CD301 mediates fusion in IL-4-driven multinucleated giant cell formation

Multinucleated giant cells (MGCs) are polykaryons of monocytic origin that are spatially associated with foreign bodies and are seen in infectious, auto-immune, neoplastic and genetic disorders (Kumar et al., 2015). MGCs are morphologically similar despite their occurrence in lesions of different pathogeneses, and arise from the fusion of discrete macrophage subtypes in pathological lesions (Takahashi et al., 1988). Classically activated M1 macrophages are associated with inflammatory processes, whereas alternatively activated M2 macrophages are associated with the resolution of inflammation (Mosser and Edwards, 2008), and characteristically produce ornithines and polyamines through the arginase pathway (MacMicking et al., 1997; Mills et al., 2000). M2 macrophages are further subdivided into various subtypes (M2a, b and c). The generation of one of these subtypes, M2a cells, is promoted by the pro-inflammatory cytokines, IL-4 or IL-13 (Lu et al., 2013).

The formation of MGCs requires cell fusion, which involves the migration of macrophages towards one another and their expression of fusogens that enable cell membrane approximation. This process requires lowering the energy barrier between lipid bilayers of approximating cells (Helming and Gordon, 2009). Currently, the characterization of specific MGC subtypes by histological analysis is challenging, as the molecular determinants and fusogens that specify their formation and function are not defined.

The formation of foreign body giant cells is thought to be regulated by lectins, which detect and bind to glycans expressed by microorganisms (Denda-Nagai et al., 2010; Zelensky and Gready, 2005). Lectins play a role in the activation of the immune system through antigen presentation and cell adhesion. Inhibition of the lectin receptor CD206 (a mannose receptor also known as MRC1), inhibits the formation of foreign body giant cells but does not entirely block fusion (McNally and Anderson, 1995), indicating that other fusogens play central roles in mediating fusion processes. Another, less well-known lectin receptor, macrophage N-acetyl-galactosamine-specific lectin (Mgl/CD301, also known as CLEC10A), which has two murine variants (CD301a/Mgl1 and CD301b/Mgl2), is expressed by macrophages and dendritic cells (Denda-Nagai et al., 2002). Although CD301 is thought to play a role in immune response signaling involving T-cells, its involvement in MGC formation has not yet been described (Denda-Nagai et al., 2010).

Identification of the molecules involved in the formation of MGCs from various monocyte- and macrophage-derived precursors could help to define the role of these cells in MGC-associated diseases, assist in elucidating their pathogeneses, and serve as potential targets for drug discovery. Here, we show that in mouse and human-derived monocytes, the differentially expressed transmembrane protein, CD301, is important for the formation of MGCs induced by IL-4, and that CD301 is expressed in MGC-containing human lesions.

We quantified the formation of MGCs from primary mouse bone marrow monocytes that were activated to become pre-osteoclasts from M1 or M2 macrophages. All nuclei were counted. In those cells with three or more nuclei, we considered that the cell had undergone fusion. The total number of cells that had undergone fusion events (i.e. with three or more nuclei) were expressed as a percentage of the overall nuclear counts (Fig. 1A). The experimental conditions that most effectively promoted cell fusion included RANKL (also known as TNFSF11), IL-1β, IL-4 and TNF, which were then selected for further investigation using tandem mass tag-mass spectrometry (TMT-MS).

Total cell lysates isolated after 5 days of treatment were tagged with stable isotopic labels for analysis by TMT-MS (Table 1). Proteins that were upregulated or downregulated by a log₂ fold change of more than 0.5 or less than −0.5 were included in the analysis. The protein MCG21506 (CD301b or Mgl2) showed markedly increased abundance after IL-4 treatment (average increased log₂ fold change of 1.0), and was chosen for subsequent and more detailed analysis because of its unexpected expression. Furthermore, we considered that if CD301 was enriched at sites adjacent to the plasma membrane, it may be involved in membrane fusion processes.

We examined Mgl1 and Mgl2 mRNA expression in RAW264.7 cells using RT-PCR and qRT-PCR. Mgl1 was constitutively expressed and was increased eightfold by IL-4 treatment; Mgl2 expression was only detectable after IL-4 treatment (Fig. 1B,C). Cells that were treated with vehicle only (i.e. none of the cytokines, IL-1, RANKL and TNF) showed no Mgl2 expression. There was also no change of basal Mgl1 levels when compared to cells without cytokine treatment (data not shown). We confirmed that these patterns of Mgl1 and Mgl2 mRNA expression were reflected in the measurements of Mgl1 and Mgl2 protein levels by immunoblotting (Fig. 1D), in which the antibody detects Mgl1 and Mgl2 (hereafter designated as CD301).

Immunolocalization studies showed that mouse CD301 (both isoforms) was expressed in RAW264.7 cells, but only after IL-4 treatment (Fig. 1E). When expressed, CD301 was particularly abundant in the cytoplasm near the plasma membrane in cells with multinucleation. We quantified CD301 localization by measuring the fluorescence intensity of CD301 in regions of interest in the cytoplasm and over the plasma membrane, and these data were expressed as a ratio (Fig. 1F). The relative abundance of CD301 staining that localized to the plasma membrane increased nearly twofold when cells with one or two nuclei, or with six to nine nuclei, were compared (P<0.0001).

