Cell wall mannan of Candida krusei mediates dendritic cell apoptosis and orchestrates Th17 polarization via TLR-2/MyD88-dependent pathway

Dendritic cells (DCs) abundantly express diverse receptors to recognize mannans in the outer surface of Candida cell wall, and these interactions dictate the host immune responses that determine disease outcomes. C. krusei prevalence in candidiasis worldwide has increased since this pathogen has developed multidrug resistance. However, little is known how the immune system responds to C. krusei. Particularly, the molecular mechanisms of the interplay between C. krusei mannan and DCs remain to be elucidated. We investigated how C. krusei mannan affected DC responses in comparison to C. albicans, C. tropicalis and C. glabrata mannan. Our results showed that only C. krusei mannan induced massive cytokine responses in DCs, and led to apoptosis. Although C. krusei mannan-activated DCs underwent apoptosis, they were still capable of initiating Th17 response. C. krusei mannan-mediated DC apoptosis was obligated to the TLR2 and MyD88 pathway. These pathways also controlled Th1/Th17 switching possibly by virtue of the production of the polarizing cytokines IL-12 and IL-6 by the C. krusei mannan activated-DCs. Our study suggests that TLR2 and MyD88 pathway in DCs are dominant for C. krusei mannan recognition, which differs from the previous reports showing a crucial role of C-type lectin receptors in Candida mannan sensing.

C.krusei infections difficult to treat and led to a high mortality rate 2,14 . Despite its increasing importance, little is known regarding the immune system response to C. krusei.
Carbohydrate constituents of Candida cell walls play a pivotal role in triggering host immune responses, which in turn either protect against the fungal infection or facilitate fungal immune evasion [15][16][17] . Mannans are mannose polymers located in the outermost part of Candida cell walls; therefore, they may be the first component to interact with the immune system. As cell wall mannans are complex structures, elaborate immune mechanisms have evolved 16,17 . While studies have shown that mannans can induce anti-fungal protective immunity [18][19][20] , other reports have revealed that mannans are a significant virulence factor associated with the severity and pathogenesis of Candida infections 21,22 . Furthermore, high levels of mannans can be detected in the blood of invasive candidiasis patients and it has been related to disease onset and progression 23,24 .
Dendritic cells (DCs) are potent antigen-presenting cells that reside in both lymphoid and non-lymphoid tissues and act as sentinels of the immune system. Interactions between invading pathogens and DCs via pathogen-associated molecular patterns (PAMPs) pattern-recognition receptors (PRRs) provide the foundation that triggers adaptive immune responses 16,25 . DCs abundantly express C-type lectin receptors (CLRs) and Toll-like receptors (TLRs), many of which can bind to Candida mannans. The activation of different types of mannan-specific receptors leads to differential DC activation that subsequently dictates distinct T cell responses 16,17,25 . Recognition of Candida mannans by CLRs and TLRs on DCs depends on mannan structure and mannosyl composition. In general, Candida N-linked mannans are recognized by dectin-2, mincle, mannose receptor (MR or CD206) and DC-SIGN (CD209), while O-linked mannans are recognized by TLR-4 17 . Furthermore, the α-mannans preferentially engage with dectin-2 and dectin-3 20,26 , while the β-mannans specifically ligate to galectin-3, which mediates TLR-2 activation 27,28 . The interactions of Candida mannans with several CLRs expressed on DCs induce Syk activation, which consequently mediates innate resistance to systemic fungal infection and orchestrates the Th17 response 19,29,30 . However, some mannose residues mediates signal transduction via the TLR/MyD88-dependent pathway, and participates in host defense against C. albicans infection [31][32][33] .
To date, the role of C. krusei mannan in DC immunity is not clear. Since mannan structures and mannosyl composition in the cell wall of Candida species are highly diverse, we compared the effects of cell wall mannans extracted from C. albicans, C. troplicalis, C. glabrata and C. krusei on DCs, and T cell responses.

