Dendritic Cell Interaction with Candida albicans Critically Depends on N-Linked Mannan^,

Abstract

The fungus Candida albicans is the most common cause of mycotic infections in immunocompromised hosts. Little is known about the initial interactions between Candida and immune cell receptors, because a detailed characterization at the structural level is lacking. Antigen-presenting dendritic cells (DCs), strategically located at mucosal surfaces and in the skin, may play an important role in anti-Candida protective immunity. However, the contribution of the various Candida-associated molecular patterns and their counter-receptors to DC function remains unknown. Here, we demonstrate that two C-type lectins, DC-SIGN and the macrophage mannose receptor, specifically mediate C. albicans binding and internalization by human DCs. Moreover, by combining a range of C. albicans glycosylation mutants with receptor-specific blocking and cytokine production assays, we determined that N-linked mannan but not O-linked or phosphomannan is the fungal carbohydrate structure specifically recognized by both C-type lectins on human DCs and directly influences the production of the proinflammatory cytokine IL-6. Better insight in the carbohydrate recognition profile of C-type lectins will ultimately provide relevant information for the development of new drugs targeting specific fungal cell wall antigens.

The fungus Candida is the most common cause of opportunistic mycotic infections in severely immunocompromised hosts, such as surgical, cancer, and transplant patients, worldwide (1). During infection, the transition from simple yeast form to multicellular filaments (hyphae) plays a fundamental role in evading the host immune defense. C. albicans remains the most common Candida species in human pathology, but the prevalence of other species including C. glabrata, C. parapsilosis, and C. dubliniensis, remains significant (2, 3).

The stimulation of proinflammatory cytokines and the subsequent activation of antifungal host defense depend on specific recognition of the invading fungus. Cells of the innate immune system recognize pathogens by identifying conserved microbial structures, called pathogen-associated molecular patterns (PAMPs),6 which are structurally distinct from molecules expressed on mammalian cells (4).

The Candida cell wall is almost exclusively composed of glycans. Glycans are synthesized as polymers of three types of monosaccharides: d-glucose, which forms β-(1,3) and β-(1,6) glucan, N-acetyl-d-glucosamine, which forms chitin, and d-mannose, giving rise to mannan (5). The outer layer of the C. albicans cell wall is enriched in mannoproteins, representing up to 30-40% of the cell wall dry weight (6), while chitin, β-(1,3)- and β-(1,6)-glucan are more prominent in the inner layer. In addition, the interplay between C. albicans glycans and the host induces modifications in C. albicans morphogenesis that are associated with changes in cell wall composition (7-9). Little is known about the initial interactions between Candida PAMPs and immune cell receptors, because a detailed characterization at the structural level is lacking.

The importance of detailed knowledge of the fungal PAMP repertoire and activity is also underscored by the development of more and more fungal glycan-based vaccines (10-13). Studies by Cutler and co-workers (14, 15) showed that monoclonal antibodies specifically directed against Candida mannan are able to enhance resistance to C. albicans infection in mice. Similarly, efforts are aimed at the development of synthetic glyco-conjugates based on fungal glycan structures that provide protection against candidiasis (16). A recent study also showed that yeast-derived mannosylation of recombinant tumor antigens enhanced immunogenicity and indeed efficiently stimulated T cell responses (17).

Whereas most C. albicans organisms are cleared by neutrophils and macrophages (18-20), the strategic location of antigen-presenting dendritic cells (DC) at mucosal surfaces and in the skin, the sites of Candida invasion, positions DCs in the first line of defense to obtain protective immunity. Despite this, the relative contribution of the various Candida PAMPs and their receptors to binding and cytokine stimulation by DCs remains unknown.

DCs express several pattern-recognition receptors (PRRs) that specifically recognize distinct PAMPs displayed on fungal surfaces including C-type lectin receptors (CLRs) (21) and Toll-like receptors (TLRs) (22). The macrophage mannose receptor (MMR), member of the CLR family, has been shown to recognize C. albicans on DCs (23), and we previously demonstrated that the DC-associated CLR DC-SIGN (CD209) is also involved in binding and internalization of C. albicans by human DCs (24).

Despite the progress in understanding the interaction of some of the fungal PAMPs with PRRs expressed on DCs, there is no integrated view of the mechanism(s) by which DCs sense C. albicans PAMPs. We recently showed that recognition of C. albicans by monocytes/macrophages is a multiple step process involving several PRRs recognizing the various layers of the outer portion of the fungal cell wall (25).

