Fungal recognition is mediated by the association of dectin-1 and galectin-3 in macrophages

Dectin-1 Expressed on Mammalian Cells Can Be Labeled Through Sortagging Without Affecting Its Functionality.

Murine dectin-1 equipped with an LPETG motif followed by an HA epitope tag at its C terminus is properly transported to the cell membrane and retains its function. Cells stably expressing tagged dectin-1 on the cell surface are capable of binding zymosan (SI Results and Fig. S1).To determine whether sortase A could act on tagged dectin-1 on intact cells, we incubated HEK 293T and RAW 264.7 cells that stably express dectin-1 with sortase A together with a biotinylated probe (18). The intact cells were then lysed, the lysates separated by SDS/PAGE, and biotinylated products revealed on blots with streptavidin-HRP. Dectin-1 was labeled selectively, because there were no other biotinylated polypeptides that originated either from cell surface or cytosolic proteins. When these lysates were subjected to immunoblotting to detect HA-tagged materials not cleaved by sortase, HA-tagged dectin-1 was present in all samples, both before and after incubation with sortase A and the biotinylated probe (Fig. 1A). Any cell-internal dectin-1 would be inaccessible to sortase and so retain its HA tag. Even if all dectin-1 were surface exposed, the sortase reaction may not proceed quantitatively, also accounting for the presence of HA-positive dectin-1. Cells that were first lysed, sortagged, and only then subjected to streptavidin-HRP immunoblotting also showed the presence of labeled dectin-1 (Fig. 1A).

Dectin-1 analyzed by immunoblotting presents itself as a doublet. Murine dectin-1 is glycosylated (4) and phosphorylated, modifications that could yield multiple electrophoretically distinct species. We sortagged dectin-1 in stably transduced HEK 293T cells with a biotinylated probe and subjected the cell lysates to digestion with the glycosidases N-Glycosidase F (PNGase F) or endoglycosidase H (EndoH), as well as with calf intestinal alkaline phosphatase (CIP). PNGase F hydrolyzes all types of N-glycans, whereas EndoH is specific for high mannose-type N-linked glycans. The presence of high mannose-type N-linked glycans is usually diagnostic of localization to the endoplasmic reticulum (ER) and/or cis Golgi. After PNGase F treatment, a single biotinylated polypeptide remained; samples treated with either EndoH or CIP still presented as two distinct polypeptides (Fig. 1B), which must therefore be the result of differential carbohydrate modifications. The exclusive presence of EndoH-resistant, biotinylated dectin-1 when intact cells were labeled indicates that sortase A installed the biotinylated probe only on dectin-1 that had undergone complex glycan modification before transit to the cell surface.

To investigate whether sortagged dectin-1 is functional on live cells, we installed an Alexa Fluor 647 probe on dectin-1 and determined whether the tagged protein retained its ability to bind zymosan. The specificity of sortagging was evaluated by labeling a mixture of dectin-1–expressing HEK 293T cells and control HEK 293T cells transfected with a plasmid encoding GFP. Treatment of the cells with sortase A installed the fluorophore only on the surface of cells that did not express GFP; none of the GFP+ cells was labeled. Furthermore, only those cells that displayed the Alexa 647 fluorophore bound zymosan (Fig. 1C).

We conclude that a fraction of dectin-1 is accessible at the cell surface and amenable to sortagging. This labeling is highly selective: no biotinylated materials are detected on empty vector controls, and the only biotinylated species detectable in cells expressing dectin-1 has the molecular characteristics of dectin-1. Moreover, surface-disposed sortagged dectin-1 retains its function.

Internalization of Dectin-1 upon Zymosan Presentation.

Because sortase A installs a label specifically on surface-exposed dectin-1, we could follow the location of sortagged dectin-1 after exposure of cells to zymosan. We first installed an Alexa Fluor 488 probe on surface-disposed dectin-1 in HEK 293T cells and then exposed the cells to zymosan for 30 min. Then an Alexa Fluor 647 probe was installed on the surface to mark any newly surface exposed dectin-1 that appeared after the first round of sortagging. Cells were then fixed and examined by fluorescence microscopy. The fluorophore installed first showed the presence of dectin-1, either bound to zymosan particles on the surface or internalized. However, dectin-1 sortagged in a second round of labeling with the Alexa Fluor 647 was found only on the cell surface (Fig. 2A).