C. krusei mannan induced DC maturation and triggered massive productions of pro-inflammatory cytokines.
To evaluate whether cell wall mannans extracted from four distinct Candida species differentially affected the phenotypic maturation of DCs, BMDCs were stimulated with various concentrations of mannans and subsequently characterized by flow cytometric analyses of the maturation markers CD40, CD80, CD86 and MHC class II (Figs. 1, S1 and S2). The DC population was first identified by gating a DC marker, CD11c (Fig. S1A), and geometric MFI of the maturation markers was assessed using a histogram analysis (Figs. 1A and S1B). BMDCs stimulated with C. albicans and C. tropicalis mannans did not undergo maturation compared to the negative control, whereas those stimulated with C. krusei and C. glabrata mannans were potently activated. C. krusei mannan upregulated expression of CD40, CD86 and MHC class II on BMDCs, and induced the highest levels of CD40, especially at the highest mannan concentration. Although, C. glabrata mannan also induced CD80, CD86 and MHC Class II expression on BMDCs, expression differed slightly from that of BMDCs stimulated with C. krusei mannan. To determine the number of DCs that underwent maturation, dot plot analyses were performed to evaluate the percentage of CD11c + CD40 + , CD11c + CD80 + , CD11c + CD86 + and CD11c + MHC class II + cells within the CD11c + fraction (Figs 1B and S2). The percentages of each DC subpopulation were consistent with their geometric MFI levels (Figs. 1A and 1B).
Next, to determine the effects of mannan stimulation on DC cytokine production, we quantitated the levels of the pro-inflammatory cytokines TNF-α, IL-1β, IL-6, IL-23, IFN-γ, IL-12 and the anti-inflammatory cytokines IL-4 and IL-10 in response to Candida mannans (Fig. 2). C. albicans and C. tropicalis mannans induced significant levels of IFN-γ production in BMDCs compared to the negative control. However, in parallel to the effect on DC maturation, C. albicans and C. tropicalis mannans did not stimulate BMDCs to produce the other cytokines. C. glabrata mannan significantly upregulated IFN-γ, and slightly promoted IL-6 and IL-23 production; however, it failed to stimulate BMDCs to produce the other cytokines, regardless of its high DC-activation ability. Of note, C. krusei mannan at all concentrations augmented massive production of TNF-α, IL-6, IL-23, IFN-γ, and IL-12. None of the Candida mannans promoted production of the anti-inflammatory cytokines IL-4 and IL-10. Our findings imply that the cell wall mannans of these four distinct Candida species differentially impact DC maturation and function, and C. krusei mannan, in particular, elicits robust inflammatory DC responses.
C. krusei mannan reduced BMDC viability by the induction of cellular apoptosis. Signal transduction via PRRs has been shown to lead to transcriptional expression of inflammatory cytokines and trigger cellular apoptosis 34,35 . On the basis of the vigorous pro-inflammatory cytokine production of BMDCs in response to C. krusei mannan stimulation, we hypothesized that C. krusei mannan may also induce cell death. To test this hypothesis, BMDCs were incubated with Candida mannans at various concentrations and cell viability was assessed by MTT assays. We found that C. krusei mannan at all concentrations markedly decreased BMDC viability, but the other Candida mannans did not (Fig. 3A). Changes in CD11c + BMDCs were also observed using flow cytometric analysis. After CD11c + cells were gated, live cells were identified based on forward scatter (FSC) and side scatter (SSC) signals (Fig. S3A). The percentage of live cells in the CD11c + population was notably reduced when BMDCs were incubated with C. krusei mannan, whereas they were unaffected by mannans from the other three Candida species (Figs. 3B and S3B). To determine whether the reduction in live cells was due to DC apoptosis, unstimulated and Candida-mannan-stimulated BMDCs were stained with CD11c, Annexin V and 7AAD. CD11c + cells that were single positive for Annexin V or double positive for Annexin V and 7AAD were identified as apoptotic DCs (Fig. 3C). Consistently, only C. krusei mannan significantly induced apoptosis in DCs (Fig. 3D). Our data imply that C. krusei mannan affects DC viability through a process of activation-induced cellular apoptosis.  