These evaluation methods were deepened using a separate labeling method to examine plasma membrane-localized CD301. We immunostained cells for CD301 and the plasma membrane label fluorescent concanavalin A (see Materials and Methods) to assess the spatial relationship of CD301 with the plasma membrane. In these preparations, CD301 was more obviously localized adjacent to the plasma membrane (before fixation; Fig. 2A) compared with single antibody immunostaining described above. By quantitative analysis, a moderate level of spatial colocalization (Pearson's correlation coefficient) between concanavalin A and CD301 was found in these cell preparations (r=0.54 to 0.64 in four different experiments; P<0.01). We extended these studies by examining plasma membrane proteins that were isolated by affinity labeling with concanavalin A (Lee et al., 2008). In these preparations, Na⁺/K⁺-ATPase was used as a plasma membrane marker protein (Fig. 2B). CD301 was expressed only in IL-4-treated cells and by densitometry we found there was 30-fold more CD301 in the plasma membrane fraction compared with the cytosolic fraction.

Purified glycans, Le^X and GalNAc, which are cognate ligands of Mgl1 and Mgl2, respectively, were bound to beads and incubated with IL-4-treated cells, or with IL-4-treated cells that had been pretreated with an inhibiting antibody (designated here as CD301 as it targets mouse Mgl1 and Mgl2 isoforms; Fig. 3A). Differential interference contrast images were overlaid with fluorescence microscopy images to show the location of the beads on the cells. IL-4 treatment increased binding of Le^X and GalNAc beads by threefold and sevenfold, respectively (Fig. 3B). Treatment with the Mgl1/2 inhibitory antibody decreased binding of Le^X and GalNAc beads by 1.5- and fourfold, respectively, when compared to IL-4 treatment alone. As the inhibiting antibody blocked Mgl1 and Mgl2 ligand binding, we examined cell fusion after antibody inhibition of CD301 function. Cells were treated with IL-4 for 6 days and then with the Mgl1/2 inhibitory antibody or an isotype control antibody, and the percentage of cells that underwent fusion was determined (Fig. 3C-E). Antibody inhibition reduced by twofold the percentage of IL-4 treated cells that underwent fusion (Fig. 3D). Furthermore, the number of multinucleated cells with three to five nuclei or with more than six nuclei per cell were quantified in the control and antibody inhibition groups. After inhibition of CD301, the number of cells with six or more nuclei per cell was reduced more than 15-fold (from 1.6±0.2 to 0.1±0.1, compared with the control group; P<0.0001; Fig. 3E).

Mgl1 knockout (KO), Mgl2 KO, and Mgl1/2 double KO (DKO) cells were generated (Fig. S1), and their survival and growth over 5 days was evaluated (Fig. S2). In some experiments, Mgl2 KO or DKO cells were stably transfected with an Mgl2 expression vector. The localization and abundance of CD301 protein expression were examined by immunofluorescence and immunoblotting after IL-4 treatment (Fig. 4A-E). An antibody to CD301 that detects Mgl1 and Mgl2 isoforms was used for immunostaining (designated as Mgl2 in immunoblot). Protein expression was readily visualized in wild-type and Mgl1 KO cells after IL-4 treatment but was not detectable in Mgl2 KO or DKO cells. The immunoblot shows that Mgl2 KO and DKO cells that were transfected with Mgl2 exhibited abundant CD301 staining without IL-4 stimulation, presumably reflecting constitutive activity of the 5′-long-terminal repeat retroviral promoter in the Mgl2 expression vector. Separate measurements of Mgl2 expression after IL-4 treatment were obtained by flow cytometry analysis of wild-type and KO cells after no treatment or IL-4 treatment (Fig. 4F). Mgl2 expression was increased after 4 days of treatment with IL-4, but the KO cells showed no detectable signal, even in the presence of IL-4 (median fluorescence intensity and relative numbers of Mgl2⁺ cells expressed as percentage of total cell population; Fig. 4G).

When wild-type, Mgl1 KO, Mgl2 KO or DKO cells were incubated with glycan-bound beads, there was decreased binding of Le^X beads (more than twofold) in Mgl1 KO and DKO cells compared with wild-type cells and Mgl2 KO cells (Fig. 5A). When these same groups of cells were incubated with GalNAc beads, there was more than fourfold reduced bead binding for Mgl2 KO or DKO cells compared with wild-type cells and Mgl1 cells. Collectively, these data show that the binding of glycan beads with their cognate partners was decreased in the expected KO cell groups.

