Dectin-1 intracellular domain determines species-specific ligand spectrum by modulating receptor sensitivity

C-type lectin receptors (CLRs) comprise a large family of immunoreceptors that recognize polysaccharide ligands exposed on pathogen surfaces and are conserved among mammals. However, interspecies differences in their ligand spectrums are not fully understood. Dectin-1 is a well-characterized CLR that recognizes β-glucan. We report here that seaweed-derived fucan activates cells expressing human Dectin-1 but not mouse Dectin-1. Low-valency β-glucan components within fucan appeared to be responsible for this activation, as the ligand activity was eliminated by β-glucanase treatment. The low-valency β-glucan laminarin also acted as an agonist for human Dectin-1 but not for mouse Dectin-1, whereas the high-valency β-glucan curdlan activated both human and mouse Dectin-1. Reciprocal mutagenesis analysis revealed that the ligand-binding domain of human Dectin-1 does not determine its unique sensitivity to low-valency β-glucan. Rather, we found that its intracellular domain renders human Dectin-1 reactive to low-valency β-glucan ligand. Substitution with two amino acids, Glu2 and Pro5, located in the human Dectin-1 intracellular domain was sufficient to confer sensitivity to low-valency β-glucan in mouse Dectin-1. Conversely, the introduction of mouse-specific amino acids, Lys2 and Ser5, to human Dectin-1 reduced the reactivity to low-valency β-glucan. Indeed, low-valency ligands induced a set of proinflammatory genes in human but not mouse dendritic cells. These results suggest that the intracellular domain, not ligand-binding domain, of Dectin-1 determines the species-specific ligand profile.

Our bodies are continuously exposed to and infected by various types of pathogens, most of which are directly recognized by pattern recognition receptors such as Toll-like receptors, RIG-I-like receptors, or NOD-like receptors on host cells (1,2). An additional fourth member of pattern recognition receptors is the emerging C-type lectin receptors (CLRs) 3 that senses pathogens or damaged tissues to trigger innate immune responses (3).
Within this family, Dectin-1 is the first immunoreceptor tyrosine-based activation motif (ITAM)-coupled CLR identified and recognizes ␤-glucans present in the cell wall of fungi (4 -6). Dectin-1 is a type II transmembrane protein expressed by myeloid cells and consists of an extracellular carbohydrate recognition domain (CRD) and a cytoplasmic domain containing an ITAM-like motif (hemITAM). Upon recognition of mul-tivalent␤-glucanviaitsCRD,Dectin-1multimerizesandisphosphorylated at a tyrosine residue in the hemITAM, providing a binding site for the Syk kinase. The recruited Syk then activates the CARD9-Bcl10-MALT1 and NF-B pathways to induce inflammatory cytokines, co-stimulatory molecules, and dendritic cell maturation, which promotes Th1 and Th17 responses to orchestrate immunity to pathogens (7,8). The CRD and hemITAM regions of Dectin-1 are conserved among mammals (5,9), suggesting the importance of this CLR for promoting acquired immune responses over a wide variety of species. Hence, Dectin-1 agonists hold potential as vaccine adjuvants that may facilitate protective immune responses against pathogens or cancer in mouse models and human patients. Given this potential, it is important to characterize in detail the function of human Dectin-1 in comparison with the evidences accumulating for the more extensively studied mouse Dectin-1 (10,11).
The inconsistency within studies on the "low-valency" Dectin-1 ligand may be partly due to interspecies difference of Dectin-1. In addition, the function of the cytoplasmic domain, other than its identity as a hemITAM, is not well-characterized.
In this study, we found that low-valency ␤-glucan can activate cells expressing human Dectin-1, but not mouse Dectin-1. Reciprocal mutagenesis studies revealed that the intracellular domain of human Dectin-1 confers this activity. Furthermore, we found that two intracellular amino acids, which are conserved in primates, play a critical role for enhancing the sensitivity of Dectin-1 independently of the hemITAM.

