. The monolayers were processed for double labeling using Alexa Fluor 568-phalloidin (red) and MAb 5B2 (green), and merged images are presented. (A) Islands of material reactive with MAb 5B2 (arrowheads) in the vicinity of C. albicans (arrows). (B) Analysis in the vertical plane shows that ß-1,2 oligomannosidic epitopes shed by C. albicans (long arrow) are present at the apical surfaces of the cells (short arrows). Bars, 5 µm.

View larger version (32K):
[in this window]
[in a new window]
FIG. 6. Double-labeling experiment with MAbs 5B2 and EBCA1 on Caco-2 monolayers incubated for 12 h with C. albicans. (A to C) The same microscopic field was examined under phase-contrast microscopy (A), immunofluorescence microscopy for reactivity of MAb 5B2 with ß-1,2 oligomannoside (green) (B), and immunofluorescence microscopy for reactivity of MAb EBCA1 with ${alpha}$ -1,2 oligomannoside (red) (C). Both MAbs react with C. albicans blastospore and germ tube and with clusters of antigenic material at the surface of the cells. (D) Merged image of panels B and C. The secondary antibodies were Alexa Fluor 488 goat anti-mouse IgM (B) and Alexa Fluor 594 goat anti-rat IgM (C). Bars, 5 µm.

DISCUSSION

Top

Discussion

Most studies of the adhesion of C. albicans to human cells were performedwith buccal epithelial cells (4, 10, 33), esophageal epithelialcells (14), endothelial cells (19), fibroblasts (32), keratinocytes(34), HeLa cells (21), or macrophages (16), and although intestinalcolonization is accepted as the main source of bloodstream dissemination (38, 40, 47),little is known about the molecular basis of attachment of C.albicans to enterocytes. Three recent studies have addressedthe C. albicans-enterocyte interaction. Weide and Ernst showedthat transcellular migration of C. albicans occurred duringa 12- to 24-h interaction (49). Wiesner et al. demonstratedthat the absence of a functional INT1 gene, a gene involvedin filamentous growth, attachment to HeLa cells, and virulence,was associated with decreased adherence of C. albicans blastosporesto enterocytes (50). In a model of undifferentiated Caco-2 cells,Timpel et al. showed that a C. albicans strain defective inPMT1, a gene encoding a mannosyltransferase involved in theO-glycosylation pathway, showed decreased adherence (44). Wechose to study the attachment of C. albicans to Caco-2 cells.This cell line derives from a human colorectal adenocarcinoma, expressescharacteristics of enterocytic differentiation upon reaching confluence(37), and has been used to study interactions of human intestinalcells with bacteria (51), virus (2), protozoa (28), and C. albicans(44, 49, 50). The limitation of this model is that Caco-2 cellsdo not produce mucins. Therefore, our system is not suitableto investigate to what extent mucins may modulate the C. albicans-Caco-2interaction (10).

C. albicans blastospores that bound anti- ${alpha}$ -1,2 or anti ß-1,2mannoside MAbs showed a 50% decrease in adhesion to Caco-2 monolayerswhen exposed to higher antibody concentrations (Fig. 2). Despitethe moderate (but statistically significant) increase in adhesionobserved with MAb 5B2 at 6 µg/ml (Fig. 2), for which wehave no explanation, the major attachment inhibition occurringupon exposure of blastospores to both anti-mannoside antibodiesat 60 µg/ml suggests that ${alpha}$ -1,2- and ß-1,2-linkedoligomannoside sequences present at the yeast cell surface areinvolved in attachment to Caco-2 cells. One limitation of theseexperiments is the use of nonagglutinated blastospores presentin supernatant fractions after mild centrifugation of the yeastpreparations. Indeed, despite a dose-dependent reactivity withanti-mannoside antibodies by immunofluorescence assay (Table 1),we cannot rule out the possibility that this blastospore populationhas adhesive properties distinct from their agglutinated counterparts.Moreover, immunoglobulins bound to the yeast surface could maskadhesive epitopes other than ${alpha}$ -1,2 and ß-1,2 oligomannosides.We therefore performed competitive experiments with synthetic ${alpha}$ -1,2and ß-1,2 tetramannosides. Previous studies of macrophagesshowed that similar inhibition was achieved by incubating cellswith native or synthetic ß-1,2 mannotetraoses (16),demonstrating that synthetic molecules with a degree of polymerizationof four mannose residues showed normal biological activity.Furthermore, ${alpha}$ -1,2 and ß-1,2 tetramannosides reactednormally with antibodies specific for ${alpha}$ - and ß-mannosidicepitopes in an enzyme immunoassay, indicating that the syntheticmolecules retained the antigenicities of their native counterparts (12).The competing effect obtained with ${alpha}$ -1,2 and ß-1,2tetramannosides is a striking result of the present investigation.The inhibitions obtained with both molecular species are dosedependent (Fig. 3). The specificity of this effect was establishedby comparing a panel of carbohydrates. None of the complex glycanstested reduced attachment at 1 mg/ml, a finding similar to thatreported in a model of the adhesion of Candida glabrata to Hep2cells (9). Other mono- and disaccharides inhibited C. albicansattachment to Caco-2 cells, with IC₅₀s ranging from 7.68 mMfor N-acetyllactosamine to 114.57 mM for glucose (Fig. 4). Thus,among all carbohydrates tested, ß-1,2 tetramannosidesexhibited the strongest inhibitory effects on the adhesion ofC. albicans to Caco-2 cells. This finding is consistent withthe dramatic reduction in gastrointestinal colonization reportedin the infant mouse model after oral administration of syntheticß-1,2 tetramannosides (12). Conversely, the inhibitoryeffect observed in our study with synthetic ${alpha}$ -1,2 tetramannosidescontrasts with their lack of activity on gastrointestinal colonizationin the mouse model (12). The reason why in vivo and in vitrodata obtained with ß-1,2 tetramannosides are correlatedwhereas those obtained with ${alpha}$ -1,2 oligomannosides are not isunclear. In macrophages, oligomannosides with distinct anomeretypes of linkage have distinct receptors, since ß-1,2 mannosidesbind galectin 3 (17) and ${alpha}$ -1,2 mannosides react with the macrophagemannose receptor (42). If ${alpha}$ -1,2 and ß-1,2 mannosidesalso bind distinct receptors at the enterocyte surface, onepossible explanation is that Caco-2 cells express both receptors,whereas infant mouse enterocytes express receptors for ß-1,2mannosides only. An alternative hypothesis could be that enterocytereceptors for ${alpha}$ -1,2 mannosides were saturated in the infant mousemodel by ubiquitous ${alpha}$ -mannosidic sequences expressed by the bacterialflora of the gastrointestinal tract in vivo.