In wild-type and KO cell lines treated with IL-4 (Fig. 5B), cell fusion was decreased by 2.2- and 4.8-fold in Mgl2 KO and DKO cells, respectively, compared with wild-type cells. Notably, there was no detectable reduction of fusion of the KO cells compared with wild-type cells after treatment with RANKL (Fig. S3). The Mgl2 KO cells that were rescued with Mgl2 and the DKO cells that were rescued with Mgl2, exhibited cell fusion with or without IL-4 treatment, and there was no difference in the number of cell fusion events between these groups (P>0.2), indicating that Mgl2 expression and not IL-4 treatment per se, promoted cell fusion.

We examined the ability of KO cells to undergo fusion if their proximity to one another was increased. This experiment was conducted by plating cells at three different densities (10³, 10⁴ and 10⁵ cells/well). After 6 days of IL-4 treatment, cells were fixed and the number of cells that underwent fusion events was analyzed (Fig. 5C). For wild type and the various KO cell lines, there were no statistically different proportions of fusions at the different plating densities (P>0.2). Furthermore, Mgl2 KO and DKO cells all showed statistically different numbers of fusion events compared with wild-type cells at all plating densities (P<0.0001).

THP-1 human monocyte-like cells were treated with pro-fusogenic cytokines to assess the formation of multinucleated cells. Treatment with human macrophage CSF (hM-CSF; also known as CSF1) with IL-4 or RANKL enhanced MGC formation, but no fusion occurred after hM-CSF treatment alone (Fig. 6A,B). CD301 mRNA expression was measured by qRT-PCR (Fig. 6C). The cells all exhibited low basal CD301 levels but CD301 expression was increased by nearly fivefold above control levels after treatment with IL-4. THP-1 cells cultured under pro-fusogenic conditions were subjected to different treatments to assess the effects of inhibition or activation of CD301 on cell fusion (Fig. 6D,E). For the activation of CD301, cultures were incubated with N-acetylgalactosamine (GalNAc), a high affinity glycan ligand for human CD301. Recombinant human CD301 (rhCD301) was added to bind glycan partners, in order to inhibit the activation of the cell-bound lectin. In cells treated with IL-4, rhCD301 reduced cell fusion by nearly tenfold compared with controls (P<0.0001). In cultures treated only with hM-CSF, there were no fused cells, but the presence of GalNAc promoted fusion to a level similar to that observed with IL-4 treatment alone. In cells treated with RANKL, there was only a small reduction of the percentage of fused cells in the presence of soluble GalNAc. Furthermore, GalNAc did not inhibit RANKL-driven cell fusion after treatment with rhCD301.

We examined whether MGCs in human pathological lesions expressed CD301. Two reactive/reparatory lesions were studied, the central giant cell granuloma (CGCG) and peripheral giant cell granuloma (PGCG). Immunohistochemistry of the two variants of the giant cell granuloma was performed, and showed that CD301 and the transmembrane protein CD68 were highly expressed by cells of the monocytic lineage (Fig. 6E). Mononucleated and multinucleated cells exhibited CD301 staining in both lesions.

We used TMT-MS to obtain new insights into the mechanisms of monocytic cell fusion, which led to the discovery of the transmembrane protein Mgl2 and its role in the formation of MGCs in mice. The human variant of CD301 (also called C-type lectin domain family 10 member) is a transmembrane protein in the group of lectin receptors known as C-type lectin receptors, to which the mannose receptor also belongs (a pro-fusogen in osteoclasts) (Zelensky and Gready, 2005). CD301, which is expressed by human dendritic cells and macrophages, and is involved in immune response signaling through interactions with effector T-cells (Higashi et al., 2002; van Vliet et al., 2006), is upregulated by IL-4 and is considered to be a marker of M2 macrophages (Raes et al., 2005). These findings are consistent with our TMT-MS, mRNA, immunostaining and immunoblot data, and the impact of the M2 activator IL-4 and its upregulation of CD301 expression (Raes et al., 2005). Notably, the other variant of the CD301 protein in mice, Mgl1, did not show differential expression, at least as assessed by our threshold for analysis by TMT-MS (i.e. more than 0.5 log₂ fold change between the various cytokine treatments). Mgl1 exhibits different glycan affinity binding profiles than Mgl2, and when knocked out using CRISPR/Cas9, there was no significant difference in the number of IL-4-induced cell fusion events compared with the number in wild type (Singh et al., 2009). Mouse Mgl2 is similar in amino acid sequence and carbohydrate specificity to human CD301 (Denda-Nagai et al., 2010). Correspondingly, we found that deletion of the Mgl2 gene in mouse macrophages led to similar findings with respect to IL-4-mediated cell fusion in human cells.

When we immunostained the two isoforms of CD301 (Mgl1 and Mgl2) in mouse monocytes with a non-isoform-specific antibody, we found that as the number of nuclei increased, CD301 became more localized to the plasma membrane. Furthermore, when another non-isoform-specific antibody was used to block IL-4-induced fusion, there were reduced overall numbers of fusion events. The numbers of cells with six or more nuclei per cell were greatly decreased but the numbers of cells with three to five nuclei were slightly increased. This finding, along with the localization of CD301 to the plasma membrane of larger cells, suggests a role for CD301 in the formation of larger MGCs and in fusion events at later stages of MGC formation. A similar observation was made after inhibition of the mannose receptor in IL-4-driven multinucleation and osteoclast formation (McNally et al., 1996; Morishima et al., 2003). In both of these fusion conditions, the formation of multinucleated cells does occur, but multinucleation occurs to a lesser extent. These parallel findings, which also involve a C-type lectin receptor, support the notion that this type of receptor is involved in later fusion events. Conceivably, another fusion molecule leads to the formation of smaller MGCs. A possible candidate protein that enables early fusion events is the dendritic cell-specific transmembrane protein (Dcstamp), as it is required for the formation of IL-4-driven multinucleation and osteoclastogenesis (Yagi et al., 2005).