These data establish that, upon contact with zymosan, dectin-1 is internalized. As expected, regardless of zymosan-induced internalization, dectin-1 continues to emerge at the cell surface via the biosynthetic pathway. To follow dectin-1 internalization in time, a biotinylated probe was installed on dectin-1 in intact HEK 293T cells. Cells were then incubated with cycloheximide to arrest protein synthesis, and the culture was divided in two. One sample received zymosan, the other served as a control, and both were then sampled at different time points. Two different fractions of dectin-1 were assayed: dectin-1 located on the cell surface as inferred from the presence of the biotin label, and intracellular dectin-1 visualized by means of the HA tag. In the presence of zymosan, the rate of internalization and degradation of biotinylated dectin-1 on the cell surface increases dramatically after 1 h of incubation (Fig. 2B). The half-life of surface-located dectin-1, estimated to be between 8 h and 12 h in the absence of zymosan, decreases to <2 h in the presence of zymosan. The rate of degradation of intracellular HA-tagged dectin-1 was higher than that of biotinylated dectin-1, always showing a rapid decrease after 1 h of incubation, regardless of the presence of zymosan (Fig. 2B). This observation suggests that a significant fraction of newly synthesized dectin-1 fails to mature and is degraded instead.

Dectin-1 Associates with Galectin-3 in Macrophages.

The installation of a biotinylated probe on surface-exposed dectin-1 and its subsequent retrieval with streptavidin beads uncovered an association between dectin-1 and galectin-3 in RAW 264.7 macrophages (SI Results and Fig. S2). We further analyzed the interaction between dectin-1 and galectin-3 in RAW 264.7 macrophages by coimmunoprecipitation. Cell lysates from RAW 264.7 cells stably expressing HA-tagged dectin-1 (and not exposed to sortase, in this case) were immunoprecipitated with an anti–galectin-3 antibody. Western blot analysis using anti-HA antibodies revealed HA-tagged dectin-1, confirming the interaction between galectin-3 and dectin-1 (Fig. 3A). We also assayed the effect of zymosan presentation on this association. Intact RAW 264.7 cells with stable HA-tagged dectin-1 expression were incubated with zymosan particles for 30 min, lysed, and then treated with the anti–galectin-3 antibody. As expected, the HA-immunoblotting analysis of the immunoprecipitates showed higher amounts of dectin-1 compared with samples without zymosan presentation. We obtained similar results when we lysed RAW 264.7 cells stably expressing dectin-1 and subsequently incubated the lysates with zymosan particles for 30 min. The anti–galectin-3 immunoprecipitates from control cells expressing HA-tagged CD74 did not show any HA-tagged protein when analyzed by immunoblotting with anti-HA antibodies (Fig. 3A), confirming the specificity of the dectin-1/galectin-3 association.

This specificity of the interaction between dectin-1 and galectin-3 was also confirmed in dectin-1–expressing HEK 293T cells transfected with a plasmid encoding 3xFLAG-tagged galectin-3. Using sortase, we labeled surface-exposed dectin-1 in intact cells with a biotinylated probe to differentiate surface disposed dectin-1 from the intracellular HA-tagged fraction. Cells were lysed, and a fraction of the lysate was then incubated with streptavidin beads to retrieve biotinylated dectin-1. Bound proteins were removed from beads, separated by SDS gel electrophoresis, and analyzed by immunoblotting. When we incubated the blot with an anti-FLAG antibody, we observed a polypeptide of the expected size for FLAG-tagged galectin-3. We also incubated a part of the cell lysates with an anti-FLAG antibody and analyzed the immunoprecipitates by streptavidin or HA-immunoblotting. Both biotin-labeled and HA-tagged dectin-1 were detected in all cases, confirming the interaction of galectin-3 with the fraction of dectin-1 located on the surface as well as the fraction inside the cell. Neither biotinylated nor HA-tagged CD74, the type II membrane protein used for comparison, from control cells interacted with galectin-3 (Fig. 3B).

Galectin-3 in Association with Dectin-1 Modulates the Induction of TNF-α Expression.