36,37 , we questioned which pathway is responsible for C. krusei mannan-mediated DC apoptosis. To elucidate the signal transduction underlying this phenomenon, downstream signaling of the CLR-coupled Syk and TLR-MyD88 pathways was individually investigated using pathway-specific inhibitors. BMDCs were pre-treated with Syk or MyD88 inhibitors, and subsequently stimulated with C. krusei mannan. Cell viability was determined using MTT assay. C. albicans β-glucan was used as a positive control since it transduces signals via the dectin-1-Syk pathway 38 , and we found that it did not affect BMDC cell viability (Fig. 4A). Inhibition of Syk in unstimulated and β-glucan-stimulated BMDCs significantly decreased cell viability, but did not abrogate C. krusei mannan-induced cell death (Fig. 4A). In a parallel experiment to examine the role of MyD88, LPS was used as the positive control for MyD88 activation. LPS stimulation resulted in a marked reduction in BMDC viability, similarly to C. krusei mannan stimulation, and inhibition of MyD88 rescued the viability of BMDCs stimulated with C. krusei mannan as well as that of those stimulated with LPS (Fig. 4B). In addition, the apoptosis staining assay demonstrated that MyD88 inhibition, to a great extent, interfered with both LPS-and C. krusei mannan-mediated apoptosis of the CD11c + population (Figs. 4C and 4D). These results imply an intriguing role of the MyD88-dependent signaling pathway in DC apoptosis in response to C. krusei mannan stimulation. DC apoptosis-mediated by C. krusei mannan was associated with TLR2 activation. Previous studies have shown that the mannosyl residuals of C. albicans cell wall can be recognized by TLR2 39 , and this recognition mediated macrophage apoptosis 40 . We therefore investigated the functional relevance of TLR2 in DC apoptosis triggered by C. krusei mannan stimulation. To confirm that the activation of TLR2 also induce the cell death, BMDCs were incubated with a specific TLR2 ligand, Pam 3 CSK 4 , and the cell viability was determined using MTT assay. As expected, BMDCs stimulated with Pam 3 CSK 4 reduced the BMDC viability (Fig. 5A). Next, C. krusei mannan-mediated DC apoptosis via TLR2 ligation was determined by using the blocking antibodies. As a positive control, Pam 3 CSK 4 , significantly enhanced DC apoptosis, and this Pam 3 CSK 4 -induced cellular apoptosis was impeded by pre-treatment with the anti-TLR2 blocking antibody. Notably, blockade with anti-TLR2 We also observed the effect of C. krusei cell wall on DC apoptosis (Fig. S4). C. krusei yeast cells were inactivated by heat so that the cells could not secrete any molecules that might have influenced the DC response. In addition, we expected that mannans would be the first component to interact with DCs because they are located in the outermost layer of Candida cell walls 16,17 . Consistent with the apoptosis induction by C. krusei mannan (Fig. 5C), BMDCs stimulated by heat-inactivated C. krusei yeast cells also displayed significantly increased apoptosis, and blockade of TLR2 abrogated apoptosis induction (Fig. S4).

C. krusei mannan reduced BMDC viability by the induction mannan mediated
Taken together, our data indicate that TLR2 plays an important role in the mechanism of apoptosis induction by the cell wall mannan of C. krusei.
C. krusei mannan-stimulated DCs orchestrated antigen-specific Th17 response. Since C. krusei mannan induced massive cytokine production and impaired DC viability, we next investigated antigen-specific T cell responses to stimulated DCs. As a large number of DCs reside in the skin 41 , mice were immunized with a mixture of ovalbumin and various Candida mannans (OVA-Candida mannans) via a subcutaneous route so that DCs could be exposed directly to the stimuli. Seven days after the final immunization, the immune cell populations (T cells, B cells, memory T cells, and Th subsets) in regional lymph nodes (RLNs) were examined by flow cytometric analysis (Figs S5 and S6). Total numbers of RLN cells were significantly increased when mice were immunized with OVA-C. albicans mannan or OVA-C. krusei mannan. However, none of the immunized mice exhibited any alteration in the proportions of immune cell populations (Fig. S6). Several lines of evidence krusei mannan, and negative control groups. On the other hand, immunization with OVA-C. glabrata mannan did not activate cytokine production. Augmented IL-17 production was observed particularly in OVA-C. krusei mannan-immunized LN cells.