Non-CRD region of hDectin-1 confers reactivity to low-valency ␤-glucan
As the CRD of Dectin-1 mediates direct ligand binding (16), we initially suspected that the slightly different amino acid sequences within the CRDs comparing hDectin-1 and mDectin-1 might determine the reactivity to soluble ␤-glucan. To address this possibility, we generated a chimeric mDectin-1 protein harboring the CRD from hDectin-1 (mDectin-1 hCRD chimera) (Fig. 4A). Contrary to our initial assumption, the reporter cells expressing the mDectin-1 hCRD chimera were not activated by laminarin similar to mDectin-1-bearing cells (Fig.  4B, mD1 hCRD ), although they showed substantial activity upon stimulation with high-valency curdlan (Fig. 4C). These results suggest that the direct ligand-binding domain, CRD, is not responsible for determining laminarin sensitivity to hDectin-1. We therefore created a hDectin-1 chimera harboring the CRD 2B4 NFAT-GFP cells expressing the indicated receptors and signaling components were left untreated (None) or stimulated with 30 g/ml of fucan (Fucan). m, mouse; r, rat; g, guinea pig; h, human. The detail of receptor constructs and co-transfected genes are described under "Experimental procedures." The expression of GFP was analyzed by flow cytometry. All data are presented as the mean Ϯ S.D., and representative results from two independent experiments with similar results are shown.

Opposite effect of low-valency ␤-glucan on mDectin-1 and hDectin-1
Given the observation that mDectin-1 binds low-valency ␤-glucan but is unable to deliver activating signaling, we next examined whether these ligands act as antagonists for mDectin-1. To assess this, we added a graded amount of fucan or laminarin to mDectin-1 cells in the presence of suboptimal concentrations of a high-valency ligand zymosan. Indeed, both ligands suppressed mDectin-1-induced reporter activation (Fig. 5A), in sharp contrast to their agonistic effects on hDectin-1 (Fig. 5B). These results indicate that low-valency ␤-glucan acts oppositely on mDectin-1 and hDectin-1.

Cytoplasmic N-terminal region of hDectin-1 determines its sensitivity to low-valency ␤-glucan
As the stalk region of hDectin-1 is dispensable for the reactivity to fucan and laminarin (supplemental Fig. S1A), we next focused on the cytoplasmic domain of hDectin-1. To this end, we generated a series of chimeric mDectin-1 proteins in which each region of the cytoplasmic domain was replaced with the corresponding region of hDectin-1 (Fig. 6A). In contrast to fulllength (WT) mDectin-1, mDectin-1 possessing 30 amino acids of the human N-terminal region (mDectin-1 hN30 ) normally responded to laminarin (Fig. 6, B and C). Further chimeric analysis revealed that as little as a 10-amino acid sequence at the N terminus of hDectin-1 (mDectin-1 hN10 ) was sufficient to confer reactivity to low-valency ␤-glucan to mDectin-1 (Fig. 6B).

Low-valency Dectin-1 ligands activate human DCs but not murine DCs
To confirm the observed hDectin-1-specific phenomenon using primary cells, we finally compared the reactivity of human and murine myeloid cells to low-valency Dectin-1 ligand in a non-biased manner by stimulating human monocyte-derived dendritic cells (hMoDCs) and mouse bone marrow-derived dendritic cells (mBMDCs) with laminarin.

Intracellular non-ITAM sequence sensitizes human Dectin-1
Laminarin potently activated hMoDCs to induce a set of inflammatory genes including IL1B, IL1A, and CLEC4E, which are reportedly induced by CLR-mediated signaling (17,18). In contrast, the majority of mouse orthologues of genes upregulated in hMoDCs did not show substantial induction in mBMDCs (Fig. 9). Note that mBMDCs constitutively expressed Dectin-1 (see GSE98814 and GSE98825), and responded normally to other stimuli, such as LPS (supplemental Fig. S5). Collectively, these data support the idea that hDectin-1, but not mDectin-1, is an activating receptor for low-valency ␤-glucan.