Here, by combining a range of C. albicans isogenic glycosylation mutants with receptor-specific blocking and cytokine production assays, we determined the structure of the fungal mannan that is specifically recognized by the CLRs on human DCs and can directly influence the production of the pro-inflammatory cytokine IL-6. Our results indicate that different mannan structures may skew the cytokine profile of DCs and differentially modulate the immune response. Therefore, better insight in the carbohydrate recognition profile of CLRs will ultimately provide relevant information for the development of new drugs targeting specific fungal cell wall antigens.

EXPERIMENTAL PROCEDURES

Reagents and Antibodies—Fluorescein isothiocyanate (FITC) was from Sigma. Recombinant gp120, mAb 1748 anti-DCIR, and polyclonal goat-anti-Dectin-1 were from R&D Systems Europe (Oxon, UK). ICAM-3Fc was produced as already described (26). The following Abs were used: mAb AZN-D1 and AZN-D2 anti-DC-SIGN (27); mAb anti-MMR from BD Pharmingen (Erenbodegem, Belgium); mAb MG38 anti-DEC-205 from eBioscience (AMDS, Malden, The Netherlands); APC-conjugated mAb anti-CD45RO from Becton Dickinson (Mountain View, CA); Alexa647-conjugated goat-anti-mouse IgG, Alexa647-conjugated goat-anti-mouse IgG1, and Alexa533 goat-anti-mouse IgG2b were from Molecular Probes (Leiden, Netherlands).

Carbohydrate Polymers—Zymosan and mannan (Lot 16H3842 from Saccharomyces cerevisiae) were from Sigma. Mannan derived from C. albicans (CA-mannan) was a kind gift from Dr. G. Kogan (Institute of Chemistry of Slovak Academy of Sciences, Bratislava, Slovakia) and was isolated as previously described (28). The CA-mannan contained no nitrogen as determined by elemental analysis. Glucomannan, chitin, and chitin-glucan were also a gift of Dr. G. Kogan (29). Glucan was isolated from C. albicans yeast cells as already reported (9). Scleroglucan was isolated and characterized as described (30). Supplemental Table S1 shows the carbohydrates used in this study and some of their characteristics. The N-linked oligosaccharides Man-9, Man-8, Man-7, Man-6, Man-5, and M3N2 (see supplemental Fig. S6 for their detailed structures) were purchased from QA-Bio (Palm Desert, CA).

Cells—Immature DCs were generated from human peripheral blood monocytes as already described (31), and their CLR expression pattern was as previously published (i.e.: positive for DC-SIGN, MMR, Dectin-1, and DCIR (24, 32, 33) and negative for DEC-205 (34)). Stable K562 transfectants expressing DC-SIGN (K-DC-SIGN) were already reported (27).

Candida Strains and Culture Conditions—C. albicans, strain UC820, a well described clinical isolate (35), was maintained on agar slants at 4 °C. Strain UC820 was inoculated into 100 ml of Sabouraud broth and cultured for 24 h at 37 °C. C. albicans serotype B wild-type (LGH 1095) and Δmnn4 were grown as reported elsewhere (36). Briefly, starter cultures were grown in 10 ml of Sabouraud broth overnight at 30 °C. 1 ml of overnight culture was inoculated into 100 ml of Sabouraud broth and cultured at 30 °C until log phase is reached. After two washes with pyrogen-free saline by centrifugation at 1500 × g, the number of yeast cells was counted in a hemocytometer and resuspended at 1 × 108 cells/ml. Heat-kill was at 56 °C for 1 h, if it is required. Clinical isolates of C. albicans, C. dubliniensis, C. glabrata, and C. parapsilosis yeast cells were used for Fig. 1A. Yeast cell suspensions were kept frozen at -80 °C until used. The homozygous null mutants in glycosylation genes were constructed in the C. albicans CAI-4 serotype A background by targeted gene disruption (37) and their detailed generation is already published (25, 38-41). Strain details are given in supplemental Fig. S3.

FIGURE 1.

DC-SIGN binds to yeast cells of different fungi. A, K-DC-SIGN cells were incubated with FITC-labeled C. albicans yeast cells in the presence or absence of blocking agents, as described under “Experimental Procedures.” The percentage of binding was calculated by flow cytometry, and indicates the amount of cells that became fluorescent upon interactions with FITC-labeled yeast cells. Data are presented as means ± S.D. Results are pooled from three independent experiments. B, DCs were incubated with FITC-labeled yeast cells of C. albicans or C. dubliniensis. The percentage of binding was calculated by flow cytometry. Four representative donors are depicted. C, DCs were incubated with FITC-labeled C. albicans or C. dubliniensis for 1 h at 37 °C to allow phagocytosis. Subsequently, samples were fixed, permeabilized, and specifically labeled with mAb against DC-SIGN. After addition of fluorescent secondary Ab, samples were analyzed by confocal microscopy (Candida, green; DC-SIGN, red). Scale bar is 5 μm.