In vivo mannan immunization may not only activate DCs but also possibly affect the response of other immune cells. Therefore, to investigate whether the distinct T cell responses were due to the direct impact of Candida mannan-stimulated DCs on CD4 T cells, we performed an in vitro co-culture assay. BMDCs were primed with Candida mannans and subsequently pulsed with OVA. These BMDCs were then co-cultured with T cells isolated from OT-II TCR transgenic mice. Secretion of IFN-γ, IL-17, IL-4 and IL-10 in the culture supernatant, representing T helper cell responses, was analyzed. Concurring with our results from the ex vivo re-stimulation assays, C. abicans and C. tropicalis mannan-stimulated DCs profoundly augmented IFN-γ production from OT-II T cells, while C. krusei mannan-stimulated DCs preferentially induced IL-17 production (Fig. 6B). IL-4 and IL-10 were not detectable in this co-culture system. Collectively, our results reveal that DCs activated with C. krusei mannan, despite undergoing apoptosis, were capable of orchestrating the Th17 response in an antigen-specific manner. Th17 induction by C. krusei mannan-stimulated DCs was via the MyD88 signaling pathway. Having demonstrated the involvement of the MyD88-dependent pathway in DC apoptotic response to C. krusei mannan (Fig. 4), we investigated whether MyD88 was also required for Th17 induction by C. krusei mannan-stimulated DCs. BMDCs were pre-treated with a MyD88 inhibitor and then primed with C. albicans mannan or C. krusei mannan. Subsequently, these BMDCs were pulsed with OVA and co-cultured with OT-II T cells. Inhibition of MyD88 in DCs stimulated with mannan from C. krusei converted the production of IL-17 to IFN-γ in the activated OT-II T cells, but this did not occur in DCs stimulated with mannan from C. albicans ( Fig. 7A). TLR-2 blockade in C. krusei mannan-stimulated DCs also resulted in reduced IL-17 production by OT-II T cells, but did not affect IFN-γ production (Fig. S7A). Consistently, MyD88 inhibition in BMDCs stimulated with C. krusei mannan significantly reduced IL-6, the key cytokine in Th17 induction 42,43 , while it substantially enhanced the production of IL-12, which is required for Th1 induction 42 (Fig. 7B). We further verified the roles of TLR2 in the cytokine production of C. krusei mannan-stimulated DCs by using blocking antibodies. TLR2 blockade significantly suppressed production of IL-6, but not that of IL-12 and IL-23, in BMDCs stimulated with Pam 3 CSK 4 and with C. krusei mannan (Fig. 7C). IL-1β is also play a role in Th17 polarization 44 , and TLR-2 stimulation has participated in the production of IL-1β 45 . However, we found that TLR-2 blockade did not alter IL-1β production in BMDCs stimulated with C. krusei mannan (Fig. S7B). This result may be due to the inability of C. krusei mannan to induce IL-1β production in BMDCs (Fig. 2).
In summary, our findings indicate that DCs recognize C. krusei mannan via a TLR2/MyD88 dependent pathway, and this interaction results in apoptosis. Likewise, signal transduction through TLR2/MyD88 also regulates IL-12 and IL-6 production in DCs, which results in a shift toward Th1 and Th17 immunity, respectively (Fig. 8).

Discussion
In this study, we provide evidence that DCs respond differently to cell wall mannans from different Candida species. We investigated the molecular mechanisms underlying the induction of DC and T-cell immunity in response to C. krusei cell wall mannan. In contrast to previous reports on mannans from other species, our results show that C.krusei mannan activates DCs and consequently guides Th cell responses through the TLR2/MyD88 pathway.
