Dectin-2 Recognizes Mannosylated O-antigens of Human Opportunistic Pathogens and Augments Lipopolysaccharide Activation of Myeloid Cells

LPS consists of a relatively conserved region of lipid A and core oligosaccharide and a highly variable region of O-antigen polysaccharide. Whereas lipid A is known to bind to the Toll-like receptor 4 (TLR4)-myeloid differentiation factor 2 (MD2) complex, the role of the O-antigen remains unclear. Here we report a novel molecular interaction between dendritic cell-associated C-type lectin-2 (Dectin-2) and mannosylated O-antigen found in a human opportunistic pathogen, Hafnia alvei PCM 1223, which has a repeating unit of [-Man-α1,3-Man-α1,2-Man-α1,2-Man-α1,2-Man-α1,3-]. H. alvei LPS induced higher levels of TNFα and IL-10 from mouse bone marrow-derived dendritic cells (BM-DCs), when compared with Salmonella enterica O66 LPS, which has a repeat of [-Gal-α1,6-Gal-α1,4-[Glc-β1,3]GalNAc-α1,3-GalNAc-β1,3-]. In a cell-based reporter assay, Dectin-2 was shown to recognize H. alvei LPS. This binding was inhibited by mannosidase treatment of H. alvei LPS and by mutations in the carbohydrate-binding domain of Dectin-2, demonstrating that H. alvei LPS is a novel glycan ligand of Dectin-2. The enhanced cytokine production by H. alvei LPS was Dectin-2-dependent, because Dectin-2 knock-out BM-DCs failed to do so. This receptor cross-talk between Dectin-2 and TLR4 involved events including spleen tyrosine kinase (Syk) activation and receptor juxtaposition. Furthermore, another mannosylated LPS from Escherichia coli O9a also bound to Dectin-2 and augmented TLR4 activation of BM-DCs. Taken together, these data indicate that mannosylated O-antigens from several Gram-negative bacteria augment TLR4 responses through interaction with Dectin-2.

In this study, we investigated the contribution of the ␣-linked mannosylated O-antigen in the LPS activation of myeloid cells. We compared DC response and Dectin-2 binding to the mannosylated LPS (Man-LPS) from H. alvei PCM 1223 and E. coli O9a with the LPS from Salmonella enterica O66 or K. pneumonia O1, which has the galactosylated O-antigen (Gal-LPS) (Fig.  1A). We observed binding between Man-LPS and Dectin-2, which led to augmentation of TLR4 response in mouse DCs and human monocytes. These results demonstrate a novel role of mannosylated O-antigen in activation of TLR4 in myeloid cells.  (36). The core oligosaccharide and lipid A of these two LPSs are relatively conserved (37)(38)(39)(40). TLR4 activation by these two types of LPS was first measured using the TLR4-MD2expressing HEK293 reporter cells, with Man-LPS being 4-fold more potent as compared with Gal-LPS (Fig. 1B). Based on this result, we standardized TLR4 activation by using a 4-fold higher concentration of Gal-LPS compared with Man-LPS in the rest of the study. Under these conditions, Man-LPS induced 2-fold more TNF␣ and IL-10 from mouse BM-DCs than Gal-LPS (Fig.  1C). Man and Gal-LPS induced the co-stimulatory molecule CD80 and mouse MHC class II molecule I-A b to a similar extent (data not shown).