Candida Binding Studies—The binding of DCs or K-DC-SIGN to Candida yeast cells was measured by flow cytometry using the FACSCalibur (BD Biosciences) and performed as already described (24). To test the effect of various reagents on ligand binding the following concentrations were used: different carbohydrates 150 μg/ml (unless otherwise indicated), EGTA 2 mm, isotype control (mouse IgG1), AZN-D1 anti-DC-SIGN, anti-MMR, anti-Dectin-1, anti-DCIR, and anti-DEC-205 (30 μg/ml). Incubations were performed in 20 mm Tris, pH 8.0, 150 mm NaCl, 1 mm CaCl₂, 2 mm MgCl₂, and 1% bovine serum albumin, as already published (24). FITC-labeled Candida was added in a cell/yeast ratio of 1:5. After 30 min of incubation at 37 °C, cell-yeast conjugates were analyzed by flow cytometry.

Phagocytosis—Immature DCs (5 × 10⁵) were allowed to adhere onto fibronectin and subsequently incubated with FITC-labeled C. albicans or C. dubliniensis yeast cells (2.5 × 10⁶) or FITC-labeled C. albicans CAI-4 wild-type and mannosylation mutants (2.5 × 10⁶) for 60 min at 37 °C. At the end of the incubation period, the samples were fixed in 4% PFA, permeabilized in Triton X-100, and labeled for DC-SIGN and/or the MMR using specific mAb and isotype-specific fluorescent secondary Abs (24). Samples were analyzed using a Zeiss LSM 510 confocal microscope.

Fluorescent Beads Adhesion Assay—The fluorescent beads adhesion assay was performed as described previously (42). The percentage of cells that bound fluorescent beads was determined by flow cytometry. In inhibition studies, the assay was performed in the presence of the inhibitors, as described above for the Candida binding studies.

NMR Spectroscopy—The proton and carbon-13 NMR spectra were collected on a JEOL Eclipse+ 600 NMR spectrometer in 5-mm OD NMR tubes at 70 °C in D₂O. Internal chemical shift reference was provided by trimethylsilyl-2,2,3,3-d₄-propionic acid (TSP) (43-45) at 0.0 and -2.78 ppm for ¹H and ¹³C spectra, respectively. Individual solutions of SC-mannan and CA-mannan were prepared at concentrations of 24 and 15 mg per 1 ml of solvent, respectively.

GPC/MALLS Methodology—The molecular weight distributions of the mannans were established by gel permeation chromatography (GPC) with online multi-angle laser light scattering photometry (MALLS) as described previously (46).

Stimulation of IL-6 Production in Human DCs—Samples of 3 × 10⁵ immature DCs in 300-μl volume were added to round-bottomed 48-well plates (Greiner Bio-One) and incubated with the various strains of C. albicans (ratio DCs/Candida = 1:10). After incubation for 6 h at 37 °C, the DCs/C. albicans cell suspensions were centrifuged, and the supernatants were collected and stored at -70 °C until assayed. IL-6 concentration was measured by commercial ELISA kits (Sanquin), as already reported (25).

RESULTS

DC-SIGN Recognizes Several Species of Candida—We previously demonstrated that DC-SIGN binds and internalizes C. albicans (24). Because several distinct Candida species are commonly found in patients with candidiasis (2, 3), we analyzed the capacity of DC-SIGN to recognize and bind a range of different Candida species (Fig. 1A). K562 cells stably expressing DC-SIGN (K-DC-SIGN) strongly bound C. albicans, C. dubliniensis, and C. glabrata, but only weakly interacted with C. parapsilosis. In addition, DC-SIGN bound S. cerevisiae-derived zymosan particles. Anti-DC-SIGN mAb and the Ca²⁺-chelating agent EGTA effectively inhibited the interactions demonstrating the specificity for DC-SIGN. DC-SIGN contributes to the binding and internalization of C. albicans by DCs (24). In Fig. 1B, we compare the binding of DCs isolated from different donors to C. albicans and C. dubliniensis and show that similarly to C. albicans, a significant percentage of the binding to C. dubliniensis was also mediated by DC-SIGN. Moreover, both C. albicans and C. dubliniensis yeast cells were found in DC-SIGN-enriched vesicles (Fig. 1C). Similar results were obtained when zymosan was used (data not shown). These results demonstrate that although the affinity of DC-SIGN varies among Candida species, in DCs DC-SIGN is involved in the uptake of all Candida species tested.