We examined the responses of DCs to the cell wall mannans of four Candida species that are commonly found as opportunistic pathogens in immunocompromised patients: C. albicans, C. tropicalis, C. glabrata and C. krusei. Several Candida species are dimorphic fungi that can switch between the yeast and filamentous forms in response to environmental conditions such as high temperature, pH, and nutrition, whereas other Candida species exist in only one form 52,53 . C. albicans and C. tropicalis have three forms with distinct morphology; yeast cell, pseudohyphal cell and hyphal cell, while C. krusei yeast cells can produce pseudohyphae but not true hyphae. On the other hand, there is no yeast-to-hyphae transition in C. glabrata 53 . Because the yeast form exists in all Candida species and is the form that initially colonizes host tissues and can circulate in the bloodstream throughout the body 17,54,55 , we focused on comparison of cell wall mannans extracted from the yeast form only. It is likely that mannans from various forms may have different effects, a possibility that would be worth exploring in future studies.
Concurring with the similarity in mannan structure and composition of these two Candida species, we observed similar effects of C. albicans and C. tropicalis mannans on DCs (Figs 1 and 2), as well as similar T cell responses (Figs 6 and 7). In contrast, the cell wall mannan of C. glabrata contains small branches with low α-mannan content and one or two units of β-1,2-linked mannose residues 36,46 (Fig. S10). Meanwhile, C. krusei mannan consists of a long chain of α-1,2-linked mannose backbone with one or two α-1,6-linked mannose residues located in the middle of the chain, and a small number of short side chains of α-1,2-linked mannose residues 48,50 (Fig. S11).
Presumably because the mannans of C. glabrata and C. krusei differ from those of C. albicans and C. tropicalis, they produce different effects on DC maturation and cytokine production ( Figs. 1 and 2).
We found that C. krusei mannan induced DC maturation and massive cytokine production, which verified the DC state of high activation, and also led to increased apoptosis 34,35 . In contrast, C. albicans and C. tropicalis mannans had no effect on DC maturation, less effect on cytokine production, and did not induce DC apoptosis. C. glabrata mannan was a potent stimulus that induced DC maturation, but it failed to promote cytokine production or DC apoptosis (Fig. 3). Thus, high levels of DC activation and cytokine production may explain the increase in DC apoptosis in response to stimulation by C. krusei mannan. In addition, C. albicans, C. tropicalis and C. glabrata mannans, unlike C. krusei mannan, induced production of IFN-γ, which may act as an autocrine signal that enables the survival of BMDCs [59][60][61] .
Ample evidence supports the central role of CLRs and Syk in recognition of Candida mannans and protection against C. albicans infection 17,29 . Our results demonstrate that inhibition of Syk reduced the cell viability of unstimulated and β-glucan stimulated BMDCs; these results are consistent with those of previous studies [62][63][64] . Signal transduction via Syk is required for cell survival, as it leads to STAT3 phosphorylation, which mediates cell growth and differentiation 64 . Therefore, the mannans of C. albicans, C. tropicalis and C. glabrata may ligate to the receptors that activate Syk 17,19 , and this could partly explain the prevention of apoptosis in BMDCs (Fig. 3). In contrast, we found that C. krusei mannan mediated BMDC apoptosis via activation of MyD88 and TLR-2, which can transduce downstream signals through Fas-associated death domain protein (FADD) and caspase 8 65 . Of note, although DC apoptosis was increased in response to C. krusei mannan stimulation, the DCs retained their immunogenic functions that initiated T cell responses (Fig. 6).