In the human monocytic cell line THP-1, we found that CD301 expression was upregulated by IL-4 treatment. With the addition of exogenous recombinant human CD301 to cultures, which reduced the availability of membrane-bound and free glycans, we found that IL-4-induced cell fusion was greatly decreased. This competitive inhibition of cell fusion by the addition of excess CD391 did not occur in RANKL-treated cells. Notably, we observed by qRT-PCR that CD301 is expressed at moderate levels when cells were treated with hM-CSF alone. When the glycan ligand for CD301 (GalNAc) was added to cells, MGC formation occurred without the need for IL-4, suggesting that constitutive expression levels of CD301 were sufficient to promote cell fusion events, at least to a limited degree. When Mgl2 was re-expressed in the KO mouse cell lines, not only was cell fusion restored, but these cells no longer required IL-4 for cell fusion to occur. Taken together, these results suggest that ligand binding to CD301 (or Mgl2) alone can stimulate monocyte fusion. The experimental interventions that reduced fusion events in these experiments suggest that it is either a lectin-glycan interaction or a CD301-GalNAc interaction that promotes cell fusion. In this context, allosteric activation of CD301 may promote ERK1/2 signaling, which is known to regulate the function of the Arp2/3 complex and to promote actin assembly in nascent lamellipodia (van Vliet et al., 2013). The formation of lamellipodia may facilitate approximation of plasma membranes in adjacent cells after an initial CD301/Mgl2-glycan interaction (Mendoza et al., 2015). As a result of subcortical actin assembly and its integration into protrusion-dependent processes leading to fusion, pores in the plasma membrane are formed, as seen in the fusion of myoblasts, which enables sharing of cytoplasmic contents between fusing (Sens et al., 2010).

Our original goal was to identify a differentially expressed fusogen in MGCs derived from cells with different origins to improve, over the long term, diagnosis of MGC-containing lesions. We found that in the CGCG and the PGCG, mononuclear CD68⁺ cells and MGCs were stained for CD301. There is a long-standing debate on whether the CGCG and PGCG are only intra- and extra-osseous versions of one another (Gupta et al., 2019; Neville et al., 2016). IL-10, a cytokine released by alternatively activated M2a cells, is a marker of anti-inflammatory macrophages (Wang et al., 2019). IL-10, along with its receptor, is thought to be equally expressed by MGCs of CGCGs and PGCGs (Syrio et al., 2011). In studies of peripheral blood of patients with CGCGs, monocytes showed increased IL-10 expression compared with lymphocytes or with control monocytes, indicating a possible M2b phenotype (Souza et al., 2005). Our findings of CD301-stained mononuclear and multinucleated cells in the PGCG and CGCG lesions supports the notion that these lesions are associated with an M2 macrophage subtype and are involved in reactive/reparative processes.

Although a linkage between M2 macrophages and MGC formation has been described previously (McNally and Anderson, 2011), a role for CD301 in cell fusion leading to MGC formation has not been reported. Previously, it was thought that CD206 is important for foreign body (or IL-4-induced) MGC formation. However, CD206 is also involved in the formation of osteoclasts (RANKL-induced MGC formation) (McNally et al., 1996; Morishima et al., 2003). Our findings from human and mouse cell lines show that CD301 and Mgl2, respectively, are required for the formation of IL-4-induced MGCs, but the formation of osteoclasts was not affected in cells from either species. This finding could be relevant for new diagnostics or therapies. Currently, CD206 is known to be expressed by two different MGC subtypes, whereas CD301 is only expressed by one, thereby providing a possible marker for differentiating between the two subtypes. CD301, also appears to be involved in the formation of MGCs with larger numbers of nuclei, those which may cause the greatest number of issues pathologically to patients. Some current therapeutic approaches have targeted CD206, or RANKL (in the formation of osteoclast-like giant cells), as both of these molecules are important for osteoclast formation (Azad et al., 2014; Lipplaa et al., 2019). However, the use of these treatments can inhibit physiological bone remodeling and increase the risk of osteonecrotic lesions (Lipplaa et al., 2019). Accordingly, an improved understanding of CD301-mediated cell fusion could provide a selective treatment for IL-4-driven MGC-containing disease processes.