DC-SIGN Specifically Interacts with Candida-derived Mannan—The Candida cell wall is primarily comprised of mannan and by a lesser content of glucan and chitin (5). DC-SIGN is known to bind high mannose moieties (47). To determine the PAMP structures specifically recognized by DC-SIGN on the C. albicans cell wall, we investigated whether purified S. cerevisiae-derived mannan (SC-mannan) would inhibit DC-SIGN-mediated binding (Fig. 2A). As expected, binding to known DC-SIGN ligands, such as ICAM-3 and gp120 (27, 48), coated onto fluorescent beads, as well as to zymosan particles, was inhibited by SC-mannan. Surprisingly, the interactions with both C. albicans and C. dubliniensis were not affected by the pretreatment with SC-mannan (Fig. 2A).

FIGURE 2.

DC-SIGN specifically recognizes mannan of C. albicans cell wall. A, K-DC-SIGN cells were incubated with different ligands in the presence or absence of anti-DC-SIGN blocking agents, as described under “Experimental Procedures.” ICAM-3- and gp120-beads indicate fluorescent beads (1 μmØ) coated with soluble ICAM-3 and gp120, respectively. The percentage of binding was calculated by flow cytometry and indicates the proportion of cells that became fluorescent upon interactions with either fluorescent beads or FITC-labeled fungi. B, K562 cells stably expressing similar levels of either DC-SIGN wild type or V351G- or E324A-DC-SIGN mutant were incubated with FITC-labeled C. albicans in a ratio of 1:5, in the presence or absence of blocking agents, as described under “Experimental Procedures.” C, K-DC-SIGN cells were incubated with FITC-labeled C. albicans in a ratio of 1:5, in the presence or absence of purified fungal cell wall glycans. CA-mannan 50 and CA-mannan 100 indicate a concentration of 50 and 100 μg/ml, respectively. All other glycans were used at a concentration of 150 μg/ml, as described under “Experimental Procedures.” Basal binding (none) is set as 100%. Data are presented as means ± S.D. Results are pooled from three independent experiments. D, DCs were incubated with C. albicans FITC-yeast cells in the presence or absence of purified glycans. Basal binding (none) is set as 100%. Data are presented as means ± S.D. Results are pooled from three independent experiments. E, DCs were incubated with C. albicans FITC-yeast cells in the presence of different CLR-blocking mAbs (30 μg/ml). Basal binding (none) is set as 100%. Data are presented as means ± S.D. Results are pooled from three independent experiments. ^*, p < 0.01 versus basal binding, as determined by the two sample t-test.

To further investigate whether the interaction between DC-SIGN and Candida occurs at the carbohydrate recognition domain (CRD) of the receptor, two mutants of DC-SIGN, each carrying a specific point mutation known to affect binding, were compared with the wild-type receptor for their ability to bind to the fungus. The amino acid residue Val³⁵¹ has been shown to be essential for DC-SIGN-mediated binding, because its mutation to Gly abrogated the interaction with ICAM-3 (49) and LewisX (50). Furthermore, it was reported that mutating the Ca²⁺-chelating residue Asp³²⁴ into Ala resulted in complete loss of ligand binding (49). The observation that the V351G mutant of DC-SIGN binds to Candida much less efficiently than wild type and that the E324A mutant had totally lost its capacity to interact with the fungus (Fig. 2B) demonstrates that the interaction between DC-SIGN and Candida is specifically mediated by the CRD. It has to be noted that the expression level of these mutated forms of DC-SIGN is comparable to the wild-type (supplemental Fig. S1).

To get further insight into the specificity of DC-SIGN recognizing mannan structures from fungi, we analyzed several purified fungal cell wall carbohydrates (supplemental Table S1) for their capacity to inhibit this interaction. Strikingly, CA-mannan was the only fungal cell wall carbohydrate that specifically inhibitedK-DC-SIGNbindingto C. albicans inaconcentration-dependent manner (Fig. 2C). The high specificity of this recognition was further demonstrated by the observation that CA-mannan but not SC-mannan was able to inhibit binding of C. dubliniensis to K-DC-SIGN (supplemental Fig. S1A). The inhibitory effect of SC-mannan on DC-SIGN binding to zymosan (Fig. 2A) suggests that SC-mannan may better represent the structure of the zymosan mannan rather than the mannan structure exposed at the cell wall of Candida.