To study the biological consequences of the dectin-1/galectin-3 interaction, we performed siRNA-based knockdowns of galectin-3 in RAW 264.7 macrophages and determined the effect of its reduced expression on the proinflammatory response after fungal presentation. A comparison of the response to Saccharomyces cerevisiae and Candida albicans required that both be in the yeast form. However, C. albicans produces filaments under the conditions of this assay. For this reason we inactivated fungi with UV (both C. albicans and S. cerevisiae) because UV-irradiated Candida cells do not present as filaments and remain in the yeast form. We first incubated galectin-3–competent RAW 264.7 cells with UV-treated S. cerevisiae or C. albicans at a ratio of 1:5 and measured TNF-α mRNA expression after 4 h. There was no significant increase in the level of induction of TNF-α expression, compared with cells without fungal presentation. These cells have very low expression levels of dectin-1 (Fig. 4A), so we next studied the response in cells that stably express dectin-1. In these cells we observed a significant increase in TNF-α mRNA levels after fungal presentation (P < 0.05). This observation confirms the key role of dectin-1 in the induction of TNF-α when confronted with either S. cerevisiae or C. albicans. We then studied the response in RAW 264.7 cells in which the level of galectin-3 was reduced by an RNA hairpin directed against galectin-3 mRNA. This hairpin successfully reduced the level of galectin-3 mRNA (Fig. 4A). Cells with and without the hairpin had similar mRNA levels of TNF-α when they were incubated with S. cerevisiae. However, when cells were incubated with C. albicans, there is a nearly fourfold decrease (P < 0.05) compared with galectin-3–competent cells (Fig. 4B). These data show that galectin-3 functions in concert with dectin-1 to modulate differentially the response to these two fungi. Significantly, the intensity of the response is much greater for the pathogen C. albicans than the nonpathogen S. cerevisiae.

Dectin-1/Galectin-3 Interaction Contributes to TNF-α Induction upon Recognition of β-glucan.

The functional consequences of the dectin-1/galectin-3 interaction could be muted by masking of the β-glucan by the outer coat of mannoproteins on yeast cell walls (21). Therefore, we analyzed the proinflammatory response using mutant yeast strains that unmask the underlying cell wall β-glucan. The S. cerevisiae gas1Δ and C. albicans phr2Δ/Δ and kre5Δ/Δstrains have increased β-glucan exposure on their surface compared with wild-type cells. The GAS1 gene encodes a β-glucan transglycosylase required for β-glucan branching and cell wall integrity. The PHR2 gene is the C. albicans homolog of the S. cerevisiae GAS1 gene. KRE5, essential for S. cerevisiae viability, is involved in the initiation of β-1,6-glucan synthesis. RAW 264.7 cells were incubated with wild-type and mutant yeast at a ratio of 1:5, and, after 12 h incubation, levels of TNF-α were quantified. When challenged with either wild-type or the mutants, macrophages that express higher levels of dectin-1 produce more TNF-α. When challenged with wild-type C. albicans, cells that have reduced galectin-3 expression have a significant decrease (P < 0.05) in TNF-α levels. In contrast, there was no significant decrease in the TNF-α levels when cells were exposed to S. cerevisiae cells (Fig. 4C).

Yeast mutants with more exposed β-glucan showed a much greater reduction in TNF-α secretion (P < 0.01) by macrophages with reduced galectin-3 expression. We observed similar results when we exposed bone marrow-derived macrophages (BMDMs) from wild-type mice and dectin-1 knockout (dectin-1KO) mice that do not express dectin-1 (Fig. 5A) to yeast. When challenged with C. albicans, wild-type BMDMs elicited TNF-α only when exposed to C. albicans kre5Δ/Δ, whereas dectin-1KO BMDMs failed to trigger a TNF-α response with either wild-type or the mutant. Wild-type BMDMs with reduced galectin-3 expression showed a significant reduction in TNF-α secretion (P < 0.01) when exposed to the C. albicans mutant. We did not detect significant changes in TNF-α secretion between BMDMs with different levels of expression of either dectin-1 or galectin-3 when challenged with S. cerevisiae (Fig. 5B). These data show (i) that the observed proinflammatory response elicited by C. albicans is dectin-1 dependent and (ii) that a decrease in galectin-3 expression drastically affects the proinflammatory response to C. albicans but not that to S. cerevisiae.