In conclusion, IL-4-induced MGC formation is evidently dependent on the expression of Mgl2 (in mouse) or CD301 (in human) in macrophages. In contrast, RANKL-induced formation of osteoclasts is independent of CD301. Notably, CD301 appears to be involved in the later stages of cell fusion events, which lead to the generation of MGCs with large numbers of nuclei. Further investigations into MGCs with broader CD301 expression profiles may show that this protein is a phenotypic marker of reactionary/reparative processes. These findings could rationalize its use as a diagnostic and therapeutic target of MGC-containing lesions.

All procedures were performed in accordance with the Guide for the Humane Use and Care of Laboratory Animals, and were approved by the University of Toronto Animal Care Committee. Bone marrow was harvested from tibia and femurs of male C57BL/6 mice between 6 and 10 weeks old. Single-cell suspensions were prepared from marrow in complete minimum essential medium alpha (αMEM), which includes 10% fetal bovine serum and 164 IU/ml of penicillin G, 50 mg/ml of gentamicin and 0.25 mg/ml of fungizone. Cells were added to T-75 cm² flasks (10⁶ cells/ml), incubated overnight, then centrifuged over Ficoll-Paque Plus, at 300 g for 30 min at room temperature, washed with αMEM, then plated at 10⁵ cells/ml with M-CSF (20 ng/ml) for 4 days before experimentation.

RAW264.7 and THP-1 cells from ATCC were cultured in complete Dulbecco's modified Eagle's medium (DMEM) or Roswell Park Memorial Institute medium with 0.05 mM β-mercaptoethanol), respectively. Cellular phenotype was examined at each subculture; pleuropneumonia-like organism infection was examined by DAPI staining at every subculture and there was no evidence of infection. Complete medium was medium that was supplemented with 10% fetal bovine serum, 164 IU/ml penicillin G, 50 µg/ml gentamicin and 0.25 µg/ml fungizone. Medium and cytokines were replenished every 48 h during experiments. Cells were seeded in 60 mm² culture dishes at 5×10⁶ cells/dish.

Mouse macrophage RAW264.7 cells were treated with RANKL (60 ng/ml), IL-1β (20 ng/ml), IL-4 (10 ng/ml) or TNF (10 ng/ml), fixed on day 5 of culture with 4% paraformaldehyde (PFA), washed with PBS, and frozen. Relative protein quantification was determined with TMT-MS using stable isotopic labels, and a comparison of peptide counts from proteins that were identified with high probability (P<0.05). Samples were fractionated into three groups, allowing for one missed cleavage. Data were analyzed with Scaffold Q+ software. Relative abundances of log₂ fold change (more than 0.5 or less than −0.5) were tabulated (overexpressed and underexpressed, respectively). Data from these experiments will be made accessible by request to the corresponding author (C.A.M.).

RAW264.7 cells were plated at 10,000 cells per well on fibronectin-coated eight-well chamber slides with complete DMEM and with vehicle-only treatment or IL-4 (10 ng/ml) for 6 days. Rat anti-mouse CD301 IgG2a monoclonal antibody (ER-MP23; 50 µg/ml, Novus Biologicals) or its isotype control (50 µg/ml, BioLegend rat IgG2a, RTK2758) were added to wells directly for antibody inhibition experiments. Medium, antibodies and cytokines were replenished every 48 h. The authenticity of the blocking antibody was assessed by comparison with the isotype control and its lack of efficacy in blocking in cells that did not express CD301.

In some experiments, primary mouse monocytes were plated in complete αMEM and with either recombinant mouse M-CSF or granulocyte M-CSF (GM-CSF) (20 ng/ml). In experiments that examined the effect of certain cytokines on the expression of CD301, the following cytokines were added to culture medium at the indicated concentrations: RANKL (60 ng/ml), LPS (10 ng/ml), INFγ (1 ng/ml), IL-1β (20 ng/ml), TNF (10 ng/ml), IL-4 (10 ng/ml) or TGF-β1 (5 ng/ml).

In experiments that examined the role of CD301 in human monocyte fusion, human monocytic THP-1 cells were plated as described above, with the addition of recombinant hM-CSF (20 ng/ml) alone, or with RANKL (60 ng/ml) or recombinant human IL-4 (10 ng/ml). For experiments in which CD301 activation or inhibition experiments were conducted using THP-1 cells, N-acetylgalactosamine (GalNAc; α-GalNAc-PAA-biotin, GlycoTech, 1 µg/ml) or recombinant human CD301 (R&D Systems, 4888-CL, 1 µg/ml) were added and replenished with the indicated cytokines. Full-length recombinant CD301 (from R&D Systems) forms trimers in solution, which enables binding to glycans and, as a result, inhibits activation of membrane-bound CD301 (Beatson et al., 2015; Jégouzo et al., 2013). After 6 days, all cells were fixed with PFA for 20 min, washed with PBS, stained with Alexa Fluor 488 Phalloidin (for actin filaments) at room temperature for 20 min, stained with DAPI (for nuclei) for 10 min, mounted and imaged by confocal microscopy. Cells were considered to have fused if there were three or more nuclei per cell. The number of fused nuclei was divided by the total nuclei and expressed as a percentage.