Discussion
In this study, we report that hDectin-1 responds to low-valency ␤-glucan and its cytoplasmic region is critical for this function.
Within the cytoplasmic N terminus, we show that Glu 2 of hDectin-1 is one of the critical residues controlling the sensitivity of hDectin-1 to low-valency ␤-glucan, as a substitution of a mouse-specific residue in this position (hDectin-1 E2K ) impaired this activity. It appears that it is important that the residue at position 2 be a "non-Lys" residue, as substitutions to other residues did not reduce the activity (hDectin-1 E2A or hDectin-1 E2D ) (supplemental Fig. S6). In line with these observations, Lys residues of mDectin-1 are reported to undergo ubiquitination, which results in the degradation and desensitization of mDectin-1 upon ligand binding (19). Thus, the inactivation of the Lys 2 residue might be one of the reasons that sensitize hDectin-1.
Meanwhile, the precise role of hDectin-1 Pro 5 , which is conserved in primates, is not yet clear. The reciprocal single substitution (mDectin-1 S5P and hDectin-1 P5S ) had no impact ( Fig.  7 and data not shown), whereas mDectin-1 carrying the double mutation (mDectin-1 K2E/S5P ) was active, suggesting that Pro 5 renders hDectin-1 sensitive, albeit it is not sufficient in and of itself. One potential explanation is that mouse Ser 5 , which is conserved in most non-primates mammals (supplemental Fig.  S7), may act to reduce the receptor signaling by promoting protein modification as Ser phosphorylation is linked to ubiquitination in several signaling molecules (20 -22). Alternatively, Ser 5 may interfere with the function of hemITAM through phosphorylation or other modifications. Indeed, phosphorylation of the Ser residue in the cytoplasmic region of FcR␥ and Ig␣ is reported to inhibit tyrosine phosphorylation of their own ITAMs, which then inhibits downstream signaling (23,24). More detailed analysis is needed to clarify whether Pro 5 (human) cancels the negative function of Ser 5 (mouse), or, alternatively, actively promotes downstream signaling.
From the analysis of the phylogenic comparisons, one could speculate that "elimination" of Lys 2 and Ser 5 by substitution (human) or truncation (rat) (Fig. 8A and supplemental Fig. S7) might be a common strategy to increase the sensitivity of Dectin-1 to low-valency ␤-glucan during evolution. In support of this idea, the 5Ј-UTR of the mRNA for rDectin-1 contains a sequence corresponding to the Ser 5 found in mice (Fig. 8A), suggesting that Ser 5 is preserved as a relic in rDectin-1, which has been inactivated by the introduction of a downstream start codon, as speculated in other immune receptors (25).
Despite the different sensitivities to low-valency ␤-glucan, the various Dectin-1 mutants used in this study retained substantial reactivities to curdlan. This suggests that Dectin-1, like other ITAM-coupled receptors such as the TCR and BCR, has the capacity to sense the quality of ligand. Recently, another CLR, mMincle and hMincle, recognize different ligands and induce distinct responses (26,27). It is tempting to speculate that the CLR family members might have modulated their sensitivities and ligand spectrums during evolution to adapt to their environment, an idea that warrants further investigation.

Intracellular non-ITAM sequence sensitizes human Dectin-1
Thus, the role of Dectin-1 in fungal infection characterized by use of gene-deficient mice (10, 11) may not fully reflect the function of human orthologue. These findings demonstrate that the therapeutic approaches targeted to Dectin-1 requires re-evaluation of its functions in vivo by generating models expressing hDectin-1 in future studies.

Reagents and antibodies
Laminarin and curdlan were purchased from InvivoGen. Zymosan (Z4250) and LPS (L4516) were purchased from Sigma. Westase (9095) was purchased from TaKaRa. Fucan from Cladosiphon novae-caledoniae Kylin was kindly provided by Daiichi Sangyo and the supernatant was collected after centrifugation at 20,000 ϫ g and used as a stimulant. Phycoerythrin (PE)-conjugated anti-HA Ab (clone 16B12) was purchased from Abcam.