Similarly, we demonstrated that CA-mannan was the only fungal cell wall-derived carbohydrate tested that specifically and significantly abrogated binding of DCs to C. albicans (Fig. 2D). The blockade of binding observed after pretreatment of DCs with CA-mannan but not with other carbohydrates also shows that no significant role is played by glucan receptors on human DCs in binding C. albicans. This notion is further supported by the finding that blocking of DC binding to C. dubliniensis was also detected only in the presence of CA-mannan (supplemental Fig. S1B). In agreement with these observations, when a panel of DC-associated CLRs was blocked, only inhibition of MMR and DC-SIGN significantly affected the recognition of Candida by DCs (Fig. 2E). These data clearly indicate that the PAMP specifically recognized by DCs in the interactions with Candida is mannan.

DC-SIGN Recognizes Structural Differences in Fungal N-Mannan—To reveal the differences between CA- and SC-mannan, we next performed a comparative structural analysis by nuclear magnetic resonance (NMR) (supplemental Fig. S2). All fungal mannans consist of a backbone of α-(1,6)-linked mannosyl repeat units with side chains of α- or β-(1,2)-linked mannosyl repeat units connected to the backbone by a α-(1,2)-linkage as well as phosphate di-mannosyl esters and α-(1,3)-linked mannosyl units at the end of the side chains (51). The α-anomer glycosidic linkage configuration was the major conformation found with a low level of β-anomer glycosidic linkage configuration. Structural characterization of the mannans indicated that CA-mannan better resembles the native mannan from the cell wall of living Candida cells than SC-mannan (supplemental Fig. S2). Molecular weights of the mannans were established by GPC/MALLS analysis. The GPC-determined molecular weight for SC-mannan was 3.7 × 10⁴ g/mol, while for CA-mannan was of 4.9 × 10⁵ g/mol. This molecular weight difference is supported in the resonance line widths observed in the C-13 NMR spectra (supplemental Fig. S2C). Schematic representations of the structural features derived from NMR and GPC/MALLS analysis are given in Fig. 3A (see also supplemental Fig. S2D for a detailed structure). This comparative NMR analysis clearly confirmed that the SC- and CA-mannan used in our studies significantly differ in terms of size and branching, suggesting that these specific structural differences between the two mannans could account for differences in affinity for the receptor and therefore in blocking capacity.

FIGURE 3.

DC-SIGN recognizes structural differences in fungal N-mannans. A, structures of SC- and CA-mannan are schematically shown according to the NMR structural analysis (see supplemental Fig. S3 for details) and follow the model of Ref. 51. B, detailed structures of the N- and O-linked mannan for different isogenic mutants of C. albicans. (i), wild-type strain NGY152 [CAI-4 plus Clp10 vector]; (ii) the pmr1 mutant (strain NGY355) presents large defects in mannosylation, characterized by absence of phosphomannan and dramatic reduction of N- and O-mannosyl chains (39); (iii) the och1 null mutant (strain NGY357) is defective in the branched outer N-linked mannosyl chain (38); (iv) the mnt1/mnt2 mutant (strain NGY337) lacks the 4 terminal O-linked α_1,2-mannosyl residues but has intact N-mannan (41); (v) the mnn4 mutant (strain CDH15) only lacks phosphomannan within the N-mannan structure (40). Man, mannosyl; β-GlcNac, β-N-acetylglucosamine. C, K-DC-SIGN cells were incubated with FITC-labeled yeast cells of C. albicans wild-type and mutated strains in the presence or absence of blocking agents for 30 min at 37 °C. The percentage of cells binding to Candida was determined by flow cytometry, as described under “Experimental Procedures.” Data are presented as means ± S.D. Results are pooled from three independent experiments. D, K-DC-SIGN cells were incubated with FITC-labeled yeast cells of C. albicans serotype B in the presence or absence of blocking agents for 30 min at 37 °C. The percentage of cells binding to Candida was determined by flow cytometry, as described under “Experimental Procedures.” The inset shows a part of serotype B N-linked mannan. The color code for the mannose residues is the same as in B. Data are presented as means ± S.D. of one experiment out of four.