For studying the spatial relationship between CD301 and the plasma membrane, we used membrane labeling with concanavalin A, which was followed by immunostaining with CD301. Briefly, RAW264.7 cells (10⁴ cells/well, passage 8) were incubated in 700 µl complete DMEM with or without recombinant murine IL-4 (PeproTech, 214-14, 10 ng/ml) in eight-well chamber slides (Falcon) for 4 days. Cell culture medium was changed at day 2. The medium was removed and cells were rinsed with prewarmed Hank's Balanced Salt Solution (HBSS, Wisent Bioproducts, 311-515-CL). Live cells were stained with HBSS (250 µl/well) containing 50 µg/ml concanavalin A-Alexa Fluor 594 (Invitrogen, C11253) at 37°C for 30 min. The cells were washed with HBSS and fixed with 4% PFA in PBS for 20 min. After being washed three times with PBS, the fixative was quenched with 100 mM glycine in PBS (pH 7.4) for 10 min. Cells were permeabilized with 0.1% Triton X-100 in PBS for 3 min and blocked with 1% bovine serum albumin (BSA) in PBS for 30 min in the dark at room temperature. Cells were incubated with rat anti-mouse CD301 antibody (Novus Biologicals, NB100-64874, 1:500) in 1% BSA/PBS overnight at 4°C. After washing with PBS, cells were stained with goat anti-rat IgG Alexa Fluor 488 (Invitrogen, A-11006, 1:1000 in 1% BSA/PBS) at 37°C for 1 h. Finally, cells were counterstained with DAPI (0.165 µM in PBS, Sigma-Aldrich) for 10 min and imaged by confocal microscopy. Analysis of co-localization of concanavalin A and CD301 was performed using the ImageJ Coloc 2 plug-in for Pearson's correlation coefficient, as described previously (Bolte and Cordelières, 2006).

A second approach was used to examine the relationship between the plasma membrane and CD301. In this instance, plasma membrane proteins were isolated using magnetic beads and immobilized concanavalin A (Lee et al., 2008). Briefly, streptavidin magnetic beads (MJS BioLynx, VECTM1003010), in a 150 µl suspension, were placed into a 1.5 ml Eppendorf tube. Beads were separated from the solution by holding the tube near a magnet; the remnant solution was discarded by pipetting. Beads were washed twice with 1 ml of distilled water, followed by 1 ml of Tris-buffered saline (TBS) containing 150 mM NaCl and 50 mM Tris HCl, pH 7.4. Beads were re-suspended in 245 μl TBS, and 35 μl biotinylated concanavalin A (5 mg/ml, BioLynx, VECTB1005) was added. This suspension was mixed on a rocker for 1 h to enable the binding of biotinylated concanavalin A to streptavidin. Beads were washed sequentially with 1 ml of TBS, 1 ml of TBS with 1% Triton X-100, and three times with 1 ml of TBS. Beads were resuspended in 450 µl of TBS and stored at 4°C. Subsequently, RAW264.7 cells (0.5×10⁶/dish, passage 8) were treated with or without IL-4 (10 ng/ml) for 4 days in five 10 cm² plates in complete DMEM (10 ml/dish). The medium was changed at day 2. After washing twice with TBS, cells were collected by scraping each plate with 1 ml of TBS. Scraped cell suspensions were collected in a 12 ml centrifuge tube and centrifuged at 325 g for 5 min to collect the cells. The supernatant was removed and the cell pellet was saved. The pellet was resuspended in 0.5 ml of TBS and homogenized in a 2 ml Dounce homogenizer. An aliquot of the homogenized total cell lysate [50 µl, diluted in 300 µl radioimmunoprecipitation assay (RIPA) buffer] was saved for later analysis. The remaining lysate (in TBS) was added to the streptavidin magnetic beads that had been prepared 1 day before, as described above, and then mixed on a rocker for 1 h. After binding, the streptavidin magnetic beads were collected by holding the tube near a magnet. The unbound flow-through fraction was also saved. Beads were washed with 1 ml of TBS (five times), and proteins that bound to the concanavalin A magnetic beads were eluted with 250 µl TBS containing 0.25 M methyl α-D-mannoside (BioLynx, VECTS9005) and 0.5% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS, Sigma-Aldrich). Protein concentrations were measured with a bicinchoninic acid (BCA) kit (Thermo Scientific). Equal amounts of protein samples (20 µg/lane) were run on a 12% SDS polyacrylamide gel for CD301 and for beta-actin. A 7% gel was used to resolve Na⁺-K⁺ ATPase (a plasma membrane protein), which was followed by immunoblotting. For analyses, goat anti-mouse Mgl2/CD301b antibody from R&D Systems (A2835, 1:200 dilution of 0.2 mg/ml), mouse monoclonal anti-β-actin antibody from Sigma-Aldrich (clone AC-15, A5441, 1:8000), and rabbit anti-α1 Na⁺-K⁺ ATPase antibody from Cell Signaling Technology (3010, 1:1000) were used. Proteins were visualized using the following secondary antibodies: IRDye 800CW donkey anti-goat IgG, IRDye 680CW goat anti-mouse IgG, or IRDye 800CW goat anti-rabbit IgG, respectively, and a Li-Cor Odyssey detection system.