Man-LPS Produced a Higher Level of TNF␣ and IL-10 from Bone Marrow-derived DCs (BM-DCs) than Gal-LPS-To
Man-LPS Is a Novel Glycan Ligand of Dectin-2-Because O-antigen of Man-LPS consists of ␣-linked mannose, a glycan ligand of Dectin-2, we tested Dectin-2 binding to the purified LPS in a cell-based reporter assay, in which the lectin-glycan interaction is monitored as ␤-galactosidase expression (41). The Dectin-2 BWZ cells were cultured in a 96-well plate coated with Man and Gal-LPS. In this assay, Dectin-2 bound to Man-LPS but not to Gal-LPS ( Fig. 2A). No binding was observed between mock BWZ cells and LPS, indicating specific binding of Dectin-2 to Man-LPS ( Fig. 2A). In addition, plant-derived galactan, ␤-linked mannan (Fig. 2B), and LPS from K. pneumoniae O1 (Fig. 2C), which has a homopolymeric Gal O-antigen (42), failed to bind to Dectin-2, confirming that Dectin-2 binding to Man-LPS is sugar composition-and linkage-specific alvei PCM 1223 has a mannosylated repeating unit, whereas Gal-LPS from S. enterica O66 has a galactosylated repeat. B, HEK293 cells stably transfected with TLR4-MD2 were cultured in the presence of LPS. The TLR4 activation was monitored by measuring alkaline phosphatase activity using the substrate. C, mouse BM-DCs were stimulated with 1 g/ml Man-LPS or 4 g/ml Gal-LPS for 7 h. The amount of TNF␣ and IL-10 in the culture supernatant was analyzed by ELISA. Data are representative of three independent experiments with similar results. Error bars, S.D. Statistical analyses were performed by one-way ANOVA followed by Tukey's test. ***, p Ͻ 0.001. rather than nonspecific binding to homopolymeric carbohydrate polymer. Furthermore, the lipid A isolated from Man-LPS failed to bind to Dectin-2 (Fig. 2D), and treatment of Man-LPS with ␣-mannosidase inhibited the binding (Fig. 2E), suggesting that binding was mediated by the mannosylated O-antigen of Man-LPS. To assess whether the binding is through the carbohydrate-recognition domain of Dectin-2, we compared the binding of WT Dectin-2 and the QPD mutant that no longer recognizes mannose (17,43). As shown in Fig. 2F, the binding was significantly reduced by the mutations. These results demonstrate that Dectin-2 recognizes the ␣-linked mannosylated O-antigen of H. alvei LPS. Because mannosylated O-antigen is found in other Gram-negative bacteria, including E. coli O9a (44), we tested whether Dectin-2 recognizes mannosylated O-antigen from E. coli O9a. We found that E. coli O9a LPS bound to Dectin-2, whereas the rough mutant LPS, which lacks the O-antigen (45), failed (Fig. 2G). We also tested the binding of Dectin-2 to H. alvei in the reporter assay. Dectin-2 bound to paraformaldehyde (PFA)-fixed H. alvei, whereas the QPD mutant did not (Fig. 2H), suggesting the role of Dectin-2 as a recognition receptor for Gram-negative bacteria with ␣-linked mannosylated O-antigens.
Man-LPS Activation Involves a Synergy between Dectin-2 and TLR4 -To assess the involvement of Dectin-2 in Man-LPS activation of immune cells, we generated BM-DCs from Dectin-2 KO mice (Fig. 3A). In contrast to WT BM-DCs, Dectin-2 KO BM-DCs were unable to enhance TNF␣ and IL-10 production in response to Man-LPS (Fig. 3B), indicating that Dectin-2 augments TLR4 activation by Man-LPS. This was reproducible when we used another Gal-LPS from K. pneumoniae O1 (data not shown). Of note, in the TLR4 KO BM-DCs, neither Man nor Gal-LPS induced the cytokine production ( Fig. 3B), suggesting that the mannosylated O-antigen is not sufficient to activate Dectin-2 in the absence of TLR4. Similarly, Dectin-2dependent enhancement in TNF␣ production by BM-DCs was seen in response to the WT E. coli O9a LPS but not to the rough LPS (Fig. 3C). IL-10 response showed a similar tendency, but this was not statistically significant between WT and rough LPS (Fig. 3C). The TLR4 reactivity of WT and rough LPS was indistinguishable (data not shown). We also assessed the contribution of Dectin-2 to the DC response toward H. alvei. IL-10 production in response to H. alvei was Dectin-2-dependent, suggesting a regulatory role of Dectin-2 in the recognition of H. alvei (Fig. 3D). Although TNF␣ production in response to Man-LPS was enhanced by Dectin-2 ( Fig. 3B), the TNF␣ response to H. alvei was similar between WT and Dectin-2 KO BM-DCs (Fig. 3D), suggesting alternative molecular mechanisms leading to TNF␣ production, such as TLR2 that recognizes bacterial cell wall components (46).
The Receptor Synergy Is Syk-dependent and Requires Receptor Juxtaposition-To investigate the intracellular signaling events involved in this process, we assessed the impact of Man-LPS on Syk, a key molecule in the Dectin-2 pathway (15,16). Syk was found to be phosphorylated upon stimulation of BM-DCs by Man-LPS but not Gal-LPS (Fig. 4A). Furthermore, treatment of BM-DCs with the Syk inhibitor R406 abrogated the augmented cytokine production in response to Man-LPS (Fig. 4B), demonstrating Syk-dependent synergy between Dectin-2 and TLR4. Next, we assessed the impact of the Syk activation by Man-LPS on the activation of NF-B and MAPK pathways, hallmark of TLR4 activation (3). The phosphorylation of p38 and degradation of IB was indistinguishable between Man and Gal-LPS, respectively, suggesting that other signaling pathways are modified by Syk activation through Dectin-2 (Fig. 4C).