We next examined by immunofluorescence the fate of C. albicans ${alpha}$ -1,2 and ß-1,2 mannosidic epitopes during coculturewith Caco-2 cells. The reactivity of MAbs 5B2 and EBCA1 withantigenic material localized at the surface of the cells (Fig.5 and 6) suggests that ${alpha}$ -1,2 and ß-1,2 oligomannosidesare released by C. albicans during interaction with Caco-2 monolayersand bind to the surfaces of the cells. No such labeling wasobserved upon incubation of monolayers with heat-killed blastospores.The distribution of ${alpha}$ -1,2 and ß-1,2 mannoside labelingon Caco-2 monolayers was heterogeneous and was localized closeto C. albicans at 30 min postinfection (Fig. 5A). Low levelsof antigen shedding relative to enterocyte surfaces may accountfor this observation, a hypothesis supported by the increasedreactivity of monolayers upon prolonged incubation with C. albicans. Double-labelingexperiments with EBCA1 and 5B2 antibodies identified domainsof the monolayers that reacted with both antibodies. In such domains,merged images showed that clusters of ${alpha}$ -1,2 and ß-1,2mannoside reactivity did not necessarily colocalize (Fig. 6D),an observation consistent with the existence of a cell-specificor nonsynchronous expression of receptors for ${alpha}$ -1,2 and ß-1,2 oligomannosides.The shedding of ß-1,2 mannoside epitopes in associationwith C. albicans PLM and adhesion to the host cell membranewere previously reported during cocultures with macrophages (25).At present, we cannot decide whether ß-1,2 mannosidicepitopes shed by C. albicans during coculture with enterocytescorrespond to PLM or to PPM. However, the fact that some ofthe shed material reacted with 5B2 but not with EBCA1 suggeststhat PLM is involved, since the carbohydrate moiety of PLM iscomposed of ß-1,2 oligomannosides only. Altogether,this body of informations highlights the need to characterizeCaco-2 receptors for ${alpha}$ -1,2 and ß-1,2 oligomannosidesof C. albicans and to determine the biological significanceof this shedding process in the Candida-Caco-2 interaction.

Finally, Timpel et al. demonstrated recently that a homozygousC. albicans mutant defective in PMT1, a gene encoding a mannosyltransferase involvedin the O-glycosylation pathway, showed reduced adherence to Caco-2cells. This observation pointed to the role of O-linked carbohydratesin the adherence of C. albicans to enterocytes (44). Interestingly,the present paper indicates that ß-1,2- and ${alpha}$ -1,2-linked oligomannosides,two components of the N-glycan moiety expressed at the surfaceof C. albicans, participate in the adhesion of blastosporesto Caco-2 cells. Therefore, and despite the fact that the monolayersemployed in the two studies were presumably not in the samestate of differentiation, both O-linked and N-linked carbohydratesseem to represent putative adhesins in the host-fungus interplaythat takes place at the intestinal mucosal surface. The characterizationof the glycosylation pathways of C. albicans, including theirregulation and response to the yeast environment, will thusbe pivotal to understanding the commensal-pathogen transitionand the physiopathological events occurring in the early stagesof invasive candidiasis.

ACKNOWLEDGMENTS

This work was supported by the "Réseau Infections Fongiques"from the Ministère de l'Education Nationale, de la Rechercheet de la Technologie (MNERT); by a grant from the UniversityHospital of Dijon (Appel d'Offre Interne 1999); by "ProgrammeHospitalier de Recherche Clinique," appel d'offre régional2001; and by "Conseil Régional de Bourgogne" (grant 02/513/CP/02/S276).

FOOTNOTES

* Corresponding author. Mailing address: Alain Bonnin, Laboratoire de Parasitologie Mycologie, Hôpital du Bocage, BP 77908, 21079 Dijon Cedex, France. Phone: 33 (0)380 29 36 03 or 33 (0)380 29 35 23. Fax: 33 (0)380 29 32 80. E-mail: alain.bonnin{at}chu-dijon.fr.

Editor: T. R. Kozel

REFERENCES

Top