Subsequently, to better understand which structural elements of the fungal mannan are essential for the interaction between Candida and DC-SIGN, we exploited the well-defined isogenic mutants of C. albicans depleted in specific mannan structures (25). In Fig. 3B, the N- and O-mannan structures of the wild-type Candida strain and the isogenic mutants are displayed. The och1 mutant lacks branched outer N-linked mannosyl chains (38), while the pmr1 mutant has defects both in N- and O-linked mannosylation (39). The mnn4 mutant only lacks phosphomannan linked to the N-mannan (40), and the double mutant mnt1/mnt2 has intact N-mannan but lacks the 4 terminal O-linked α-(1,2)-mannosyl residues (41).

We used these mutants to analyze the interaction between K-DC-SIGN and Candida mannan (Fig. 3C). Binding to DC-SIGN was severely reduced in the pmr1 mutant as well as in the och1 strain. In contrast, the absence of mannosylphosphate (mnn4) or O-linked mannan (mnt1/mnt2) had no effect on the interaction between DC-SIGN and Candida. Similarly, when C. dubliniensis glycosylation mutant och17 was used, no binding of K-DC-SIGN was detected (supplemental Fig. 4A), indicating that the N-linked mannan is the general structure recognized by DC-SIGN in all Candida species. These data unequivocally demonstrate that N-mannan is the only structure required for the recognition of Candida by DC-SIGN and that no O-linked or phosphomannan structures are recognized.

To further define the N-mannan epitope responsible for this interaction, C. albicans serotype B was used. Serotype B specifically lacks the terminating β-1,2-linked mannose units in the acid-stable region, but still has the β-1,2 mannose attached to the phosphomannan (52). As shown in Fig. 3D, DC-SIGN is able to bind to serotype B Candida, suggesting that the terminating β-1,2 mannose units are not required for binding. K-DC-SIGN cells were also able to bind to serotype B mnn4Δ mutant, which lacks all β-1,2 mannose residues (36). This further indicates that no β-1,2 mannose is involved in the interaction of DC-SIGN with Candida (data not shown).

N-Mannosylation Mediates Binding, Phagocytosis, and Cytokine Production by Dendritic Cells—Next, exploiting the same series of Candida glycosylation mutants, we determined which mannan structures specifically were involved in recognition of Candida by DCs expressing both DC-SIGN and MMR (Fig. 4A). The mannan structures recognized by DC-SIGN on the transfected K562 cell line were also required to mediate the interaction between the DCs and Candida. Wild-type N-mannan, with or without phosphomannan, is therefore essential for interaction of C. albicans with DCs. Similarly to what observed for K-DC-SIGN cells, DCs also bind to C. albicans serotype B wild-type as well as Δmnn4 (supplemental Fig. S5), indicating that no terminating β-1,2-mannose residues are involved in the binding of Candida to DCs. Finally, binding of DCs to C. dubliniensis is also depending on the presence of the N-linked mannan, as shown by the lack of interaction between DCs and C. dubliniensis och1 mutant (supplemental Fig. S4B), further supporting the important role played by N-linked mannan in the binding of Candida to DCs. Subsequently, we analyzed the individual contribution of DC-SIGN and MMR to the binding of DCs to the Candida wild-type, mnt1/mnt2, and mnn4 mutants (Fig. 4B). The partial block observed in the presence of the anti-DC-SIGN blocking mAb is in agreement with previous observations, where MMR and DC-SIGN were responsible for ∼70 and ∼30% of the binding to Candida, respectively (24). Moreover, similarly to what was observed upon preincubation with CA-mannan, inhibition of both MMR and DC-SIGN almost completely abrogated the interaction of the various mutant strains with DCs (Fig. 4B).

FIGURE 4.

N-mannosylation mediates binding, phagocytosis, and IL-6 production by DCs. A, DCs were incubated with FITC-labeled C. albicans wild-type and mutated strains for 30 min at 37 °C, as described under “Experimental Procedures.” Data are presented as means ± S.D. Results are pooled from three independent experiments with six different donors. B, DCs were incubated with FITC-labeled C. albicans wild-type and mutated strains in the presence or absence of blocking mAb for 30 min at 37 °C, as described under “Experimental Procedures.” The percentage of binding was calculated by flow cytometry: basal binding was set as 100%. Data are presented as means ± S.D. of a representative experiment out of 5 performed in duplicate. C, DCs were allowed to adhere onto fibronectin-coated glass coverslips for 30 min at 37 °C and subsequently incubated with FITC-labeled C. albicans wild-type and mutated strains (green) for 90 min at 37 °C to allow phagocytosis. After washing of unbound Candida, fixation, and permeabilization, DCs were fluorescently labeled with mAbs specific for DC-SIGN (red) and the MMR (blue). Scale bars, 10 μm. D, DCs were stimulated with the various C. albicans strains: the wild-type strain CAI-4, the pmr1 mutant defective in both N- and O-mannosylation, the och1 null mutant lacking the branched outer N-linked mannosyl chain, the mnt1/mnt2 mutant lacking only the 4 terminal O-linked α_1,2-mannosyl residues, and the mnn4 mutant defective in phosphomannan. After incubation for 6 h at 37 °C, supernatants were collected, and IL-6 expression was determined by ELISA. Results (mean ± S.D.) are pooled data from three separate experiments with three different donors. ^*, p < 0.01 versus wild-type CAI-4, as determined by the two sample t-test.