Because Man-LPS has the binding epitopes for both Dectin-2 and TLR4, we hypothesized that receptor juxtaposition by Man-LPS is the mechanism underpinning the synergy. To test this hypothesis, BM-DCs were stimulated with Gal-LPS as a TLR4 ligand in the presence of yeast ␣-linked mannan, a known Dectin-2 ligand. As shown in Fig. 5, the addition of yeast mannan was not sufficient to enhance cytokine production, compared with the Gal-LPS, demonstrating that receptor juxtaposition is required to achieve synergy.
Human Monocytes Recapitulate the Enhanced Cytokine Production in Response to Man-LPS-Man-LPS activation was also tested on human myeloid cells expressing Dectin-2. We found that human peripheral blood monocytes expressed Dectin-2 at a high level, and blood DCs and monocyte-derived DCs (Mo-DCs) expressed at a negligible level (Fig. 6A), which is consistent with previous reports analyzing human Dectin-2 mRNA

Dectin-2 Augments TLR4 Signaling
expression among human immune cells (11,12). Human monocytes produced higher levels of TNF␣ and IL-10 in response to Man-LPS compared with Gal-LPS (Fig. 6B). The anti-Dectin-2 antibody (Ab), however, failed to block the enhanced cytokine production (data not shown), implying the potential involvement of other lectins.

Discussion
In line with our findings, recent reports show interactions between mammalian lectins and O-antigens. Several strains of H. alvei are targeted by Ficolin-3, a complement-associated soluble lectin (47,48). Langerin, a C-type lectin specific to ␣-linked mannose, is suggested to recognize the internal Man-␣1,2-Man repeat found in O-antigens from E. coli O106 and Shigella boydii B10 in the pathogen glycan array (49), implying that C-type lectins are capable of interacting with the internal glycan epitopes. At this moment, it is unclear whether the binding of Dectin-2 to the mannosylated O-antigen is mediated by the terminal mannose residue at the non-reducing end, the internal ␣-linked mannose repeats, or both.
Our findings have identified novel carbohydrate ligands of Dectin-2 and provided a deeper understanding in host-microbe interactions mediated by Dectin-2. Previous studies demonstrate that Dectin-2 plays a key role in fungal and mycobacterial infection and house dust allergy (14,16,17,50). In this report, we have revealed a group of Gram-negative bacteria recognized by Dectin-2. LPS from H. alvei PCM 1223 and E. coli O9a bound to Dectin-2 and enhanced TLR4 responses of BM-DCs in a Dectin-2-dependent manner, suggesting a novel role of Dectin-2 in the interaction between host and Gram-negative bacteria bearing ␣-linked mannosylated O-antigen. Of note, H. alvei LPS was more potent than E. coli O9a LPS in Dectin-2 engagement in BM-DCs (Fig. 3, B and C); this may be due to the   AUGUST 19, 2016 • VOLUME 291 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 17633
different O-chain length because the sugar structure is identical between these two bacteria (44).
Our findings that Dectin-2 interacts with TLR4 upon Man-LPS stimulation strengthen the proposed model that lectins are capable of regulating TLR pathways in various ways. For instance, stimulation of DCs and macrophages with ␤-glucan in the presence of various TLR ligands allowed enhancement of TNF␣ and IL-10 production (51-53). Co-stimulation of DCs with DC-SIGN and TLR ligands selectively enhances IL-10 production (54,55). Although these studies clearly demonstrate the cross-talk between C-type lectins and TLRs, the molecular mechanisms underpinning such receptor cross-talk remain elusive. One potential mechanism is receptor juxtaposition, as previously proposed for the ITAM-coupled lectin Siglec-H; positioning Siglec-H in close proximity to TLR9 in the endosome would enhance the TLR9 activation (56). Here we showed that such synergy was observed between Dectin-2 and TLR4. Whether this is a common strategy for the modulation of TLR functions by membrane-bound lectins remains to be demonstrated.