Mgl1 KO and Mgl2 KO RAW264.7 cell lines were generated. Mgl1 and Mgl2 DKO RAW264.7 cell lines were also generated with sequential KO of Mgl1 from an Mgl2 KO cell line by Applied StemCell. Two guide RNAs (gRNAs) were chosen for each gene: g1 5′-ACGAAAACCTCCAGAACTCA-3′ and g2 5′-CCTCCAGAACTCAAGGATCG-3′ targeting exon 2 of mouse Mgl1; and g1 5′-AAACTTCCAGAACTTGGAGC-3′ and g2 5′-GGAGCGGGAAGAGAAAAACC-3′ targeting exon 2 of mouse Mgl2 (Fig. S1). Based on endonuclease activity, g2 for Mgl1 and g1 for Mgl2 were chosen, and cloned into a gRNA/Cas9 expression vector and introduced to RAW264.7 cells. Mutagenesis within the targeted regions in a single homozygous cell clone from the transfected cells was identified by PCR (Table S1). Sanger sequencing confirmed multiple clones of Mgl1 KO, Mgl2 KO, and sequentially, Mgl2, then Mgl1 DKO. Two Mgl1 KO clones, A1 (−2, −1 bp) and C1 (−8 bp), and three Mgl2 KO clones, A2 (−2 bp), B3 (−7 bp) and B8 (−17 bp), were analyzed. The Mgl2 KO clone B8 was used for sequential Mgl1 KO; three clones were obtained, G9 (−7 bp), G10 (−56 bp) and F10 (−40 bp). Mgl1 KO clone A1, Mgl2 KO clone B8, and Mgl1/2 double knockout (DKO) clone G10 were utilized for experiments. The growth of selected clones in T-75 flasks was observed over 7 days by phase contrast microscopy.

An In-Fusion Cloning kit (Takara) was used for Mgl2 gene cloning and amplification of Myc-DDK-tagged mouse Mgl2 (NM_145137) cDNA (OriGene Technologies), using the primers described in Table S1. The amplified cDNA (2 µg) was inserted into BglII/XhoI-linearized pMSCVpuro and the insertion was confirmed by sequencing. pMSCVpuro-Mgl2-Myc-DDK was co-transfected with pVSV-G into a retroviral packaging cell line GP-293 to produce retrovirus-expressing Mgl2. Mgl2 KO and Mgl1/2 DKO cells were virus infected, and selected for 2 weeks with 7 µg/ml puromycin and maintained in cell culture medium with 3 µg/ml puromycin.

RAW264.7 wild-type and KO cells treated with cytokines were detached from culture dishes using EDTA, and fixed with fresh methanol-free formaldehyde (1.6% final concentration) for 15 min on ice before processing. Cells (5×10⁵ cells/tube) were resuspended (50 µl; 5×10⁵ cells/tube) in fluorescence-activated cell sorting buffer (1× HBSS^−/−, 2 mM EDTA, 1% BSA), blocked with mouse IgG (2 µl mg; Sigma-Aldrich) and rat serum (1 µl; Sigma-Aldrich), then labeled with APC-tagged rat anti-mouse Mgl2 antibody (1:50, BioLegend; clone LOM-14; Rat IgG2b, κ). Mgl2 null cells did not show significant staining with this antibody. Sample acquisition was performed using a Sony flow cytometer (Spectral Analyzer). At least 2×10⁵ gated events were acquired for each sample. Data were analyzed using FCS Express software (Version 7; De Novo Software).

Total RNA was extracted (RNeasy Mini Kit, Qiagen) and residual genomic DNA was digested (RNase-Free DNase Kit, Qiagen). The RNA concentration was determined using a NanoDrop spectrophotometer. RNA quality was analyzed with an Agilent Technologies Bioanalyzer. Total RNA (1 µg) was reverse-transcribed into cDNA using Oligo-dT18 VN primers (ACGT Corp). qRT-PCR was performed in triplicate (Bio-Rad CFX96 real-time system). CFX Manager Software (Version 1.0) was used to analyze PCR results; mRNA expression was normalized to the housekeeping gene GAPDH (Table S1). For detection of CD301 in THP-1 cells, each 20 µl reaction contained 5 µl of 1:10 diluted cDNA, 200 nM of forward and reverse primers and 10 µl of the Advanced qPCR master mix (Wisent). RNA expression was normalized to the housekeeping gene beta-actin. PCR primers used for human CD301 detection were designed using mRNA sequence previously deposited in GenBank (NM_182906.4) (Ostrop et al., 2015).