In Fig. 4C, we show that in addition to binding, DCs also phagocytose C. albicans mutants, and that both MMR and DC-SIGN are found around the Candida containing phagosomes. While the och1 and pmr1 mutants lacking N-linked mannan chains were not phagocytosed, mutants with an intact N-mannan but with defects in O-mannosylation and phosphomannan biosynthesis-such as mnt1/mnt2 and mnn4 were readily bound and internalized. Three-dimensional Z-sectioning of DCs in the confocal microscope unequivocally proved that the Candida yeast cells were not only bound but indeed ingested by DCs.

Finally, we investigated the role of N- and O-linked mannosylation in stimulating cytokine production in human DCs by C. albicans by comparing the production of the prototype proinflammatory cytokine IL-6, induced after 6 h of exposure to the various mannan defective mutants. The absence of O-linked mannosylation (mnt1/mnt2, mnn4) did not have any significant effect on IL-6 expression, whereas the mutants defective in N-linked mannosyl residues (och1, pmr1) produced only very low levels of cytokine (Fig. 4D). The same pattern was observed after 20 h of incubation (data not shown). Altogether these results clearly demonstrate that the presence of N-linked mannosyl residues is essential for interaction, recognition, and internalization of Candida yeast cells as well as IL-6 production by DCs.

DISCUSSION

Disseminated infections with C. albicans cause significant morbidity and mortality among immunocompromised individuals, such as HIV patients, transplant recipients, and cancer patients (1). Therefore, a better understanding of the specific receptors involved in the interactions between invading fungi and immune cells is necessary.

We demonstrated previously that DCs bind and internalize C. albicans through the C-type lectin DC-SIGN (24). Here we show that besides C. albicans, DC-SIGN also binds to yeast cells of a range of Candida species as well as to zymosan particles. Furthermore, we determined that the PAMP bound by DC-SIGN on Candida cell wall is mannan, and that N-linked mannosyl residues are essential for this interaction. Finally, we demonstrate that the N-mannosylation is specifically required for the binding, phagocytosis, and immune sensing of C. albicans by the DCs.

Interactions between host and fungal pathogens occur at the level of the cell wall (7), which is composed of glycoproteins embedded within a polysaccharide matrix. Therefore, carbohydrates are the immediate point of interaction with host tissues. A detailed characterization of these initial interactions at a molecular level has been lacking. The findings in this report demonstrate that the interaction between DC-SIGN and C. albicans is mediated by mannan rather than other fungal cell wall components such as glucan, chitin, or chitin-glucan complex. However, the SC-mannan, purified from S. cerevisiae cell wall, was not able to block the interaction between DC-SIGN and Candida, whereas CA-mannan, purified from C. albicans, could specifically inhibit binding of this fungus to DC-SIGN. This observation together with structural data obtained by the NMR and GPC-MALLS analysis indicate that SC-mannan and CA-mannan differ in their average structures, with SC-mannan being a smaller polymer without a phosphate diester linkage, while the CA-mannan is a larger polymer containing the phosphate diester linkage. SC-mannan is isolated by a procedure involving acetolysis, which selectively cleaves α-(1,6)-linkages in the backbone, while CA-mannan isolation did not involve acetolysis (53, 54). Therefore, different isolation protocols could be responsible for the different structural properties. However, the recent work of Shibata et al. (55) shows the chemical structure of the cell-wall mannan of C. albicans after acetolysis that is overall similar to the CA-mannan structure used in this study. These differences in mannan structure may account for the observed differences in receptor interaction and binding efficiency.