We have identified Syk as a key molecule for cross-talk between Dectin-2 and TLR4. Syk has been proposed as a regulator of TLR4 signaling. Several studies have reported that in DCs and macrophages, Syk gets phosphorylated upon LPS stimulation (57), and Syk deficiency results in enhanced TNF␣ and reduced IL-10 (58). Of note, most of the studies employed LPS from E. coli O111:B4 (59). In our study, because Gal-LPS failed to induce phosphorylation of Syk, it is likely that O-antigen structure influences Syk activation. Thus, to address the role of Syk in TLR4 signaling, it would be essential to test whether reported Syk activation by E. coli O111 LPS involves Dectin-2 or other lectins.
Although our findings have revealed a novel function of mannosylated O-antigen in LPS activation of innate immune cells (Fig. 7), the role of core oligosaccharides in the TLR4 activation remains elusive. In this regard, SIGNR1 is reported to recognize the core oligosaccharide of E. coli LPS and augment cytokine responses (60). Further studies of lectin recognition of both highly variable O-antigen and the conserved core oligosaccharide are warranted to gain a better understanding of LPS recognition by innate immune cells.

Experimental Procedures
Mice-C57BL/6J (WT), TLR4 KO (a gift from Dr. J. S. Frick, University of Tuebingen), and Dectin-2 KO mice were maintained in the specific pathogen-free animal facilities at the University of East Anglia, University of Tuebingen, and Chiba University, respectively. Animal use in this study was in accordance with United Kingdom Home Office guidelines, the Regierungspraesidium Tuebingen, and the ethics committee of Chiba University.
Cells and Bacteria-BWZ.36 cells harboring IL-2-driven ␤-galactosidase cassette (63) and the retrovirus-packaging cell line Plat-E were obtained from Dr. N. Shastri (University of California, Berkeley, CA) and Dr. T. Kitamura (University of Tokyo), respectively, and maintained as described before (64). Flow Cytometry-Cells were washed with Hanks' balanced saline solution (Lonza) containing 0.1% BSA, 2 mM EDTA (FACS buffer); blocked with anti-mouse Fc block Ab (Biolegend) for 5 min at 25°C; and stained with the Abs for 30 min at 4°C. Stained cells were washed once with FACS buffer and analyzed by Fortessa (BD Biosciences). For dead cell exclusion, propidium iodide (PI) was added to the sample at a final concentration of 0.33 g/ml before the analysis. Acquired data were analyzed with FlowJo (Tree Star).
TLR4 Reporter Assay-The TLR4 reporter assay was performed using the HEK-Blue human TLR4 reporter cells that produce alkaline phosphatase in response to LPS, according to the manufacturer's instructions (InvivoGen). Briefly, the TLR4-HEK293 cells were cultured in the HEK-Blue detection medium containing the substrate for alkaline phosphatase in the presence of the LPS for 20 h. After incubation, the absorbance at 620 nm was measured.
Establishment of Mouse Dectin-2 Reporter Cells-The Dectin-2 reporter cells were established as described previously (64). Briefly, the extracellular domain of mouse Dectin-2 (Gln 42 -Leu 209 ) was cloned into the retrovirus vector pMXs-IRES-EGFP-Ly49A-CD3 harboring the transmembrane region of the mouse Ly49A and the cytoplasmic domain of the mouse CD3 (64). The pMXs-IRES-EGFP-Dectin-2-Ly49A-CD3 vector and the parental vector were used to establish Dectin-2 and mock BWZ cells, respectively, for the retrovirus transduction using Plat-E cells (65). To establish reporter cells expressing the carbohydrate-binding incompetent mutant of Dectin-2 (Dectin-2 QPD ), two missense mutations (G502C and A508G) in the mannose recognition domain were introduced, which results in amino acid substitutions E168Q and N170D. The DNA fragment encoding the extracellular domain of Dectin-2 with the two missense mutations was synthesized (GenScript) and cloned into the pMXs-IRES-EGFP-Ly49A-CD3 vector and used as described above.
Dectin-2 Reporter Assay-A reporter assay was performed as described previously (64). Briefly, the 96-well flat-bottom ELISA plate (MaxiSorp, Thermo Scientific Nunc) was coated with 100 l containing 40 ng of LPS or polysaccharides in 100 mM sodium bicarbonate buffer, pH 9.5, for 16 h at 4°C. Dectin-2 or mock BWZ cells were then cultured in the well, and ␤-galactosidase activity was monitored as described previously (64). For the mannosidase treatment, a 96-well plate was coated with 1.6 ng of Man-LPS. After discarding the solution, wells were washed once with 200 l of PBS and blocked with 100 l of 4% BSA in PBS for 1 h at 25°C. Blocked wells were washed, and 50 units of ␣1-2,3 mannosidase (New England BioLabs) was added to the wells and incubated at 37°C for 14 h. The reporter cells were added to test the binding. For the reporter assay with bacteria, 1.0 ϫ 10 6 of PFA-fixed H. alvei were immobilized on the 96-well plate as described above.