C. albicans and C. tropicalis mannans induced high IFN-γ production in BMDCs (Fig. 2) and consequently skewed Th1 responses (Fig. 6). Both murine and human DCs are capable of producing IFN-γ upon the maturation 61,66 . IFN-γ is generally considered a key cytokine for Th1 induction since it epigenetically controls the expression and function of T-bet, a master transcription factor of Th1. IFN-γ also induces and maintains IL-12 receptor expression on T cells 67 . Thus, although a low level of IL-12 was detected in BMDCs stimulated with C. albicans and C. tropicalis mannans, the regulation of IL-12 receptor expression by IFN-γ probably helps Th1 polarization. In contrast to our finding, a previous study reported that ligation of C. albicans mannan to dectin-2 or mannose receptor promoted IL-6 and IL-23 production in APCs, and hence mediated Th17 response 20,68 . These contradicting results could be due to differences in the mannosyl composition of yeast cells grown under different conditions. The previous study cultured yeasts under specific conditions with limited carbon, low pH, and low temperature; the yeasts could not synthesize β-linked mannose 20 . In contrast, we grew the yeasts in an enriched media at 30 °C, which may have allowed them to synthesize the entire spectrum of αand β-linked mannose components. It is likely that the presence of β-linked mannose in C. albicans mannan reduces inflammatory cytokine production in DCs 69 . In addition, the intracellular signal through DC-SIGN, which is a CLR that also recognizes C. albicans N-linked mannan 70 , appears to inhibit dectin-1-dependent Th17 generation, instead of favoring Th1 response, upon Mycobacterium tuberculosis infection 71 . Therefore, our observations support the hypothesis that cell wall mannan may play a role in C. albicans immune evasion by shifting the Th1/Th17 balance 72,73 .
TLR2 also serve as immune sensors that recognize specific mannose residues of Candida species 16 , and they are essential for host defense against C. albicans infection 31,32 . TLR2 directly engages with phospholipomannans 74 , which contain phospholipid and β-1,2-linked mannose residues. In addition, recognition of β-linked mannose by galectin-3 is associated with TLR2 activation 56 . The structure and composition of C. krusei mannan differ from these previously identified carbohydrate ligands of TLR2, but our results clearly show that C. krusei mannan was recognized by TLR2. The distinct structure of C. krusei mannan, which contains the interlaced α-1,2and α-1,6-linked mannose residues 48,50 , possibly facilitates the specificity of the ligand-receptor binding.
Signal transduction via TLR2 and MyD88 in DCs has been shown to be involved in Th17 polarization 75,76 . Consistently, our findings demonstrate that activation of MyD88 in DCs by C. krusei mannan controlled Th1/ Th17 switching by virtue of the polarizing cytokines IL-12 and IL-6 ( Figs. 7 and 8). Although the roles of TLR2 and MyD88 in the C. krusei mannan-mediated Th17 response remain unclear, they may, at least in part, trigger IL-6, the key cytokine for Th17 generation 77,78 .
The in vitro C. krusei mannan-stimulated BMDCs exhibited the high production of both Th17 and Th1 polarizing cytokines, IL-6, IL-23, IFN-γ and IL-12 (Fig. 2), while the antigen-specific T cell responses to C. krusei mannan-activated DCs contributed toward the Th17 shift (Fig. 6). These results may be explained by the negative regulatory effect of IL-6 and IL-23 on Th1 differentiation. Previous report demonstrated that in the presence of IL-6 together with IFN-γ, CD4 T cells preferred Th17 differentiation. This response was resulted from the IL-6-mediated suppressor of cytokine signaling 1 (SOCS1) upregulation 79 , which consequently leaded to the interference in IFN-γ signaling in T cells. IL-23 also showed the suppressive effect on IFN-γ production from CD4 T cells, but via the inhibition of IL-12R signaling 80 . Therefore, it is possible that IL-6 and IL-23 interfere with IFN-γ and IL-12 signaling in T cells, and concomitantly regulate the switch from Th1 to Th17. Our results imply that the cell wall mannan of C. krusei may help the pathogen to evade host defenses by augmenting DC apoptosis, but may also activate Th17 responses. Previous studies have shown that while Th17 plays a crucial role in protective immunity against Candida infections 25 , it can also cause severe pathology as it is a major driver of hyperinflammation and tissue damage [81][82][83] . Furthermore, alteration of the Th17 phenotype and function may be affected by changes in the microenvironment at various stages and progression of the fungal infection 81,84 . Thus, it is still uncertain whether the Th17 response to C. krusei mannan stimulation has a positive or negative effect on the host, and this warrants further investigation.