Cells
2B4-NFAT-GFP reporter cells expressing various CLRs were prepared as previously described (26,28). For FcR␥or DAP12coupled receptors, reporter cells were co-transfected with FcR␥ or DAP12, respectively. hemITAM-harbored receptors were co-expressed with Syk. Receptors that are not coupled to ITAM signaling were expressed as chimeric receptors by fusing to CD3. mBMDCs were prepared as previously described (29). Briefly, BM cells from a WT C57BL/6J mouse were suspended in RPMI1640 medium supplemented with 10% FBS, antibiotics, and ␤-mercaptoethanol at a density of 5 ϫ 10 5 cells/ml in the presence of culture supernatant of MGM-5 (provided by Dr. S.

Intracellular non-ITAM sequence sensitizes human Dectin-1
Nagata) as a source of GM-CSF containing conditioned medium, and cultured for 7 days at 37°C. hMoDCs were also generated as previously described (30). Briefly, peripheral blood mononuclear cells (PBMCs) from a healthy donor were isolated by Lymphocyte Separation Solution (d ϭ 1.077) (Nacalai Tesque) for gradient centrifugation. Human CD14 ϩ monocytes were purified from PBMCs using anti-human CD14 MicroBeads (Miltenyi Biotech), and cultured in RPMI1640 supplemented with 10% FBS, non-essential amino acid, antibiotics, 10 ng/ml of human GM-CSF (PeproTech), and 10 ng/ml of human IL-4 (PeproTech) for 7 days at 37°C. The collection and use of human PBMCs were approved by the institutional review boards of Research Institute for Microbial Diseases, Osaka University .

In vitro stimulation
The reporter cells were stimulated with curdlan, zymosan, fucan, or laminarin for 18 h at 37°C. The GFP expression of reporter cells was evaluated by FACS Calibur flow cytometer (BD Biosciences).

Construction of chimeric Dectin-1 receptors
For mouse/human Dectin-1 chimeras, constructs were generated by overlapping extension PCR. Primers used were listed on supplemental Table S1. The resulting constructs were cloned into pMX-IRES-hCD8 or pMX-puro retroviral vector containing HA tag at the C terminus and delivered into 2B4 NFAT-GFP cells expressing Syk as previously described (28,29).

Microarray analysis
1 ϫ 10 6 hMoDCs from a healthy volunteer or 1 ϫ 10 6 mBMDCs from a WT mouse were left untreated or stimulated with 500 g/ml of laminarin for 8 h at 37°C. Total RNA was isolated by TRIzol (Thermo Scientific). DNA microarray analysis was performed using Human Gene 1.0 ST array (Affimetrix) or Mouse Gene 1.0 ST array (Affimetrix). A Z-score was calculated for each gene between each sample. Genes with both a Z-score exceeding (or equal to) 2 and a sample/reference ratio exceeding (or equal to) 1.5 were defined as up-regulated. The array data were deposited in the Gene Expression Omnibus (accession number GSE98826).

Real-time PCR
1 ϫ 10 6 hMoDCs from a healthy volunteer or 1 ϫ 10 6 mBMDCs from a WT mouse were stimulated or left untreated for 8 h at 37°C. Total RNA was prepared using Sepasol RNA I Super G (Nacalai Tesque) and used to generate cDNA templates with ReverTra Ace (TOYOBO). Quantitative PCR was performed by using THUNDERBIRD SYBR qPCR mix (TOYOBO) and ABI PRISM 7000 (Applied Biosystems). Human and mouse ␤-actin mRNA were used for normalization. All primers for specific target genes are listed in supplemental Table S1.  Genes are listed in the order of sample/reference ratio. A heat map of the mouse orthologues of the human genes is displayed on the right. mBMDCs were left untreated or stimulated with laminarin. Columns of genes whose mouse orthologues are not conserved are displayed as gray.