To investigate the role of cell wall mannosyl groups in the recognition of C. albicans yeast cells by DCs and DC-SIGN in particular, specific isogenic mutant strains of C. albicans were used with specific defects in the mannosylation of cell wall proteins. Gross defects in both N-linked and O-linked mannosylation were investigated using C. albicans pmr1 mutant (39). The role of O-linked mannosylation was assessed using the mnt1/mnt2 mutant that lacks two α_1,2-mannosyl transferases required for the addition of mannose residues to a linear oligomannoside (41), while the role of phosphomannan was studied in the mnn4 mutant (40). Finally, the importance of N-linked mannosylation was indicated in experiments using the och1 mutant strain, which is unable to synthesize the branched outer mannan chains (38). We demonstrated that for the interaction with C. albicans, DC-SIGN required wild type N-linked mannan, but was not dependent on the presence of phosphomannan, or β-1,2 mannose or O-linked mannosyl residues. Thus altogether our binding studies indicate that the epitope necessary for binding of Candida to DC-SIGN-expressing cells must be within the α-1,2 branched mannose residues present in the acid-stable part of the N-linked mannan (Fig. 5).

FIGURE 5.

The mannan epitope is essential for the interaction between C. albicans and DC-SIGN-expressing cells. DC-SIGN-expressing cells bind to Candida lacking the O-linked mannan (mnt1/mnt2), to Candida lacking the phosphomannan (mnn4) and to Candida lacking the terminating β-1,2 mannose residues (serotype B wild-type as well as Δmnn4), indicating that these moieties do not represent the essential binding epitope. On the contrary, DC-SIGN-expressing cells are unable to bind to Candida that has a dramatic reduction (pmr1) or a complete loss (och1) of the branched acid-stable N-linked mannan part. Mannosyl residues not essential for binding are shown in gray.

The binding of C. albicans specifically involves the CRD of the lectin receptor, since mutations that altered amino acid residues known to be essential for ligand binding impaired the binding of DC-SIGN to Candida. Moreover, the V351G mutant does not bind ICAM-3 (49) nor LewisX (50), but does bind gp120 (49). Therefore, its lower but detectable binding to C. albicans might suggest that the binding site of DC-SIGN responsible for the interaction with CA-mannan partially overlaps with that of gp120.

On the other hand, preincubation of K-DC-SIGN cells with well-defined N-linked oligosaccharides did not affect the binding to C. albicans (supplemental Fig. S6), suggesting that the complex CA-mannan moiety might not bind to DC-SIGN as conventional high mannose oligosaccharides. The binding of DC-SIGN to synthetic fragments of high mannose oligosaccharides has been shown to occur via multiple modes that may involve different amino acids and further enhance the receptor avidity (56, 57). Moreover, the intrinsic oligomerization of DC-SIGN (58) and its tendency to form nanoclusters (59) are likely to provide additional regulation levels for the interaction of DC-SIGN with this type of fungal mannan arrays.

DCs bind to C. albicans via two CLRs, MMR and DC-SIGN. Interestingly, although the MMR recognizes single terminal mannose residues and DC-SIGN interacts with more complex mannose residues in specific conformations, the same mannosylation mutations that hamper DC-SIGN binding to Candida abrogate binding and subsequent phagocytosis by DCs. This suggests that despite the differences in the receptors, immune sensing of Candida by DCs requires N-linked mannan oligosaccharides. In addition to binding and phagocytosis, interaction of C. albicans with DCs also led to production and release of the prototype proinflammatory cytokine IL-6, a process crucial for the activation and modulation of the adaptive immune response. When the various C. albicans mutant strains were used to stimulate IL-6 production, it appeared that the absence of N-linked mannosylation, but not O-linked mannans, led to markedly reduced production of IL-6. The residual 20-25% IL-6 production observed in absence of N-linked mannosylation could be induced by other fungal cell wall components present also in the och1 and pmr1 mutants. Studies aimed at dissecting the role of various PRRs involved in the signaling cascade leading to cytokine production on human DCs stimulated with Candida are ongoing.

This study represents the first comprehensive attempt to understand the mechanism of Candida recognition by human DCs. We conclude that recognition of N-linked mannans by DC-SIGN and MMR is vitally important for binding and phagocytosis of C. albicans by human DCs.

To understand how CLRs expressed on DCs discriminate between endogenous and pathogen-associated antigens, detailed knowledge of the specific carbohydrate structures recognized is essential. Especially the use of mutants that differ in their carbohydrate make-up are a valuable tool because we test an intact pathogen rather than isolated structures. This will allow us to get better insight in how the integrated signals from both CLRs and TLRs act on the immune system and how carbohydrate recognition profile analyses can ultimately lead to the development of new drugs targeting specific microbial carbohydrate antigens.