Generation of Anti-Dectin-2 mAb-The anti-Dectin-2 mAb was generated as described previously (66). Briefly, two female Lewis rats (Japan SLC) were immunized with Dectin-2-BWZ cells emulsified with complete Freund's adjuvant (Difco). Following two injections of the cells emulsified with incomplete Freund's adjuvant (Difco Laboratories), the immunized rats were sacrificed, and common iliac lymph nodes were harvested to generate hybridomas as described previously (66). The established hybridomas were screened by the reporter assay described previously (67). The animal experiments were performed in accordance with the institutional animal ethics committee at the University of Tokyo. The established hybridoma clone 2B4 produced anti-mouse Dectin-2 mAb, a rat IgG2a, chain determined by flow cytometry (66). Monoclonal Ab 2B4 was purified from the culture supernatant and labeled with Alexa Fluor-647 (Life Technologies, Inc.) as described previously (66).
Stimulation of BM-DCs-BM-DCs (1 ϫ 10 5 cells) were cultured in a 96-well plate using the RPMI medium described above without GM-CSF. Man and Gal-LPS were added to the culture at a final concentration of 1 or 4 g/ml, respectively, and incubated for 7 h at 37°C. The concentration was standardized for their reactivity toward TLR4 (Fig. 1B). For the cytokine analysis, culture supernatant was harvested, and TNF␣ and IL-10 were measured by ELISA. When the Syk inhibitor R406 was used, it was added to the cell culture at a final concentration of 1 M and incubated for 30 min before the addition of LPS. For CD80 and I-A b expression analysis, stimulated BM-DCs were harvested and analyzed by flow cytometry as described above.
Western Blotting of Intracellular Proteins-One million BM-DCs were stimulated with 1 or 4 g/ml Man-and Gal-LPS, respectively, for the indicated time period. Cells were then processed as described previously with adaptation of lysis buffer volume to 75 l (69). Cell lysate equivalent to two hundred thousand cells (15 l) was subjected to SDS-PAGE using a 4 -15% gradient TGX minigel (Bio-Rad) for 90 min at 100 V. Proteins were transferred onto nitrocellulose membrane (GE Healthcare) at 100 V for 30 min. The membranes were blocked with 5% nonfat milk (Lonza) in PBS containing 0.05% Tween 20 (PBS-T) for 1 h at 25°C. The blocked membranes were washed four times by incubating in PBS-T for 5 min each. Membranes were then incubated with the primary antibodies in PBS containing 1% bovine serum albumin for 1 h at 25°C at a dilution of 1:1000 for Syk, phospho-Syk, IB, and ␤-actin and 1:5000 for p38 and phospho-p38. Membranes were washed as above and probed with anti-rabbit IgG conjugated with horseradish peroxidase (Cell Signaling) in 5% nonfat milk in PBS-T at a dilution of 1:3000 for 1 h at 25°C. Membranes were washed as above and incubated with the ECL detection reagent (GE Healthcare) and imaged using Fluorochem E (ProteinSimple).
Human Blood Monocytes-Human peripheral blood was obtained from the hemochromatosis patients undergoing a therapeutic venesection at the Norfolk and Norwich University Hospital (Norwich, UK). Human blood monocytes were isolated as described previously (70). The Dectin-2 expression on the monocytes was analyzed by flow cytometry. The freshly isolated monocytes were stimulated with 1 ng/ml Man and 4 ng/ml Gal-LPS for 20 h at 37°C. Human TNF␣ and IL-10 production was monitored by ELISA. Mo-DCs were generated by in vitro culture of human blood monocytes as described previously (70). Blood collection in this study was approved by the ethics committee at the Faculty of Medicine and Health Sciences at the University of East Anglia (reference number 2013/ 2014-14HT). Importantly, the monocytes in the patients with iron overload have been shown to respond to LPS, although the response was lower when compared with that of monocytes from healthy donors (71), justifying the use of monocytes from the patients to study human Dectin-2 function.
Statistical Analysis-Student's t test and one-way ANOVA followed by Tukey's test were used for statistical analysis on Prism software (GraphPad). p Ͻ 0.05 was considered as statistically significant.