In conclusion, our study implies that the structure and composition of cell wall mannans from different Candida species crucially influence the regulation of host protective immunity and fungal immune evasion through differential activation of DCs. In this study, we selected a single strain of each species to examine immune responses related to inter-species mannan variation. However, Candida displays intra-species diversity in cell wall mannans, which may lead to the differential response of DCs, as well as adaptive immune responses. Therefore, DC responses to the mannans of other Candida strains, including clinical isolates, should be further investigated to reveal the host-Candida interaction in-depth. A thorough comprehension of Candida recognition by PRRs and subsequent induction of adaptive immunity may bring important contributions to the development of new diagnostic and therapeutic applications.

Materials and Methods
Animals. Six to eight-week-old female BALB/c and C57BL/6 mice were purchased from National Laboratory Animal Center, Mahidol University) and were housed at Chulalongkorn University Lab Animal Center. All animal experiments were performed in accordance with the protocol approved by the Institutional Animal Use and Care Committee of Chulalongkorn University Lab Animal Center (Protocol number 1573005). BALB/c mice were used for the overall experiments and C57BL/6 mice were used for OT-II experiments.
Preparation of Candida cell wall mannan. C. krusei (ATCC 6258), C. albicans (ATCC 24433), C. glabrata (ATCC 2001) and C. tropicalis (ATCC 750) were selected as these strains are reference strains for quality control and antifungal drug susceptibility testing. All Candida were grown in YPD medium (Ajax Finechem, NSW, New Zealand) at 30 °C for 8 h with 200 rpm shaking. The fungal cultures were then diluted to OD 600 of 0.05, and cultured at 30 °C for 14 h with 145 rpm shaking. With this culture condition, all Candida species grow as budding yeast-like cells [85][86][87] .
Mannan was extracted from Candida cell wall following the previously described method 88 . Briefly, Candida yeast cell pellets (100 g) were resuspended in 250 ml of citrate buffer (0.02 M, pH 7.0) and autoclaved at 121 °C for 90 min. The supernatant was collected by centrifugation, and the residual sediment was re-extracted with the same procedures. An equal volume of Fehling's solution was added into the combined supernatants, and the mixture was stirred at 4 °C for overnight. The precipitate of mannan-copper complex was collected and dissolved in 6-8 ml of 3 N HCl, and the copper was then removed by washing in methanol-acetic acid (8:1, v/v) solution. After centrifugation, the mannan precipitate was collected and dissolved in sterile water. Subsequently, the mannan solution was dialyzed in sterile water for 48 h, and was further lyophilized for long-term storage. The amount of mannan was measured by phenol-sulfuric acid method 89 . All yeast culture and mannan preparation procedures were performed using endotoxin free water and containers. In addition, since the mannan extraction procedure required high temperature for a long period, nucleic acid and proteins are denatured and degraded. Therefore, there would be very less or no contamination of other PAMPs such as RNA and DNA.
Generation and stimulation of BMDCs. BMDCs were generated following the previously described protocol 90 . Briefly, bone marrow cells were collected from femurs and tibias. The cells (1 × 10 6 cells) were cultured in 24-well plates in 1 ml of RPMI 1640 (GIBCO, ThermoFisher Scientific, NY, USA) supplemented with 10% heat-inactivated fetal bovine serum (GIBCO), 0.2 mM Glutamax (GIBCO), 100 U/ml penicillin and 100 mg/ml streptomycin (HyClone, UT, USA), 10 ng/ml recombinant murine GM-CSF and 10 ng/ml recombinant murine IL-4 (Peprotech, NJ, USA). The cell culture were incubated under humidified atmosphere of 5% CO 2 at 37 °C for 7 days. Half of culture media was replaced every 2 days. On day 7, BMDCs were stimulated with mannan at the concentrations of 12.5, 25 and 50 μg/ml for 24-48 h. After stimulation, the culture supernatant was collected for cytokine measurement and the cells were harvested for flow cytometric analyses. Unstimulated BMDCs were used as the negative control, and BMDCs stimulated with 0.5 μg/ml of LPS (Sigma-Aldrich, MO, USA) were used as the positive control.