After 5 days of culture, cells were lysed (RIPA buffer containing protease inhibitors), sedimented and the supernatants were collected. Protein was quantified using a BCA kit; 30 μg of protein in Laemmli sample buffer was boiled for 10 min, separated by 10% SDS-PAGE and transferred to nitrocellulose membranes for 90 min. Membranes were blocked with 5% milk, incubated overnight at 4°C with goat anti-mouse CD301 (Novus Biologicals, AF4888, 1:200) and mouse anti-β-actin (Sigma-Aldrich, A5441, 1:800) antibodies with gentle shaking. The authenticity of the antibody was evaluated through the knockdown experiments described in the Results. Membranes were washed with Tris-buffered saline with 0.2% Tween 20, incubated with secondary antibodies (IRDye 800CW donkey anti-goat IgG and IRDye 680CW goat anti-mouse IgG, both 1:1000) and blot densities quantified (Li-Cor Odyssey). Densitometry analysis was performed using ImageJ software; CD301 was normalized to beta-actin expression.

RAW264.7 cells were plated (100,000 cells/well) on fibronectin-coated eight-well chamber slides with complete DMEM containing vehicle, RANKL (60 ng/ml), IL-1β (20 ng/ml), IL-4 (10 ng/ml) or TNF (10 ng/ml) for 4 days. Medium and cytokines were replenished every 48 h. Cells were fixed with 4% PFA for 20 min, quenched, permeabilized (0.1% Triton X-100 in PBS) and blocked with BSA (1% in PBS containing 0.1% Triton X-100). Cells were incubated with rat anti-mouse CD301 IgG2a monoclonal antibody (1:500, Novus Biologicals) at 4°C overnight, stained with Alexa Fluor 488 goat anti-rat IgG (H+L) antibody (1:500, Invitrogen) at 37°C for 1 h, then Phalloidin-Texas Red-stained, and DAPI stained before mounting and imaging using a Leica SP8 confocal microscope.

Magnetic streptavidin-bound 2.8 µm beads (Invitrogen Dynabeads) were prepared by incubation with biotinylated glycans GalNAc and Sialyl-Lewis X (Le^X) (alpha-GalNAc-PAA-biotin; Le^X-PAA-biotin, GlycoTech; 3:1 ratio); then added to cells directly after washing. Wild-type or KO cells were plated at 50,000 cells/well with complete DMEM with no treatment or IL-4 (10 ng/ml) for 3 days. Medium and cytokines were replenished after 48 h. For antibody inhibition experiments, rat anti-mouse CD301 IgG2a monoclonal antibody (50 µg/ml, Novus Biologicals) or isotype control (50 µg/ml, BioLegend rat IgG2a, RTK2758) were added to the wells 1 h before bead addition. Unbound, GalNAc- or Le^X-bound beads (6.7×10⁸ beads/ml) were incubated with cells for 1 h, PFA fixed, washed, then Phalloidin-488- and DAPI-stained. Slides were mounted and imaged using a confocal microscope. The number of beads bound to membranes and unbound beads were counted, and their ratio was expressed as a percentage.

With approval from the University of Toronto Research Ethics Board (#35822) (which included obtaining appropriate consent), samples of peripheral and central giant cell granulomas were obtained from the Toronto Oral Pathology Service, Faculty of Dentistry, University of Toronto. All clinical investigations were conducted in accord with the principles expressed in the Declaration of Helsinki. Sections (5 µm) from paraffin-embedded biopsy specimens were mounted on glass slides, heated for 30 min at 60°C, antigen retrieved (citrate buffer, pH 6.0, Abcam), H₂O washed, permeabilized (Triton X-100), blocked (Sea Block Serum, Abcam), immunostained with mouse monoclonal anti-human CD68 Alexa Fluor 405 (Novus Biologicals, NB600-985AF405, 1:200) and mouse anti-human CD301 Alexa Fluor 546 (Novus Biologicals, DDX0010A546-100, 1:200), counterstained with DRAQ7 (Abcam, ab109202, 3 µM), imaged using a confocal microscope on the same day, and analyzed using Volocity 3D Image Analysis Software (PerkinElmer). Normal human oral mucosa and liver paraffin sections (Amsbio, HP-314) were likewise stained as CD301-negative and positive controls, respectively.

Mean and s.e.m. were calculated for all continuous variables. An unpaired Student's t-test was used for comparisons of two samples; one- and two-way ANOVA analyses with Tukey's post-hoc test were employed for analysis of multiple samples. Numerical data are expressed as mean±s.e.m., and for all experiments, each experiment was repeated on different days at least in triplicate and in some instances up to six trials, with each experiment comprising a minimum of three separate culture dishes. For TMT-MS, duplicate assays were performed. For all analyses employing flow cytometry, a minimum of 10,000 events were analysed. Statistical significance is indicated as *P<0.05, **P<0.01, ***P<0.001, and ****P<0.0001. We found in initial experiments that with sample sizes of a minimum of three separate experiments conducted on different days, and with each experiment containing a minimum of three samples, that our ability to detect statistically significant differences (P<0.05) was not affected by insufficient power.

We are grateful to Dhaarmini Rajshanhkar, Wilson Lee and Aiman Ali for technical support with imaging and immunostaining optimization. We are particularly grateful to Grace Bradley and the University of Toronto Oral Pathology Service for the selection and provision of human biopsy specimens.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.248864.reviewer-comments.pdf