Assessment of cell viability. BMDCs (1 × 10 5 cells/well) were stimulated with Candida mannan in 96-well plates for 48 h. Next, the BMDCs were further incubated in the media containing 0.5 mg/ml of MTT (Life technologies, ThermoFisher Scientific, OR, USA) at 37 °C for 3 h. After removal of the culture media, the incorporated formazan crystals in viable cells were solubilized in dimethyl sulfoxide (DMSO; AMRESCO, OH, USA). The absorbance was measured using a microplate reader (Synergy H1, BioTek, VT, USA) at 570 nm. The percent of cell viability was calculated by normalization with the negative control. For  In vivo immunization and ex vivo re-stimulation assay. Mice were subcutaneously injection in the scruff of the neck with the mixture of Candida mannan (50 μg mannan per 1 g of body weight) and OVA (30 μg per animal; Sigma Aldrich) in 200 μl PBS at day 0 and day 7. On day 14, the cervical, axillary and brachial LNs were excised, digested with collagenase type IV (Sigma Aldrich) at 37°C for 30 min, and subsequently incubated with DNase I (Sigma Aldrich) at room temperature for 10 min. The cells were washed, and resuspended in RPMI 1640 supplemented with 10% heat-inactivated FBS, 0.2 mM Glutamax, 100 U/ml penicillin and 100 mg/ml streptomycin, and 55 μM 2ME (GIBCO). For the negative control, mice were immunized with OVA in PBS.

Treatment with inhibitors and TLR antagonists.
For ex vivo re-stimulation assay, LN cells (4 × 10 6 cells) were cultured in 24-well plates in 1 ml media at the presence of 250 μg/ml OVA. The culture supernatant was collected at 48 h and 72 h after OVA stimulation.
In vitro OT-II T cell stimulation. T cells from spleens of OT-II mice were enriched by immunomagnetic beads (Pan T Cell Isolation Kit II, mouse; Miltenyi Biotec, CA, USA). BMDCs were stimulated with Candida mannan for 12 h and pulsed with 500 μg/ml whole OVA protein overnight, and washed twice with the media to remove the mannan and the free OVA. Then, OT-II T cells were co-cultured with the OVA-pulsed BMDCs at T:DC ratio of 10:1. The supernatant was collected at 48 h and 72 h for cytokine determination.
In the MyD88 inhibitor and TLR-2 blockade experiment, BMDCs were pre-treated with control peptide, MyD88, control IgG or anti-mouse TLR2 mAbs as described above. The cells were then stimulated with 25 μg/ml of C. albicans or C. krusei mannan for 12 h, and pulsed with 500 μg/ml whole OVA protein overnight. Thereafter, all cells were washed twice with RPMI media to remove the inhibitor, Candida mannan and the free OVA. Then, OT-II T cells were co-cultured with the OVA-pulsed BMDCs.

Measurement of cytokines.
Cytokines in the supernatant collected from BMDC, LN cell and OT-II T cell cultures were quantitated by standard sandwich ELISA using commercially available paired antibody sets for IL-1β, IL-4, IL-6, IL-10, IL-12, IL-17, IL-23, IFN-γ and TNF-α. The procedures were performed according to the manufacturer's instructions (Biolegend and eBioscience).

Statistical Analysis.
All data values were expressed as mean ± SD, and the sample size was indicated in each figure legend. The statistical analysis was performed using one-way ANOVA with post-hoc Turkey HSD test for the comparison of 3-5 groups, and using Student's T test for the comparison between 2 groups. Values of p < 0.05 were considered significant.